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;
  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         }
 151
 152         /**
 153          * Reduce memory usage to reduce the impact of circular references
 154          */
 155         function __destruct() {
 156                 if ( isset( $this->mLinkHolders ) ) {
 157                         $this->mLinkHolders->__destruct();
 158                 }
 159                 foreach ( $this as $name => $value ) {
 160                         unset( $this->$name );
 161                 }
 162         }
 163
 164         /**
 165          * Do various kinds of initialisation on the first call of the parser
 166          */
 167         function firstCallInit() {
 168                 if ( !$this->mFirstCall ) {
 169                         return;
 170                 }
 171                 $this->mFirstCall = false;
 172
 173                 wfProfileIn( __METHOD__ );
 174
 175                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 176                 CoreParserFunctions::register( $this );
 177                 $this->initialiseVariables();
 178
 179                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 180                 wfProfileOut( __METHOD__ );
 181         }
 182
 183         /**
 184          * Clear Parser state
 185          *
 186          * @private
 187          */
 188         function clearState() {
 189                 wfProfileIn( __METHOD__ );
 190                 if ( $this->mFirstCall ) {
 191                         $this->firstCallInit();
 192                 }
 193                 $this->mOutput = new ParserOutput;
 194                 $this->mAutonumber = 0;
 195                 $this->mLastSection = '';
 196                 $this->mDTopen = false;
 197                 $this->mIncludeCount = array();
 198                 $this->mStripState = new StripState;
 199                 $this->mArgStack = false;
 200                 $this->mInPre = false;
 201                 $this->mLinkHolders = new LinkHolderArray( $this );
 202                 $this->mLinkID = 0;
 203                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 204                 $this->mVarCache = array();
 205
 206                 /**
 207                  * Prefix for temporary replacement strings for the multipass parser.
 208                  * \x07 should never appear in input as it's disallowed in XML.
 209                  * Using it at the front also gives us a little extra robustness
 210                  * since it shouldn't match when butted up against identifier-like
 211                  * string constructs.
 212                  *
 213                  * Must not consist of all title characters, or else it will change
 214                  * the behaviour of <nowiki> in a link.
 215                  */
 216                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 217                 # Changed to \x7f to allow XML double-parsing -- TS
 218                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 219
 220
 221                 # Clear these on every parse, bug 4549
 222                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 223
 224                 $this->mShowToc = true;
 225                 $this->mForceTocPosition = false;
 226                 $this->mIncludeSizes = array(
 227                         'post-expand' => 0,
 228                         'arg' => 0,
 229                 );
 230                 $this->mPPNodeCount = 0;
 231                 $this->mDefaultSort = false;
 232                 $this->mHeadings = array();
 233                 $this->mDoubleUnderscores = array();
 234                 $this->mExpensiveFunctionCount = 0;
 235
 236                 # Fix cloning
 237                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 238                         $this->mPreprocessor = null;
 239                 }
 240
 241                 wfRunHooks( 'ParserClearState', array( &$this ) );
 242                 wfProfileOut( __METHOD__ );
 243         }
 244
 245         function setOutputType( $ot ) {
 246                 $this->mOutputType = $ot;
 247                 // Shortcut alias
 248                 $this->ot = array(
 249                         'html' => $ot == self::OT_HTML,
 250                         'wiki' => $ot == self::OT_WIKI,
 251                         'pre' => $ot == self::OT_PREPROCESS,
 252                 );
 253         }
 254
 255         /**
 256          * Set the context title
 257          */
 258         function setTitle( $t ) {
 259                 if ( !$t || $t instanceof FakeTitle ) {
 260                         $t = Title::newFromText( 'NO TITLE' );
 261                 }
 262
 263                 if ( strval( $t->getFragment() ) !== '' ) {
 264                         # Strip the fragment to avoid various odd effects
 265                         $this->mTitle = clone $t;
 266                         $this->mTitle->setFragment( '' );
 267                 } else {
 268                         $this->mTitle = $t;
 269                 }
 270         }
 271
 272         /**
 273          * Accessor for mUniqPrefix.
 274          *
 275          * @public
 276          */
 277         function uniqPrefix() {
 278                 if( !isset( $this->mUniqPrefix ) ) {
 279                         // @todo Fixme: this is probably *horribly wrong*
 280                         // LanguageConverter seems to want $wgParser's uniqPrefix, however
 281                         // if this is called for a parser cache hit, the parser may not
 282                         // have ever been initialized in the first place.
 283                         // Not really sure what the heck is supposed to be going on here.
 284                         return '';
 285                         //throw new MWException( "Accessing uninitialized mUniqPrefix" );
 286                 }
 287                 return $this->mUniqPrefix;
 288         }
 289
 290         /**
 291          * Convert wikitext to HTML
 292          * Do not call this function recursively.
 293          *
 294          * @param $text String: text we want to parse
 295          * @param $title A title object
 296          * @param $options ParserOptions
 297          * @param $linestart boolean
 298          * @param $clearState boolean
 299          * @param $revid Int: number to pass in {{REVISIONID}}
 300          * @return ParserOutput a ParserOutput
 301          */
 302         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 303                 /**
 304                  * First pass--just handle <nowiki> sections, pass the rest off
 305                  * to internalParse() which does all the real work.
 306                  */
 307
 308                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang, $wgDisableLangConversion;
 309                 $fname = __METHOD__.'-' . wfGetCaller();
 310                 wfProfileIn( __METHOD__ );
 311                 wfProfileIn( $fname );
 312
 313                 if ( $clearState ) {
 314                         $this->clearState();
 315                 }
 316
 317                 $this->mOptions = $options;
 318                 $this->setTitle( $title ); // Page title has to be set for the pre-processor
 319
 320                 $oldRevisionId = $this->mRevisionId;
 321                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 322                 if( $revid !== null ) {
 323                         $this->mRevisionId = $revid;
 324                         $this->mRevisionTimestamp = null;
 325                 }
 326                 $this->setOutputType( self::OT_HTML );
 327                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 328                 # No more strip!
 329                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 330                 $text = $this->internalParse( $text );
 331
 332                 $text = $this->mStripState->unstripGeneral( $text );
 333
 334                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 335                 $fixtags = array(
 336                         # french spaces, last one Guillemet-left
 337                         # only if there is something before the space
 338                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 339                         # french spaces, Guillemet-right
 340                         '/(\\302\\253) /' => '\\1&nbsp;',
 341                         '/&nbsp;(!\s*important)/' => ' \\1', #Beware of CSS magic word !important, bug #11874.
 342                 );
 343                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 344
 345                 $text = $this->doBlockLevels( $text, $linestart );
 346
 347                 $this->replaceLinkHolders( $text );
 348
 349                 // The position of the convert() call should not be changed. it
 350                 // assumes that the links are all replaced and the only thing left
 351                 // is the <nowiki> mark.
 352                 if ( !( $wgDisableLangConversion
 353                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 354                                 || $this->mTitle->isTalkPage()
 355                                 || $this->mTitle->isConversionTable() ) ) {
 356                         $text = $wgContLang->convert( $text );
 357                 }
 358
 359                 // A title may have been set in a conversion rule.
 360                 // Note that if a user tries to set a title in a conversion
 361                 // rule but content conversion was not done, then the parser
 362                 // won't pick it up.  This is probably expected behavior.
 363                 if ( $wgContLang->getConvRuleTitle() ) {
 364                         $this->mOutput->setTitleText( $wgContLang->getConvRuleTitle() );
 365                 } elseif ( !( $wgDisableLangConversion
 366                                           || isset( $this->mDoubleUnderscores['notitleconvert'] ) ) ) {
 367                         $this->mOutput->setTitleText( $wgContLang->convert( $title ) );
 368                 }
 369
 370                 $text = $this->mStripState->unstripNoWiki( $text );
 371
 372                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 373
 374 //!JF Move to its own function
 375
 376                 $uniq_prefix = $this->mUniqPrefix;
 377                 $matches = array();
 378                 $elements = array_keys( $this->mTransparentTagHooks );
 379                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 380
 381                 foreach( $matches as $marker => $data ) {
 382                         list( $element, $content, $params, $tag ) = $data;
 383                         $tagName = strtolower( $element );
 384                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 385                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 386                                         array( $content, $params, $this ) );
 387                         } else {
 388                                 $output = $tag;
 389                         }
 390                         $this->mStripState->general->setPair( $marker, $output );
 391                 }
 392                 $text = $this->mStripState->unstripGeneral( $text );
 393
 394                 $text = Sanitizer::normalizeCharReferences( $text );
 395
 396                 if ( ( $wgUseTidy && $this->mOptions->mTidy ) || $wgAlwaysUseTidy ) {
 397                         $text = MWTidy::tidy( $text );
 398                 } else {
 399                         # attempt to sanitize at least some nesting problems
 400                         # (bug #2702 and quite a few others)
 401                         $tidyregs = array(
 402                                 # ''Something [http://www.cool.com cool''] -->
 403                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 404                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 405                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 406                                 # fix up an anchor inside another anchor, only
 407                                 # at least for a single single nested link (bug 3695)
 408                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 409                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 410                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 411                                 # contains a div, so fix it up here; replace
 412                                 # div with escaped text
 413                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 414                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 415                                 # remove empty italic or bold tag pairs, some
 416                                 # introduced by rules above
 417                                 '/<([bi])><\/\\1>/' => '',
 418                         );
 419
 420                         $text = preg_replace(
 421                                 array_keys( $tidyregs ),
 422                                 array_values( $tidyregs ),
 423                                 $text );
 424                 }
 425                 global $wgExpensiveParserFunctionLimit;
 426                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 427                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 428                 }
 429
 430                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 431
 432                 # Information on include size limits, for the benefit of users who try to skirt them
 433                 if ( $this->mOptions->getEnableLimitReport() ) {
 434                         global $wgExpensiveParserFunctionLimit;
 435                         $max = $this->mOptions->getMaxIncludeSize();
 436                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 437                         $limitReport =
 438                                 "NewPP limit report\n" .
 439                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 440                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 441                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 442                                 $PFreport;
 443                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 444                         $text .= "\n<!-- \n$limitReport-->\n";
 445                 }
 446                 $this->mOutput->setText( $text );
 447
 448                 $this->mRevisionId = $oldRevisionId;
 449                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 450                 wfProfileOut( $fname );
 451                 wfProfileOut( __METHOD__ );
 452
 453                 return $this->mOutput;
 454         }
 455
 456         /**
 457          * Recursive parser entry point that can be called from an extension tag
 458          * hook.
 459          *
 460          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 461          *
 462          * @param $text String: text extension wants to have parsed
 463          * @param PPFrame $frame: The frame to use for expanding any template variables
 464          */
 465         function recursiveTagParse( $text, $frame=false ) {
 466                 wfProfileIn( __METHOD__ );
 467                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 468                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 469                 $text = $this->internalParse( $text, false, $frame );
 470                 wfProfileOut( __METHOD__ );
 471                 return $text;
 472         }
 473
 474         /**
 475          * Expand templates and variables in the text, producing valid, static wikitext.
 476          * Also removes comments.
 477          */
 478         function preprocess( $text, $title, $options, $revid = null ) {
 479                 wfProfileIn( __METHOD__ );
 480                 $this->clearState();
 481                 $this->setOutputType( self::OT_PREPROCESS );
 482                 $this->mOptions = $options;
 483                 $this->setTitle( $title );
 484                 if( $revid !== null ) {
 485                         $this->mRevisionId = $revid;
 486                 }
 487                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 488                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 489                 $text = $this->replaceVariables( $text );
 490                 $text = $this->mStripState->unstripBoth( $text );
 491                 wfProfileOut( __METHOD__ );
 492                 return $text;
 493         }
 494
 495         /**
 496          * Get a random string
 497          *
 498          * @private
 499          * @static
 500          */
 501         function getRandomString() {
 502                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 503         }
 504
 505         function &getTitle() { return $this->mTitle; }
 506         function getOptions() { return $this->mOptions; }
 507         function getRevisionId() { return $this->mRevisionId; }
 508         function getOutput() { return $this->mOutput; }
 509         function nextLinkID() { return $this->mLinkID++; }
 510
 511         function getFunctionLang() {
 512                 global $wgLang, $wgContLang;
 513
 514                 $target = $this->mOptions->getTargetLanguage();
 515                 if ( $target !== null ) {
 516                         return $target;
 517                 } else {
 518                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 519                 }
 520         }
 521
 522         /**
 523          * Get a preprocessor object
 524          */
 525         function getPreprocessor() {
 526                 if ( !isset( $this->mPreprocessor ) ) {
 527                         $class = $this->mPreprocessorClass;
 528                         $this->mPreprocessor = new $class( $this );
 529                 }
 530                 return $this->mPreprocessor;
 531         }
 532
 533         /**
 534          * Replaces all occurrences of HTML-style comments and the given tags
 535          * in the text with a random marker and returns the next text. The output
 536          * parameter $matches will be an associative array filled with data in
 537          * the form:
 538          *   'UNIQ-xxxxx' => array(
 539          *     'element',
 540          *     'tag content',
 541          *     array( 'param' => 'x' ),
 542          *     '<element param="x">tag content</element>' ) )
 543          *
 544          * @param $elements list of element names. Comments are always extracted.
 545          * @param $text Source text string.
 546          * @param $uniq_prefix
 547          *
 548          * @public
 549          * @static
 550          */
 551         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 552                 static $n = 1;
 553                 $stripped = '';
 554                 $matches = array();
 555
 556                 $taglist = implode( '|', $elements );
 557                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 558
 559                 while ( $text != '' ) {
 560                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 561                         $stripped .= $p[0];
 562                         if( count( $p ) < 5 ) {
 563                                 break;
 564                         }
 565                         if( count( $p ) > 5 ) {
 566                                 // comment
 567                                 $element    = $p[4];
 568                                 $attributes = '';
 569                                 $close      = '';
 570                                 $inside     = $p[5];
 571                         } else {
 572                                 // tag
 573                                 $element    = $p[1];
 574                                 $attributes = $p[2];
 575                                 $close      = $p[3];
 576                                 $inside     = $p[4];
 577                         }
 578
 579                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . self::MARKER_SUFFIX;
 580                         $stripped .= $marker;
 581
 582                         if ( $close === '/>' ) {
 583                                 // Empty element tag, <tag />
 584                                 $content = null;
 585                                 $text = $inside;
 586                                 $tail = null;
 587                         } else {
 588                                 if( $element === '!--' ) {
 589                                         $end = '/(-->)/';
 590                                 } else {
 591                                         $end = "/(<\\/$element\\s*>)/i";
 592                                 }
 593                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 594                                 $content = $q[0];
 595                                 if( count( $q ) < 3 ) {
 596                                         # No end tag -- let it run out to the end of the text.
 597                                         $tail = '';
 598                                         $text = '';
 599                                 } else {
 600                                         $tail = $q[1];
 601                                         $text = $q[2];
 602                                 }
 603                         }
 604
 605                         $matches[$marker] = array( $element,
 606                                 $content,
 607                                 Sanitizer::decodeTagAttributes( $attributes ),
 608                                 "<$element$attributes$close$content$tail" );
 609                 }
 610                 return $stripped;
 611         }
 612
 613         /**
 614          * Get a list of strippable XML-like elements
 615          */
 616         function getStripList() {
 617                 global $wgRawHtml;
 618                 $elements = $this->mStripList;
 619                 if( $wgRawHtml ) {
 620                         $elements[] = 'html';
 621                 }
 622                 if( $this->mOptions->getUseTeX() ) {
 623                         $elements[] = 'math';
 624                 }
 625                 return $elements;
 626         }
 627
 628         /**
 629          * @deprecated use replaceVariables
 630          */
 631         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 632                 return $text;
 633         }
 634
 635         /**
 636          * Restores pre, math, and other extensions removed by strip()
 637          *
 638          * always call unstripNoWiki() after this one
 639          * @private
 640          * @deprecated use $this->mStripState->unstrip()
 641          */
 642         function unstrip( $text, $state ) {
 643                 return $state->unstripGeneral( $text );
 644         }
 645
 646         /**
 647          * Always call this after unstrip() to preserve the order
 648          *
 649          * @private
 650          * @deprecated use $this->mStripState->unstrip()
 651          */
 652         function unstripNoWiki( $text, $state ) {
 653                 return $state->unstripNoWiki( $text );
 654         }
 655
 656         /**
 657          * @deprecated use $this->mStripState->unstripBoth()
 658          */
 659         function unstripForHTML( $text ) {
 660                 return $this->mStripState->unstripBoth( $text );
 661         }
 662
 663         /**
 664          * Add an item to the strip state
 665          * Returns the unique tag which must be inserted into the stripped text
 666          * The tag will be replaced with the original text in unstrip()
 667          *
 668          * @private
 669          */
 670         function insertStripItem( $text ) {
 671                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 672                 $this->mMarkerIndex++;
 673                 $this->mStripState->general->setPair( $rnd, $text );
 674                 return $rnd;
 675         }
 676
 677         /**
 678          * Interface with html tidy
 679          * @deprecated Use MWTidy::tidy()
 680          */
 681         public static function tidy( $text ) {
 682                 wfDeprecated( __METHOD__ );
 683                 return MWTidy::tidy( $text );
 684         }
 685
 686         /**
 687          * parse the wiki syntax used to render tables
 688          *
 689          * @private
 690          */
 691         function doTableStuff ( $text ) {
 692                 wfProfileIn( __METHOD__ );
 693
 694                 $lines = StringUtils::explode( "\n", $text );
 695                 $out = '';
 696                 $td_history = array (); // Is currently a td tag open?
 697                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 698                 $tr_history = array (); // Is currently a tr tag open?
 699                 $tr_attributes = array (); // history of tr attributes
 700                 $has_opened_tr = array(); // Did this table open a <tr> element?
 701                 $indent_level = 0; // indent level of the table
 702
 703                 foreach ( $lines as $outLine ) {
 704                         $line = trim( $outLine );
 705
 706                         if( $line == '' ) { // empty line, go to next line
 707                                 $out .= $outLine."\n";
 708                                 continue;
 709                         }
 710                         $first_character = $line[0];
 711                         $matches = array();
 712
 713                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 714                                 // First check if we are starting a new table
 715                                 $indent_level = strlen( $matches[1] );
 716
 717                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 718                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 719
 720                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 721                                 array_push( $td_history , false );
 722                                 array_push( $last_tag_history , '' );
 723                                 array_push( $tr_history , false );
 724                                 array_push( $tr_attributes , '' );
 725                                 array_push( $has_opened_tr , false );
 726                         } else if ( count ( $td_history ) == 0 ) {
 727                                 // Don't do any of the following
 728                                 $out .= $outLine."\n";
 729                                 continue;
 730                         } else if ( substr ( $line , 0 , 2 ) === '|}' ) {
 731                                 // We are ending a table
 732                                 $line = '</table>' . substr ( $line , 2 );
 733                                 $last_tag = array_pop ( $last_tag_history );
 734
 735                                 if ( !array_pop ( $has_opened_tr ) ) {
 736                                         $line = "<tr><td></td></tr>{$line}";
 737                                 }
 738
 739                                 if ( array_pop ( $tr_history ) ) {
 740                                         $line = "</tr>{$line}";
 741                                 }
 742
 743                                 if ( array_pop ( $td_history ) ) {
 744                                         $line = "</{$last_tag}>{$line}";
 745                                 }
 746                                 array_pop ( $tr_attributes );
 747                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 748                         } else if ( substr ( $line , 0 , 2 ) === '|-' ) {
 749                                 // Now we have a table row
 750                                 $line = preg_replace( '#^\|-+#', '', $line );
 751
 752                                 // Whats after the tag is now only attributes
 753                                 $attributes = $this->mStripState->unstripBoth( $line );
 754                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 755                                 array_pop( $tr_attributes );
 756                                 array_push( $tr_attributes, $attributes );
 757
 758                                 $line = '';
 759                                 $last_tag = array_pop ( $last_tag_history );
 760                                 array_pop ( $has_opened_tr );
 761                                 array_push ( $has_opened_tr , true );
 762
 763                                 if ( array_pop ( $tr_history ) ) {
 764                                         $line = '</tr>';
 765                                 }
 766
 767                                 if ( array_pop ( $td_history ) ) {
 768                                         $line = "</{$last_tag}>{$line}";
 769                                 }
 770
 771                                 $outLine = $line;
 772                                 array_push ( $tr_history , false );
 773                                 array_push ( $td_history , false );
 774                                 array_push ( $last_tag_history , '' );
 775                         }
 776                         else if ( $first_character === '|' || $first_character === '!' || substr ( $line , 0 , 2 )  === '|+' ) {
 777                                 // This might be cell elements, td, th or captions
 778                                 if ( substr ( $line , 0 , 2 ) === '|+' ) {
 779                                         $first_character = '+';
 780                                         $line = substr ( $line , 1 );
 781                                 }
 782
 783                                 $line = substr ( $line , 1 );
 784
 785                                 if ( $first_character === '!' ) {
 786                                         $line = str_replace ( '!!' , '||' , $line );
 787                                 }
 788
 789                                 // Split up multiple cells on the same line.
 790                                 // FIXME : This can result in improper nesting of tags processed
 791                                 // by earlier parser steps, but should avoid splitting up eg
 792                                 // attribute values containing literal "||".
 793                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 794
 795                                 $outLine = '';
 796
 797                                 // Loop through each table cell
 798                                 foreach ( $cells as $cell )
 799                                 {
 800                                         $previous = '';
 801                                         if ( $first_character !== '+' )
 802                                         {
 803                                                 $tr_after = array_pop ( $tr_attributes );
 804                                                 if ( !array_pop ( $tr_history ) ) {
 805                                                         $previous = "<tr{$tr_after}>\n";
 806                                                 }
 807                                                 array_push ( $tr_history , true );
 808                                                 array_push ( $tr_attributes , '' );
 809                                                 array_pop ( $has_opened_tr );
 810                                                 array_push ( $has_opened_tr , true );
 811                                         }
 812
 813                                         $last_tag = array_pop ( $last_tag_history );
 814
 815                                         if ( array_pop ( $td_history ) ) {
 816                                                 $previous = "</{$last_tag}>{$previous}";
 817                                         }
 818
 819                                         if ( $first_character === '|' ) {
 820                                                 $last_tag = 'td';
 821                                         } else if ( $first_character === '!' ) {
 822                                                 $last_tag = 'th';
 823                                         } else if ( $first_character === '+' ) {
 824                                                 $last_tag = 'caption';
 825                                         } else {
 826                                                 $last_tag = '';
 827                                         }
 828
 829                                         array_push ( $last_tag_history , $last_tag );
 830
 831                                         // A cell could contain both parameters and data
 832                                         $cell_data = explode ( '|' , $cell , 2 );
 833
 834                                         // Bug 553: Note that a '|' inside an invalid link should not
 835                                         // be mistaken as delimiting cell parameters
 836                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 837                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 838                                         } else if ( count ( $cell_data ) == 1 )
 839                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 840                                         else {
 841                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 842                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 843                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 844                                         }
 845
 846                                         $outLine .= $cell;
 847                                         array_push ( $td_history , true );
 848                                 }
 849                         }
 850                         $out .= $outLine . "\n";
 851                 }
 852
 853                 // Closing open td, tr && table
 854                 while ( count ( $td_history ) > 0 )
 855                 {
 856                         if ( array_pop ( $td_history ) ) {
 857                                 $out .= "</td>\n";
 858                         }
 859                         if ( array_pop ( $tr_history ) ) {
 860                                 $out .= "</tr>\n";
 861                         }
 862                         if ( !array_pop ( $has_opened_tr ) ) {
 863                                 $out .= "<tr><td></td></tr>\n" ;
 864                         }
 865
 866                         $out .= "</table>\n";
 867                 }
 868
 869                 // Remove trailing line-ending (b/c)
 870                 if ( substr( $out, -1 ) === "\n" ) {
 871                         $out = substr( $out, 0, -1 );
 872                 }
 873
 874                 // special case: don't return empty table
 875                 if( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 876                         $out = '';
 877                 }
 878
 879                 wfProfileOut( __METHOD__ );
 880
 881                 return $out;
 882         }
 883
 884         /**
 885          * Helper function for parse() that transforms wiki markup into
 886          * HTML. Only called for $mOutputType == self::OT_HTML.
 887          *
 888          * @private
 889          */
 890         function internalParse( $text, $isMain = true, $frame=false ) {
 891                 wfProfileIn( __METHOD__ );
 892
 893                 $origText = $text;
 894
 895                 # Hook to suspend the parser in this state
 896                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 897                         wfProfileOut( __METHOD__ );
 898                         return $text ;
 899                 }
 900
 901                 // if $frame is provided, then use $frame for replacing any variables
 902                 if ($frame) {
 903                         // use frame depth to infer how include/noinclude tags should be handled
 904                         // depth=0 means this is the top-level document; otherwise it's an included document
 905                         if( !$frame->depth )
 906                                 $flag = 0;
 907                         else
 908                                 $flag = Parser::PTD_FOR_INCLUSION;
 909                         $dom = $this->preprocessToDom( $text, $flag );
 910                         $text = $frame->expand( $dom );
 911                 }
 912                 // if $frame is not provided, then use old-style replaceVariables
 913                 else {
 914                         $text = $this->replaceVariables( $text );
 915                 }
 916
 917                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 918                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 919
 920                 // Tables need to come after variable replacement for things to work
 921                 // properly; putting them before other transformations should keep
 922                 // exciting things like link expansions from showing up in surprising
 923                 // places.
 924                 $text = $this->doTableStuff( $text );
 925
 926                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 927
 928                 $text = $this->doDoubleUnderscore( $text );
 929
 930                 $text = $this->doHeadings( $text );
 931                 if( $this->mOptions->getUseDynamicDates() ) {
 932                         $df = DateFormatter::getInstance();
 933                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 934                 }
 935                 $text = $this->doAllQuotes( $text );
 936                 $text = $this->replaceInternalLinks( $text );
 937                 $text = $this->replaceExternalLinks( $text );
 938
 939                 # replaceInternalLinks may sometimes leave behind
 940                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 941                 $text = str_replace($this->mUniqPrefix.'NOPARSE', '', $text);
 942
 943                 $text = $this->doMagicLinks( $text );
 944                 $text = $this->formatHeadings( $text, $origText, $isMain );
 945
 946                 wfProfileOut( __METHOD__ );
 947                 return $text;
 948         }
 949
 950         /**
 951          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 952          * magic external links.
 953          *
 954          * DML
 955          * @private
 956          */
 957         function doMagicLinks( $text ) {
 958                 wfProfileIn( __METHOD__ );
 959                 $prots = $this->mUrlProtocols;
 960                 $urlChar = self::EXT_LINK_URL_CLASS;
 961                 $text = preg_replace_callback(
 962                         '!(?:                           # Start cases
 963                                 (<a.*?</a>) |               # m[1]: Skip link text
 964                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
 965                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
 966                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
 967                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
 968                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 969                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 970                                     [0-9Xx]                 # check digit
 971                                     \b)
 972                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 973                 wfProfileOut( __METHOD__ );
 974                 return $text;
 975         }
 976
 977         function magicLinkCallback( $m ) {
 978                 if ( isset( $m[1] ) && $m[1] !== '' ) {
 979                         # Skip anchor
 980                         return $m[0];
 981                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
 982                         # Skip HTML element
 983                         return $m[0];
 984                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
 985                         # Free external link
 986                         return $this->makeFreeExternalLink( $m[0] );
 987                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
 988                         # RFC or PMID
 989                         $CssClass = '';
 990                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
 991                                 $keyword = 'RFC';
 992                                 $urlmsg = 'rfcurl';
 993                                 $CssClass = 'mw-magiclink-rfc';
 994                                 $id = $m[4];
 995                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
 996                                 $keyword = 'PMID';
 997                                 $urlmsg = 'pubmedurl';
 998                                 $CssClass = 'mw-magiclink-pmid';
 999                                 $id = $m[4];
1000                         } else {
1001                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1002                                         substr($m[0], 0, 20 ) . '"' );
1003                         }
1004                         $url = wfMsg( $urlmsg, $id);
1005                         $sk = $this->mOptions->getSkin();
1006                         $la = $sk->getExternalLinkAttributes( "external $CssClass" );
1007                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1008                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1009                         # ISBN
1010                         $isbn = $m[5];
1011                         $num = strtr( $isbn, array(
1012                                 '-' => '',
1013                                 ' ' => '',
1014                                 'x' => 'X',
1015                         ));
1016                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1017                         return'<a href="' .
1018                                 $titleObj->escapeLocalUrl() .
1019                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1020                 } else {
1021                         return $m[0];
1022                 }
1023         }
1024
1025         /**
1026          * Make a free external link, given a user-supplied URL
1027          * @return HTML
1028          * @private
1029          */
1030         function makeFreeExternalLink( $url ) {
1031                 global $wgContLang;
1032                 wfProfileIn( __METHOD__ );
1033
1034                 $sk = $this->mOptions->getSkin();
1035                 $trail = '';
1036
1037                 # The characters '<' and '>' (which were escaped by
1038                 # removeHTMLtags()) should not be included in
1039                 # URLs, per RFC 2396.
1040                 $m2 = array();
1041                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1042                         $trail = substr($url, $m2[0][1]) . $trail;
1043                         $url = substr($url, 0, $m2[0][1]);
1044                 }
1045
1046                 # Move trailing punctuation to $trail
1047                 $sep = ',;\.:!?';
1048                 # If there is no left bracket, then consider right brackets fair game too
1049                 if ( strpos( $url, '(' ) === false ) {
1050                         $sep .= ')';
1051                 }
1052
1053                 $numSepChars = strspn( strrev( $url ), $sep );
1054                 if ( $numSepChars ) {
1055                         $trail = substr( $url, -$numSepChars ) . $trail;
1056                         $url = substr( $url, 0, -$numSepChars );
1057                 }
1058
1059                 $url = Sanitizer::cleanUrl( $url );
1060
1061                 # Is this an external image?
1062                 $text = $this->maybeMakeExternalImage( $url );
1063                 if ( $text === false ) {
1064                         # Not an image, make a link
1065                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1066                                 $this->getExternalLinkAttribs( $url ) );
1067                         # Register it in the output object...
1068                         # Replace unnecessary URL escape codes with their equivalent characters
1069                         $pasteurized = self::replaceUnusualEscapes( $url );
1070                         $this->mOutput->addExternalLink( $pasteurized );
1071                 }
1072                 wfProfileOut( __METHOD__ );
1073                 return $text . $trail;
1074         }
1075
1076
1077         /**
1078          * Parse headers and return html
1079          *
1080          * @private
1081          */
1082         function doHeadings( $text ) {
1083                 wfProfileIn( __METHOD__ );
1084                 for ( $i = 6; $i >= 1; --$i ) {
1085                         $h = str_repeat( '=', $i );
1086                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1087                           "<h$i>\\1</h$i>", $text );
1088                 }
1089                 wfProfileOut( __METHOD__ );
1090                 return $text;
1091         }
1092
1093         /**
1094          * Replace single quotes with HTML markup
1095          * @private
1096          * @return string the altered text
1097          */
1098         function doAllQuotes( $text ) {
1099                 wfProfileIn( __METHOD__ );
1100                 $outtext = '';
1101                 $lines = StringUtils::explode( "\n", $text );
1102                 foreach ( $lines as $line ) {
1103                         $outtext .= $this->doQuotes( $line ) . "\n";
1104                 }
1105                 $outtext = substr($outtext, 0,-1);
1106                 wfProfileOut( __METHOD__ );
1107                 return $outtext;
1108         }
1109
1110         /**
1111          * Processes bolds and italics on a single line.
1112          * Helper function for doAllQuotes()
1113          */
1114         public function doQuotes( $text ) {
1115                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1116                 if ( count( $arr ) == 1 )
1117                         return $text;
1118                 else
1119                 {
1120                         # First, do some preliminary work. This may shift some apostrophes from
1121                         # being mark-up to being text. It also counts the number of occurrences
1122                         # of bold and italics mark-ups.
1123                         $i = 0;
1124                         $numbold = 0;
1125                         $numitalics = 0;
1126                         foreach ( $arr as $r )
1127                         {
1128                                 if ( ( $i % 2 ) == 1 )
1129                                 {
1130                                         # If there are ever four apostrophes, assume the first is supposed to
1131                                         # be text, and the remaining three constitute mark-up for bold text.
1132                                         if ( strlen( $arr[$i] ) == 4 )
1133                                         {
1134                                                 $arr[$i-1] .= "'";
1135                                                 $arr[$i] = "'''";
1136                                         }
1137                                         # If there are more than 6 apostrophes in a row, assume they're all
1138                                         # text except for the last 6.
1139                                         else if ( strlen( $arr[$i] ) > 6 )
1140                                         {
1141                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 6 );
1142                                                 $arr[$i] = "''''''";
1143                                         }
1144                                         # Count the number of occurrences of bold and italics mark-ups.
1145                                         # We are not counting sequences of five apostrophes.
1146                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1147                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1148                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1149                                         else if ( strlen( $arr[$i] ) == 6 ) { $numbold+=2; }
1150                                 }
1151                                 $i++;
1152                         }
1153
1154                         # If there is an odd number of both bold and italics, it is likely
1155                         # that one of the bold ones was meant to be an apostrophe followed
1156                         # by italics. Which one we cannot know for certain, but it is more
1157                         # likely to be one that has a single-letter word before it.
1158                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1159                         {
1160                                 $i = 0;
1161
1162                                 #These are indexes to the /next/ array entry than the
1163                                 #one holding the text matching the condition.
1164                                 $firstsingleletterword = -1;
1165                                 $firstmultiletterword = -1;
1166                                 $firstspace = -1;
1167
1168                                 foreach ( $arr as $r )
1169                                 {
1170                                         #Filter the "'''". Separators are on odd positions.
1171                                         #$arr[0] will be an empty string if needed.
1172                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1173                                         {
1174                                                 $x1 = substr ($arr[$i-1], -1);
1175                                                 $x2 = substr ($arr[$i-1], -2, 1);
1176                                                 if ($x1 === ' ') {
1177                                                         if ($firstspace == -1) $firstspace = $i;
1178                                                 } else if ($x2 === ' ') {
1179                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1180                                                 } else if ($arr[$i-1] != "") {
1181                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1182                                                 }
1183                                         }
1184                                         $i++;
1185                                 }
1186
1187                                 # If there is a single-letter word, use it!
1188                                 if ($firstsingleletterword > -1)
1189                                 {
1190                                         $arr [ $firstsingleletterword ] = "''";
1191                                         $arr [ $firstsingleletterword-1 ] .= "'";
1192                                 }
1193                                 # If not, but there's a multi-letter word, use that one.
1194                                 else if ($firstmultiletterword > -1)
1195                                 {
1196                                         $arr [ $firstmultiletterword ] = "''";
1197                                         $arr [ $firstmultiletterword-1 ] .= "'";
1198                                 }
1199                                 # ... otherwise use the first one that has neither.
1200                                 # (notice that it is possible for all three to be -1 if, for example,
1201                                 # there is only one pentuple-apostrophe in the line)
1202                                 else if ($firstspace > -1)
1203                                 {
1204                                         $arr [ $firstspace ] = "''";
1205                                         $arr [ $firstspace-1 ] .= "'";
1206                                 }
1207                         }
1208
1209                         # Now let's actually convert our apostrophic mush to HTML!
1210                         $output = ''; #Processed text
1211                         $buffer = ''; #Content if $state is 'both'
1212                         $state = '';  #Flags with the order of open tags: '|b|i|bi|ib|both'
1213                         $i = 0;
1214                         foreach ($arr as $r)
1215                         {
1216                                 if (($i % 2) == 0)
1217                                 {
1218                                         if ($state === 'both')
1219                                                 $buffer .= $r;
1220                                         else
1221                                                 $output .= $r;
1222                                 }
1223                                 else
1224                                 {
1225                                         if (strlen ($r) == 2)
1226                                         {
1227                                                 if ($state === 'i')
1228                                                 { $output .= '</i>'; $state = ''; }
1229                                                 else if ($state === 'bi')
1230                                                 { $output .= '</i>'; $state = 'b'; }
1231                                                 else if ($state === 'ib')
1232                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1233                                                 else if ($state === 'both')
1234                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1235                                                 else # $state can be 'b' or ''
1236                                                 { $output .= '<i>'; $state .= 'i'; }
1237                                         }
1238                                         else if (strlen ($r) == 3)
1239                                         {
1240                                                 if ($state === 'b')
1241                                                 { $output .= '</b>'; $state = ''; }
1242                                                 else if ($state === 'bi')
1243                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1244                                                 else if ($state === 'ib')
1245                                                 { $output .= '</b>'; $state = 'i'; }
1246                                                 else if ($state === 'both')
1247                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1248                                                 else # $state can be 'i' or ''
1249                                                 { $output .= '<b>'; $state .= 'b'; }
1250                                         }
1251                                         else if (strlen ($r) == 5)
1252                                         {
1253                                                 if ($state === 'b')
1254                                                 { $output .= '</b><i>'; $state = 'i'; }
1255                                                 else if ($state === 'i')
1256                                                 { $output .= '</i><b>'; $state = 'b'; }
1257                                                 else if ($state === 'bi')
1258                                                 { $output .= '</i></b>'; $state = ''; }
1259                                                 else if ($state === 'ib')
1260                                                 { $output .= '</b></i>'; $state = ''; }
1261                                                 else if ($state === 'both')
1262                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1263                                                 else # ($state == '')
1264                                                 { $buffer = ''; $state = 'both'; }
1265                                         }
1266                                         else if (strlen ($r) == 6)
1267                                         {
1268                                                 if ($state === 'b')
1269                                                 { $output .= '</b><b>'; $state = 'b'; }
1270                                                 else if ($state === 'i')
1271                                                 { $output .= '\'</i><b>'; $state = 'b'; }
1272                                                 else if ($state === 'bi')
1273                                                 { $output .= '\'</i></b>'; $state = ''; }
1274                                                 else if ($state === 'ib')
1275                                                 { $output .= '\'</b></i>'; $state = ''; }
1276                                                 else if ($state === 'both')
1277                                                 { $output .= '<i><b>'.$buffer.'</b><b>'; $state = 'ib'; }
1278                                                 else # ($state == '')
1279                                                 { $buffer = ''; $state = ''; }
1280                                         }
1281                                 }
1282                                 $i++;
1283                         }
1284                         # Now close all remaining tags.  Notice that the order is important.
1285                         if ($state === 'b' || $state === 'ib')
1286                                 $output .= '</b>';
1287                         if ($state === 'i' || $state === 'bi' || $state === 'ib')
1288                                 $output .= '</i>';
1289                         if ($state === 'bi')
1290                                 $output .= '</b>';
1291                         # There might be lonely ''''', so make sure we have a buffer
1292                         if ($state === 'both' && $buffer)
1293                                 $output .= '<b><i>'.$buffer.'</i></b>';
1294                         return $output;
1295                 }
1296         }
1297
1298         /**
1299          * Replace external links (REL)
1300          *
1301          * Note: this is all very hackish and the order of execution matters a lot.
1302          * Make sure to run maintenance/parserTests.php if you change this code.
1303          *
1304          * @private
1305          */
1306         function replaceExternalLinks( $text ) {
1307                 global $wgContLang;
1308                 wfProfileIn( __METHOD__ );
1309
1310                 $sk = $this->mOptions->getSkin();
1311
1312                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1313                 $s = array_shift( $bits );
1314
1315                 $i = 0;
1316                 while ( $i<count( $bits ) ) {
1317                         $url = $bits[$i++];
1318                         $protocol = $bits[$i++];
1319                         $text = $bits[$i++];
1320                         $trail = $bits[$i++];
1321
1322                         # The characters '<' and '>' (which were escaped by
1323                         # removeHTMLtags()) should not be included in
1324                         # URLs, per RFC 2396.
1325                         $m2 = array();
1326                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1327                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1328                                 $url = substr($url, 0, $m2[0][1]);
1329                         }
1330
1331                         # If the link text is an image URL, replace it with an <img> tag
1332                         # This happened by accident in the original parser, but some people used it extensively
1333                         $img = $this->maybeMakeExternalImage( $text );
1334                         if ( $img !== false ) {
1335                                 $text = $img;
1336                         }
1337
1338                         $dtrail = '';
1339
1340                         # Set linktype for CSS - if URL==text, link is essentially free
1341                         $linktype = ($text === $url) ? 'free' : 'text';
1342
1343                         # No link text, e.g. [http://domain.tld/some.link]
1344                         if ( $text == '' ) {
1345                                 # Autonumber if allowed. See bug #5918
1346                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1347                                         $langObj = $this->getFunctionLang();
1348                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1349                                         $linktype = 'autonumber';
1350                                 } else {
1351                                         # Otherwise just use the URL
1352                                         $text = htmlspecialchars( $url );
1353                                         $linktype = 'free';
1354                                 }
1355                         } else {
1356                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1357                                 # Check for trail
1358                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1359                         }
1360
1361                         $text = $wgContLang->markNoConversion($text);
1362
1363                         $url = Sanitizer::cleanUrl( $url );
1364
1365                         # Use the encoded URL
1366                         # This means that users can paste URLs directly into the text
1367                         # Funny characters like &ouml; aren't valid in URLs anyway
1368                         # This was changed in August 2004
1369                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1370                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1371
1372                         # Register link in the output object.
1373                         # Replace unnecessary URL escape codes with the referenced character
1374                         # This prevents spammers from hiding links from the filters
1375                         $pasteurized = self::replaceUnusualEscapes( $url );
1376                         $this->mOutput->addExternalLink( $pasteurized );
1377                 }
1378
1379                 wfProfileOut( __METHOD__ );
1380                 return $s;
1381         }
1382
1383         /**
1384          * Get an associative array of additional HTML attributes appropriate for a
1385          * particular external link.  This currently may include rel => nofollow
1386          * (depending on configuration, namespace, and the URL's domain) and/or a
1387          * target attribute (depending on configuration).
1388          *
1389          * @param string $url Optional URL, to extract the domain from for rel =>
1390          *   nofollow if appropriate
1391          * @return array Associative array of HTML attributes
1392          */
1393         function getExternalLinkAttribs( $url = false ) {
1394                 $attribs = array();
1395                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1396                 $ns = $this->mTitle->getNamespace();
1397                 if( $wgNoFollowLinks && !in_array($ns, $wgNoFollowNsExceptions) ) {
1398                         $attribs['rel'] = 'nofollow';
1399
1400                         global $wgNoFollowDomainExceptions;
1401                         if ( $wgNoFollowDomainExceptions ) {
1402                                 $bits = wfParseUrl( $url );
1403                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1404                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1405                                                 if( substr( $bits['host'], -strlen( $domain ) )
1406                                                 == $domain ) {
1407                                                         unset( $attribs['rel'] );
1408                                                         break;
1409                                                 }
1410                                         }
1411                                 }
1412                         }
1413                 }
1414                 if ( $this->mOptions->getExternalLinkTarget() ) {
1415                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1416                 }
1417                 return $attribs;
1418         }
1419
1420
1421         /**
1422          * Replace unusual URL escape codes with their equivalent characters
1423          * @param string
1424          * @return string
1425          * @static
1426          * @todo  This can merge genuinely required bits in the path or query string,
1427          *        breaking legit URLs. A proper fix would treat the various parts of
1428          *        the URL differently; as a workaround, just use the output for
1429          *        statistical records, not for actual linking/output.
1430          */
1431         static function replaceUnusualEscapes( $url ) {
1432                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1433                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1434         }
1435
1436         /**
1437          * Callback function used in replaceUnusualEscapes().
1438          * Replaces unusual URL escape codes with their equivalent character
1439          * @static
1440          * @private
1441          */
1442         private static function replaceUnusualEscapesCallback( $matches ) {
1443                 $char = urldecode( $matches[0] );
1444                 $ord = ord( $char );
1445                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1446                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1447                         // No, shouldn't be escaped
1448                         return $char;
1449                 } else {
1450                         // Yes, leave it escaped
1451                         return $matches[0];
1452                 }
1453         }
1454
1455         /**
1456          * make an image if it's allowed, either through the global
1457          * option, through the exception, or through the on-wiki whitelist
1458          * @private
1459          */
1460         function maybeMakeExternalImage( $url ) {
1461                 $sk = $this->mOptions->getSkin();
1462                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1463                 $imagesexception = !empty($imagesfrom);
1464                 $text = false;
1465                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1466                 if( $imagesexception && is_array( $imagesfrom ) ) {
1467                         $imagematch = false;
1468                         foreach( $imagesfrom as $match ) {
1469                                 if( strpos( $url, $match ) === 0 ) {
1470                                         $imagematch = true;
1471                                         break;
1472                                 }
1473                         }
1474                 } elseif( $imagesexception ) {
1475                         $imagematch = (strpos( $url, $imagesfrom ) === 0);
1476                 } else {
1477                         $imagematch = false;
1478                 }
1479                 if ( $this->mOptions->getAllowExternalImages()
1480                      || ( $imagesexception && $imagematch ) ) {
1481                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1482                                 # Image found
1483                                 $text = $sk->makeExternalImage( $url );
1484                         }
1485                 }
1486                 if( !$text && $this->mOptions->getEnableImageWhitelist()
1487                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1488                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1489                         foreach( $whitelist as $entry ) {
1490                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1491                                 if( strpos( $entry, '#' ) === 0 || $entry === '' )
1492                                         continue;
1493                                 if( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1494                                         # Image matches a whitelist entry
1495                                         $text = $sk->makeExternalImage( $url );
1496                                         break;
1497                                 }
1498                         }
1499                 }
1500                 return $text;
1501         }
1502
1503         /**
1504          * Process [[ ]] wikilinks
1505          * @return processed text
1506          *
1507          * @private
1508          */
1509         function replaceInternalLinks( $s ) {
1510                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1511                 return $s;
1512         }
1513
1514         /**
1515          * Process [[ ]] wikilinks (RIL)
1516          * @return LinkHolderArray
1517          *
1518          * @private
1519          */
1520         function replaceInternalLinks2( &$s ) {
1521                 global $wgContLang;
1522
1523                 wfProfileIn( __METHOD__ );
1524
1525                 wfProfileIn( __METHOD__.'-setup' );
1526                 static $tc = FALSE, $e1, $e1_img;
1527                 # the % is needed to support urlencoded titles as well
1528                 if ( !$tc ) {
1529                         $tc = Title::legalChars() . '#%';
1530                         # Match a link having the form [[namespace:link|alternate]]trail
1531                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1532                         # Match cases where there is no "]]", which might still be images
1533                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1534                 }
1535
1536                 $sk = $this->mOptions->getSkin();
1537                 $holders = new LinkHolderArray( $this );
1538
1539                 #split the entire text string on occurences of [[
1540                 $a = StringUtils::explode( '[[', ' ' . $s );
1541                 #get the first element (all text up to first [[), and remove the space we added
1542                 $s = $a->current();
1543                 $a->next();
1544                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1545                 $s = substr( $s, 1 );
1546
1547                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1548                 $e2 = null;
1549                 if ( $useLinkPrefixExtension ) {
1550                         # Match the end of a line for a word that's not followed by whitespace,
1551                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1552                         $e2 = wfMsgForContent( 'linkprefix' );
1553                 }
1554
1555                 if( is_null( $this->mTitle ) ) {
1556                         wfProfileOut( __METHOD__.'-setup' );
1557                         wfProfileOut( __METHOD__ );
1558                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1559                 }
1560                 $nottalk = !$this->mTitle->isTalkPage();
1561
1562                 if ( $useLinkPrefixExtension ) {
1563                         $m = array();
1564                         if ( preg_match( $e2, $s, $m ) ) {
1565                                 $first_prefix = $m[2];
1566                         } else {
1567                                 $first_prefix = false;
1568                         }
1569                 } else {
1570                         $prefix = '';
1571                 }
1572
1573                 if($wgContLang->hasVariants()) {
1574                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1575                 } else {
1576                         $selflink = array($this->mTitle->getPrefixedText());
1577                 }
1578                 $useSubpages = $this->areSubpagesAllowed();
1579                 wfProfileOut( __METHOD__.'-setup' );
1580
1581                 # Loop for each link
1582                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1583                         # Check for excessive memory usage
1584                         if ( $holders->isBig() ) {
1585                                 # Too big
1586                                 # Do the existence check, replace the link holders and clear the array
1587                                 $holders->replace( $s );
1588                                 $holders->clear();
1589                         }
1590
1591                         if ( $useLinkPrefixExtension ) {
1592                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1593                                 if ( preg_match( $e2, $s, $m ) ) {
1594                                         $prefix = $m[2];
1595                                         $s = $m[1];
1596                                 } else {
1597                                         $prefix='';
1598                                 }
1599                                 # first link
1600                                 if($first_prefix) {
1601                                         $prefix = $first_prefix;
1602                                         $first_prefix = false;
1603                                 }
1604                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1605                         }
1606
1607                         $might_be_img = false;
1608
1609                         wfProfileIn( __METHOD__."-e1" );
1610                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1611                                 $text = $m[2];
1612                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1613                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1614                                 # the real problem is with the $e1 regex
1615                                 # See bug 1300.
1616                                 #
1617                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1618                                 # and no image is in sight. See bug 2095.
1619                                 #
1620                                 if( $text !== '' &&
1621                                         substr( $m[3], 0, 1 ) === ']' &&
1622                                         strpos($text, '[') !== false
1623                                 )
1624                                 {
1625                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1626                                         $m[3] = substr( $m[3], 1 );
1627                                 }
1628                                 # fix up urlencoded title texts
1629                                 if( strpos( $m[1], '%' ) !== false ) {
1630                                         # Should anchors '#' also be rejected?
1631                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1632                                 }
1633                                 $trail = $m[3];
1634                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1635                                 $might_be_img = true;
1636                                 $text = $m[2];
1637                                 if ( strpos( $m[1], '%' ) !== false ) {
1638                                         $m[1] = urldecode($m[1]);
1639                                 }
1640                                 $trail = "";
1641                         } else { # Invalid form; output directly
1642                                 $s .= $prefix . '[[' . $line ;
1643                                 wfProfileOut( __METHOD__."-e1" );
1644                                 continue;
1645                         }
1646                         wfProfileOut( __METHOD__."-e1" );
1647                         wfProfileIn( __METHOD__."-misc" );
1648
1649                         # Don't allow internal links to pages containing
1650                         # PROTO: where PROTO is a valid URL protocol; these
1651                         # should be external links.
1652                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1653                                 $s .= $prefix . '[[' . $line ;
1654                                 wfProfileOut( __METHOD__."-misc" );
1655                                 continue;
1656                         }
1657
1658                         # Make subpage if necessary
1659                         if ( $useSubpages ) {
1660                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1661                         } else {
1662                                 $link = $m[1];
1663                         }
1664
1665                         $noforce = (substr( $m[1], 0, 1 ) !== ':');
1666                         if (!$noforce) {
1667                                 # Strip off leading ':'
1668                                 $link = substr( $link, 1 );
1669                         }
1670
1671                         wfProfileOut( __METHOD__."-misc" );
1672                         wfProfileIn( __METHOD__."-title" );
1673                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1674                         if ( $nt === null ) {
1675                                 $s .= $prefix . '[[' . $line;
1676                                 wfProfileOut( __METHOD__."-title" );
1677                                 continue;
1678                         }
1679
1680                         $ns = $nt->getNamespace();
1681                         $iw = $nt->getInterWiki();
1682                         wfProfileOut( __METHOD__."-title" );
1683
1684                         if ( $might_be_img ) { # if this is actually an invalid link
1685                                 wfProfileIn( __METHOD__."-might_be_img" );
1686                                 if ( $ns == NS_FILE && $noforce ) { #but might be an image
1687                                         $found = false;
1688                                         while ( true ) {
1689                                                 #look at the next 'line' to see if we can close it there
1690                                                 $a->next();
1691                                                 $next_line = $a->current();
1692                                                 if ( $next_line === false || $next_line === null ) {
1693                                                         break;
1694                                                 }
1695                                                 $m = explode( ']]', $next_line, 3 );
1696                                                 if ( count( $m ) == 3 ) {
1697                                                         # the first ]] closes the inner link, the second the image
1698                                                         $found = true;
1699                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1700                                                         $trail = $m[2];
1701                                                         break;
1702                                                 } elseif ( count( $m ) == 2 ) {
1703                                                         #if there's exactly one ]] that's fine, we'll keep looking
1704                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1705                                                 } else {
1706                                                         #if $next_line is invalid too, we need look no further
1707                                                         $text .= '[[' . $next_line;
1708                                                         break;
1709                                                 }
1710                                         }
1711                                         if ( !$found ) {
1712                                                 # we couldn't find the end of this imageLink, so output it raw
1713                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1714                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1715                                                 $s .= "{$prefix}[[$link|$text";
1716                                                 # note: no $trail, because without an end, there *is* no trail
1717                                                 wfProfileOut( __METHOD__."-might_be_img" );
1718                                                 continue;
1719                                         }
1720                                 } else { #it's not an image, so output it raw
1721                                         $s .= "{$prefix}[[$link|$text";
1722                                         # note: no $trail, because without an end, there *is* no trail
1723                                         wfProfileOut( __METHOD__."-might_be_img" );
1724                                         continue;
1725                                 }
1726                                 wfProfileOut( __METHOD__."-might_be_img" );
1727                         }
1728
1729                         $wasblank = ( $text  == '' );
1730                         if ( $wasblank ) $text = $link;
1731
1732                         # Link not escaped by : , create the various objects
1733                         if ( $noforce ) {
1734
1735                                 # Interwikis
1736                                 wfProfileIn( __METHOD__."-interwiki" );
1737                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1738                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1739                                         $s = rtrim($s . $prefix);
1740                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1741                                         wfProfileOut( __METHOD__."-interwiki" );
1742                                         continue;
1743                                 }
1744                                 wfProfileOut( __METHOD__."-interwiki" );
1745
1746                                 if ( $ns == NS_FILE ) {
1747                                         wfProfileIn( __METHOD__."-image" );
1748                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1749                                                 if ( $wasblank ) {
1750                                                         # if no parameters were passed, $text
1751                                                         # becomes something like "File:Foo.png",
1752                                                         # which we don't want to pass on to the
1753                                                         # image generator
1754                                                         $text = '';
1755                                                 } else {
1756                                                         # recursively parse links inside the image caption
1757                                                         # actually, this will parse them in any other parameters, too,
1758                                                         # but it might be hard to fix that, and it doesn't matter ATM
1759                                                         $text = $this->replaceExternalLinks($text);
1760                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1761                                                 }
1762                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1763                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1764                                         }
1765                                         $this->mOutput->addImage( $nt->getDBkey() );
1766                                         wfProfileOut( __METHOD__."-image" );
1767                                         continue;
1768
1769                                 }
1770
1771                                 if ( $ns == NS_CATEGORY ) {
1772                                         wfProfileIn( __METHOD__."-category" );
1773                                         $s = rtrim($s . "\n"); # bug 87
1774
1775                                         if ( $wasblank ) {
1776                                                 $sortkey = $this->getDefaultSort();
1777                                         } else {
1778                                                 $sortkey = $text;
1779                                         }
1780                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1781                                         $sortkey = str_replace( "\n", '', $sortkey );
1782                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1783                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1784
1785                                         /**
1786                                          * Strip the whitespace Category links produce, see bug 87
1787                                          * @todo We might want to use trim($tmp, "\n") here.
1788                                          */
1789                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1790
1791                                         wfProfileOut( __METHOD__."-category" );
1792                                         continue;
1793                                 }
1794                         }
1795
1796                         # Self-link checking
1797                         if( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1798                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1799                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1800                                         continue;
1801                                 }
1802                         }
1803
1804                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1805                         # FIXME: Should do batch file existence checks, see comment below
1806                         if( $ns == NS_MEDIA ) {
1807                                 wfProfileIn( __METHOD__."-media" );
1808                                 # Give extensions a chance to select the file revision for us
1809                                 $skip = $time = false;
1810                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1811                                 if ( $skip ) {
1812                                         $link = $sk->link( $nt );
1813                                 } else {
1814                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1815                                 }
1816                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1817                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1818                                 $this->mOutput->addImage( $nt->getDBkey() );
1819                                 wfProfileOut( __METHOD__."-media" );
1820                                 continue;
1821                         }
1822
1823                         wfProfileIn( __METHOD__."-always_known" );
1824                         # Some titles, such as valid special pages or files in foreign repos, should
1825                         # be shown as bluelinks even though they're not included in the page table
1826                         #
1827                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1828                         # batch file existence checks for NS_FILE and NS_MEDIA
1829                         if( $iw == '' && $nt->isAlwaysKnown() ) {
1830                                 $this->mOutput->addLink( $nt );
1831                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1832                         } else {
1833                                 # Links will be added to the output link list after checking
1834                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1835                         }
1836                         wfProfileOut( __METHOD__."-always_known" );
1837                 }
1838                 wfProfileOut( __METHOD__ );
1839                 return $holders;
1840         }
1841
1842         /**
1843          * Make a link placeholder. The text returned can be later resolved to a real link with
1844          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1845          * parsing of interwiki links, and secondly to allow all existence checks and
1846          * article length checks (for stub links) to be bundled into a single query.
1847          *
1848          * @deprecated
1849          */
1850         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1851                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1852         }
1853
1854         /**
1855          * Render a forced-blue link inline; protect against double expansion of
1856          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1857          * Since this little disaster has to split off the trail text to avoid
1858          * breaking URLs in the following text without breaking trails on the
1859          * wiki links, it's been made into a horrible function.
1860          *
1861          * @param Title $nt
1862          * @param string $text
1863          * @param string $query
1864          * @param string $trail
1865          * @param string $prefix
1866          * @return string HTML-wikitext mix oh yuck
1867          */
1868         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1869                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1870                 $sk = $this->mOptions->getSkin();
1871                 // FIXME: use link() instead of deprecated makeKnownLinkObj()
1872                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1873                 return $this->armorLinks( $link ) . $trail;
1874         }
1875
1876         /**
1877          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1878          * going to go through further parsing steps before inline URL expansion.
1879          *
1880          * Not needed quite as much as it used to be since free links are a bit
1881          * more sensible these days. But bracketed links are still an issue.
1882          *
1883          * @param string more-or-less HTML
1884          * @return string less-or-more HTML with NOPARSE bits
1885          */
1886         function armorLinks( $text ) {
1887                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1888                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1889         }
1890
1891         /**
1892          * Return true if subpage links should be expanded on this page.
1893          * @return bool
1894          */
1895         function areSubpagesAllowed() {
1896                 # Some namespaces don't allow subpages
1897                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1898         }
1899
1900         /**
1901          * Handle link to subpage if necessary
1902          * @param string $target the source of the link
1903          * @param string &$text the link text, modified as necessary
1904          * @return string the full name of the link
1905          * @private
1906          */
1907         function maybeDoSubpageLink($target, &$text) {
1908                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
1909         }
1910
1911         /**#@+
1912          * Used by doBlockLevels()
1913          * @private
1914          */
1915         /* private */ function closeParagraph() {
1916                 $result = '';
1917                 if ( $this->mLastSection != '' ) {
1918                         $result = '</' . $this->mLastSection  . ">\n";
1919                 }
1920                 $this->mInPre = false;
1921                 $this->mLastSection = '';
1922                 return $result;
1923         }
1924         # getCommon() returns the length of the longest common substring
1925         # of both arguments, starting at the beginning of both.
1926         #
1927         /* private */ function getCommon( $st1, $st2 ) {
1928                 $fl = strlen( $st1 );
1929                 $shorter = strlen( $st2 );
1930                 if ( $fl < $shorter ) { $shorter = $fl; }
1931
1932                 for ( $i = 0; $i < $shorter; ++$i ) {
1933                         if ( $st1{$i} != $st2{$i} ) { break; }
1934                 }
1935                 return $i;
1936         }
1937         # These next three functions open, continue, and close the list
1938         # element appropriate to the prefix character passed into them.
1939         #
1940         /* private */ function openList( $char ) {
1941                 $result = $this->closeParagraph();
1942
1943                 if ( '*' === $char ) { $result .= '<ul><li>'; }
1944                 elseif ( '#' === $char ) { $result .= '<ol><li>'; }
1945                 elseif ( ':' === $char ) { $result .= '<dl><dd>'; }
1946                 elseif ( ';' === $char ) {
1947                         $result .= '<dl><dt>';
1948                         $this->mDTopen = true;
1949                 }
1950                 else { $result = '<!-- ERR 1 -->'; }
1951
1952                 return $result;
1953         }
1954
1955         /* private */ function nextItem( $char ) {
1956                 if ( '*' === $char || '#' === $char ) { return '</li><li>'; }
1957                 elseif ( ':' === $char || ';' === $char ) {
1958                         $close = '</dd>';
1959                         if ( $this->mDTopen ) { $close = '</dt>'; }
1960                         if ( ';' === $char ) {
1961                                 $this->mDTopen = true;
1962                                 return $close . '<dt>';
1963                         } else {
1964                                 $this->mDTopen = false;
1965                                 return $close . '<dd>';
1966                         }
1967                 }
1968                 return '<!-- ERR 2 -->';
1969         }
1970
1971         /* private */ function closeList( $char ) {
1972                 if ( '*' === $char ) { $text = '</li></ul>'; }
1973                 elseif ( '#' === $char ) { $text = '</li></ol>'; }
1974                 elseif ( ':' === $char ) {
1975                         if ( $this->mDTopen ) {
1976                                 $this->mDTopen = false;
1977                                 $text = '</dt></dl>';
1978                         } else {
1979                                 $text = '</dd></dl>';
1980                         }
1981                 }
1982                 else {  return '<!-- ERR 3 -->'; }
1983                 return $text."\n";
1984         }
1985         /**#@-*/
1986
1987         /**
1988          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
1989          *
1990          * @param $linestart bool whether or not this is at the start of a line.
1991          * @private
1992          * @return string the lists rendered as HTML
1993          */
1994         function doBlockLevels( $text, $linestart ) {
1995                 wfProfileIn( __METHOD__ );
1996
1997                 # Parsing through the text line by line.  The main thing
1998                 # happening here is handling of block-level elements p, pre,
1999                 # and making lists from lines starting with * # : etc.
2000                 #
2001                 $textLines = StringUtils::explode( "\n", $text );
2002
2003                 $lastPrefix = $output = '';
2004                 $this->mDTopen = $inBlockElem = false;
2005                 $prefixLength = 0;
2006                 $paragraphStack = false;
2007
2008                 foreach ( $textLines as $oLine ) {
2009                         # Fix up $linestart
2010                         if ( !$linestart ) {
2011                                 $output .= $oLine;
2012                                 $linestart = true;
2013                                 continue;
2014                         }
2015                         // * = ul
2016                         // # = ol
2017                         // ; = dt
2018                         // : = dd
2019
2020                         $lastPrefixLength = strlen( $lastPrefix );
2021                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2022                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2023                         // If not in a <pre> element, scan for and figure out what prefixes are there.
2024                         if ( !$this->mInPre ) {
2025                                 # Multiple prefixes may abut each other for nested lists.
2026                                 $prefixLength = strspn( $oLine, '*#:;' );
2027                                 $prefix = substr( $oLine, 0, $prefixLength );
2028
2029                                 # eh?
2030                                 // ; and : are both from definition-lists, so they're equivalent
2031                                 //  for the purposes of determining whether or not we need to open/close
2032                                 //  elements.
2033                                 $prefix2 = str_replace( ';', ':', $prefix );
2034                                 $t = substr( $oLine, $prefixLength );
2035                                 $this->mInPre = (bool)$preOpenMatch;
2036                         } else {
2037                                 # Don't interpret any other prefixes in preformatted text
2038                                 $prefixLength = 0;
2039                                 $prefix = $prefix2 = '';
2040                                 $t = $oLine;
2041                         }
2042
2043                         # List generation
2044                         if( $prefixLength && $lastPrefix === $prefix2 ) {
2045                                 # Same as the last item, so no need to deal with nesting or opening stuff
2046                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2047                                 $paragraphStack = false;
2048
2049                                 if ( substr( $prefix, -1 ) === ';') {
2050                                         # The one nasty exception: definition lists work like this:
2051                                         # ; title : definition text
2052                                         # So we check for : in the remainder text to split up the
2053                                         # title and definition, without b0rking links.
2054                                         $term = $t2 = '';
2055                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2056                                                 $t = $t2;
2057                                                 $output .= $term . $this->nextItem( ':' );
2058                                         }
2059                                 }
2060                         } elseif( $prefixLength || $lastPrefixLength ) {
2061                                 // We need to open or close prefixes, or both.
2062
2063                                 # Either open or close a level...
2064                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2065                                 $paragraphStack = false;
2066
2067                                 // Close all the prefixes which aren't shared.
2068                                 while( $commonPrefixLength < $lastPrefixLength ) {
2069                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2070                                         --$lastPrefixLength;
2071                                 }
2072
2073                                 // Continue the current prefix if appropriate.
2074                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2075                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2076                                 }
2077
2078                                 // Open prefixes where appropriate.
2079                                 while ( $prefixLength > $commonPrefixLength ) {
2080                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2081                                         $output .= $this->openList( $char );
2082
2083                                         if ( ';' === $char ) {
2084                                                 # FIXME: This is dupe of code above
2085                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2086                                                         $t = $t2;
2087                                                         $output .= $term . $this->nextItem( ':' );
2088                                                 }
2089                                         }
2090                                         ++$commonPrefixLength;
2091                                 }
2092                                 $lastPrefix = $prefix2;
2093                         }
2094
2095                         // If we have no prefixes, go to paragraph mode.
2096                         if( 0 == $prefixLength ) {
2097                                 wfProfileIn( __METHOD__."-paragraph" );
2098                                 # No prefix (not in list)--go to paragraph mode
2099                                 // XXX: use a stack for nestable elements like span, table and div
2100                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2101                                 $closematch = preg_match(
2102                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2103                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2104                                 if ( $openmatch or $closematch ) {
2105                                         $paragraphStack = false;
2106                                         # TODO bug 5718: paragraph closed
2107                                         $output .= $this->closeParagraph();
2108                                         if ( $preOpenMatch and !$preCloseMatch ) {
2109                                                 $this->mInPre = true;
2110                                         }
2111                                         if ( $closematch ) {
2112                                                 $inBlockElem = false;
2113                                         } else {
2114                                                 $inBlockElem = true;
2115                                         }
2116                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2117                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' or trim($t) != '' ) ) {
2118                                                 // pre
2119                                                 if ($this->mLastSection !== 'pre') {
2120                                                         $paragraphStack = false;
2121                                                         $output .= $this->closeParagraph().'<pre>';
2122                                                         $this->mLastSection = 'pre';
2123                                                 }
2124                                                 $t = substr( $t, 1 );
2125                                         } else {
2126                                                 // paragraph
2127                                                 if ( trim($t) == '' ) {
2128                                                         if ( $paragraphStack ) {
2129                                                                 $output .= $paragraphStack.'<br />';
2130                                                                 $paragraphStack = false;
2131                                                                 $this->mLastSection = 'p';
2132                                                         } else {
2133                                                                 if ($this->mLastSection !== 'p' ) {
2134                                                                         $output .= $this->closeParagraph();
2135                                                                         $this->mLastSection = '';
2136                                                                         $paragraphStack = '<p>';
2137                                                                 } else {
2138                                                                         $paragraphStack = '</p><p>';
2139                                                                 }
2140                                                         }
2141                                                 } else {
2142                                                         if ( $paragraphStack ) {
2143                                                                 $output .= $paragraphStack;
2144                                                                 $paragraphStack = false;
2145                                                                 $this->mLastSection = 'p';
2146                                                         } else if ($this->mLastSection !== 'p') {
2147                                                                 $output .= $this->closeParagraph().'<p>';
2148                                                                 $this->mLastSection = 'p';
2149                                                         }
2150                                                 }
2151                                         }
2152                                 }
2153                                 wfProfileOut( __METHOD__."-paragraph" );
2154                         }
2155                         // somewhere above we forget to get out of pre block (bug 785)
2156                         if($preCloseMatch && $this->mInPre) {
2157                                 $this->mInPre = false;
2158                         }
2159                         if ($paragraphStack === false) {
2160                                 $output .= $t."\n";
2161                         }
2162                 }
2163                 while ( $prefixLength ) {
2164                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2165                         --$prefixLength;
2166                 }
2167                 if ( $this->mLastSection != '' ) {
2168                         $output .= '</' . $this->mLastSection . '>';
2169                         $this->mLastSection = '';
2170                 }
2171
2172                 wfProfileOut( __METHOD__ );
2173                 return $output;
2174         }
2175
2176         /**
2177          * Split up a string on ':', ignoring any occurences inside tags
2178          * to prevent illegal overlapping.
2179          * @param string $str the string to split
2180          * @param string &$before set to everything before the ':'
2181          * @param string &$after set to everything after the ':'
2182          * return string the position of the ':', or false if none found
2183          */
2184         function findColonNoLinks($str, &$before, &$after) {
2185                 wfProfileIn( __METHOD__ );
2186
2187                 $pos = strpos( $str, ':' );
2188                 if( $pos === false ) {
2189                         // Nothing to find!
2190                         wfProfileOut( __METHOD__ );
2191                         return false;
2192                 }
2193
2194                 $lt = strpos( $str, '<' );
2195                 if( $lt === false || $lt > $pos ) {
2196                         // Easy; no tag nesting to worry about
2197                         $before = substr( $str, 0, $pos );
2198                         $after = substr( $str, $pos+1 );
2199                         wfProfileOut( __METHOD__ );
2200                         return $pos;
2201                 }
2202
2203                 // Ugly state machine to walk through avoiding tags.
2204                 $state = self::COLON_STATE_TEXT;
2205                 $stack = 0;
2206                 $len = strlen( $str );
2207                 for( $i = 0; $i < $len; $i++ ) {
2208                         $c = $str{$i};
2209
2210                         switch( $state ) {
2211                         // (Using the number is a performance hack for common cases)
2212                         case 0: // self::COLON_STATE_TEXT:
2213                                 switch( $c ) {
2214                                 case "<":
2215                                         // Could be either a <start> tag or an </end> tag
2216                                         $state = self::COLON_STATE_TAGSTART;
2217                                         break;
2218                                 case ":":
2219                                         if( $stack == 0 ) {
2220                                                 // We found it!
2221                                                 $before = substr( $str, 0, $i );
2222                                                 $after = substr( $str, $i + 1 );
2223                                                 wfProfileOut( __METHOD__ );
2224                                                 return $i;
2225                                         }
2226                                         // Embedded in a tag; don't break it.
2227                                         break;
2228                                 default:
2229                                         // Skip ahead looking for something interesting
2230                                         $colon = strpos( $str, ':', $i );
2231                                         if( $colon === false ) {
2232                                                 // Nothing else interesting
2233                                                 wfProfileOut( __METHOD__ );
2234                                                 return false;
2235                                         }
2236                                         $lt = strpos( $str, '<', $i );
2237                                         if( $stack === 0 ) {
2238                                                 if( $lt === false || $colon < $lt ) {
2239                                                         // We found it!
2240                                                         $before = substr( $str, 0, $colon );
2241                                                         $after = substr( $str, $colon + 1 );
2242                                                         wfProfileOut( __METHOD__ );
2243                                                         return $i;
2244                                                 }
2245                                         }
2246                                         if( $lt === false ) {
2247                                                 // Nothing else interesting to find; abort!
2248                                                 // We're nested, but there's no close tags left. Abort!
2249                                                 break 2;
2250                                         }
2251                                         // Skip ahead to next tag start
2252                                         $i = $lt;
2253                                         $state = self::COLON_STATE_TAGSTART;
2254                                 }
2255                                 break;
2256                         case 1: // self::COLON_STATE_TAG:
2257                                 // In a <tag>
2258                                 switch( $c ) {
2259                                 case ">":
2260                                         $stack++;
2261                                         $state = self::COLON_STATE_TEXT;
2262                                         break;
2263                                 case "/":
2264                                         // Slash may be followed by >?
2265                                         $state = self::COLON_STATE_TAGSLASH;
2266                                         break;
2267                                 default:
2268                                         // ignore
2269                                 }
2270                                 break;
2271                         case 2: // self::COLON_STATE_TAGSTART:
2272                                 switch( $c ) {
2273                                 case "/":
2274                                         $state = self::COLON_STATE_CLOSETAG;
2275                                         break;
2276                                 case "!":
2277                                         $state = self::COLON_STATE_COMMENT;
2278                                         break;
2279                                 case ">":
2280                                         // Illegal early close? This shouldn't happen D:
2281                                         $state = self::COLON_STATE_TEXT;
2282                                         break;
2283                                 default:
2284                                         $state = self::COLON_STATE_TAG;
2285                                 }
2286                                 break;
2287                         case 3: // self::COLON_STATE_CLOSETAG:
2288                                 // In a </tag>
2289                                 if( $c === ">" ) {
2290                                         $stack--;
2291                                         if( $stack < 0 ) {
2292                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2293                                                 wfProfileOut( __METHOD__ );
2294                                                 return false;
2295                                         }
2296                                         $state = self::COLON_STATE_TEXT;
2297                                 }
2298                                 break;
2299                         case self::COLON_STATE_TAGSLASH:
2300                                 if( $c === ">" ) {
2301                                         // Yes, a self-closed tag <blah/>
2302                                         $state = self::COLON_STATE_TEXT;
2303                                 } else {
2304                                         // Probably we're jumping the gun, and this is an attribute
2305                                         $state = self::COLON_STATE_TAG;
2306                                 }
2307                                 break;
2308                         case 5: // self::COLON_STATE_COMMENT:
2309                                 if( $c === "-" ) {
2310                                         $state = self::COLON_STATE_COMMENTDASH;
2311                                 }
2312                                 break;
2313                         case self::COLON_STATE_COMMENTDASH:
2314                                 if( $c === "-" ) {
2315                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2316                                 } else {
2317                                         $state = self::COLON_STATE_COMMENT;
2318                                 }
2319                                 break;
2320                         case self::COLON_STATE_COMMENTDASHDASH:
2321                                 if( $c === ">" ) {
2322                                         $state = self::COLON_STATE_TEXT;
2323                                 } else {
2324                                         $state = self::COLON_STATE_COMMENT;
2325                                 }
2326                                 break;
2327                         default:
2328                                 throw new MWException( "State machine error in " . __METHOD__ );
2329                         }
2330                 }
2331                 if( $stack > 0 ) {
2332                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2333                         return false;
2334                 }
2335                 wfProfileOut( __METHOD__ );
2336                 return false;
2337         }
2338
2339         /**
2340          * Return value of a magic variable (like PAGENAME)
2341          *
2342          * @private
2343          */
2344         function getVariableValue( $index, $frame=false ) {
2345                 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2346                 global $wgScriptPath, $wgStylePath;
2347
2348                 /**
2349                  * Some of these require message or data lookups and can be
2350                  * expensive to check many times.
2351                  */
2352                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2353                         if ( isset( $this->mVarCache[$index] ) ) {
2354                                 return $this->mVarCache[$index];
2355                         }
2356                 }
2357
2358                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2359                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2360
2361                 # Use the time zone
2362                 global $wgLocaltimezone;
2363                 if ( isset( $wgLocaltimezone ) ) {
2364                         $oldtz = date_default_timezone_get();
2365                         date_default_timezone_set( $wgLocaltimezone );
2366                 }
2367
2368                 $localTimestamp = date( 'YmdHis', $ts );
2369                 $localMonth = date( 'm', $ts );
2370                 $localMonth1 = date( 'n', $ts );
2371                 $localMonthName = date( 'n', $ts );
2372                 $localDay = date( 'j', $ts );
2373                 $localDay2 = date( 'd', $ts );
2374                 $localDayOfWeek = date( 'w', $ts );
2375                 $localWeek = date( 'W', $ts );
2376                 $localYear = date( 'Y', $ts );
2377                 $localHour = date( 'H', $ts );
2378                 if ( isset( $wgLocaltimezone ) ) {
2379                         date_default_timezone_set( $oldtz );
2380                 }
2381
2382                 switch ( $index ) {
2383                         case 'currentmonth':
2384                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2385                                 break;
2386                         case 'currentmonth1':
2387                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2388                                 break;
2389                         case 'currentmonthname':
2390                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2391                                 break;
2392                         case 'currentmonthnamegen':
2393                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2394                                 break;
2395                         case 'currentmonthabbrev':
2396                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2397                                 break;
2398                         case 'currentday':
2399                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2400                                 break;
2401                         case 'currentday2':
2402                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2403                                 break;
2404                         case 'localmonth':
2405                                 $value = $wgContLang->formatNum( $localMonth );
2406                                 break;
2407                         case 'localmonth1':
2408                                 $value = $wgContLang->formatNum( $localMonth1 );
2409                                 break;
2410                         case 'localmonthname':
2411                                 $value = $wgContLang->getMonthName( $localMonthName );
2412                                 break;
2413                         case 'localmonthnamegen':
2414                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2415                                 break;
2416                         case 'localmonthabbrev':
2417                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2418                                 break;
2419                         case 'localday':
2420                                 $value = $wgContLang->formatNum( $localDay );
2421                                 break;
2422                         case 'localday2':
2423                                 $value = $wgContLang->formatNum( $localDay2 );
2424                                 break;
2425                         case 'pagename':
2426                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2427                                 break;
2428                         case 'pagenamee':
2429                                 $value = $this->mTitle->getPartialURL();
2430                                 break;
2431                         case 'fullpagename':
2432                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2433                                 break;
2434                         case 'fullpagenamee':
2435                                 $value = $this->mTitle->getPrefixedURL();
2436                                 break;
2437                         case 'subpagename':
2438                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2439                                 break;
2440                         case 'subpagenamee':
2441                                 $value = $this->mTitle->getSubpageUrlForm();
2442                                 break;
2443                         case 'basepagename':
2444                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2445                                 break;
2446                         case 'basepagenamee':
2447                                 $value = wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2448                                 break;
2449                         case 'talkpagename':
2450                                 if( $this->mTitle->canTalk() ) {
2451                                         $talkPage = $this->mTitle->getTalkPage();
2452                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2453                                 } else {
2454                                         $value = '';
2455                                 }
2456                                 break;
2457                         case 'talkpagenamee':
2458                                 if( $this->mTitle->canTalk() ) {
2459                                         $talkPage = $this->mTitle->getTalkPage();
2460                                         $value = $talkPage->getPrefixedUrl();
2461                                 } else {
2462                                         $value = '';
2463                                 }
2464                                 break;
2465                         case 'subjectpagename':
2466                                 $subjPage = $this->mTitle->getSubjectPage();
2467                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2468                                 break;
2469                         case 'subjectpagenamee':
2470                                 $subjPage = $this->mTitle->getSubjectPage();
2471                                 $value = $subjPage->getPrefixedUrl();
2472                                 break;
2473                         case 'revisionid':
2474                                 // Let the edit saving system know we should parse the page
2475                                 // *after* a revision ID has been assigned.
2476                                 $this->mOutput->setFlag( 'vary-revision' );
2477                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2478                                 $value = $this->mRevisionId;
2479                                 break;
2480                         case 'revisionday':
2481                                 // Let the edit saving system know we should parse the page
2482                                 // *after* a revision ID has been assigned. This is for null edits.
2483                                 $this->mOutput->setFlag( 'vary-revision' );
2484                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2485                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2486                                 break;
2487                         case 'revisionday2':
2488                                 // Let the edit saving system know we should parse the page
2489                                 // *after* a revision ID has been assigned. This is for null edits.
2490                                 $this->mOutput->setFlag( 'vary-revision' );
2491                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2492                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2493                                 break;
2494                         case 'revisionmonth':
2495                                 // Let the edit saving system know we should parse the page
2496                                 // *after* a revision ID has been assigned. This is for null edits.
2497                                 $this->mOutput->setFlag( 'vary-revision' );
2498                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2499                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2500                                 break;
2501                         case 'revisionyear':
2502                                 // Let the edit saving system know we should parse the page
2503                                 // *after* a revision ID has been assigned. This is for null edits.
2504                                 $this->mOutput->setFlag( 'vary-revision' );
2505                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2506                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2507                                 break;
2508                         case 'revisiontimestamp':
2509                                 // Let the edit saving system know we should parse the page
2510                                 // *after* a revision ID has been assigned. This is for null edits.
2511                                 $this->mOutput->setFlag( 'vary-revision' );
2512                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2513                                 $value = $this->getRevisionTimestamp();
2514                                 break;
2515                         case 'revisionuser':
2516                                 // Let the edit saving system know we should parse the page
2517                                 // *after* a revision ID has been assigned. This is for null edits.
2518                                 $this->mOutput->setFlag( 'vary-revision' );
2519                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2520                                 $value = $this->getRevisionUser();
2521                                 break;
2522                         case 'namespace':
2523                                 $value = str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2524                                 break;
2525                         case 'namespacee':
2526                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2527                                 break;
2528                         case 'talkspace':
2529                                 $value = $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2530                                 break;
2531                         case 'talkspacee':
2532                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2533                                 break;
2534                         case 'subjectspace':
2535                                 $value = $this->mTitle->getSubjectNsText();
2536                                 break;
2537                         case 'subjectspacee':
2538                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2539                                 break;
2540                         case 'currentdayname':
2541                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2542                                 break;
2543                         case 'currentyear':
2544                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2545                                 break;
2546                         case 'currenttime':
2547                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2548                                 break;
2549                         case 'currenthour':
2550                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2551                                 break;
2552                         case 'currentweek':
2553                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2554                                 // int to remove the padding
2555                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2556                                 break;
2557                         case 'currentdow':
2558                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2559                                 break;
2560                         case 'localdayname':
2561                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2562                                 break;
2563                         case 'localyear':
2564                                 $value = $wgContLang->formatNum( $localYear, true );
2565                                 break;
2566                         case 'localtime':
2567                                 $value = $wgContLang->time( $localTimestamp, false, false );
2568                                 break;
2569                         case 'localhour':
2570                                 $value = $wgContLang->formatNum( $localHour, true );
2571                                 break;
2572                         case 'localweek':
2573                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2574                                 // int to remove the padding
2575                                 $value = $wgContLang->formatNum( (int)$localWeek );
2576                                 break;
2577                         case 'localdow':
2578                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2579                                 break;
2580                         case 'numberofarticles':
2581                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2582                                 break;
2583                         case 'numberoffiles':
2584                                 $value = $wgContLang->formatNum( SiteStats::images() );
2585                                 break;
2586                         case 'numberofusers':
2587                                 $value = $wgContLang->formatNum( SiteStats::users() );
2588                                 break;
2589                         case 'numberofactiveusers':
2590                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2591                                 break;
2592                         case 'numberofpages':
2593                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2594                                 break;
2595                         case 'numberofadmins':
2596                                 $value = $wgContLang->formatNum( SiteStats::numberingroup('sysop') );
2597                                 break;
2598                         case 'numberofedits':
2599                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2600                                 break;
2601                         case 'numberofviews':
2602                                 $value = $wgContLang->formatNum( SiteStats::views() );
2603                                 break;
2604                         case 'currenttimestamp':
2605                                 $value = wfTimestamp( TS_MW, $ts );
2606                                 break;
2607                         case 'localtimestamp':
2608                                 $value = $localTimestamp;
2609                                 break;
2610                         case 'currentversion':
2611                                 $value = SpecialVersion::getVersion();
2612                                 break;
2613                         case 'sitename':
2614                                 return $wgSitename;
2615                         case 'server':
2616                                 return $wgServer;
2617                         case 'servername':
2618                                 return $wgServerName;
2619                         case 'scriptpath':
2620                                 return $wgScriptPath;
2621                         case 'stylepath':
2622                                 return $wgStylePath;
2623                         case 'directionmark':
2624                                 return $wgContLang->getDirMark();
2625                         case 'contentlanguage':
2626                                 global $wgContLanguageCode;
2627                                 return $wgContLanguageCode;
2628                         default:
2629                                 $ret = null;
2630                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) )
2631                                         return $ret;
2632                                 else
2633                                         return null;
2634                 }
2635
2636                 if ( $index )
2637                         $this->mVarCache[$index] = $value;
2638
2639                 return $value;
2640         }
2641
2642         /**
2643          * initialise the magic variables (like CURRENTMONTHNAME)
2644          *
2645          * @private
2646          */
2647         function initialiseVariables() {
2648                 wfProfileIn( __METHOD__ );
2649                 $variableIDs = MagicWord::getVariableIDs();
2650
2651                 $this->mVariables = new MagicWordArray( $variableIDs );
2652                 wfProfileOut( __METHOD__ );
2653         }
2654
2655         /**
2656          * Preprocess some wikitext and return the document tree.
2657          * This is the ghost of replace_variables().
2658          *
2659          * @param string $text The text to parse
2660          * @param integer flags Bitwise combination of:
2661          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2662          *                                     included. Default is to assume a direct page view.
2663          *
2664          * The generated DOM tree must depend only on the input text and the flags.
2665          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2666          *
2667          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2668          * change in the DOM tree for a given text, must be passed through the section identifier
2669          * in the section edit link and thus back to extractSections().
2670          *
2671          * The output of this function is currently only cached in process memory, but a persistent
2672          * cache may be implemented at a later date which takes further advantage of these strict
2673          * dependency requirements.
2674          *
2675          * @private
2676          */
2677         function preprocessToDom ( $text, $flags = 0 ) {
2678                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2679                 return $dom;
2680         }
2681
2682         /*
2683          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2684          */
2685         public static function splitWhitespace( $s ) {
2686                 $ltrimmed = ltrim( $s );
2687                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2688                 $trimmed = rtrim( $ltrimmed );
2689                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2690                 if ( $diff > 0 ) {
2691                         $w2 = substr( $ltrimmed, -$diff );
2692                 } else {
2693                         $w2 = '';
2694                 }
2695                 return array( $w1, $trimmed, $w2 );
2696         }
2697
2698         /**
2699          * Replace magic variables, templates, and template arguments
2700          * with the appropriate text. Templates are substituted recursively,
2701          * taking care to avoid infinite loops.
2702          *
2703          * Note that the substitution depends on value of $mOutputType:
2704          *  self::OT_WIKI: only {{subst:}} templates
2705          *  self::OT_PREPROCESS: templates but not extension tags
2706          *  self::OT_HTML: all templates and extension tags
2707          *
2708          * @param string $tex The text to transform
2709          * @param PPFrame $frame Object describing the arguments passed to the template.
2710          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2711          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2712          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2713          * @private
2714          */
2715         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2716                 # Is there any text? Also, Prevent too big inclusions!
2717                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2718                         return $text;
2719                 }
2720                 wfProfileIn( __METHOD__ );
2721
2722                 if ( $frame === false ) {
2723                         $frame = $this->getPreprocessor()->newFrame();
2724                 } elseif ( !( $frame instanceof PPFrame ) ) {
2725                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2726                         $frame = $this->getPreprocessor()->newCustomFrame($frame);
2727                 }
2728
2729                 $dom = $this->preprocessToDom( $text );
2730                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2731                 $text = $frame->expand( $dom, $flags );
2732
2733                 wfProfileOut( __METHOD__ );
2734                 return $text;
2735         }
2736
2737         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2738         static function createAssocArgs( $args ) {
2739                 $assocArgs = array();
2740                 $index = 1;
2741                 foreach( $args as $arg ) {
2742                         $eqpos = strpos( $arg, '=' );
2743                         if ( $eqpos === false ) {
2744                                 $assocArgs[$index++] = $arg;
2745                         } else {
2746                                 $name = trim( substr( $arg, 0, $eqpos ) );
2747                                 $value = trim( substr( $arg, $eqpos+1 ) );
2748                                 if ( $value === false ) {
2749                                         $value = '';
2750                                 }
2751                                 if ( $name !== false ) {
2752                                         $assocArgs[$name] = $value;
2753                                 }
2754                         }
2755                 }
2756
2757                 return $assocArgs;
2758         }
2759
2760         /**
2761          * Warn the user when a parser limitation is reached
2762          * Will warn at most once the user per limitation type
2763          *
2764          * @param string $limitationType, should be one of:
2765          *   'expensive-parserfunction' (corresponding messages: 'expensive-parserfunction-warning', 'expensive-parserfunction-category')
2766          *   'post-expand-template-argument' (corresponding messages: 'post-expand-template-argument-warning', 'post-expand-template-argument-category')
2767          *   'post-expand-template-inclusion' (corresponding messages: 'post-expand-template-inclusion-warning', 'post-expand-template-inclusion-category')
2768          * @params int $current, $max When an explicit limit has been
2769          *       exceeded, provide the values (optional)
2770          */
2771         function limitationWarn( $limitationType, $current=null, $max=null) {
2772                 //does no harm if $current and $max are present but are unnecessary for the message
2773                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
2774                 $this->mOutput->addWarning( $warning );
2775                 $this->addTrackingCategory( "$limitationType-category" );
2776         }
2777
2778         /**
2779          * Return the text of a template, after recursively
2780          * replacing any variables or templates within the template.
2781          *
2782          * @param array $piece The parts of the template
2783          *  $piece['title']: the title, i.e. the part before the |
2784          *  $piece['parts']: the parameter array
2785          *  $piece['lineStart']: whether the brace was at the start of a line
2786          * @param PPFrame The current frame, contains template arguments
2787          * @return string the text of the template
2788          * @private
2789          */
2790         function braceSubstitution( $piece, $frame ) {
2791                 global $wgContLang, $wgNonincludableNamespaces;
2792                 wfProfileIn( __METHOD__ );
2793                 wfProfileIn( __METHOD__.'-setup' );
2794
2795                 # Flags
2796                 $found = false;             # $text has been filled
2797                 $nowiki = false;            # wiki markup in $text should be escaped
2798                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2799                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2800                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2801                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2802
2803                 # Title object, where $text came from
2804                 $title = null;
2805
2806                 # $part1 is the bit before the first |, and must contain only title characters.
2807                 # Various prefixes will be stripped from it later.
2808                 $titleWithSpaces = $frame->expand( $piece['title'] );
2809                 $part1 = trim( $titleWithSpaces );
2810                 $titleText = false;
2811
2812                 # Original title text preserved for various purposes
2813                 $originalTitle = $part1;
2814
2815                 # $args is a list of argument nodes, starting from index 0, not including $part1
2816                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2817                 wfProfileOut( __METHOD__.'-setup' );
2818
2819                 # SUBST
2820                 wfProfileIn( __METHOD__.'-modifiers' );
2821                 if ( !$found ) {
2822                         $mwSubst = MagicWord::get( 'subst' );
2823                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2824                                 # One of two possibilities is true:
2825                                 # 1) Found SUBST but not in the PST phase
2826                                 # 2) Didn't find SUBST and in the PST phase
2827                                 # In either case, return without further processing
2828                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2829                                 $isLocalObj = true;
2830                                 $found = true;
2831                         }
2832                 }
2833
2834                 # Variables
2835                 if ( !$found && $args->getLength() == 0 ) {
2836                         $id = $this->mVariables->matchStartToEnd( $part1 );
2837                         if ( $id !== false ) {
2838                                 $text = $this->getVariableValue( $id, $frame );
2839                                 if (MagicWord::getCacheTTL($id)>-1)
2840                                         $this->mOutput->mContainsOldMagic = true;
2841                                 $found = true;
2842                         }
2843                 }
2844
2845                 # MSG, MSGNW and RAW
2846                 if ( !$found ) {
2847                         # Check for MSGNW:
2848                         $mwMsgnw = MagicWord::get( 'msgnw' );
2849                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2850                                 $nowiki = true;
2851                         } else {
2852                                 # Remove obsolete MSG:
2853                                 $mwMsg = MagicWord::get( 'msg' );
2854                                 $mwMsg->matchStartAndRemove( $part1 );
2855                         }
2856
2857                         # Check for RAW:
2858                         $mwRaw = MagicWord::get( 'raw' );
2859                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2860                                 $forceRawInterwiki = true;
2861                         }
2862                 }
2863                 wfProfileOut( __METHOD__.'-modifiers' );
2864
2865                 # Parser functions
2866                 if ( !$found ) {
2867                         wfProfileIn( __METHOD__ . '-pfunc' );
2868
2869                         $colonPos = strpos( $part1, ':' );
2870                         if ( $colonPos !== false ) {
2871                                 # Case sensitive functions
2872                                 $function = substr( $part1, 0, $colonPos );
2873                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2874                                         $function = $this->mFunctionSynonyms[1][$function];
2875                                 } else {
2876                                         # Case insensitive functions
2877                                         $function = $wgContLang->lc( $function );
2878                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2879                                                 $function = $this->mFunctionSynonyms[0][$function];
2880                                         } else {
2881                                                 $function = false;
2882                                         }
2883                                 }
2884                                 if ( $function ) {
2885                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2886                                         $initialArgs = array( &$this );
2887                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2888                                         if ( $flags & SFH_OBJECT_ARGS ) {
2889                                                 # Add a frame parameter, and pass the arguments as an array
2890                                                 $allArgs = $initialArgs;
2891                                                 $allArgs[] = $frame;
2892                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2893                                                         $funcArgs[] = $args->item( $i );
2894                                                 }
2895                                                 $allArgs[] = $funcArgs;
2896                                         } else {
2897                                                 # Convert arguments to plain text
2898                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2899                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2900                                                 }
2901                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2902                                         }
2903
2904                                         # Workaround for PHP bug 35229 and similar
2905                                         if ( !is_callable( $callback ) ) {
2906                                                 wfProfileOut( __METHOD__ . '-pfunc' );
2907                                                 wfProfileOut( __METHOD__ );
2908                                                 throw new MWException( "Tag hook for $function is not callable\n" );
2909                                         }
2910                                         $result = call_user_func_array( $callback, $allArgs );
2911                                         $found = true;
2912                                         $noparse = true;
2913                                         $preprocessFlags = 0;
2914
2915                                         if ( is_array( $result ) ) {
2916                                                 if ( isset( $result[0] ) ) {
2917                                                         $text = $result[0];
2918                                                         unset( $result[0] );
2919                                                 }
2920
2921                                                 // Extract flags into the local scope
2922                                                 // This allows callers to set flags such as nowiki, found, etc.
2923                                                 extract( $result );
2924                                         } else {
2925                                                 $text = $result;
2926                                         }
2927                                         if ( !$noparse ) {
2928                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
2929                                                 $isChildObj = true;
2930                                         }
2931                                 }
2932                         }
2933                         wfProfileOut( __METHOD__ . '-pfunc' );
2934                 }
2935
2936                 # Finish mangling title and then check for loops.
2937                 # Set $title to a Title object and $titleText to the PDBK
2938                 if ( !$found ) {
2939                         $ns = NS_TEMPLATE;
2940                         # Split the title into page and subpage
2941                         $subpage = '';
2942                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2943                         if ($subpage !== '') {
2944                                 $ns = $this->mTitle->getNamespace();
2945                         }
2946                         $title = Title::newFromText( $part1, $ns );
2947                         if ( $title ) {
2948                                 $titleText = $title->getPrefixedText();
2949                                 # Check for language variants if the template is not found
2950                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
2951                                         $wgContLang->findVariantLink( $part1, $title, true );
2952                                 }
2953                                 # Do recursion depth check
2954                                 $limit = $this->mOptions->getMaxTemplateDepth();
2955                                 if ( $frame->depth >= $limit ) {
2956                                         $found = true;
2957                                         $text = '<span class="error">' . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit ) . '</span>';
2958                                 }
2959                         }
2960                 }
2961
2962                 # Load from database
2963                 if ( !$found && $title ) {
2964                         wfProfileIn( __METHOD__ . '-loadtpl' );
2965                         if ( !$title->isExternal() ) {
2966                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
2967                                         $text = SpecialPage::capturePath( $title );
2968                                         if ( is_string( $text ) ) {
2969                                                 $found = true;
2970                                                 $isHTML = true;
2971                                                 $this->disableCache();
2972                                         }
2973                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
2974                                         $found = false; //access denied
2975                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
2976                                 } else {
2977                                         list( $text, $title ) = $this->getTemplateDom( $title );
2978                                         if ( $text !== false ) {
2979                                                 $found = true;
2980                                                 $isChildObj = true;
2981                                         }
2982                                 }
2983
2984                                 # If the title is valid but undisplayable, make a link to it
2985                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2986                                         $text = "[[:$titleText]]";
2987                                         $found = true;
2988                                 }
2989                         } elseif ( $title->isTrans() ) {
2990                                 // Interwiki transclusion
2991                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
2992                                         $text = $this->interwikiTransclude( $title, 'render' );
2993                                         $isHTML = true;
2994                                 } else {
2995                                         $text = $this->interwikiTransclude( $title, 'raw' );
2996                                         // Preprocess it like a template
2997                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
2998                                         $isChildObj = true;
2999                                 }
3000                                 $found = true;
3001                         }
3002
3003                         # Do infinite loop check
3004                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3005                         if ( !$frame->loopCheck( $title ) ) {
3006                                 $found = true;
3007                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3008                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3009                         }
3010                         wfProfileOut( __METHOD__ . '-loadtpl' );
3011                 }
3012
3013                 # If we haven't found text to substitute by now, we're done
3014                 # Recover the source wikitext and return it
3015                 if ( !$found ) {
3016                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3017                         wfProfileOut( __METHOD__ );
3018                         return array( 'object' => $text );
3019                 }
3020
3021                 # Expand DOM-style return values in a child frame
3022                 if ( $isChildObj ) {
3023                         # Clean up argument array
3024                         $newFrame = $frame->newChild( $args, $title );
3025
3026                         if ( $nowiki ) {
3027                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3028                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3029                                 # Expansion is eligible for the empty-frame cache
3030                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3031                                         $text = $this->mTplExpandCache[$titleText];
3032                                 } else {
3033                                         $text = $newFrame->expand( $text );
3034                                         $this->mTplExpandCache[$titleText] = $text;
3035                                 }
3036                         } else {
3037                                 # Uncached expansion
3038                                 $text = $newFrame->expand( $text );
3039                         }
3040                 }
3041                 if ( $isLocalObj && $nowiki ) {
3042                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3043                         $isLocalObj = false;
3044                 }
3045
3046                 # Replace raw HTML by a placeholder
3047                 # Add a blank line preceding, to prevent it from mucking up
3048                 # immediately preceding headings
3049                 if ( $isHTML ) {
3050                         $text = "\n\n" . $this->insertStripItem( $text );
3051                 }
3052                 # Escape nowiki-style return values
3053                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3054                         $text = wfEscapeWikiText( $text );
3055                 }
3056                 # Bug 529: if the template begins with a table or block-level
3057                 # element, it should be treated as beginning a new line.
3058                 # This behaviour is somewhat controversial.
3059                 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3060                         $text = "\n" . $text;
3061                 }
3062
3063                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3064                         # Error, oversize inclusion
3065                         $text = "[[$originalTitle]]" .
3066                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3067                         $this->limitationWarn( 'post-expand-template-inclusion' );
3068                 }
3069
3070                 if ( $isLocalObj ) {
3071                         $ret = array( 'object' => $text );
3072                 } else {
3073                         $ret = array( 'text' => $text );
3074                 }
3075
3076                 wfProfileOut( __METHOD__ );
3077                 return $ret;
3078         }
3079
3080         /**
3081          * Get the semi-parsed DOM representation of a template with a given title,
3082          * and its redirect destination title. Cached.
3083          */
3084         function getTemplateDom( $title ) {
3085                 $cacheTitle = $title;
3086                 $titleText = $title->getPrefixedDBkey();
3087
3088                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3089                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3090                         $title = Title::makeTitle( $ns, $dbk );
3091                         $titleText = $title->getPrefixedDBkey();
3092                 }
3093                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3094                         return array( $this->mTplDomCache[$titleText], $title );
3095                 }
3096
3097                 // Cache miss, go to the database
3098                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3099
3100                 if ( $text === false ) {
3101                         $this->mTplDomCache[$titleText] = false;
3102                         return array( false, $title );
3103                 }
3104
3105                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3106                 $this->mTplDomCache[ $titleText ] = $dom;
3107
3108                 if (! $title->equals($cacheTitle)) {
3109                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3110                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3111                 }
3112
3113                 return array( $dom, $title );
3114         }
3115
3116         /**
3117          * Fetch the unparsed text of a template and register a reference to it.
3118          */
3119         function fetchTemplateAndTitle( $title ) {
3120                 $templateCb = $this->mOptions->getTemplateCallback();
3121                 $stuff = call_user_func( $templateCb, $title, $this );
3122                 $text = $stuff['text'];
3123                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3124                 if ( isset( $stuff['deps'] ) ) {
3125                         foreach ( $stuff['deps'] as $dep ) {
3126                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3127                         }
3128                 }
3129                 return array($text,$finalTitle);
3130         }
3131
3132         function fetchTemplate( $title ) {
3133                 $rv = $this->fetchTemplateAndTitle($title);
3134                 return $rv[0];
3135         }
3136
3137         /**
3138          * Static function to get a template
3139          * Can be overridden via ParserOptions::setTemplateCallback().
3140          */
3141         static function statelessFetchTemplate( $title, $parser=false ) {
3142                 $text = $skip = false;
3143                 $finalTitle = $title;
3144                 $deps = array();
3145
3146                 // Loop to fetch the article, with up to 1 redirect
3147                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3148                         # Give extensions a chance to select the revision instead
3149                         $id = false; // Assume current
3150                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3151
3152                         if( $skip ) {
3153                                 $text = false;
3154                                 $deps[] = array(
3155                                         'title' => $title,
3156                                         'page_id' => $title->getArticleID(),
3157                                         'rev_id' => null );
3158                                 break;
3159                         }
3160                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3161                         $rev_id = $rev ? $rev->getId() : 0;
3162                         // If there is no current revision, there is no page
3163                         if( $id === false && !$rev ) {
3164                                 $linkCache = LinkCache::singleton();
3165                                 $linkCache->addBadLinkObj( $title );
3166                         }
3167
3168                         $deps[] = array(
3169                                 'title' => $title,
3170                                 'page_id' => $title->getArticleID(),
3171                                 'rev_id' => $rev_id );
3172
3173                         if( $rev ) {
3174                                 $text = $rev->getText();
3175                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3176                                 global $wgContLang;
3177                                 $message = $wgContLang->lcfirst( $title->getText() );
3178                                 $text = wfMsgForContentNoTrans( $message );
3179                                 if( wfEmptyMsg( $message, $text ) ) {
3180                                         $text = false;
3181                                         break;
3182                                 }
3183                         } else {
3184                                 break;
3185                         }
3186                         if ( $text === false ) {
3187                                 break;
3188                         }
3189                         // Redirect?
3190                         $finalTitle = $title;
3191                         $title = Title::newFromRedirect( $text );
3192                 }
3193                 return array(
3194                         'text' => $text,
3195                         'finalTitle' => $finalTitle,
3196                         'deps' => $deps );
3197         }
3198
3199         /**
3200          * Transclude an interwiki link.
3201          */
3202         function interwikiTransclude( $title, $action ) {
3203                 global $wgEnableScaryTranscluding;
3204
3205                 if (!$wgEnableScaryTranscluding)
3206                         return wfMsg('scarytranscludedisabled');
3207
3208                 $url = $title->getFullUrl( "action=$action" );
3209
3210                 if (strlen($url) > 255)
3211                         return wfMsg('scarytranscludetoolong');
3212                 return $this->fetchScaryTemplateMaybeFromCache($url);
3213         }
3214
3215         function fetchScaryTemplateMaybeFromCache($url) {
3216                 global $wgTranscludeCacheExpiry;
3217                 $dbr = wfGetDB(DB_SLAVE);
3218                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3219                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3220                                 array('tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3221                 if ($obj) {
3222                         return $obj->tc_contents;
3223                 }
3224
3225                 $text = Http::get($url);
3226                 if (!$text)
3227                         return wfMsg('scarytranscludefailed', $url);
3228
3229                 $dbw = wfGetDB(DB_MASTER);
3230                 $dbw->replace('transcache', array('tc_url'), array(
3231                         'tc_url' => $url,
3232                         'tc_time' => $dbw->timestamp( time() ),
3233                         'tc_contents' => $text));
3234                 return $text;
3235         }
3236
3237
3238         /**
3239          * Triple brace replacement -- used for template arguments
3240          * @private
3241          */
3242         function argSubstitution( $piece, $frame ) {
3243                 wfProfileIn( __METHOD__ );
3244
3245                 $error = false;
3246                 $parts = $piece['parts'];
3247                 $nameWithSpaces = $frame->expand( $piece['title'] );
3248                 $argName = trim( $nameWithSpaces );
3249                 $object = false;
3250                 $text = $frame->getArgument( $argName );
3251                 if (  $text === false && $parts->getLength() > 0
3252                   && (
3253                     $this->ot['html']
3254                     || $this->ot['pre']
3255                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3256                   )
3257                 ) {
3258                         # No match in frame, use the supplied default
3259                         $object = $parts->item( 0 )->getChildren();
3260                 }
3261                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3262                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3263                         $this->limitationWarn( 'post-expand-template-argument' );
3264                 }
3265
3266                 if ( $text === false && $object === false ) {
3267                         # No match anywhere
3268                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3269                 }
3270                 if ( $error !== false ) {
3271                         $text .= $error;
3272                 }
3273                 if ( $object !== false ) {
3274                         $ret = array( 'object' => $object );
3275                 } else {
3276                         $ret = array( 'text' => $text );
3277                 }
3278
3279                 wfProfileOut( __METHOD__ );
3280                 return $ret;
3281         }
3282
3283         /**
3284          * Return the text to be used for a given extension tag.
3285          * This is the ghost of strip().
3286          *
3287          * @param array $params Associative array of parameters:
3288          *     name       PPNode for the tag name
3289          *     attr       PPNode for unparsed text where tag attributes are thought to be
3290          *     attributes Optional associative array of parsed attributes
3291          *     inner      Contents of extension element
3292          *     noClose    Original text did not have a close tag
3293          * @param PPFrame $frame
3294          */
3295         function extensionSubstitution( $params, $frame ) {
3296                 global $wgRawHtml, $wgContLang;
3297
3298                 $name = $frame->expand( $params['name'] );
3299                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3300                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3301
3302                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $this->mMarkerIndex++) . self::MARKER_SUFFIX;
3303
3304                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3305                         ( $this->ot['html'] || $this->ot['pre'] );
3306                 if ( $this->ot['html'] || $isFunctionTag ) {
3307                         $name = strtolower( $name );
3308                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3309                         if ( isset( $params['attributes'] ) ) {
3310                                 $attributes = $attributes + $params['attributes'];
3311                         }
3312                         switch ( $name ) {
3313                                 case 'html':
3314                                         if( $wgRawHtml ) {
3315                                                 $output = $content;
3316                                                 break;
3317                                         } else {
3318                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3319                                         }
3320                                 case 'nowiki':
3321                                         $content = strtr($content, array('-{' => '-&#123;', '}-' => '&#125;-'));
3322                                         $output = Xml::escapeTagsOnly( $content );
3323                                         break;
3324                                 case 'gallery':
3325                                         $output = $this->renderImageGallery( $content, $attributes );
3326                                         break;
3327                                 case 'a':
3328                                         $output = $this->renderHyperlink( $content, $attributes, $frame );
3329                                         break;
3330                                 case 'math':
3331                                         if ( $this->mOptions->getUseTeX() ) {
3332                                                 $output = $wgContLang->armourMath(
3333                                                         MathRenderer::renderMath( $content, $attributes ) );
3334                                                 break;
3335                                         }
3336                                         /* else let a tag hook handle it (bug 21222) */
3337                                 default:
3338                                         if( isset( $this->mTagHooks[$name] ) ) {
3339                                                 # Workaround for PHP bug 35229 and similar
3340                                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3341                                                         throw new MWException( "Tag hook for $name is not callable\n" );
3342                                                 }
3343                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3344                                                         array( $content, $attributes, $this, $frame ) );
3345                                         } elseif( isset( $this->mFunctionTagHooks[$name] ) ) {
3346                                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3347                                                 if( !is_callable( $callback ) )
3348                                                         throw new MWException( "Tag hook for $name is not callable\n" );
3349
3350                                                 $output = call_user_func_array( $callback,
3351                                                         array( &$this, $frame, $content, $attributes ) );
3352                                         } else {
3353                                                 $output = '<span class="error">Invalid tag extension name: ' .
3354                                                         htmlspecialchars( $name ) . '</span>';
3355                                         }
3356                         }
3357                 } else {
3358                         if ( is_null( $attrText ) ) {
3359                                 $attrText = '';
3360                         }
3361                         if ( isset( $params['attributes'] ) ) {
3362                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3363                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3364                                                 htmlspecialchars( $attrValue ) . '"';
3365                                 }
3366                         }
3367                         if ( $content === null ) {
3368                                 $output = "<$name$attrText/>";
3369                         } else {
3370                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3371                                 $output = "<$name$attrText>$content$close";
3372                         }
3373                 }
3374
3375                 if( $isFunctionTag ) {
3376                         return $output;
3377                 } elseif ( $name === 'html' || $name === 'nowiki' ) {
3378                         $this->mStripState->nowiki->setPair( $marker, $output );
3379                 } else {
3380                         $this->mStripState->general->setPair( $marker, $output );
3381                 }
3382                 return $marker;
3383         }
3384
3385         /**
3386          * Increment an include size counter
3387          *
3388          * @param string $type The type of expansion
3389          * @param integer $size The size of the text
3390          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3391          */
3392         function incrementIncludeSize( $type, $size ) {
3393                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3394                         return false;
3395                 } else {
3396                         $this->mIncludeSizes[$type] += $size;
3397                         return true;
3398                 }
3399         }
3400
3401         /**
3402          * Increment the expensive function count
3403          *
3404          * @return boolean False if the limit has been exceeded
3405          */
3406         function incrementExpensiveFunctionCount() {
3407                 global $wgExpensiveParserFunctionLimit;
3408                 $this->mExpensiveFunctionCount++;
3409                 if($this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit) {
3410                         return true;
3411                 }
3412                 return false;
3413         }
3414
3415         /**
3416          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3417          * Fills $this->mDoubleUnderscores, returns the modified text
3418          */
3419         function doDoubleUnderscore( $text ) {
3420                 wfProfileIn( __METHOD__ );
3421
3422                 // The position of __TOC__ needs to be recorded
3423                 $mw = MagicWord::get( 'toc' );
3424                 if( $mw->match( $text ) ) {
3425                         $this->mShowToc = true;
3426                         $this->mForceTocPosition = true;
3427
3428                         // Set a placeholder. At the end we'll fill it in with the TOC.
3429                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3430
3431                         // Only keep the first one.
3432                         $text = $mw->replace( '', $text );
3433                 }
3434
3435                 // Now match and remove the rest of them
3436                 $mwa = MagicWord::getDoubleUnderscoreArray();
3437                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3438
3439                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3440                         $this->mOutput->mNoGallery = true;
3441                 }
3442                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3443                         $this->mShowToc = false;
3444                 }
3445                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3446                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3447                         $this->addTrackingCategory( 'hidden-category-category' );
3448                 }
3449                 # (bug 8068) Allow control over whether robots index a page.
3450                 #
3451                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3452                 # is not desirable, the last one on the page should win.
3453                 if( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3454                         $this->mOutput->setIndexPolicy( 'noindex' );
3455                         $this->addTrackingCategory( 'noindex-category' );
3456                 }
3457                 if( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ){
3458                         $this->mOutput->setIndexPolicy( 'index' );
3459                         $this->addTrackingCategory( 'index-category' );
3460                 }
3461
3462                 wfProfileOut( __METHOD__ );
3463                 return $text;
3464         }
3465
3466         /**
3467          * Add a tracking category, getting the title from a system message,
3468          * or print a debug message if the title is invalid.
3469          * @param $msg String message key
3470          * @return Bool whether the addition was successful
3471          */
3472         protected function addTrackingCategory( $msg ){
3473                 $cat = wfMsgForContent( $msg );
3474
3475                 # Allow tracking categories to be disabled by setting them to "-"
3476                 if( $cat === '-' ) return false;
3477
3478                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3479                 if ( $containerCategory ) {
3480                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3481                         return true;
3482                 } else {
3483                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3484                         return false;
3485                 }
3486         }
3487
3488         /**
3489          * This function accomplishes several tasks:
3490          * 1) Auto-number headings if that option is enabled
3491          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3492          * 3) Add a Table of contents on the top for users who have enabled the option
3493          * 4) Auto-anchor headings
3494          *
3495          * It loops through all headlines, collects the necessary data, then splits up the
3496          * string and re-inserts the newly formatted headlines.
3497          *
3498          * @param string $text
3499          * @param string $origText Original, untouched wikitext
3500          * @param boolean $isMain
3501          * @private
3502          */
3503         function formatHeadings( $text, $origText, $isMain=true ) {
3504                 global $wgMaxTocLevel, $wgContLang, $wgEnforceHtmlIds;
3505
3506                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3507                 $showEditLink = $this->mOptions->getEditSection();
3508
3509                 // Do not call quickUserCan unless necessary
3510                 if( $showEditLink && !$this->mTitle->quickUserCan( 'edit' ) ) {
3511                         $showEditLink = 0;
3512                 }
3513
3514                 # Inhibit editsection links if requested in the page
3515                 if ( isset( $this->mDoubleUnderscores['noeditsection'] )  || $this->mOptions->getIsPrintable() ) {
3516                         $showEditLink = 0;
3517                 }
3518
3519                 # Get all headlines for numbering them and adding funky stuff like [edit]
3520                 # links - this is for later, but we need the number of headlines right now
3521                 $matches = array();
3522                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3523
3524                 # if there are fewer than 4 headlines in the article, do not show TOC
3525                 # unless it's been explicitly enabled.
3526                 $enoughToc = $this->mShowToc &&
3527                         (($numMatches >= 4) || $this->mForceTocPosition);
3528
3529                 # Allow user to stipulate that a page should have a "new section"
3530                 # link added via __NEWSECTIONLINK__
3531                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3532                         $this->mOutput->setNewSection( true );
3533                 }
3534
3535                 # Allow user to remove the "new section"
3536                 # link via __NONEWSECTIONLINK__
3537                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3538                         $this->mOutput->hideNewSection( true );
3539                 }
3540
3541                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3542                 # override above conditions and always show TOC above first header
3543                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3544                         $this->mShowToc = true;
3545                         $enoughToc = true;
3546                 }
3547
3548                 # We need this to perform operations on the HTML
3549                 $sk = $this->mOptions->getSkin();
3550
3551                 # headline counter
3552                 $headlineCount = 0;
3553                 $numVisible = 0;
3554
3555                 # Ugh .. the TOC should have neat indentation levels which can be
3556                 # passed to the skin functions. These are determined here
3557                 $toc = '';
3558                 $full = '';
3559                 $head = array();
3560                 $sublevelCount = array();
3561                 $levelCount = array();
3562                 $toclevel = 0;
3563                 $level = 0;
3564                 $prevlevel = 0;
3565                 $toclevel = 0;
3566                 $prevtoclevel = 0;
3567                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3568                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3569                 $oldType = $this->mOutputType;
3570                 $this->setOutputType( self::OT_WIKI );
3571                 $frame = $this->getPreprocessor()->newFrame();
3572                 $root = $this->preprocessToDom( $origText );
3573                 $node = $root->getFirstChild();
3574                 $byteOffset = 0;
3575                 $tocraw = array();
3576
3577                 foreach( $matches[3] as $headline ) {
3578                         $isTemplate = false;
3579                         $titleText = false;
3580                         $sectionIndex = false;
3581                         $numbering = '';
3582                         $markerMatches = array();
3583                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3584                                 $serial = $markerMatches[1];
3585                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3586                                 $isTemplate = ($titleText != $baseTitleText);
3587                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3588                         }
3589
3590                         if( $toclevel ) {
3591                                 $prevlevel = $level;
3592                                 $prevtoclevel = $toclevel;
3593                         }
3594                         $level = $matches[1][$headlineCount];
3595
3596                         if ( $level > $prevlevel ) {
3597                                 # Increase TOC level
3598                                 $toclevel++;
3599                                 $sublevelCount[$toclevel] = 0;
3600                                 if( $toclevel<$wgMaxTocLevel ) {
3601                                         $prevtoclevel = $toclevel;
3602                                         $toc .= $sk->tocIndent();
3603                                         $numVisible++;
3604                                 }
3605                         }
3606                         elseif ( $level < $prevlevel && $toclevel > 1 ) {
3607                                 # Decrease TOC level, find level to jump to
3608
3609                                 for ($i = $toclevel; $i > 0; $i--) {
3610                                         if ( $levelCount[$i] == $level ) {
3611                                                 # Found last matching level
3612                                                 $toclevel = $i;
3613                                                 break;
3614                                         }
3615                                         elseif ( $levelCount[$i] < $level ) {
3616                                                 # Found first matching level below current level
3617                                                 $toclevel = $i + 1;
3618                                                 break;
3619                                         }
3620                                 }
3621                                 if( $i == 0 ) $toclevel = 1;
3622                                 if( $toclevel<$wgMaxTocLevel ) {
3623                                         if($prevtoclevel < $wgMaxTocLevel) {
3624                                                 # Unindent only if the previous toc level was shown :p
3625                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3626                                                 $prevtoclevel = $toclevel;
3627                                         } else {
3628                                                 $toc .= $sk->tocLineEnd();
3629                                         }
3630                                 }
3631                         }
3632                         else {
3633                                 # No change in level, end TOC line
3634                                 if( $toclevel<$wgMaxTocLevel ) {
3635                                         $toc .= $sk->tocLineEnd();
3636                                 }
3637                         }
3638
3639                         $levelCount[$toclevel] = $level;
3640
3641                         # count number of headlines for each level
3642                         @$sublevelCount[$toclevel]++;
3643                         $dot = 0;
3644                         for( $i = 1; $i <= $toclevel; $i++ ) {
3645                                 if( !empty( $sublevelCount[$i] ) ) {
3646                                         if( $dot ) {
3647                                                 $numbering .= '.';
3648                                         }
3649                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3650                                         $dot = 1;
3651                                 }
3652                         }
3653
3654                         # The safe header is a version of the header text safe to use for links
3655                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3656                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3657
3658                         # Remove link placeholders by the link text.
3659                         #     <!--LINK number-->
3660                         # turns into
3661                         #     link text with suffix
3662                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3663
3664                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3665                         $tocline = preg_replace(
3666                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3667                                 array( '',                          '<$1>'),
3668                                 $safeHeadline
3669                         );
3670                         $tocline = trim( $tocline );
3671
3672                         # For the anchor, strip out HTML-y stuff period
3673                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3674                         $safeHeadline = preg_replace( '/[ _]+/', ' ', $safeHeadline );
3675                         $safeHeadline = trim( $safeHeadline );
3676
3677                         # Save headline for section edit hint before it's escaped
3678                         $headlineHint = $safeHeadline;
3679
3680                         if ( $wgEnforceHtmlIds ) {
3681                                 $legacyHeadline = false;
3682                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3683                                         'noninitial' );
3684                         } else {
3685                                 # For reverse compatibility, provide an id that's
3686                                 # HTML4-compatible, like we used to.
3687                                 #
3688                                 # It may be worth noting, academically, that it's possible for
3689                                 # the legacy anchor to conflict with a non-legacy headline
3690                                 # anchor on the page.  In this case likely the "correct" thing
3691                                 # would be to either drop the legacy anchors or make sure
3692                                 # they're numbered first.  However, this would require people
3693                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3694                                 # manually, so let's not bother worrying about it.
3695                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3696                                         'noninitial' );
3697                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline, 'xml' );
3698
3699                                 if ( $legacyHeadline == $safeHeadline ) {
3700                                         # No reason to have both (in fact, we can't)
3701                                         $legacyHeadline = false;
3702                                 } elseif ( $legacyHeadline != Sanitizer::escapeId(
3703                                 $legacyHeadline, 'xml' ) ) {
3704                                         # The legacy id is invalid XML.  We used to allow this, but
3705                                         # there's no reason to do so anymore.  Backward
3706                                         # compatibility will fail slightly in this case, but it's
3707                                         # no big deal.
3708                                         $legacyHeadline = false;
3709                                 }
3710                         }
3711
3712                         # HTML names must be case-insensitively unique (bug 10721).  FIXME:
3713                         # Does this apply to Unicode characters?  Because we aren't
3714                         # handling those here.
3715                         $arrayKey = strtolower( $safeHeadline );
3716                         if ( $legacyHeadline === false ) {
3717                                 $legacyArrayKey = false;
3718                         } else {
3719                                 $legacyArrayKey = strtolower( $legacyHeadline );
3720                         }
3721
3722                         # count how many in assoc. array so we can track dupes in anchors
3723                         if ( isset( $refers[$arrayKey] ) ) {
3724                                 $refers[$arrayKey]++;
3725                         } else {
3726                                 $refers[$arrayKey] = 1;
3727                         }
3728                         if ( isset( $refers[$legacyArrayKey] ) ) {
3729                                 $refers[$legacyArrayKey]++;
3730                         } else {
3731                                 $refers[$legacyArrayKey] = 1;
3732                         }
3733
3734                         # Don't number the heading if it is the only one (looks silly)
3735                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3736                                 # the two are different if the line contains a link
3737                                 $headline=$numbering . ' ' . $headline;
3738                         }
3739
3740                         # Create the anchor for linking from the TOC to the section
3741                         $anchor = $safeHeadline;
3742                         $legacyAnchor = $legacyHeadline;
3743                         if ( $refers[$arrayKey] > 1 ) {
3744                                 $anchor .= '_' . $refers[$arrayKey];
3745                         }
3746                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3747                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3748                         }
3749                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3750                                 $toc .= $sk->tocLine($anchor, $tocline,
3751                                         $numbering, $toclevel, ($isTemplate ? false : $sectionIndex));
3752                         }
3753
3754                         # Add the section to the section tree
3755                         # Find the DOM node for this header
3756                         while ( $node && !$isTemplate ) {
3757                                 if ( $node->getName() === 'h' ) {
3758                                         $bits = $node->splitHeading();
3759                                         if ( $bits['i'] == $sectionIndex )
3760                                                 break;
3761                                 }
3762                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
3763                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
3764                                 $node = $node->getNextSibling();
3765                         }
3766                         $tocraw[] = array(
3767                                 'toclevel' => $toclevel,
3768                                 'level' => $level,
3769                                 'line' => $tocline,
3770                                 'number' => $numbering,
3771                                 'index' => ($isTemplate ? 'T-' : '' ) . $sectionIndex,
3772                                 'fromtitle' => $titleText,
3773                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
3774                                 'anchor' => $anchor,
3775                         );
3776
3777                         # give headline the correct <h#> tag
3778                         if( $showEditLink && $sectionIndex !== false ) {
3779                                 if( $isTemplate ) {
3780                                         # Put a T flag in the section identifier, to indicate to extractSections()
3781                                         # that sections inside <includeonly> should be counted.
3782                                         $editlink = $sk->doEditSectionLink(Title::newFromText( $titleText ), "T-$sectionIndex");
3783                                 } else {
3784                                         $editlink = $sk->doEditSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3785                                 }
3786                         } else {
3787                                 $editlink = '';
3788                         }
3789                         $head[$headlineCount] = $sk->makeHeadline( $level,
3790                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3791                                 $editlink, $legacyAnchor );
3792
3793                         $headlineCount++;
3794                 }
3795
3796                 $this->setOutputType( $oldType );
3797
3798                 # Never ever show TOC if no headers
3799                 if( $numVisible < 1 ) {
3800                         $enoughToc = false;
3801                 }
3802
3803                 if( $enoughToc ) {
3804                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3805                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3806                         }
3807                         $toc = $sk->tocList( $toc );
3808                         $this->mOutput->setTOCHTML( $toc );
3809                 }
3810
3811                 if ( $isMain ) {
3812                         $this->mOutput->setSections( $tocraw );
3813                 }
3814
3815                 # split up and insert constructed headlines
3816
3817                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3818                 $i = 0;
3819
3820                 foreach( $blocks as $block ) {
3821                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3822                                 # This is the [edit] link that appears for the top block of text when
3823                                 # section editing is enabled
3824
3825                                 # Disabled because it broke block formatting
3826                                 # For example, a bullet point in the top line
3827                                 # $full .= $sk->editSectionLink(0);
3828                         }
3829                         $full .= $block;
3830                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3831                                 # Top anchor now in skin
3832                                 $full = $full.$toc;
3833                         }
3834
3835                         if( !empty( $head[$i] ) ) {
3836                                 $full .= $head[$i];
3837                         }
3838                         $i++;
3839                 }
3840                 if( $this->mForceTocPosition ) {
3841                         return str_replace( '<!--MWTOC-->', $toc, $full );
3842                 } else {
3843                         return $full;
3844                 }
3845         }
3846
3847         /**
3848          * Merge $tree2 into $tree1 by replacing the section with index
3849          * $section in $tree1 and its descendants with the sections in $tree2.
3850          * Note that in the returned section tree, only the 'index' and
3851          * 'byteoffset' fields are guaranteed to be correct.
3852          * @param $tree1 array Section tree from ParserOutput::getSectons()
3853         *  @param $tree2 array Section tree
3854          * @param $section int Section index
3855          * @param $title Title Title both section trees come from
3856          * @param $len2 int Length of the original wikitext for $tree2
3857          * @return array Merged section tree
3858          */
3859         public static function mergeSectionTrees( $tree1, $tree2, $section, $title, $len2 ) {
3860                 global $wgContLang;
3861                 $newTree = array();
3862                 $targetLevel = false;
3863                 $merged = false;
3864                 $lastLevel = 1;
3865                 $nextIndex = 1;
3866                 $numbering = array( 0 );
3867                 $titletext = $title->getPrefixedDBkey();
3868                 foreach ( $tree1 as $s ) {
3869                         if ( $targetLevel !== false ) {
3870                                 if ( $s['level'] <= $targetLevel )
3871                                         // We've skipped enough
3872                                         $targetLevel = false;
3873                                 else
3874                                         continue;
3875                         }
3876                         if ( $s['index'] != $section ||
3877                                         $s['fromtitle'] != $titletext ) {
3878                                 self::incrementNumbering( $numbering,
3879                                         $s['toclevel'], $lastLevel );
3880
3881                                 // Rewrite index, byteoffset and number
3882                                 if ( $s['fromtitle'] == $titletext ) {
3883                                         $s['index'] = $nextIndex++;
3884                                         if ( $merged )
3885                                                 $s['byteoffset'] += $len2;
3886                                 }
3887                                 $s['number']  = implode( '.', array_map(
3888                                         array( $wgContLang, 'formatnum' ),
3889                                         $numbering ) );
3890                                 $lastLevel = $s['toclevel'];
3891                                 $newTree[] = $s;
3892                         } else {
3893                                 // We're at $section
3894                                 // Insert sections from $tree2 here
3895                                 foreach ( $tree2 as $s2 ) {
3896                                         // Rewrite the fields in $s2
3897                                         // before inserting it
3898                                         $s2['toclevel'] += $s['toclevel'] - 1;
3899                                         $s2['level'] += $s['level'] - 1;
3900                                         $s2['index'] = $nextIndex++;
3901                                         $s2['byteoffset'] += $s['byteoffset'];
3902
3903                                         self::incrementNumbering( $numbering,
3904                                                 $s2['toclevel'], $lastLevel );
3905                                         $s2['number']  = implode( '.', array_map(
3906                                                 array( $wgContLang, 'formatnum' ),
3907                                                 $numbering ) );
3908                                         $lastLevel = $s2['toclevel'];
3909                                         $newTree[] = $s2;
3910                                 }
3911                                 // Skip all descendants of $section in $tree1
3912                                 $targetLevel = $s['level'];
3913                                 $merged = true;
3914                         }
3915                 }
3916                 return $newTree;
3917         }
3918
3919         /**
3920          * Increment a section number. Helper function for mergeSectionTrees()
3921          * @param $number array Array representing a section number
3922          * @param $level int Current TOC level (depth)
3923          * @param $lastLevel int Level of previous TOC entry
3924          */
3925         private static function incrementNumbering( &$number, $level, $lastLevel ) {
3926                 if ( $level > $lastLevel )
3927                         $number[$level - 1] = 1;
3928                 else if ( $level < $lastLevel ) {
3929                         foreach ( $number as $key => $unused )
3930                                 if ( $key >= $level )
3931                                         unset( $number[$key] );
3932                         $number[$level - 1]++;
3933                 } else
3934                         $number[$level - 1]++;
3935         }
3936
3937         /**
3938          * Transform wiki markup when saving a page by doing \r\n -> \n
3939          * conversion, substitting signatures, {{subst:}} templates, etc.
3940          *
3941          * @param string $text the text to transform
3942          * @param Title &$title the Title object for the current article
3943          * @param User $user the User object describing the current user
3944          * @param ParserOptions $options parsing options
3945          * @param bool $clearState whether to clear the parser state first
3946          * @return string the altered wiki markup
3947          * @public
3948          */
3949         function preSaveTransform( $text, Title $title, $user, $options, $clearState = true ) {
3950                 $this->mOptions = $options;
3951                 $this->setTitle( $title );
3952                 $this->setOutputType( self::OT_WIKI );
3953
3954                 if ( $clearState ) {
3955                         $this->clearState();
3956                 }
3957
3958                 $pairs = array(
3959                         "\r\n" => "\n",
3960                 );
3961                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3962                 $text = $this->pstPass2( $text, $user );
3963                 $text = $this->mStripState->unstripBoth( $text );
3964                 return $text;
3965         }
3966
3967         /**
3968          * Pre-save transform helper function
3969          * @private
3970          */
3971         function pstPass2( $text, $user ) {
3972                 global $wgContLang, $wgLocaltimezone;
3973
3974                 /* Note: This is the timestamp saved as hardcoded wikitext to
3975                  * the database, we use $wgContLang here in order to give
3976                  * everyone the same signature and use the default one rather
3977                  * than the one selected in each user's preferences.
3978                  *
3979                  * (see also bug 12815)
3980                  */
3981                 $ts = $this->mOptions->getTimestamp();
3982                 if ( isset( $wgLocaltimezone ) ) {
3983                         $tz = $wgLocaltimezone;
3984                 } else {
3985                         $tz = date_default_timezone_get();
3986                 }
3987
3988                 $unixts = wfTimestamp( TS_UNIX, $ts );
3989                 $oldtz = date_default_timezone_get();
3990                 date_default_timezone_set( $tz );
3991                 $ts = date( 'YmdHis', $unixts );
3992                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
3993
3994                 /* Allow translation of timezones trough wiki. date() can return
3995                  * whatever crap the system uses, localised or not, so we cannot
3996                  * ship premade translations.
3997                  */
3998                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
3999                 $value = wfMsgForContent( $key );
4000                 if ( !wfEmptyMsg( $key, $value ) ) $tzMsg = $value;
4001
4002                 date_default_timezone_set( $oldtz );
4003
4004                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4005
4006                 # Variable replacement
4007                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4008                 $text = $this->replaceVariables( $text );
4009
4010                 # Signatures
4011                 $sigText = $this->getUserSig( $user );
4012                 $text = strtr( $text, array(
4013                         '~~~~~' => $d,
4014                         '~~~~' => "$sigText $d",
4015                         '~~~' => $sigText
4016                 ) );
4017
4018                 # Context links: [[|name]] and [[name (context)|]]
4019                 #
4020                 global $wgLegalTitleChars;
4021                 $tc = "[$wgLegalTitleChars]";
4022                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4023
4024                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4025                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
4026                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4027                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4028
4029                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4030                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4031                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4032                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4033
4034                 $t = $this->mTitle->getText();
4035                 $m = array();
4036                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4037                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4038                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4039                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4040                 } else {
4041                         # if there's no context, don't bother duplicating the title
4042                         $text = preg_replace( $p2, '[[\\1]]', $text );
4043                 }
4044
4045                 # Trim trailing whitespace
4046                 $text = rtrim( $text );
4047
4048                 return $text;
4049         }
4050
4051         /**
4052          * Fetch the user's signature text, if any, and normalize to
4053          * validated, ready-to-insert wikitext.
4054          * If you have pre-fetched the nickname or the fancySig option, you can
4055          * specify them here to save a database query.
4056          *
4057          * @param User $user
4058          * @return string
4059          */
4060         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4061                 global $wgMaxSigChars;
4062
4063                 $username = $user->getName();
4064
4065                 // If not given, retrieve from the user object.
4066                 if ( $nickname === false )
4067                         $nickname = $user->getOption( 'nickname' );
4068
4069                 if ( is_null( $fancySig ) )
4070                         $fancySig = $user->getBoolOption( 'fancysig' );
4071
4072                 $nickname = $nickname == null ? $username : $nickname;
4073
4074                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4075                         $nickname = $username;
4076                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4077                 } elseif( $fancySig !== false ) {
4078                         # Sig. might contain markup; validate this
4079                         if( $this->validateSig( $nickname ) !== false ) {
4080                                 # Validated; clean up (if needed) and return it
4081                                 return $this->cleanSig( $nickname, true );
4082                         } else {
4083                                 # Failed to validate; fall back to the default
4084                                 $nickname = $username;
4085                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4086                         }
4087                 }
4088
4089                 // Make sure nickname doesnt get a sig in a sig
4090                 $nickname = $this->cleanSigInSig( $nickname );
4091
4092                 # If we're still here, make it a link to the user page
4093                 $userText = wfEscapeWikiText( $username );
4094                 $nickText = wfEscapeWikiText( $nickname );
4095                 if ( $user->isAnon() )  {
4096                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4097                 } else {
4098                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4099                 }
4100         }
4101
4102         /**
4103          * Check that the user's signature contains no bad XML
4104          *
4105          * @param string $text
4106          * @return mixed An expanded string, or false if invalid.
4107          */
4108         function validateSig( $text ) {
4109                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4110         }
4111
4112         /**
4113          * Clean up signature text
4114          *
4115          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4116          * 2) Substitute all transclusions
4117          *
4118          * @param string $text
4119          * @param $parsing Whether we're cleaning (preferences save) or parsing
4120          * @return string Signature text
4121          */
4122         function cleanSig( $text, $parsing = false ) {
4123                 if ( !$parsing ) {
4124                         global $wgTitle;
4125                         $this->clearState();
4126                         $this->setTitle( $wgTitle );
4127                         $this->mOptions = new ParserOptions;
4128                         $this->setOutputType = self::OT_PREPROCESS;
4129                 }
4130
4131                 # Option to disable this feature
4132                 if ( !$this->mOptions->getCleanSignatures() ) {
4133                         return $text;
4134                 }
4135
4136                 # FIXME: regex doesn't respect extension tags or nowiki
4137                 #  => Move this logic to braceSubstitution()
4138                 $substWord = MagicWord::get( 'subst' );
4139                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4140                 $substText = '{{' . $substWord->getSynonym( 0 );
4141
4142                 $text = preg_replace( $substRegex, $substText, $text );
4143                 $text = $this->cleanSigInSig( $text );
4144                 $dom = $this->preprocessToDom( $text );
4145                 $frame = $this->getPreprocessor()->newFrame();
4146                 $text = $frame->expand( $dom );
4147
4148                 if ( !$parsing ) {
4149                         $text = $this->mStripState->unstripBoth( $text );
4150                 }
4151
4152                 return $text;
4153         }
4154
4155         /**
4156          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4157          * @param string $text
4158          * @return string Signature text with /~{3,5}/ removed
4159          */
4160         function cleanSigInSig( $text ) {
4161                 $text = preg_replace( '/~{3,5}/', '', $text );
4162                 return $text;
4163         }
4164
4165         /**
4166          * Set up some variables which are usually set up in parse()
4167          * so that an external function can call some class members with confidence
4168          * @public
4169          */
4170         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4171                 $this->setTitle( $title );
4172                 $this->mOptions = $options;
4173                 $this->setOutputType( $outputType );
4174                 if ( $clearState ) {
4175                         $this->clearState();
4176                 }
4177         }
4178
4179         /**
4180          * Wrapper for preprocess()
4181          *
4182          * @param string $text the text to preprocess
4183          * @param ParserOptions $options  options
4184          * @return string
4185          * @public
4186          */
4187         function transformMsg( $text, $options ) {
4188                 global $wgTitle;
4189                 static $executing = false;
4190
4191                 # Guard against infinite recursion
4192                 if ( $executing ) {
4193                         return $text;
4194                 }
4195                 $executing = true;
4196
4197                 wfProfileIn(__METHOD__);
4198                 $text = $this->preprocess( $text, $wgTitle, $options );
4199
4200                 $executing = false;
4201                 wfProfileOut(__METHOD__);
4202                 return $text;
4203         }
4204
4205         /**
4206          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4207          * The callback should have the following form:
4208          *    function myParserHook( $text, $params, &$parser ) { ... }
4209          *
4210          * Transform and return $text. Use $parser for any required context, e.g. use
4211          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4212          *
4213          * @public
4214          *
4215          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4216          * @param mixed $callback The callback function (and object) to use for the tag
4217          *
4218          * @return The old value of the mTagHooks array associated with the hook
4219          */
4220         function setHook( $tag, $callback ) {
4221                 $tag = strtolower( $tag );
4222                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4223                 $this->mTagHooks[$tag] = $callback;
4224                 if( !in_array( $tag, $this->mStripList ) ) {
4225                         $this->mStripList[] = $tag;
4226                 }
4227
4228                 return $oldVal;
4229         }
4230
4231         function setTransparentTagHook( $tag, $callback ) {
4232                 $tag = strtolower( $tag );
4233                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4234                 $this->mTransparentTagHooks[$tag] = $callback;
4235
4236                 return $oldVal;
4237         }
4238
4239         /**
4240          * Remove all tag hooks
4241          */
4242         function clearTagHooks() {
4243                 $this->mTagHooks = array();
4244                 $this->mStripList = $this->mDefaultStripList;
4245         }
4246
4247         /**
4248          * Create a function, e.g. {{sum:1|2|3}}
4249          * The callback function should have the form:
4250          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4251          *
4252          * Or with SFH_OBJECT_ARGS:
4253          *    function myParserFunction( $parser, $frame, $args ) { ... }
4254          *
4255          * The callback may either return the text result of the function, or an array with the text
4256          * in element 0, and a number of flags in the other elements. The names of the flags are
4257          * specified in the keys. Valid flags are:
4258          *   found                     The text returned is valid, stop processing the template. This
4259          *                             is on by default.
4260          *   nowiki                    Wiki markup in the return value should be escaped
4261          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4262          *
4263          * @public
4264          *
4265          * @param string $id The magic word ID
4266          * @param mixed $callback The callback function (and object) to use
4267          * @param integer $flags a combination of the following flags:
4268          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4269          *
4270          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4271          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4272          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4273          *     the arguments, and to control the way they are expanded.
4274          *
4275          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4276          *     arguments, for instance:
4277          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4278          *
4279          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4280          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4281          *     working if/when this is changed.
4282          *
4283          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4284          *     expansion.
4285          *
4286          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4287          *     about the methods available in PPFrame and PPNode.
4288          *
4289          * @return The old callback function for this name, if any
4290          */
4291         function setFunctionHook( $id, $callback, $flags = 0 ) {
4292                 global $wgContLang;
4293
4294                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4295                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4296
4297                 # Add to function cache
4298                 $mw = MagicWord::get( $id );
4299                 if( !$mw )
4300                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4301
4302                 $synonyms = $mw->getSynonyms();
4303                 $sensitive = intval( $mw->isCaseSensitive() );
4304
4305                 foreach ( $synonyms as $syn ) {
4306                         # Case
4307                         if ( !$sensitive ) {
4308                                 $syn = $wgContLang->lc( $syn );
4309                         }
4310                         # Add leading hash
4311                         if ( !( $flags & SFH_NO_HASH ) ) {
4312                                 $syn = '#' . $syn;
4313                         }
4314                         # Remove trailing colon
4315                         if ( substr( $syn, -1, 1 ) === ':' ) {
4316                                 $syn = substr( $syn, 0, -1 );
4317                         }
4318                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4319                 }
4320                 return $oldVal;
4321         }
4322
4323         /**
4324          * Get all registered function hook identifiers
4325          *
4326          * @return array
4327          */
4328         function getFunctionHooks() {
4329                 return array_keys( $this->mFunctionHooks );
4330         }
4331
4332         /**
4333          * Create a tag function, e.g. <test>some stuff</test>.
4334          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4335          * Unlike parser functions, their content is not preprocessed.
4336          */
4337         function setFunctionTagHook( $tag, $callback, $flags ) {
4338                 $tag = strtolower( $tag );
4339                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4340                         $this->mFunctionTagHooks[$tag] : null;
4341                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4342
4343                 if( !in_array( $tag, $this->mStripList ) ) {
4344                         $this->mStripList[] = $tag;
4345                 }
4346
4347                 return $old;
4348         }
4349
4350         /**
4351          * FIXME: update documentation. makeLinkObj() is deprecated.
4352          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4353          * Placeholders created in Skin::makeLinkObj()
4354          * Returns an array of link CSS classes, indexed by PDBK.
4355          */
4356         function replaceLinkHolders( &$text, $options = 0 ) {
4357                 return $this->mLinkHolders->replace( $text );
4358         }
4359
4360         /**
4361          * Replace <!--LINK--> link placeholders with plain text of links
4362          * (not HTML-formatted).
4363          * @param string $text
4364          * @return string
4365          */
4366         function replaceLinkHoldersText( $text ) {
4367                 return $this->mLinkHolders->replaceText( $text );
4368         }
4369
4370         /**
4371          * Tag hook handler for 'pre'.
4372          */
4373         function renderPreTag( $text, $attribs ) {
4374                 // Backwards-compatibility hack
4375                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4376
4377                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4378                 return Xml::openElement( 'pre', $attribs ) .
4379                         Xml::escapeTagsOnly( $content ) .
4380                         '</pre>';
4381         }
4382
4383         /**
4384         * Tag hook handler for 'a'. Renders a HTML &lt;a&gt; tag, allowing most attributes, filtering href against
4385         * allowed protocols and spam blacklist.
4386         **/
4387         function renderHyperlink( $text, $params, $frame = false ) {
4388                 foreach ( $params as $name => $value ) {
4389                         $params[ $name ] = $this->replaceVariables( $value, $frame );
4390                 }
4391
4392                 $whitelist = Sanitizer::attributeWhitelist( 'a' );
4393                 $params = Sanitizer::validateAttributes( $params, $whitelist );
4394
4395                 $content = $this->recursiveTagParse( trim( $text ), $frame );
4396
4397                 if ( isset( $params[ 'href' ] ) ) {
4398                         $href = $params[ 'href' ];
4399                         $this->mOutput->addExternalLink( $href );
4400                         unset( $params[ 'href' ] );
4401                 } else {
4402                         # Non-link <a> tag
4403                         return Xml::openElement( 'a', $params ) . $content . Xml::closeElement( 'a' );
4404                 }
4405
4406                 $sk = $this->mOptions->getSkin();
4407                 $html = $sk->makeExternalLink( $href, $content, false, '', $params );
4408
4409                 return $html;
4410         }
4411
4412         /**
4413          * Renders an image gallery from a text with one line per image.
4414          * text labels may be given by using |-style alternative text. E.g.
4415          *   Image:one.jpg|The number "1"
4416          *   Image:tree.jpg|A tree
4417          * given as text will return the HTML of a gallery with two images,
4418          * labeled 'The number "1"' and
4419          * 'A tree'.
4420          */
4421         function renderImageGallery( $text, $params ) {
4422                 $ig = new ImageGallery();
4423                 $ig->setContextTitle( $this->mTitle );
4424                 $ig->setShowBytes( false );
4425                 $ig->setShowFilename( false );
4426                 $ig->setParser( $this );
4427                 $ig->setHideBadImages();
4428                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4429                 $ig->useSkin( $this->mOptions->getSkin() );
4430                 $ig->mRevisionId = $this->mRevisionId;
4431
4432                 if( isset( $params['caption'] ) ) {
4433                         $caption = $params['caption'];
4434                         $caption = htmlspecialchars( $caption );
4435                         $caption = $this->replaceInternalLinks( $caption );
4436                         $ig->setCaptionHtml( $caption );
4437                 }
4438                 if( isset( $params['perrow'] ) ) {
4439                         $ig->setPerRow( $params['perrow'] );
4440                 }
4441                 if( isset( $params['widths'] ) ) {
4442                         $ig->setWidths( $params['widths'] );
4443                 }
4444                 if( isset( $params['heights'] ) ) {
4445                         $ig->setHeights( $params['heights'] );
4446                 }
4447
4448                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4449
4450                 $lines = StringUtils::explode( "\n", $text );
4451                 foreach ( $lines as $line ) {
4452                         # match lines like these:
4453                         # Image:someimage.jpg|This is some image
4454                         $matches = array();
4455                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4456                         # Skip empty lines
4457                         if ( count( $matches ) == 0 ) {
4458                                 continue;
4459                         }
4460
4461                         if ( strpos( $matches[0], '%' ) !== false )
4462                                 $matches[1] = urldecode( $matches[1] );
4463                         $tp = Title::newFromText( $matches[1]/*, NS_FILE*/ );
4464                         $nt =& $tp;
4465                         if( is_null( $nt ) ) {
4466                                 # Bogus title. Ignore these so we don't bomb out later.
4467                                 continue;
4468                         }
4469                         if ( isset( $matches[3] ) ) {
4470                                 $label = $matches[3];
4471                         } else {
4472                                 $label = '';
4473                         }
4474
4475                         $html = $this->recursiveTagParse( trim( $label ) );
4476
4477                         $ig->add( $nt, $html );
4478
4479                         # Only add real images (bug #5586)
4480                         if ( $nt->getNamespace() == NS_FILE ) {
4481                                 $this->mOutput->addImage( $nt->getDBkey() );
4482                         }
4483                 }
4484                 return $ig->toHTML();
4485         }
4486
4487         function getImageParams( $handler ) {
4488                 if ( $handler ) {
4489                         $handlerClass = get_class( $handler );
4490                 } else {
4491                         $handlerClass = '';
4492                 }
4493                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4494                         // Initialise static lists
4495                         static $internalParamNames = array(
4496                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4497                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4498                                         'bottom', 'text-bottom' ),
4499                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4500                                         'upright', 'border', 'link', 'alt' ),
4501                         );
4502                         static $internalParamMap;
4503                         if ( !$internalParamMap ) {
4504                                 $internalParamMap = array();
4505                                 foreach ( $internalParamNames as $type => $names ) {
4506                                         foreach ( $names as $name ) {
4507                                                 $magicName = str_replace( '-', '_', "img_$name" );
4508                                                 $internalParamMap[$magicName] = array( $type, $name );
4509                                         }
4510                                 }
4511                         }
4512
4513                         // Add handler params
4514                         $paramMap = $internalParamMap;
4515                         if ( $handler ) {
4516                                 $handlerParamMap = $handler->getParamMap();
4517                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4518                                         $paramMap[$magic] = array( 'handler', $paramName );
4519                                 }
4520                         }
4521                         $this->mImageParams[$handlerClass] = $paramMap;
4522                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4523                 }
4524                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4525         }
4526
4527         /**
4528          * Parse image options text and use it to make an image
4529          * @param Title $title
4530          * @param string $options
4531          * @param LinkHolderArray $holders
4532          */
4533         function makeImage( $title, $options, $holders = false ) {
4534                 # Check if the options text is of the form "options|alt text"
4535                 # Options are:
4536                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4537                 #  * left       no resizing, just left align. label is used for alt= only
4538                 #  * right      same, but right aligned
4539                 #  * none       same, but not aligned
4540                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4541                 #  * center     center the image
4542                 #  * frame      Keep original image size, no magnify-button.
4543                 #  * framed     Same as "frame"
4544                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4545                 #  * upright    reduce width for upright images, rounded to full __0 px
4546                 #  * border     draw a 1px border around the image
4547                 #  * alt        Text for HTML alt attribute (defaults to empty)
4548                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4549                 # vertical-align values (no % or length right now):
4550                 #  * baseline
4551                 #  * sub
4552                 #  * super
4553                 #  * top
4554                 #  * text-top
4555                 #  * middle
4556                 #  * bottom
4557                 #  * text-bottom
4558
4559                 $parts = StringUtils::explode( "|", $options );
4560                 $sk = $this->mOptions->getSkin();
4561
4562                 # Give extensions a chance to select the file revision for us
4563                 $skip = $time = $descQuery = false;
4564                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4565
4566                 if ( $skip ) {
4567                         return $sk->link( $title );
4568                 }
4569
4570                 # Get the file
4571                 $imagename = $title->getDBkey();
4572                 $file = wfFindFile( $title, array( 'time' => $time ) );
4573                 # Get parameter map
4574                 $handler = $file ? $file->getHandler() : false;
4575
4576                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4577
4578                 # Process the input parameters
4579                 $caption = '';
4580                 $params = array( 'frame' => array(), 'handler' => array(),
4581                         'horizAlign' => array(), 'vertAlign' => array() );
4582                 foreach( $parts as $part ) {
4583                         $part = trim( $part );
4584                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4585                         $validated = false;
4586                         if( isset( $paramMap[$magicName] ) ) {
4587                                 list( $type, $paramName ) = $paramMap[$magicName];
4588
4589                                 // Special case; width and height come in one variable together
4590                                 if( $type === 'handler' && $paramName === 'width' ) {
4591                                         $m = array();
4592                                         # (bug 13500) In both cases (width/height and width only),
4593                                         # permit trailing "px" for backward compatibility.
4594                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4595                                                 $width = intval( $m[1] );
4596                                                 $height = intval( $m[2] );
4597                                                 if ( $handler->validateParam( 'width', $width ) ) {
4598                                                         $params[$type]['width'] = $width;
4599                                                         $validated = true;
4600                                                 }
4601                                                 if ( $handler->validateParam( 'height', $height ) ) {
4602                                                         $params[$type]['height'] = $height;
4603                                                         $validated = true;
4604                                                 }
4605                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4606                                                 $width = intval( $value );
4607                                                 if ( $handler->validateParam( 'width', $width ) ) {
4608                                                         $params[$type]['width'] = $width;
4609                                                         $validated = true;
4610                                                 }
4611                                         } // else no validation -- bug 13436
4612                                 } else {
4613                                         if ( $type === 'handler' ) {
4614                                                 # Validate handler parameter
4615                                                 $validated = $handler->validateParam( $paramName, $value );
4616                                         } else {
4617                                                 # Validate internal parameters
4618                                                 switch( $paramName ) {
4619                                                 case 'manualthumb':
4620                                                 case 'alt':
4621                                                         // @todo Fixme: possibly check validity here for
4622                                                         // manualthumb? downstream behavior seems odd with
4623                                                         // missing manual thumbs.
4624                                                         $validated = true;
4625                                                         $value = $this->stripAltText( $value, $holders );
4626                                                         break;
4627                                                 case 'link':
4628                                                         $chars = self::EXT_LINK_URL_CLASS;
4629                                                         $prots = $this->mUrlProtocols;
4630                                                         if ( $value === '' ) {
4631                                                                 $paramName = 'no-link';
4632                                                                 $value = true;
4633                                                                 $validated = true;
4634                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4635                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4636                                                                         $paramName = 'link-url';
4637                                                                         $this->mOutput->addExternalLink( $value );
4638                                                                         $validated = true;
4639                                                                 }
4640                                                         } else {
4641                                                                 $linkTitle = Title::newFromText( $value );
4642                                                                 if ( $linkTitle ) {
4643                                                                         $paramName = 'link-title';
4644                                                                         $value = $linkTitle;
4645                                                                         $this->mOutput->addLink( $linkTitle );
4646                                                                         $validated = true;
4647                                                                 }
4648                                                         }
4649                                                         break;
4650                                                 default:
4651                                                         // Most other things appear to be empty or numeric...
4652                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4653                                                 }
4654                                         }
4655
4656                                         if ( $validated ) {
4657                                                 $params[$type][$paramName] = $value;
4658                                         }
4659                                 }
4660                         }
4661                         if ( !$validated ) {
4662                                 $caption = $part;
4663                         }
4664                 }
4665
4666                 # Process alignment parameters
4667                 if ( $params['horizAlign'] ) {
4668                         $params['frame']['align'] = key( $params['horizAlign'] );
4669                 }
4670                 if ( $params['vertAlign'] ) {
4671                         $params['frame']['valign'] = key( $params['vertAlign'] );
4672                 }
4673
4674                 $params['frame']['caption'] = $caption;
4675
4676                 # Will the image be presented in a frame, with the caption below?
4677                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4678                                  isset( $params['frame']['framed'] ) ||
4679                                  isset( $params['frame']['thumbnail'] ) ||
4680                                  isset( $params['frame']['manualthumb'] );
4681
4682                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4683                 # came to also set the caption, ordinary text after the image -- which
4684                 # makes no sense, because that just repeats the text multiple times in
4685                 # screen readers.  It *also* came to set the title attribute.
4686                 #
4687                 # Now that we have an alt attribute, we should not set the alt text to
4688                 # equal the caption: that's worse than useless, it just repeats the
4689                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4690                 # use the unnamed parameter for alt text as well, just for the time be-
4691                 # ing, if the unnamed param is set and the alt param is not.
4692                 #
4693                 # For the future, we need to figure out if we want to tweak this more,
4694                 # e.g., introducing a title= parameter for the title; ignoring the un-
4695                 # named parameter entirely for images without a caption; adding an ex-
4696                 # plicit caption= parameter and preserving the old magic unnamed para-
4697                 # meter for BC; ...
4698                 if ( $imageIsFramed ) { # Framed image
4699                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4700                                 # No caption or alt text, add the filename as the alt text so
4701                                 # that screen readers at least get some description of the image
4702                                 $params['frame']['alt'] = $title->getText();
4703                         }
4704                         # Do not set $params['frame']['title'] because tooltips don't make sense
4705                         # for framed images
4706                 } else { # Inline image
4707                         if ( !isset( $params['frame']['alt'] ) ) {
4708                                 # No alt text, use the "caption" for the alt text
4709                                 if ( $caption !== '') {
4710                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4711                                 } else {
4712                                         # No caption, fall back to using the filename for the
4713                                         # alt text
4714                                         $params['frame']['alt'] = $title->getText();
4715                                 }
4716                         }
4717                         # Use the "caption" for the tooltip text
4718                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4719                 }
4720
4721                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4722
4723                 # Linker does the rest
4724                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4725
4726                 # Give the handler a chance to modify the parser object
4727                 if ( $handler ) {
4728                         $handler->parserTransformHook( $this, $file );
4729                 }
4730
4731                 return $ret;
4732         }
4733
4734         protected function stripAltText( $caption, $holders ) {
4735                 # Strip bad stuff out of the title (tooltip).  We can't just use
4736                 # replaceLinkHoldersText() here, because if this function is called
4737                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4738                 if ( $holders ) {
4739                         $tooltip = $holders->replaceText( $caption );
4740                 } else {
4741                         $tooltip = $this->replaceLinkHoldersText( $caption );
4742                 }
4743
4744                 # make sure there are no placeholders in thumbnail attributes
4745                 # that are later expanded to html- so expand them now and
4746                 # remove the tags
4747                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4748                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4749
4750                 return $tooltip;
4751         }
4752
4753         /**
4754          * Set a flag in the output object indicating that the content is dynamic and
4755          * shouldn't be cached.
4756          */
4757         function disableCache() {
4758                 wfDebug( "Parser output marked as uncacheable.\n" );
4759                 $this->mOutput->mCacheTime = -1;
4760         }
4761
4762         /**#@+
4763          * Callback from the Sanitizer for expanding items found in HTML attribute
4764          * values, so they can be safely tested and escaped.
4765          * @param string $text
4766          * @param PPFrame $frame
4767          * @return string
4768          * @private
4769          */
4770         function attributeStripCallback( &$text, $frame = false ) {
4771                 $text = $this->replaceVariables( $text, $frame );
4772                 $text = $this->mStripState->unstripBoth( $text );
4773                 return $text;
4774         }
4775
4776         /**#@-*/
4777
4778         /**#@+
4779          * Accessor/mutator
4780          */
4781         function Title( $x = null ) { return wfSetVar( $this->mTitle, $x ); }
4782         function Options( $x = null ) { return wfSetVar( $this->mOptions, $x ); }
4783         function OutputType( $x = null ) { return wfSetVar( $this->mOutputType, $x ); }
4784         /**#@-*/
4785
4786         /**#@+
4787          * Accessor
4788          */
4789         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4790         /**#@-*/
4791
4792
4793         /**
4794          * Break wikitext input into sections, and either pull or replace
4795          * some particular section's text.
4796          *
4797          * External callers should use the getSection and replaceSection methods.
4798          *
4799          * @param string $text Page wikitext
4800          * @param string $section A section identifier string of the form:
4801          *   <flag1> - <flag2> - ... - <section number>
4802          *
4803          * Currently the only recognised flag is "T", which means the target section number
4804          * was derived during a template inclusion parse, in other words this is a template
4805          * section edit link. If no flags are given, it was an ordinary section edit link.
4806          * This flag is required to avoid a section numbering mismatch when a section is
4807          * enclosed by <includeonly> (bug 6563).
4808          *
4809          * The section number 0 pulls the text before the first heading; other numbers will
4810          * pull the given section along with its lower-level subsections. If the section is
4811          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4812          *
4813          * @param string $mode One of "get" or "replace"
4814          * @param string $newText Replacement text for section data.
4815          * @return string for "get", the extracted section text.
4816          *                for "replace", the whole page with the section replaced.
4817          */
4818         private function extractSections( $text, $section, $mode, $newText='' ) {
4819                 global $wgTitle;
4820                 $this->clearState();
4821                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4822                 $this->mOptions = new ParserOptions;
4823                 $this->setOutputType( self::OT_WIKI );
4824                 $outText = '';
4825                 $frame = $this->getPreprocessor()->newFrame();
4826
4827                 // Process section extraction flags
4828                 $flags = 0;
4829                 $sectionParts = explode( '-', $section );
4830                 $sectionIndex = array_pop( $sectionParts );
4831                 foreach ( $sectionParts as $part ) {
4832                         if ( $part === 'T' ) {
4833                                 $flags |= self::PTD_FOR_INCLUSION;
4834                         }
4835                 }
4836                 // Preprocess the text
4837                 $root = $this->preprocessToDom( $text, $flags );
4838
4839                 // <h> nodes indicate section breaks
4840                 // They can only occur at the top level, so we can find them by iterating the root's children
4841                 $node = $root->getFirstChild();
4842
4843                 // Find the target section
4844                 if ( $sectionIndex == 0 ) {
4845                         // Section zero doesn't nest, level=big
4846                         $targetLevel = 1000;
4847                 } else {
4848             while ( $node ) {
4849                 if ( $node->getName() === 'h' ) {
4850                     $bits = $node->splitHeading();
4851                                         if ( $bits['i'] == $sectionIndex ) {
4852                                         $targetLevel = $bits['level'];
4853                                                 break;
4854                                         }
4855                                 }
4856                                 if ( $mode === 'replace' ) {
4857                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4858                                 }
4859                                 $node = $node->getNextSibling();
4860                         }
4861                 }
4862
4863                 if ( !$node ) {
4864                         // Not found
4865                         if ( $mode === 'get' ) {
4866                                 return $newText;
4867                         } else {
4868                                 return $text;
4869                         }
4870                 }
4871
4872                 // Find the end of the section, including nested sections
4873                 do {
4874                         if ( $node->getName() === 'h' ) {
4875                                 $bits = $node->splitHeading();
4876                                 $curLevel = $bits['level'];
4877                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4878                                         break;
4879                                 }
4880                         }
4881                         if ( $mode === 'get' ) {
4882                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4883                         }
4884                         $node = $node->getNextSibling();
4885                 } while ( $node );
4886
4887                 // Write out the remainder (in replace mode only)
4888                 if ( $mode === 'replace' ) {
4889                         // Output the replacement text
4890                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4891                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4892                         // Only add trailing whitespace if there is newText
4893                         if($newText != "") {
4894                                 $outText .= $newText . "\n\n";
4895                         }
4896
4897                         while ( $node ) {
4898                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4899                                 $node = $node->getNextSibling();
4900                         }
4901                 }
4902
4903                 if ( is_string( $outText ) ) {
4904                         // Re-insert stripped tags
4905                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4906                 }
4907
4908                 return $outText;
4909         }
4910
4911         /**
4912          * This function returns the text of a section, specified by a number ($section).
4913          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4914          * the first section before any such heading (section 0).
4915          *
4916          * If a section contains subsections, these are also returned.
4917          *
4918          * @param string $text text to look in
4919          * @param string $section section identifier
4920          * @param string $deftext default to return if section is not found
4921          * @return string text of the requested section
4922          */
4923         public function getSection( $text, $section, $deftext='' ) {
4924                 return $this->extractSections( $text, $section, "get", $deftext );
4925         }
4926
4927         public function replaceSection( $oldtext, $section, $text ) {
4928                 return $this->extractSections( $oldtext, $section, "replace", $text );
4929         }
4930
4931         /**
4932          * Get the timestamp associated with the current revision, adjusted for
4933          * the default server-local timestamp
4934          */
4935         function getRevisionTimestamp() {
4936                 if ( is_null( $this->mRevisionTimestamp ) ) {
4937                         wfProfileIn( __METHOD__ );
4938                         global $wgContLang;
4939                         $dbr = wfGetDB( DB_SLAVE );
4940                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4941                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4942
4943                         // Normalize timestamp to internal MW format for timezone processing.
4944                         // This has the added side-effect of replacing a null value with
4945                         // the current time, which gives us more sensible behavior for
4946                         // previews.
4947                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4948
4949                         // The cryptic '' timezone parameter tells to use the site-default
4950                         // timezone offset instead of the user settings.
4951                         //
4952                         // Since this value will be saved into the parser cache, served
4953                         // to other users, and potentially even used inside links and such,
4954                         // it needs to be consistent for all visitors.
4955                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4956
4957                         wfProfileOut( __METHOD__ );
4958                 }
4959                 return $this->mRevisionTimestamp;
4960         }
4961
4962         /**
4963          * Get the name of the user that edited the last revision
4964          */
4965         function getRevisionUser() {
4966                 // if this template is subst: the revision id will be blank,
4967                 // so just use the current user's name
4968                 if( $this->mRevisionId ) {
4969                         $revision = Revision::newFromId( $this->mRevisionId );
4970                         $revuser = $revision->getUserText();
4971                 } else {
4972                         global $wgUser;
4973                         $revuser = $wgUser->getName();
4974                 }
4975                 return $revuser;
4976         }
4977
4978         /**
4979          * Mutator for $mDefaultSort
4980          *
4981          * @param $sort New value
4982          */
4983         public function setDefaultSort( $sort ) {
4984                 $this->mDefaultSort = $sort;
4985         }
4986
4987         /**
4988          * Accessor for $mDefaultSort
4989          * Will use the title/prefixed title if none is set
4990          *
4991          * @return string
4992          */
4993         public function getDefaultSort() {
4994                 global $wgCategoryPrefixedDefaultSortkey;
4995                 if( $this->mDefaultSort !== false ) {
4996                         return $this->mDefaultSort;
4997                 } elseif ($this->mTitle->getNamespace() == NS_CATEGORY ||
4998                         !$wgCategoryPrefixedDefaultSortkey) {
4999                         return $this->mTitle->getText();
5000                 } else {
5001                         return $this->mTitle->getPrefixedText();
5002                 }
5003         }
5004
5005         /**
5006          * Accessor for $mDefaultSort
5007          * Unlike getDefaultSort(), will return false if none is set
5008          *
5009          * @return string or false
5010          */
5011         public function getCustomDefaultSort() {
5012                 return $this->mDefaultSort;
5013         }
5014
5015         /**
5016          * Try to guess the section anchor name based on a wikitext fragment
5017          * presumably extracted from a heading, for example "Header" from
5018          * "== Header ==".
5019          */
5020         public function guessSectionNameFromWikiText( $text ) {
5021                 # Strip out wikitext links(they break the anchor)
5022                 $text = $this->stripSectionName( $text );
5023                 $headline = Sanitizer::decodeCharReferences( $text );
5024                 # strip out HTML
5025                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
5026                 $headline = trim( $headline );
5027                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
5028                 $replacearray = array(
5029                         '%3A' => ':',
5030                         '%' => '.'
5031                 );
5032                 return str_replace(
5033                         array_keys( $replacearray ),
5034                         array_values( $replacearray ),
5035                         $sectionanchor );
5036         }
5037
5038         /**
5039          * Strips a text string of wikitext for use in a section anchor
5040          *
5041          * Accepts a text string and then removes all wikitext from the
5042          * string and leaves only the resultant text (i.e. the result of
5043          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5044          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5045          * to create valid section anchors by mimicing the output of the
5046          * parser when headings are parsed.
5047          *
5048          * @param $text string Text string to be stripped of wikitext
5049          * for use in a Section anchor
5050          * @return Filtered text string
5051          */
5052         public function stripSectionName( $text ) {
5053                 # Strip internal link markup
5054                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
5055                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
5056
5057                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5058                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5059                 # on how many empty links there are on the page - need to figure that out.
5060                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
5061
5062                 # Parse wikitext quotes (italics & bold)
5063                 $text = $this->doQuotes($text);
5064
5065                 # Strip HTML tags
5066                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5067                 return $text;
5068         }
5069
5070         function srvus( $text ) {
5071                 return $this->testSrvus( $text, $this->mOutputType );
5072         }
5073
5074         /**
5075          * strip/replaceVariables/unstrip for preprocessor regression testing
5076          */
5077         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
5078                 $this->clearState();
5079                 if ( ! ( $title instanceof Title ) ) {
5080                         $title = Title::newFromText( $title );
5081                 }
5082                 $this->mTitle = $title;
5083                 $this->mOptions = $options;
5084                 $this->setOutputType( $outputType );
5085                 $text = $this->replaceVariables( $text );
5086                 $text = $this->mStripState->unstripBoth( $text );
5087                 $text = Sanitizer::removeHTMLtags( $text );
5088                 return $text;
5089         }
5090
5091         function testPst( $text, $title, $options ) {
5092                 global $wgUser;
5093                 if ( ! ( $title instanceof Title ) ) {
5094                         $title = Title::newFromText( $title );
5095                 }
5096                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5097         }
5098
5099         function testPreprocess( $text, $title, $options ) {
5100                 if ( ! ( $title instanceof Title ) ) {
5101                         $title = Title::newFromText( $title );
5102                 }
5103                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5104         }
5105
5106         function markerSkipCallback( $s, $callback ) {
5107                 $i = 0;
5108                 $out = '';
5109                 while ( $i < strlen( $s ) ) {
5110                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5111                         if ( $markerStart === false ) {
5112                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5113                                 break;
5114                         } else {
5115                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5116                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5117                                 if ( $markerEnd === false ) {
5118                                         $out .= substr( $s, $markerStart );
5119                                         break;
5120                                 } else {
5121                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5122                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5123                                         $i = $markerEnd;
5124                                 }
5125                         }
5126                 }
5127                 return $out;
5128         }
5129
5130         function serialiseHalfParsedText( $text ) {
5131                 $data = array();
5132                 $data['text'] = $text;
5133
5134                 // First, find all strip markers, and store their
5135                 //  data in an array.
5136                 $stripState = new StripState;
5137                 $pos = 0;
5138                 while( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) ) && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) ) {
5139                         $end_pos += strlen( self::MARKER_SUFFIX );
5140                         $marker = substr( $text, $start_pos, $end_pos-$start_pos );
5141
5142                         if ( !empty( $this->mStripState->general->data[$marker] ) ) {
5143                                 $replaceArray = $stripState->general;
5144                                 $stripText = $this->mStripState->general->data[$marker];
5145                         } elseif ( !empty( $this->mStripState->nowiki->data[$marker] ) ) {
5146                                 $replaceArray = $stripState->nowiki;
5147                                 $stripText = $this->mStripState->nowiki->data[$marker];
5148                         } else {
5149                                 throw new MWException( "Hanging strip marker: '$marker'." );
5150                         }
5151
5152                         $replaceArray->setPair( $marker, $stripText );
5153                         $pos = $end_pos;
5154                 }
5155                 $data['stripstate'] = $stripState;
5156
5157                 // Now, find all of our links, and store THEIR
5158                 //  data in an array! :)
5159                 $links = array( 'internal' => array(), 'interwiki' => array() );
5160                 $pos = 0;
5161
5162                 // Internal links
5163                 while( ( $start_pos = strpos( $text, '<!--LINK ', $pos ) ) ) {
5164                         list( $ns, $trail ) = explode( ':', substr( $text, $start_pos + strlen( '<!--LINK ' ) ), 2 );
5165
5166                         $ns = trim($ns);
5167                         if (empty( $links['internal'][$ns] )) {
5168                                 $links['internal'][$ns] = array();
5169                         }
5170
5171                         $key = trim( substr( $trail, 0, strpos( $trail, '-->' ) ) );
5172                         $links['internal'][$ns][] = $this->mLinkHolders->internals[$ns][$key];
5173                         $pos = $start_pos + strlen( "<!--LINK $ns:$key-->" );
5174                 }
5175
5176                 $pos = 0;
5177
5178                 // Interwiki links
5179                 while( ( $start_pos = strpos( $text, '<!--IWLINK ', $pos ) ) ) {
5180                         $data = substr( $text, $start_pos );
5181                         $key = trim( substr( $data, 0, strpos( $data, '-->' ) ) );
5182                         $links['interwiki'][] = $this->mLinkHolders->interwiki[$key];
5183                         $pos = $start_pos + strlen( "<!--IWLINK $key-->" );
5184                 }
5185
5186                 $data['linkholder'] = $links;
5187
5188                 return $data;
5189         }
5190
5191         function unserialiseHalfParsedText( $data, $intPrefix = null /* Unique identifying prefix */ ) {
5192                 if (!$intPrefix)
5193                         $intPrefix = $this->getRandomString();
5194
5195                 // First, extract the strip state.
5196                 $stripState = $data['stripstate'];
5197                 $this->mStripState->general->merge( $stripState->general );
5198                 $this->mStripState->nowiki->merge( $stripState->nowiki );
5199
5200                 // Now, extract the text, and renumber links
5201                 $text = $data['text'];
5202                 $links = $data['linkholder'];
5203
5204                 // Internal...
5205                 foreach( $links['internal'] as $ns => $nsLinks ) {
5206                         foreach( $nsLinks as $key => $entry ) {
5207                                 $newKey = $intPrefix . '-' . $key;
5208                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
5209
5210                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
5211                         }
5212                 }
5213
5214                 // Interwiki...
5215                 foreach( $links['interwiki'] as $key => $entry ) {
5216                         $newKey = "$intPrefix-$key";
5217                         $this->mLinkHolders->interwikis[$newKey] = $entry;
5218
5219                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
5220                 }
5221
5222                 // Should be good to go.
5223                 return $text;
5224         }
5225 }
5226
5227 /**
5228  * @todo document, briefly.
5229  * @ingroup Parser
5230  */
5231 class StripState {
5232         var $general, $nowiki;
5233
5234         function __construct() {
5235                 $this->general = new ReplacementArray;
5236                 $this->nowiki = new ReplacementArray;
5237         }
5238
5239         function unstripGeneral( $text ) {
5240                 wfProfileIn( __METHOD__ );
5241                 do {
5242                         $oldText = $text;
5243                         $text = $this->general->replace( $text );
5244                 } while ( $text !== $oldText );
5245                 wfProfileOut( __METHOD__ );
5246                 return $text;
5247         }
5248
5249         function unstripNoWiki( $text ) {
5250                 wfProfileIn( __METHOD__ );
5251                 do {
5252                         $oldText = $text;
5253                         $text = $this->nowiki->replace( $text );
5254                 } while ( $text !== $oldText );
5255                 wfProfileOut( __METHOD__ );
5256                 return $text;
5257         }
5258
5259         function unstripBoth( $text ) {
5260                 wfProfileIn( __METHOD__ );
5261                 do {
5262                         $oldText = $text;
5263                         $text = $this->general->replace( $text );
5264                         $text = $this->nowiki->replace( $text );
5265                 } while ( $text !== $oldText );
5266                 wfProfileOut( __METHOD__ );
5267                 return $text;
5268         }
5269 }
5270
5271 /**
5272  * @todo document, briefly.
5273  * @ingroup Parser
5274  */
5275 class OnlyIncludeReplacer {
5276         var $output = '';
5277
5278         function replace( $matches ) {
5279                 if ( substr( $matches[1], -1 ) === "\n" ) {
5280                         $this->output .= substr( $matches[1], 0, -1 );
5281                 } else {
5282                         $this->output .= $matches[1];
5283                 }
5284         }
5285 }