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  * getPreloadText()
  30  *     Removes <noinclude> sections, and <includeonly> tags.
  31  *
  32  * Globals used:
  33  *    objects:   $wgLang, $wgContLang
  34  *
  35  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  36  *
  37  * settings:
  38  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  39  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  40  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  41  *  $wgMaxArticleSize*
  42  *
  43  *  * only within ParserOptions
  44  * </pre>
  45  *
  46  * @ingroup Parser
  47  */
  48 class Parser {
  49         /**
  50          * Update this version number when the ParserOutput format
  51          * changes in an incompatible way, so the parser cache
  52          * can automatically discard old data.
  53          */
  54         const VERSION = '1.6.4';
  55
  56         # Flags for Parser::setFunctionHook
  57         # Also available as global constants from Defines.php
  58         const SFH_NO_HASH = 1;
  59         const SFH_OBJECT_ARGS = 2;
  60
  61         # Constants needed for external link processing
  62         # Everything except bracket, space, or control characters
  63         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  64         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  65                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  66
  67         # State constants for the definition list colon extraction
  68         const COLON_STATE_TEXT = 0;
  69         const COLON_STATE_TAG = 1;
  70         const COLON_STATE_TAGSTART = 2;
  71         const COLON_STATE_CLOSETAG = 3;
  72         const COLON_STATE_TAGSLASH = 4;
  73         const COLON_STATE_COMMENT = 5;
  74         const COLON_STATE_COMMENTDASH = 6;
  75         const COLON_STATE_COMMENTDASHDASH = 7;
  76
  77         # Flags for preprocessToDom
  78         const PTD_FOR_INCLUSION = 1;
  79
  80         # Allowed values for $this->mOutputType
  81         # Parameter to startExternalParse().
  82         const OT_HTML = 1; # like parse()
  83         const OT_WIKI = 2; # like preSaveTransform()
  84         const OT_PREPROCESS = 3; # like preprocess()
  85         const OT_MSG = 3;
  86         const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
  87
  88         # Marker Suffix needs to be accessible staticly.
  89         const MARKER_SUFFIX = "-QINU\x7f";
  90
  91         # Persistent:
  92         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables;
  93         var $mSubstWords, $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerIndex;
  94         var $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols, $mDefaultStripList;
  95         var $mVarCache, $mConf, $mFunctionTagHooks;
  96
  97
  98         # Cleared with clearState():
  99         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 100         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 101         var $mLinkHolders, $mLinkID;
 102         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 103         var $mTplExpandCache; # empty-frame expansion cache
 104         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 105         var $mExpensiveFunctionCount; # number of expensive parser function calls
 106
 107         # Temporary
 108         # These are variables reset at least once per parse regardless of $clearState
 109         var $mOptions;      # ParserOptions object
 110         var $mTitle;        # Title context, used for self-link rendering and similar things
 111         var $mOutputType;   # Output type, one of the OT_xxx constants
 112         var $ot;            # Shortcut alias, see setOutputType()
 113         var $mRevisionId;   # ID to display in {{REVISIONID}} tags
 114         var $mRevisionTimestamp; # The timestamp of the specified revision ID
 115         var $mRevIdForTs;   # The revision ID which was used to fetch the timestamp
 116
 117         /**
 118          * Constructor
 119          *
 120          * @public
 121          */
 122         function __construct( $conf = array() ) {
 123                 $this->mConf = $conf;
 124                 $this->mTagHooks = array();
 125                 $this->mTransparentTagHooks = array();
 126                 $this->mFunctionHooks = array();
 127                 $this->mFunctionTagHooks = array();
 128                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 129                 $this->mDefaultStripList = $this->mStripList = array();
 130                 $this->mUrlProtocols = wfUrlProtocols();
 131                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 132                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 133                 $this->mVarCache = array();
 134                 if ( isset( $conf['preprocessorClass'] ) ) {
 135                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 136                 } elseif ( extension_loaded( 'domxml' ) ) {
 137                         # PECL extension that conflicts with the core DOM extension (bug 13770)
 138                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 139                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 140                 } elseif ( extension_loaded( 'dom' ) ) {
 141                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 142                 } else {
 143                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 144                 }
 145                 $this->mMarkerIndex = 0;
 146                 $this->mFirstCall = true;
 147         }
 148
 149         /**
 150          * Reduce memory usage to reduce the impact of circular references
 151          */
 152         function __destruct() {
 153                 if ( isset( $this->mLinkHolders ) ) {
 154                         $this->mLinkHolders->__destruct();
 155                 }
 156                 foreach ( $this as $name => $value ) {
 157                         unset( $this->$name );
 158                 }
 159         }
 160
 161         /**
 162          * Do various kinds of initialisation on the first call of the parser
 163          */
 164         function firstCallInit() {
 165                 if ( !$this->mFirstCall ) {
 166                         return;
 167                 }
 168                 $this->mFirstCall = false;
 169
 170                 wfProfileIn( __METHOD__ );
 171
 172                 CoreParserFunctions::register( $this );
 173                 CoreTagHooks::register( $this );
 174                 $this->initialiseVariables();
 175
 176                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 177                 wfProfileOut( __METHOD__ );
 178         }
 179
 180         /**
 181          * Clear Parser state
 182          *
 183          * @private
 184          */
 185         function clearState() {
 186                 wfProfileIn( __METHOD__ );
 187                 if ( $this->mFirstCall ) {
 188                         $this->firstCallInit();
 189                 }
 190                 $this->mOutput = new ParserOutput;
 191                 $this->mAutonumber = 0;
 192                 $this->mLastSection = '';
 193                 $this->mDTopen = false;
 194                 $this->mIncludeCount = array();
 195                 $this->mStripState = new StripState;
 196                 $this->mArgStack = false;
 197                 $this->mInPre = false;
 198                 $this->mLinkHolders = new LinkHolderArray( $this );
 199                 $this->mLinkID = 0;
 200                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 201                 $this->mVarCache = array();
 202
 203                 /**
 204                  * Prefix for temporary replacement strings for the multipass parser.
 205                  * \x07 should never appear in input as it's disallowed in XML.
 206                  * Using it at the front also gives us a little extra robustness
 207                  * since it shouldn't match when butted up against identifier-like
 208                  * string constructs.
 209                  *
 210                  * Must not consist of all title characters, or else it will change
 211                  * the behaviour of <nowiki> in a link.
 212                  */
 213                 # $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 214                 # Changed to \x7f to allow XML double-parsing -- TS
 215                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 216
 217
 218                 # Clear these on every parse, bug 4549
 219                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 220
 221                 $this->mShowToc = true;
 222                 $this->mForceTocPosition = false;
 223                 $this->mIncludeSizes = array(
 224                         'post-expand' => 0,
 225                         'arg' => 0,
 226                 );
 227                 $this->mPPNodeCount = 0;
 228                 $this->mDefaultSort = false;
 229                 $this->mHeadings = array();
 230                 $this->mDoubleUnderscores = array();
 231                 $this->mExpensiveFunctionCount = 0;
 232
 233                 # Fix cloning
 234                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 235                         $this->mPreprocessor = null;
 236                 }
 237
 238                 wfRunHooks( 'ParserClearState', array( &$this ) );
 239                 wfProfileOut( __METHOD__ );
 240         }
 241
 242         function setOutputType( $ot ) {
 243                 $this->mOutputType = $ot;
 244                 # Shortcut alias
 245                 $this->ot = array(
 246                         'html' => $ot == self::OT_HTML,
 247                         'wiki' => $ot == self::OT_WIKI,
 248                         'pre' => $ot == self::OT_PREPROCESS,
 249                         'plain' => $ot == self::OT_PLAIN,
 250                 );
 251         }
 252
 253         /**
 254          * Set the context title
 255          */
 256         function setTitle( $t ) {
 257                 if ( !$t || $t instanceof FakeTitle ) {
 258                         $t = Title::newFromText( 'NO TITLE' );
 259                 }
 260
 261                 if ( strval( $t->getFragment() ) !== '' ) {
 262                         # Strip the fragment to avoid various odd effects
 263                         $this->mTitle = clone $t;
 264                         $this->mTitle->setFragment( '' );
 265                 } else {
 266                         $this->mTitle = $t;
 267                 }
 268         }
 269
 270         /**
 271          * Accessor for mUniqPrefix.
 272          *
 273          * @public
 274          */
 275         function uniqPrefix() {
 276                 if ( !isset( $this->mUniqPrefix ) ) {
 277                         # @todo Fixme: this is probably *horribly wrong*
 278                         # LanguageConverter seems to want $wgParser's uniqPrefix, however
 279                         # if this is called for a parser cache hit, the parser may not
 280                         # have ever been initialized in the first place.
 281                         # Not really sure what the heck is supposed to be going on here.
 282                         return '';
 283                         # throw new MWException( "Accessing uninitialized mUniqPrefix" );
 284                 }
 285                 return $this->mUniqPrefix;
 286         }
 287
 288         /**
 289          * Convert wikitext to HTML
 290          * Do not call this function recursively.
 291          *
 292          * @param $text String: text we want to parse
 293          * @param $title A title object
 294          * @param $options ParserOptions
 295          * @param $linestart boolean
 296          * @param $clearState boolean
 297          * @param $revid Int: number to pass in {{REVISIONID}}
 298          * @return ParserOutput a ParserOutput
 299          */
 300         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 301                 /**
 302                  * First pass--just handle <nowiki> sections, pass the rest off
 303                  * to internalParse() which does all the real work.
 304                  */
 305
 306                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang, $wgDisableLangConversion;
 307                 $fname = __METHOD__.'-' . wfGetCaller();
 308                 wfProfileIn( __METHOD__ );
 309                 wfProfileIn( $fname );
 310
 311                 if ( $clearState ) {
 312                         $this->clearState();
 313                 }
 314
 315                 $this->mOptions = $options;
 316                 $this->setTitle( $title ); # Page title has to be set for the pre-processor
 317
 318                 $oldRevisionId = $this->mRevisionId;
 319                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 320                 if ( $revid !== null ) {
 321                         $this->mRevisionId = $revid;
 322                         $this->mRevisionTimestamp = null;
 323                 }
 324                 $this->setOutputType( self::OT_HTML );
 325                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 326                 # No more strip!
 327                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 328                 $text = $this->internalParse( $text );
 329
 330                 $text = $this->mStripState->unstripGeneral( $text );
 331
 332                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 333                 $fixtags = array(
 334                         # french spaces, last one Guillemet-left
 335                         # only if there is something before the space
 336                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 337                         # french spaces, Guillemet-right
 338                         '/(\\302\\253) /' => '\\1&nbsp;',
 339                         '/&nbsp;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
 340                 );
 341                 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
 342
 343                 $text = $this->doBlockLevels( $text, $linestart );
 344
 345                 $this->replaceLinkHolders( $text );
 346
 347                 /**
 348                  * The page doesn't get language converted if
 349                  * a) It's disabled
 350                  * b) Titles aren't converted
 351          * c) Content isn't converted and this is not a talk page
 352          * d) It's a conversion table
 353                  */
 354                 if ( !( $wgDisableLangConversion
 355                                 || isset( $this->mDoubleUnderscores['notitleconvert'] )
 356                                 || ( isset( $this->mDoubleUnderscores['nocontentconvert'] ) && !$this->mTitle->isTalkPage() )
 357                                 || $this->mTitle->isConversionTable() ) ) {
 358
 359                         # The position of the convert() call should not be changed. it
 360                         # assumes that the links are all replaced and the only thing left
 361                         # is the <nowiki> mark.
 362
 363                         $text = $wgContLang->convert( $text );
 364                 }
 365
 366                 # A title may have been set in a conversion rule.
 367                 # Note that if a user tries to set a title in a conversion
 368                 # rule but content conversion was not done, then the parser
 369                 # won't pick it up.  This is probably expected behavior.
 370                 if ( $wgContLang->getConvRuleTitle() ) {
 371                         $this->mOutput->setTitleText( $wgContLang->getConvRuleTitle() );
 372                 }
 373
 374                 $text = $this->mStripState->unstripNoWiki( $text );
 375
 376                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 377
 378 //!JF Move to its own function
 379
 380                 $uniq_prefix = $this->mUniqPrefix;
 381                 $matches = array();
 382                 $elements = array_keys( $this->mTransparentTagHooks );
 383                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 384
 385                 foreach ( $matches as $marker => $data ) {
 386                         list( $element, $content, $params, $tag ) = $data;
 387                         $tagName = strtolower( $element );
 388                         if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 389                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
 390                         } else {
 391                                 $output = $tag;
 392                         }
 393                         $this->mStripState->general->setPair( $marker, $output );
 394                 }
 395                 $text = $this->mStripState->unstripGeneral( $text );
 396
 397                 $text = Sanitizer::normalizeCharReferences( $text );
 398
 399                 if ( ( $wgUseTidy && $this->mOptions->mTidy ) || $wgAlwaysUseTidy ) {
 400                         $text = MWTidy::tidy( $text );
 401                 } else {
 402                         # attempt to sanitize at least some nesting problems
 403                         # (bug #2702 and quite a few others)
 404                         $tidyregs = array(
 405                                 # ''Something [http://www.cool.com cool''] -->
 406                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 407                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 408                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 409                                 # fix up an anchor inside another anchor, only
 410                                 # at least for a single single nested link (bug 3695)
 411                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 412                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 413                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 414                                 # contains a div, so fix it up here; replace
 415                                 # div with escaped text
 416                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 417                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 418                                 # remove empty italic or bold tag pairs, some
 419                                 # introduced by rules above
 420                                 '/<([bi])><\/\\1>/' => '',
 421                         );
 422
 423                         $text = preg_replace(
 424                                 array_keys( $tidyregs ),
 425                                 array_values( $tidyregs ),
 426                                 $text );
 427                 }
 428                 global $wgExpensiveParserFunctionLimit;
 429                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 430                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 431                 }
 432
 433                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 434
 435                 # Information on include size limits, for the benefit of users who try to skirt them
 436                 if ( $this->mOptions->getEnableLimitReport() ) {
 437                         $max = $this->mOptions->getMaxIncludeSize();
 438                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 439                         $limitReport =
 440                                 "NewPP limit report\n" .
 441                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 442                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 443                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 444                                 $PFreport;
 445                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 446                         $text .= "\n<!-- \n$limitReport-->\n";
 447                 }
 448                 $this->mOutput->setText( $text );
 449
 450                 $this->mRevisionId = $oldRevisionId;
 451                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 452                 wfProfileOut( $fname );
 453                 wfProfileOut( __METHOD__ );
 454
 455                 return $this->mOutput;
 456         }
 457
 458         /**
 459          * Recursive parser entry point that can be called from an extension tag
 460          * hook.
 461          *
 462          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 463          *
 464          * @param $text String: text extension wants to have parsed
 465          * @param PPFrame $frame: The frame to use for expanding any template variables
 466          */
 467         function recursiveTagParse( $text, $frame=false ) {
 468                 wfProfileIn( __METHOD__ );
 469                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 470                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 471                 $text = $this->internalParse( $text, false, $frame );
 472                 wfProfileOut( __METHOD__ );
 473                 return $text;
 474         }
 475
 476         /**
 477          * Expand templates and variables in the text, producing valid, static wikitext.
 478          * Also removes comments.
 479          */
 480         function preprocess( $text, $title, $options, $revid = null ) {
 481                 wfProfileIn( __METHOD__ );
 482                 $this->clearState();
 483                 $this->setOutputType( self::OT_PREPROCESS );
 484                 $this->mOptions = $options;
 485                 $this->setTitle( $title );
 486                 if ( $revid !== null ) {
 487                         $this->mRevisionId = $revid;
 488                 }
 489                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 490                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 491                 $text = $this->replaceVariables( $text );
 492                 $text = $this->mStripState->unstripBoth( $text );
 493                 wfProfileOut( __METHOD__ );
 494                 return $text;
 495         }
 496
 497         /**
 498          * Process the wikitext for the ?preload= feature. (bug 5210)
 499          *
 500          * <noinclude>, <includeonly> etc. are parsed as for template transclusion,
 501          * comments, templates, arguments, tags hooks and parser functions are untouched.
 502          */
 503         public function getPreloadText( $text, $title, $options ) {
 504                 # Parser (re)initialisation
 505                 $this->clearState();
 506                 $this->setOutputType( self::OT_PLAIN );
 507                 $this->mOptions = $options;
 508                 $this->setTitle( $title );
 509
 510                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 511                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
 512                 return $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
 513         }
 514
 515         /**
 516          * Get a random string
 517          *
 518          * @private
 519          * @static
 520          */
 521         function getRandomString() {
 522                 return dechex( mt_rand( 0, 0x7fffffff ) ) . dechex( mt_rand( 0, 0x7fffffff ) );
 523         }
 524
 525         function &getTitle() { return $this->mTitle; }
 526         function getOptions() { return $this->mOptions; }
 527         function getRevisionId() { return $this->mRevisionId; }
 528         function getOutput() { return $this->mOutput; }
 529         function nextLinkID() { return $this->mLinkID++; }
 530
 531         function getFunctionLang() {
 532                 global $wgLang, $wgContLang;
 533
 534                 $target = $this->mOptions->getTargetLanguage();
 535                 if ( $target !== null ) {
 536                         return $target;
 537                 } else {
 538                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 539                 }
 540         }
 541
 542         /**
 543          * Get a preprocessor object
 544          */
 545         function getPreprocessor() {
 546                 if ( !isset( $this->mPreprocessor ) ) {
 547                         $class = $this->mPreprocessorClass;
 548                         $this->mPreprocessor = new $class( $this );
 549                 }
 550                 return $this->mPreprocessor;
 551         }
 552
 553         /**
 554          * Replaces all occurrences of HTML-style comments and the given tags
 555          * in the text with a random marker and returns the next text. The output
 556          * parameter $matches will be an associative array filled with data in
 557          * the form:
 558          *   'UNIQ-xxxxx' => array(
 559          *     'element',
 560          *     'tag content',
 561          *     array( 'param' => 'x' ),
 562          *     '<element param="x">tag content</element>' ) )
 563          *
 564          * @param $elements list of element names. Comments are always extracted.
 565          * @param $text Source text string.
 566          * @param $uniq_prefix
 567          *
 568          * @public
 569          * @static
 570          */
 571         function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
 572                 static $n = 1;
 573                 $stripped = '';
 574                 $matches = array();
 575
 576                 $taglist = implode( '|', $elements );
 577                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 578
 579                 while ( $text != '' ) {
 580                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 581                         $stripped .= $p[0];
 582                         if ( count( $p ) < 5 ) {
 583                                 break;
 584                         }
 585                         if ( count( $p ) > 5 ) {
 586                                 # comment
 587                                 $element    = $p[4];
 588                                 $attributes = '';
 589                                 $close      = '';
 590                                 $inside     = $p[5];
 591                         } else {
 592                                 # tag
 593                                 $element    = $p[1];
 594                                 $attributes = $p[2];
 595                                 $close      = $p[3];
 596                                 $inside     = $p[4];
 597                         }
 598
 599                         $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
 600                         $stripped .= $marker;
 601
 602                         if ( $close === '/>' ) {
 603                                 # Empty element tag, <tag />
 604                                 $content = null;
 605                                 $text = $inside;
 606                                 $tail = null;
 607                         } else {
 608                                 if ( $element === '!--' ) {
 609                                         $end = '/(-->)/';
 610                                 } else {
 611                                         $end = "/(<\\/$element\\s*>)/i";
 612                                 }
 613                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 614                                 $content = $q[0];
 615                                 if ( count( $q ) < 3 ) {
 616                                         # No end tag -- let it run out to the end of the text.
 617                                         $tail = '';
 618                                         $text = '';
 619                                 } else {
 620                                         $tail = $q[1];
 621                                         $text = $q[2];
 622                                 }
 623                         }
 624
 625                         $matches[$marker] = array( $element,
 626                                 $content,
 627                                 Sanitizer::decodeTagAttributes( $attributes ),
 628                                 "<$element$attributes$close$content$tail" );
 629                 }
 630                 return $stripped;
 631         }
 632
 633         /**
 634          * Get a list of strippable XML-like elements
 635          */
 636         function getStripList() {
 637                 return $this->mStripList;
 638         }
 639
 640         /**
 641          * @deprecated use replaceVariables
 642          */
 643         function strip( $text, $state, $stripcomments = false , $dontstrip = array() ) {
 644                 return $text;
 645         }
 646
 647         /**
 648          * Restores pre, math, and other extensions removed by strip()
 649          *
 650          * always call unstripNoWiki() after this one
 651          * @private
 652          * @deprecated use $this->mStripState->unstrip()
 653          */
 654         function unstrip( $text, $state ) {
 655                 return $state->unstripGeneral( $text );
 656         }
 657
 658         /**
 659          * Always call this after unstrip() to preserve the order
 660          *
 661          * @private
 662          * @deprecated use $this->mStripState->unstrip()
 663          */
 664         function unstripNoWiki( $text, $state ) {
 665                 return $state->unstripNoWiki( $text );
 666         }
 667
 668         /**
 669          * @deprecated use $this->mStripState->unstripBoth()
 670          */
 671         function unstripForHTML( $text ) {
 672                 return $this->mStripState->unstripBoth( $text );
 673         }
 674
 675         /**
 676          * Add an item to the strip state
 677          * Returns the unique tag which must be inserted into the stripped text
 678          * The tag will be replaced with the original text in unstrip()
 679          *
 680          * @private
 681          */
 682         function insertStripItem( $text ) {
 683                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 684                 $this->mMarkerIndex++;
 685                 $this->mStripState->general->setPair( $rnd, $text );
 686                 return $rnd;
 687         }
 688
 689         /**
 690          * Interface with html tidy
 691          * @deprecated Use MWTidy::tidy()
 692          */
 693         public static function tidy( $text ) {
 694                 wfDeprecated( __METHOD__ );
 695                 return MWTidy::tidy( $text );
 696         }
 697
 698         /**
 699          * parse the wiki syntax used to render tables
 700          *
 701          * @private
 702          */
 703         function doTableStuff( $text ) {
 704                 wfProfileIn( __METHOD__ );
 705
 706                 $lines = StringUtils::explode( "\n", $text );
 707                 $out = '';
 708                 $td_history = array(); # Is currently a td tag open?
 709                 $last_tag_history = array(); # Save history of last lag activated (td, th or caption)
 710                 $tr_history = array(); # Is currently a tr tag open?
 711                 $tr_attributes = array(); # history of tr attributes
 712                 $has_opened_tr = array(); # Did this table open a <tr> element?
 713                 $indent_level = 0; # indent level of the table
 714
 715                 foreach ( $lines as $outLine ) {
 716                         $line = trim( $outLine );
 717
 718                         if ( $line == '' ) { # empty line, go to next line
 719                                 $out .= $outLine."\n";
 720                                 continue;
 721                         }
 722                         $first_character = $line[0];
 723                         $matches = array();
 724
 725                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 726                                 # First check if we are starting a new table
 727                                 $indent_level = strlen( $matches[1] );
 728
 729                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 730                                 $attributes = Sanitizer::fixTagAttributes( $attributes , 'table' );
 731
 732                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 733                                 array_push( $td_history , false );
 734                                 array_push( $last_tag_history , '' );
 735                                 array_push( $tr_history , false );
 736                                 array_push( $tr_attributes , '' );
 737                                 array_push( $has_opened_tr , false );
 738                         } elseif ( count( $td_history ) == 0 ) {
 739                                 # Don't do any of the following
 740                                 $out .= $outLine."\n";
 741                                 continue;
 742                         } elseif ( substr( $line , 0 , 2 ) === '|}' ) {
 743                                 # We are ending a table
 744                                 $line = '</table>' . substr( $line , 2 );
 745                                 $last_tag = array_pop( $last_tag_history );
 746
 747                                 if ( !array_pop( $has_opened_tr ) ) {
 748                                         $line = "<tr><td></td></tr>{$line}";
 749                                 }
 750
 751                                 if ( array_pop( $tr_history ) ) {
 752                                         $line = "</tr>{$line}";
 753                                 }
 754
 755                                 if ( array_pop( $td_history ) ) {
 756                                         $line = "</{$last_tag}>{$line}";
 757                                 }
 758                                 array_pop( $tr_attributes );
 759                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 760                         } elseif ( substr( $line , 0 , 2 ) === '|-' ) {
 761                                 # Now we have a table row
 762                                 $line = preg_replace( '#^\|-+#', '', $line );
 763
 764                                 # Whats after the tag is now only attributes
 765                                 $attributes = $this->mStripState->unstripBoth( $line );
 766                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 767                                 array_pop( $tr_attributes );
 768                                 array_push( $tr_attributes, $attributes );
 769
 770                                 $line = '';
 771                                 $last_tag = array_pop( $last_tag_history );
 772                                 array_pop( $has_opened_tr );
 773                                 array_push( $has_opened_tr , true );
 774
 775                                 if ( array_pop( $tr_history ) ) {
 776                                         $line = '</tr>';
 777                                 }
 778
 779                                 if ( array_pop( $td_history ) ) {
 780                                         $line = "</{$last_tag}>{$line}";
 781                                 }
 782
 783                                 $outLine = $line;
 784                                 array_push( $tr_history , false );
 785                                 array_push( $td_history , false );
 786                                 array_push( $last_tag_history , '' );
 787                         } elseif ( $first_character === '|' || $first_character === '!' || substr( $line , 0 , 2 )  === '|+' ) {
 788                                 # This might be cell elements, td, th or captions
 789                                 if ( substr( $line , 0 , 2 ) === '|+' ) {
 790                                         $first_character = '+';
 791                                         $line = substr( $line , 1 );
 792                                 }
 793
 794                                 $line = substr( $line , 1 );
 795
 796                                 if ( $first_character === '!' ) {
 797                                         $line = str_replace( '!!' , '||' , $line );
 798                                 }
 799
 800                                 # Split up multiple cells on the same line.
 801                                 # FIXME : This can result in improper nesting of tags processed
 802                                 # by earlier parser steps, but should avoid splitting up eg
 803                                 # attribute values containing literal "||".
 804                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 805
 806                                 $outLine = '';
 807
 808                                 # Loop through each table cell
 809                                 foreach ( $cells as $cell ) {
 810                                         $previous = '';
 811                                         if ( $first_character !== '+' ) {
 812                                                 $tr_after = array_pop( $tr_attributes );
 813                                                 if ( !array_pop( $tr_history ) ) {
 814                                                         $previous = "<tr{$tr_after}>\n";
 815                                                 }
 816                                                 array_push( $tr_history , true );
 817                                                 array_push( $tr_attributes , '' );
 818                                                 array_pop( $has_opened_tr );
 819                                                 array_push( $has_opened_tr , true );
 820                                         }
 821
 822                                         $last_tag = array_pop( $last_tag_history );
 823
 824                                         if ( array_pop( $td_history ) ) {
 825                                                 $previous = "</{$last_tag}>{$previous}";
 826                                         }
 827
 828                                         if ( $first_character === '|' ) {
 829                                                 $last_tag = 'td';
 830                                         } elseif ( $first_character === '!' ) {
 831                                                 $last_tag = 'th';
 832                                         } elseif ( $first_character === '+' ) {
 833                                                 $last_tag = 'caption';
 834                                         } else {
 835                                                 $last_tag = '';
 836                                         }
 837
 838                                         array_push( $last_tag_history , $last_tag );
 839
 840                                         # A cell could contain both parameters and data
 841                                         $cell_data = explode( '|' , $cell , 2 );
 842
 843                                         # Bug 553: Note that a '|' inside an invalid link should not
 844                                         # be mistaken as delimiting cell parameters
 845                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 846                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 847                                         } elseif ( count( $cell_data ) == 1 ) {
 848                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 849                                         } else {
 850                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 851                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 852                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 853                                         }
 854
 855                                         $outLine .= $cell;
 856                                         array_push( $td_history , true );
 857                                 }
 858                         }
 859                         $out .= $outLine . "\n";
 860                 }
 861
 862                 # Closing open td, tr && table
 863                 while ( count( $td_history ) > 0 ) {
 864                         if ( array_pop( $td_history ) ) {
 865                                 $out .= "</td>\n";
 866                         }
 867                         if ( array_pop( $tr_history ) ) {
 868                                 $out .= "</tr>\n";
 869                         }
 870                         if ( !array_pop( $has_opened_tr ) ) {
 871                                 $out .= "<tr><td></td></tr>\n" ;
 872                         }
 873
 874                         $out .= "</table>\n";
 875                 }
 876
 877                 # Remove trailing line-ending (b/c)
 878                 if ( substr( $out, -1 ) === "\n" ) {
 879                         $out = substr( $out, 0, -1 );
 880                 }
 881
 882                 # special case: don't return empty table
 883                 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 884                         $out = '';
 885                 }
 886
 887                 wfProfileOut( __METHOD__ );
 888
 889                 return $out;
 890         }
 891
 892         /**
 893          * Helper function for parse() that transforms wiki markup into
 894          * HTML. Only called for $mOutputType == self::OT_HTML.
 895          *
 896          * @private
 897          */
 898         function internalParse( $text, $isMain = true, $frame=false ) {
 899                 wfProfileIn( __METHOD__ );
 900
 901                 $origText = $text;
 902
 903                 # Hook to suspend the parser in this state
 904                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 905                         wfProfileOut( __METHOD__ );
 906                         return $text ;
 907                 }
 908
 909                 # if $frame is provided, then use $frame for replacing any variables
 910                 if ( $frame ) {
 911                         # use frame depth to infer how include/noinclude tags should be handled
 912                         # depth=0 means this is the top-level document; otherwise it's an included document
 913                         if ( !$frame->depth ) {
 914                                 $flag = 0;
 915                         } else {
 916                                 $flag = Parser::PTD_FOR_INCLUSION;
 917                         $dom = $this->preprocessToDom( $text, $flag );
 918                         $text = $frame->expand( $dom );
 919                         }
 920                 } else {
 921                         # if $frame is not provided, then use old-style replaceVariables
 922                         $text = $this->replaceVariables( $text );
 923                 }
 924
 925                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 926                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 927
 928                 # Tables need to come after variable replacement for things to work
 929                 # properly; putting them before other transformations should keep
 930                 # exciting things like link expansions from showing up in surprising
 931                 # places.
 932                 $text = $this->doTableStuff( $text );
 933
 934                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 935
 936                 $text = $this->doDoubleUnderscore( $text );
 937
 938                 $text = $this->doHeadings( $text );
 939                 if ( $this->mOptions->getUseDynamicDates() ) {
 940                         $df = DateFormatter::getInstance();
 941                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 942                 }
 943                 $text = $this->doAllQuotes( $text );
 944                 $text = $this->replaceInternalLinks( $text );
 945                 $text = $this->replaceExternalLinks( $text );
 946
 947                 # replaceInternalLinks may sometimes leave behind
 948                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 949                 $text = str_replace( $this->mUniqPrefix.'NOPARSE', '', $text );
 950
 951                 $text = $this->doMagicLinks( $text );
 952                 $text = $this->formatHeadings( $text, $origText, $isMain );
 953
 954                 wfProfileOut( __METHOD__ );
 955                 return $text;
 956         }
 957
 958         /**
 959          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 960          * magic external links.
 961          *
 962          * DML
 963          * @private
 964          */
 965         function doMagicLinks( $text ) {
 966                 wfProfileIn( __METHOD__ );
 967                 $prots = $this->mUrlProtocols;
 968                 $urlChar = self::EXT_LINK_URL_CLASS;
 969                 $text = preg_replace_callback(
 970                         '!(?:                           # Start cases
 971                                 (<a[ \t\r\n>].*?</a>) |     # m[1]: Skip link text
 972                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
 973                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
 974                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
 975                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
 976                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 977                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 978                                     [0-9Xx]                 # check digit
 979                                     \b)
 980                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 981                 wfProfileOut( __METHOD__ );
 982                 return $text;
 983         }
 984
 985         function magicLinkCallback( $m ) {
 986                 if ( isset( $m[1] ) && $m[1] !== '' ) {
 987                         # Skip anchor
 988                         return $m[0];
 989                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
 990                         # Skip HTML element
 991                         return $m[0];
 992                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
 993                         # Free external link
 994                         return $this->makeFreeExternalLink( $m[0] );
 995                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
 996                         # RFC or PMID
 997                         $CssClass = '';
 998                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
 999                                 $keyword = 'RFC';
1000                                 $urlmsg = 'rfcurl';
1001                                 $CssClass = 'mw-magiclink-rfc';
1002                                 $id = $m[4];
1003                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1004                                 $keyword = 'PMID';
1005                                 $urlmsg = 'pubmedurl';
1006                                 $CssClass = 'mw-magiclink-pmid';
1007                                 $id = $m[4];
1008                         } else {
1009                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1010                                         substr( $m[0], 0, 20 ) . '"' );
1011                         }
1012                         $url = wfMsg( $urlmsg, $id);
1013                         $sk = $this->mOptions->getSkin();
1014                         $la = $sk->getExternalLinkAttributes( "external $CssClass" );
1015                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1016                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1017                         # ISBN
1018                         $isbn = $m[5];
1019                         $num = strtr( $isbn, array(
1020                                 '-' => '',
1021                                 ' ' => '',
1022                                 'x' => 'X',
1023                         ));
1024                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1025                         return'<a href="' .
1026                                 $titleObj->escapeLocalUrl() .
1027                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1028                 } else {
1029                         return $m[0];
1030                 }
1031         }
1032
1033         /**
1034          * Make a free external link, given a user-supplied URL
1035          * @return HTML
1036          * @private
1037          */
1038         function makeFreeExternalLink( $url ) {
1039                 global $wgContLang;
1040                 wfProfileIn( __METHOD__ );
1041
1042                 $sk = $this->mOptions->getSkin();
1043                 $trail = '';
1044
1045                 # The characters '<' and '>' (which were escaped by
1046                 # removeHTMLtags()) should not be included in
1047                 # URLs, per RFC 2396.
1048                 $m2 = array();
1049                 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1050                         $trail = substr( $url, $m2[0][1] ) . $trail;
1051                         $url = substr( $url, 0, $m2[0][1] );
1052                 }
1053
1054                 # Move trailing punctuation to $trail
1055                 $sep = ',;\.:!?';
1056                 # If there is no left bracket, then consider right brackets fair game too
1057                 if ( strpos( $url, '(' ) === false ) {
1058                         $sep .= ')';
1059                 }
1060
1061                 $numSepChars = strspn( strrev( $url ), $sep );
1062                 if ( $numSepChars ) {
1063                         $trail = substr( $url, -$numSepChars ) . $trail;
1064                         $url = substr( $url, 0, -$numSepChars );
1065                 }
1066
1067                 $url = Sanitizer::cleanUrl( $url );
1068
1069                 # Is this an external image?
1070                 $text = $this->maybeMakeExternalImage( $url );
1071                 if ( $text === false ) {
1072                         # Not an image, make a link
1073                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1074                                 $this->getExternalLinkAttribs( $url ) );
1075                         # Register it in the output object...
1076                         # Replace unnecessary URL escape codes with their equivalent characters
1077                         $pasteurized = self::replaceUnusualEscapes( $url );
1078                         $this->mOutput->addExternalLink( $pasteurized );
1079                 }
1080                 wfProfileOut( __METHOD__ );
1081                 return $text . $trail;
1082         }
1083
1084
1085         /**
1086          * Parse headers and return html
1087          *
1088          * @private
1089          */
1090         function doHeadings( $text ) {
1091                 wfProfileIn( __METHOD__ );
1092                 for ( $i = 6; $i >= 1; --$i ) {
1093                         $h = str_repeat( '=', $i );
1094                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1095                           "<h$i>\\1</h$i>", $text );
1096                 }
1097                 wfProfileOut( __METHOD__ );
1098                 return $text;
1099         }
1100
1101         /**
1102          * Replace single quotes with HTML markup
1103          * @private
1104          * @return string the altered text
1105          */
1106         function doAllQuotes( $text ) {
1107                 wfProfileIn( __METHOD__ );
1108                 $outtext = '';
1109                 $lines = StringUtils::explode( "\n", $text );
1110                 foreach ( $lines as $line ) {
1111                         $outtext .= $this->doQuotes( $line ) . "\n";
1112                 }
1113                 $outtext = substr( $outtext, 0,-1 );
1114                 wfProfileOut( __METHOD__ );
1115                 return $outtext;
1116         }
1117
1118         /**
1119          * Helper function for doAllQuotes()
1120          */
1121         public function doQuotes( $text ) {
1122                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1123                 if ( count( $arr ) == 1 ) {
1124                         return $text;
1125                 } else {
1126                         # First, do some preliminary work. This may shift some apostrophes from
1127                         # being mark-up to being text. It also counts the number of occurrences
1128                         # of bold and italics mark-ups.
1129                         $i = 0;
1130                         $numbold = 0;
1131                         $numitalics = 0;
1132                         foreach ( $arr as $r ) {
1133                                 if ( ( $i % 2 ) == 1 ) {
1134                                         # If there are ever four apostrophes, assume the first is supposed to
1135                                         # be text, and the remaining three constitute mark-up for bold text.
1136                                         if ( strlen( $arr[$i] ) == 4 ) {
1137                                                 $arr[$i-1] .= "'";
1138                                                 $arr[$i] = "'''";
1139                                         } elseif ( strlen( $arr[$i] ) > 5 ) {
1140                                                 # If there are more than 5 apostrophes in a row, assume they're all
1141                                                 # text except for the last 5.
1142                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1143                                                 $arr[$i] = "'''''";
1144                                         }
1145                                         # Count the number of occurrences of bold and italics mark-ups.
1146                                         # We are not counting sequences of five apostrophes.
1147                                         if ( strlen( $arr[$i] ) == 2 ) {
1148                                                 $numitalics++;
1149                                         } elseif ( strlen( $arr[$i] ) == 3 ) {
1150                                                 $numbold++;
1151                                         } elseif ( strlen( $arr[$i] ) == 5 ) {
1152                                                 $numitalics++;
1153                                                 $numbold++;
1154                                         }
1155                                 }
1156                                 $i++;
1157                         }
1158
1159                         # If there is an odd number of both bold and italics, it is likely
1160                         # that one of the bold ones was meant to be an apostrophe followed
1161                         # by italics. Which one we cannot know for certain, but it is more
1162                         # likely to be one that has a single-letter word before it.
1163                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1164                                 $i = 0;
1165                                 $firstsingleletterword = -1;
1166                                 $firstmultiletterword = -1;
1167                                 $firstspace = -1;
1168                                 foreach ( $arr as $r ) {
1169                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1170                                                 $x1 = substr( $arr[$i-1], -1 );
1171                                                 $x2 = substr( $arr[$i-1], -2, 1 );
1172                                                 if ( $x1 === ' ' ) {
1173                                                         if ( $firstspace == -1 ) {
1174                                                                 $firstspace = $i;
1175                                                         }
1176                                                 } elseif ( $x2 === ' ') {
1177                                                         if ( $firstsingleletterword == -1 ) {
1178                                                                 $firstsingleletterword = $i;
1179                                                         }
1180                                                 } else {
1181                                                         if ( $firstmultiletterword == -1 ) {
1182                                                                 $firstmultiletterword = $i;
1183                                                         }
1184                                                 }
1185                                         }
1186                                         $i++;
1187                                 }
1188
1189                                 # If there is a single-letter word, use it!
1190                                 if ( $firstsingleletterword > -1 ) {
1191                                         $arr[$firstsingleletterword] = "''";
1192                                         $arr[$firstsingleletterword-1] .= "'";
1193                                 } elseif ( $firstmultiletterword > -1 ) {
1194                                         # If not, but there's a multi-letter word, use that one.
1195                                         $arr[$firstmultiletterword] = "''";
1196                                         $arr[$firstmultiletterword-1] .= "'";
1197                                 } elseif ( $firstspace > -1 ) {
1198                                         # ... otherwise use the first one that has neither.
1199                                         # (notice that it is possible for all three to be -1 if, for example,
1200                                         # there is only one pentuple-apostrophe in the line)
1201                                         $arr[$firstspace] = "''";
1202                                         $arr[$firstspace-1] .= "'";
1203                                 }
1204                         }
1205
1206                         # Now let's actually convert our apostrophic mush to HTML!
1207                         $output = '';
1208                         $buffer = '';
1209                         $state = '';
1210                         $i = 0;
1211                         foreach ( $arr as $r ) {
1212                                 if ( ( $i % 2 ) == 0 ) {
1213                                         if ( $state === 'both' ) {
1214                                                 $buffer .= $r;
1215                                         } else {
1216                                                 $output .= $r;
1217                                         }
1218                                 } else {
1219                                         if ( strlen( $r ) == 2 ) {
1220                                                 if ( $state === 'i' ) {
1221                                                         $output .= '</i>'; $state = '';
1222                                                 } elseif ( $state === 'bi' ) {
1223                                                         $output .= '</i>'; $state = 'b';
1224                                                 } elseif ( $state === 'ib' ) {
1225                                                         $output .= '</b></i><b>'; $state = 'b';
1226                                                 } elseif ( $state === 'both' ) {
1227                                                         $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1228                                                 } else { # $state can be 'b' or ''
1229                                                         $output .= '<i>'; $state .= 'i';
1230                                                 }
1231                                         } elseif ( strlen( $r ) == 3 ) {
1232                                                 if ( $state === 'b' ) {
1233                                                         $output .= '</b>'; $state = '';
1234                                                 } elseif ( $state === 'bi' ) {
1235                                                         $output .= '</i></b><i>'; $state = 'i';
1236                                                 } elseif ( $state === 'ib' ) {
1237                                                         $output .= '</b>'; $state = 'i';
1238                                                 } elseif ( $state === 'both' ) {
1239                                                         $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1240                                                 } else { # $state can be 'i' or ''
1241                                                         $output .= '<b>'; $state .= 'b';
1242                                                 }
1243                                         } elseif ( strlen( $r ) == 5 ) {
1244                                                 if ( $state === 'b' ) {
1245                                                         $output .= '</b><i>'; $state = 'i';
1246                                                 } elseif ( $state === 'i' ) {
1247                                                         $output .= '</i><b>'; $state = 'b';
1248                                                 } elseif ( $state === 'bi' ) {
1249                                                         $output .= '</i></b>'; $state = '';
1250                                                 } elseif ( $state === 'ib' ) {
1251                                                         $output .= '</b></i>'; $state = '';
1252                                                 } elseif ( $state === 'both' ) {
1253                                                         $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1254                                                 } else { # ($state == '')
1255                                                         $buffer = ''; $state = 'both';
1256                                                 }
1257                                         }
1258                                 }
1259                                 $i++;
1260                         }
1261                         # Now close all remaining tags.  Notice that the order is important.
1262                         if ( $state === 'b' || $state === 'ib' ) {
1263                                 $output .= '</b>';
1264                         }
1265                         if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1266                                 $output .= '</i>';
1267                         }
1268                         if ( $state === 'bi' ) {
1269                                 $output .= '</b>';
1270                         }
1271                         # There might be lonely ''''', so make sure we have a buffer
1272                         if ( $state === 'both' && $buffer ) {
1273                                 $output .= '<b><i>'.$buffer.'</i></b>';
1274                         }
1275                         return $output;
1276                 }
1277         }
1278
1279         /**
1280          * Replace external links (REL)
1281          *
1282          * Note: this is all very hackish and the order of execution matters a lot.
1283          * Make sure to run maintenance/parserTests.php if you change this code.
1284          *
1285          * @private
1286          */
1287         function replaceExternalLinks( $text ) {
1288                 global $wgContLang;
1289                 wfProfileIn( __METHOD__ );
1290
1291                 $sk = $this->mOptions->getSkin();
1292
1293                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1294                 $s = array_shift( $bits );
1295
1296                 $i = 0;
1297                 while ( $i<count( $bits ) ) {
1298                         $url = $bits[$i++];
1299                         $protocol = $bits[$i++];
1300                         $text = $bits[$i++];
1301                         $trail = $bits[$i++];
1302
1303                         # The characters '<' and '>' (which were escaped by
1304                         # removeHTMLtags()) should not be included in
1305                         # URLs, per RFC 2396.
1306                         $m2 = array();
1307                         if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1308                                 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1309                                 $url = substr( $url, 0, $m2[0][1] );
1310                         }
1311
1312                         # If the link text is an image URL, replace it with an <img> tag
1313                         # This happened by accident in the original parser, but some people used it extensively
1314                         $img = $this->maybeMakeExternalImage( $text );
1315                         if ( $img !== false ) {
1316                                 $text = $img;
1317                         }
1318
1319                         $dtrail = '';
1320
1321                         # Set linktype for CSS - if URL==text, link is essentially free
1322                         $linktype = ( $text === $url ) ? 'free' : 'text';
1323
1324                         # No link text, e.g. [http://domain.tld/some.link]
1325                         if ( $text == '' ) {
1326                                 # Autonumber if allowed. See bug #5918
1327                                 if ( strpos( wfUrlProtocols(), substr( $protocol, 0, strpos( $protocol, ':' ) ) ) !== false ) {
1328                                         $langObj = $this->getFunctionLang();
1329                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1330                                         $linktype = 'autonumber';
1331                                 } else {
1332                                         # Otherwise just use the URL
1333                                         $text = htmlspecialchars( $url );
1334                                         $linktype = 'free';
1335                                 }
1336                         } else {
1337                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1338                                 # Check for trail
1339                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1340                         }
1341
1342                         $text = $wgContLang->markNoConversion( $text );
1343
1344                         $url = Sanitizer::cleanUrl( $url );
1345
1346                         # Use the encoded URL
1347                         # This means that users can paste URLs directly into the text
1348                         # Funny characters like &ouml; aren't valid in URLs anyway
1349                         # This was changed in August 2004
1350                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1351                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1352
1353                         # Register link in the output object.
1354                         # Replace unnecessary URL escape codes with the referenced character
1355                         # This prevents spammers from hiding links from the filters
1356                         $pasteurized = self::replaceUnusualEscapes( $url );
1357                         $this->mOutput->addExternalLink( $pasteurized );
1358                 }
1359
1360                 wfProfileOut( __METHOD__ );
1361                 return $s;
1362         }
1363
1364         /**
1365          * Get an associative array of additional HTML attributes appropriate for a
1366          * particular external link.  This currently may include rel => nofollow
1367          * (depending on configuration, namespace, and the URL's domain) and/or a
1368          * target attribute (depending on configuration).
1369          *
1370          * @param string $url Optional URL, to extract the domain from for rel =>
1371          *   nofollow if appropriate
1372          * @return array Associative array of HTML attributes
1373          */
1374         function getExternalLinkAttribs( $url = false ) {
1375                 $attribs = array();
1376                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1377                 $ns = $this->mTitle->getNamespace();
1378                 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) ) {
1379                         $attribs['rel'] = 'nofollow';
1380
1381                         global $wgNoFollowDomainExceptions;
1382                         if ( $wgNoFollowDomainExceptions ) {
1383                                 $bits = wfParseUrl( $url );
1384                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1385                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1386                                                 if ( substr( $bits['host'], -strlen( $domain ) ) == $domain ) {
1387                                                         unset( $attribs['rel'] );
1388                                                         break;
1389                                                 }
1390                                         }
1391                                 }
1392                         }
1393                 }
1394                 if ( $this->mOptions->getExternalLinkTarget() ) {
1395                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1396                 }
1397                 return $attribs;
1398         }
1399
1400
1401         /**
1402          * Replace unusual URL escape codes with their equivalent characters
1403          * @param string
1404          * @return string
1405          * @static
1406          * @todo  This can merge genuinely required bits in the path or query string,
1407          *        breaking legit URLs. A proper fix would treat the various parts of
1408          *        the URL differently; as a workaround, just use the output for
1409          *        statistical records, not for actual linking/output.
1410          */
1411         static function replaceUnusualEscapes( $url ) {
1412                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1413                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1414         }
1415
1416         /**
1417          * Callback function used in replaceUnusualEscapes().
1418          * Replaces unusual URL escape codes with their equivalent character
1419          * @static
1420          * @private
1421          */
1422         private static function replaceUnusualEscapesCallback( $matches ) {
1423                 $char = urldecode( $matches[0] );
1424                 $ord = ord( $char );
1425                 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1426                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1427                         # No, shouldn't be escaped
1428                         return $char;
1429                 } else {
1430                         # Yes, leave it escaped
1431                         return $matches[0];
1432                 }
1433         }
1434
1435         /**
1436          * make an image if it's allowed, either through the global
1437          * option, through the exception, or through the on-wiki whitelist
1438          * @private
1439          */
1440         function maybeMakeExternalImage( $url ) {
1441                 $sk = $this->mOptions->getSkin();
1442                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1443                 $imagesexception = !empty( $imagesfrom );
1444                 $text = false;
1445                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1446                 if ( $imagesexception && is_array( $imagesfrom ) ) {
1447                         $imagematch = false;
1448                         foreach ( $imagesfrom as $match ) {
1449                                 if ( strpos( $url, $match ) === 0 ) {
1450                                         $imagematch = true;
1451                                         break;
1452                                 }
1453                         }
1454                 } elseif ( $imagesexception ) {
1455                         $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1456                 } else {
1457                         $imagematch = false;
1458                 }
1459                 if ( $this->mOptions->getAllowExternalImages()
1460                      || ( $imagesexception && $imagematch ) ) {
1461                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1462                                 # Image found
1463                                 $text = $sk->makeExternalImage( $url );
1464                         }
1465                 }
1466                 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1467                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1468                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1469                         foreach ( $whitelist as $entry ) {
1470                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1471                                 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1472                                         continue;
1473                                 }
1474                                 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1475                                         # Image matches a whitelist entry
1476                                         $text = $sk->makeExternalImage( $url );
1477                                         break;
1478                                 }
1479                         }
1480                 }
1481                 return $text;
1482         }
1483
1484         /**
1485          * Process [[ ]] wikilinks
1486          * @return processed text
1487          *
1488          * @private
1489          */
1490         function replaceInternalLinks( $s ) {
1491                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1492                 return $s;
1493         }
1494
1495         /**
1496          * Process [[ ]] wikilinks (RIL)
1497          * @return LinkHolderArray
1498          *
1499          * @private
1500          */
1501         function replaceInternalLinks2( &$s ) {
1502                 global $wgContLang;
1503
1504                 wfProfileIn( __METHOD__ );
1505
1506                 wfProfileIn( __METHOD__.'-setup' );
1507                 static $tc = FALSE, $e1, $e1_img;
1508                 # the % is needed to support urlencoded titles as well
1509                 if ( !$tc ) {
1510                         $tc = Title::legalChars() . '#%';
1511                         # Match a link having the form [[namespace:link|alternate]]trail
1512                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1513                         # Match cases where there is no "]]", which might still be images
1514                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1515                 }
1516
1517                 $sk = $this->mOptions->getSkin();
1518                 $holders = new LinkHolderArray( $this );
1519
1520                 # split the entire text string on occurences of [[
1521                 $a = StringUtils::explode( '[[', ' ' . $s );
1522                 # get the first element (all text up to first [[), and remove the space we added
1523                 $s = $a->current();
1524                 $a->next();
1525                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1526                 $s = substr( $s, 1 );
1527
1528                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1529                 $e2 = null;
1530                 if ( $useLinkPrefixExtension ) {
1531                         # Match the end of a line for a word that's not followed by whitespace,
1532                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1533                         $e2 = wfMsgForContent( 'linkprefix' );
1534                 }
1535
1536                 if ( is_null( $this->mTitle ) ) {
1537                         wfProfileOut( __METHOD__.'-setup' );
1538                         wfProfileOut( __METHOD__ );
1539                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1540                 }
1541                 $nottalk = !$this->mTitle->isTalkPage();
1542
1543                 if ( $useLinkPrefixExtension ) {
1544                         $m = array();
1545                         if ( preg_match( $e2, $s, $m ) ) {
1546                                 $first_prefix = $m[2];
1547                         } else {
1548                                 $first_prefix = false;
1549                         }
1550                 } else {
1551                         $prefix = '';
1552                 }
1553
1554                 if ( $wgContLang->hasVariants() ) {
1555                         $selflink = $wgContLang->convertLinkToAllVariants( $this->mTitle->getPrefixedText() );
1556                 } else {
1557                         $selflink = array( $this->mTitle->getPrefixedText() );
1558                 }
1559                 $useSubpages = $this->areSubpagesAllowed();
1560                 wfProfileOut( __METHOD__.'-setup' );
1561
1562                 # Loop for each link
1563                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1564                         # Check for excessive memory usage
1565                         if ( $holders->isBig() ) {
1566                                 # Too big
1567                                 # Do the existence check, replace the link holders and clear the array
1568                                 $holders->replace( $s );
1569                                 $holders->clear();
1570                         }
1571
1572                         if ( $useLinkPrefixExtension ) {
1573                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1574                                 if ( preg_match( $e2, $s, $m ) ) {
1575                                         $prefix = $m[2];
1576                                         $s = $m[1];
1577                                 } else {
1578                                         $prefix='';
1579                                 }
1580                                 # first link
1581                                 if ( $first_prefix ) {
1582                                         $prefix = $first_prefix;
1583                                         $first_prefix = false;
1584                                 }
1585                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1586                         }
1587
1588                         $might_be_img = false;
1589
1590                         wfProfileIn( __METHOD__."-e1" );
1591                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1592                                 $text = $m[2];
1593                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1594                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1595                                 # the real problem is with the $e1 regex
1596                                 # See bug 1300.
1597                                 #
1598                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1599                                 # and no image is in sight. See bug 2095.
1600                                 #
1601                                 if ( $text !== '' &&
1602                                         substr( $m[3], 0, 1 ) === ']' &&
1603                                         strpos( $text, '[' ) !== false
1604                                 )
1605                                 {
1606                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1607                                         $m[3] = substr( $m[3], 1 );
1608                                 }
1609                                 # fix up urlencoded title texts
1610                                 if ( strpos( $m[1], '%' ) !== false ) {
1611                                         # Should anchors '#' also be rejected?
1612                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode( $m[1] ) );
1613                                 }
1614                                 $trail = $m[3];
1615                         } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1616                                 $might_be_img = true;
1617                                 $text = $m[2];
1618                                 if ( strpos( $m[1], '%' ) !== false ) {
1619                                         $m[1] = urldecode( $m[1] );
1620                                 }
1621                                 $trail = "";
1622                         } else { # Invalid form; output directly
1623                                 $s .= $prefix . '[[' . $line ;
1624                                 wfProfileOut( __METHOD__."-e1" );
1625                                 continue;
1626                         }
1627                         wfProfileOut( __METHOD__."-e1" );
1628                         wfProfileIn( __METHOD__."-misc" );
1629
1630                         # Don't allow internal links to pages containing
1631                         # PROTO: where PROTO is a valid URL protocol; these
1632                         # should be external links.
1633                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1634                                 $s .= $prefix . '[[' . $line ;
1635                                 wfProfileOut( __METHOD__."-misc" );
1636                                 continue;
1637                         }
1638
1639                         # Make subpage if necessary
1640                         if ( $useSubpages ) {
1641                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1642                         } else {
1643                                 $link = $m[1];
1644                         }
1645
1646                         $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1647                         if ( !$noforce ) {
1648                                 # Strip off leading ':'
1649                                 $link = substr( $link, 1 );
1650                         }
1651
1652                         wfProfileOut( __METHOD__."-misc" );
1653                         wfProfileIn( __METHOD__."-title" );
1654                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1655                         if ( $nt === null ) {
1656                                 $s .= $prefix . '[[' . $line;
1657                                 wfProfileOut( __METHOD__."-title" );
1658                                 continue;
1659                         }
1660
1661                         $ns = $nt->getNamespace();
1662                         $iw = $nt->getInterWiki();
1663                         wfProfileOut( __METHOD__."-title" );
1664
1665                         if ( $might_be_img ) { # if this is actually an invalid link
1666                                 wfProfileIn( __METHOD__."-might_be_img" );
1667                                 if ( $ns == NS_FILE && $noforce ) { # but might be an image
1668                                         $found = false;
1669                                         while ( true ) {
1670                                                 # look at the next 'line' to see if we can close it there
1671                                                 $a->next();
1672                                                 $next_line = $a->current();
1673                                                 if ( $next_line === false || $next_line === null ) {
1674                                                         break;
1675                                                 }
1676                                                 $m = explode( ']]', $next_line, 3 );
1677                                                 if ( count( $m ) == 3 ) {
1678                                                         # the first ]] closes the inner link, the second the image
1679                                                         $found = true;
1680                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1681                                                         $trail = $m[2];
1682                                                         break;
1683                                                 } elseif ( count( $m ) == 2 ) {
1684                                                         # if there's exactly one ]] that's fine, we'll keep looking
1685                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1686                                                 } else {
1687                                                         # if $next_line is invalid too, we need look no further
1688                                                         $text .= '[[' . $next_line;
1689                                                         break;
1690                                                 }
1691                                         }
1692                                         if ( !$found ) {
1693                                                 # we couldn't find the end of this imageLink, so output it raw
1694                                                 # but don't ignore what might be perfectly normal links in the text we've examined
1695                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1696                                                 $s .= "{$prefix}[[$link|$text";
1697                                                 # note: no $trail, because without an end, there *is* no trail
1698                                                 wfProfileOut( __METHOD__."-might_be_img" );
1699                                                 continue;
1700                                         }
1701                                 } else { # it's not an image, so output it raw
1702                                         $s .= "{$prefix}[[$link|$text";
1703                                         # note: no $trail, because without an end, there *is* no trail
1704                                         wfProfileOut( __METHOD__."-might_be_img" );
1705                                         continue;
1706                                 }
1707                                 wfProfileOut( __METHOD__."-might_be_img" );
1708                         }
1709
1710                         $wasblank = ( $text  == '' );
1711                         if ( $wasblank ) {
1712                                 $text = $link;
1713                         }
1714
1715                         # Link not escaped by : , create the various objects
1716                         if ( $noforce ) {
1717
1718                                 # Interwikis
1719                                 wfProfileIn( __METHOD__."-interwiki" );
1720                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1721                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1722                                         $s = rtrim( $s . $prefix );
1723                                         $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
1724                                         wfProfileOut( __METHOD__."-interwiki" );
1725                                         continue;
1726                                 }
1727                                 wfProfileOut( __METHOD__."-interwiki" );
1728
1729                                 if ( $ns == NS_FILE ) {
1730                                         wfProfileIn( __METHOD__."-image" );
1731                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1732                                                 if ( $wasblank ) {
1733                                                         # if no parameters were passed, $text
1734                                                         # becomes something like "File:Foo.png",
1735                                                         # which we don't want to pass on to the
1736                                                         # image generator
1737                                                         $text = '';
1738                                                 } else {
1739                                                         # recursively parse links inside the image caption
1740                                                         # actually, this will parse them in any other parameters, too,
1741                                                         # but it might be hard to fix that, and it doesn't matter ATM
1742                                                         $text = $this->replaceExternalLinks( $text );
1743                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1744                                                 }
1745                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1746                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1747                                         } else {
1748                                                 $s .= $prefix . $trail;
1749                                         }
1750                                         $this->mOutput->addImage( $nt->getDBkey() );
1751                                         wfProfileOut( __METHOD__."-image" );
1752                                         continue;
1753
1754                                 }
1755
1756                                 if ( $ns == NS_CATEGORY ) {
1757                                         wfProfileIn( __METHOD__."-category" );
1758                                         $s = rtrim( $s . "\n" ); # bug 87
1759
1760                                         if ( $wasblank ) {
1761                                                 $sortkey = $this->getDefaultSort();
1762                                         } else {
1763                                                 $sortkey = $text;
1764                                         }
1765                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1766                                         $sortkey = str_replace( "\n", '', $sortkey );
1767                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1768                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1769
1770                                         /**
1771                                          * Strip the whitespace Category links produce, see bug 87
1772                                          * @todo We might want to use trim($tmp, "\n") here.
1773                                          */
1774                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '': $prefix . $trail;
1775
1776                                         wfProfileOut( __METHOD__."-category" );
1777                                         continue;
1778                                 }
1779                         }
1780
1781                         # Self-link checking
1782                         if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1783                                 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1784                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1785                                         continue;
1786                                 }
1787                         }
1788
1789                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1790                         # FIXME: Should do batch file existence checks, see comment below
1791                         if ( $ns == NS_MEDIA ) {
1792                                 wfProfileIn( __METHOD__."-media" );
1793                                 # Give extensions a chance to select the file revision for us
1794                                 $skip = $time = false;
1795                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1796                                 if ( $skip ) {
1797                                         $link = $sk->link( $nt );
1798                                 } else {
1799                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1800                                 }
1801                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1802                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1803                                 $this->mOutput->addImage( $nt->getDBkey() );
1804                                 wfProfileOut( __METHOD__."-media" );
1805                                 continue;
1806                         }
1807
1808                         wfProfileIn( __METHOD__."-always_known" );
1809                         # Some titles, such as valid special pages or files in foreign repos, should
1810                         # be shown as bluelinks even though they're not included in the page table
1811                         #
1812                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1813                         # batch file existence checks for NS_FILE and NS_MEDIA
1814                         if ( $iw == '' && $nt->isAlwaysKnown() ) {
1815                                 $this->mOutput->addLink( $nt );
1816                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1817                         } else {
1818                                 # Links will be added to the output link list after checking
1819                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1820                         }
1821                         wfProfileOut( __METHOD__."-always_known" );
1822                 }
1823                 wfProfileOut( __METHOD__ );
1824                 return $holders;
1825         }
1826
1827         /**
1828          * Make a link placeholder. The text returned can be later resolved to a real link with
1829          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1830          * parsing of interwiki links, and secondly to allow all existence checks and
1831          * article length checks (for stub links) to be bundled into a single query.
1832          *
1833          * @deprecated
1834          */
1835         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1836                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1837         }
1838
1839         /**
1840          * Render a forced-blue link inline; protect against double expansion of
1841          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1842          * Since this little disaster has to split off the trail text to avoid
1843          * breaking URLs in the following text without breaking trails on the
1844          * wiki links, it's been made into a horrible function.
1845          *
1846          * @param Title $nt
1847          * @param string $text
1848          * @param string $query
1849          * @param string $trail
1850          * @param string $prefix
1851          * @return string HTML-wikitext mix oh yuck
1852          */
1853         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1854                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1855                 $sk = $this->mOptions->getSkin();
1856                 # FIXME: use link() instead of deprecated makeKnownLinkObj()
1857                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1858                 return $this->armorLinks( $link ) . $trail;
1859         }
1860
1861         /**
1862          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1863          * going to go through further parsing steps before inline URL expansion.
1864          *
1865          * Not needed quite as much as it used to be since free links are a bit
1866          * more sensible these days. But bracketed links are still an issue.
1867          *
1868          * @param string more-or-less HTML
1869          * @return string less-or-more HTML with NOPARSE bits
1870          */
1871         function armorLinks( $text ) {
1872                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1873                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1874         }
1875
1876         /**
1877          * Return true if subpage links should be expanded on this page.
1878          * @return bool
1879          */
1880         function areSubpagesAllowed() {
1881                 # Some namespaces don't allow subpages
1882                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1883         }
1884
1885         /**
1886          * Handle link to subpage if necessary
1887          * @param string $target the source of the link
1888          * @param string &$text the link text, modified as necessary
1889          * @return string the full name of the link
1890          * @private
1891          */
1892         function maybeDoSubpageLink( $target, &$text ) {
1893                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
1894         }
1895
1896         /**#@+
1897          * Used by doBlockLevels()
1898          * @private
1899          */
1900         function closeParagraph() {
1901                 $result = '';
1902                 if ( $this->mLastSection != '' ) {
1903                         $result = '</' . $this->mLastSection  . ">\n";
1904                 }
1905                 $this->mInPre = false;
1906                 $this->mLastSection = '';
1907                 return $result;
1908         }
1909         /**
1910          * getCommon() returns the length of the longest common substring
1911          * of both arguments, starting at the beginning of both.
1912          * @private
1913          */
1914         function getCommon( $st1, $st2 ) {
1915                 $fl = strlen( $st1 );
1916                 $shorter = strlen( $st2 );
1917                 if ( $fl < $shorter ) {
1918                         $shorter = $fl;
1919                 }
1920
1921                 for ( $i = 0; $i < $shorter; ++$i ) {
1922                         if ( $st1{$i} != $st2{$i} ) {
1923                                 break;
1924                         }
1925                 }
1926                 return $i;
1927         }
1928         /**
1929          * These next three functions open, continue, and close the list
1930          * element appropriate to the prefix character passed into them.
1931          * @private
1932          */
1933         function openList( $char ) {
1934                 $result = $this->closeParagraph();
1935
1936                 if ( '*' === $char ) {
1937                         $result .= '<ul><li>';
1938                 } elseif ( '#' === $char ) {
1939                         $result .= '<ol><li>';
1940                 } elseif ( ':' === $char ) {
1941                         $result .= '<dl><dd>';
1942                 } elseif ( ';' === $char ) {
1943                         $result .= '<dl><dt>';
1944                         $this->mDTopen = true;
1945                 } else {
1946                         $result = '<!-- ERR 1 -->';
1947                 }
1948
1949                 return $result;
1950         }
1951
1952         /**
1953          * TODO: document
1954          * @param $char String
1955          * @private
1956          */
1957         function nextItem( $char ) {
1958                 if ( '*' === $char || '#' === $char ) {
1959                         return '</li><li>';
1960                 } elseif ( ':' === $char || ';' === $char ) {
1961                         $close = '</dd>';
1962                         if ( $this->mDTopen ) {
1963                                 $close = '</dt>';
1964                         }
1965                         if ( ';' === $char ) {
1966                                 $this->mDTopen = true;
1967                                 return $close . '<dt>';
1968                         } else {
1969                                 $this->mDTopen = false;
1970                                 return $close . '<dd>';
1971                         }
1972                 }
1973                 return '<!-- ERR 2 -->';
1974         }
1975
1976         /**
1977          * TODO: document
1978          * @param $char String
1979          * @private
1980          */
1981         function closeList( $char ) {
1982                 if ( '*' === $char ) {
1983                         $text = '</li></ul>';
1984                 } elseif ( '#' === $char ) {
1985                         $text = '</li></ol>';
1986                 } elseif ( ':' === $char ) {
1987                         if ( $this->mDTopen ) {
1988                                 $this->mDTopen = false;
1989                                 $text = '</dt></dl>';
1990                         } else {
1991                                 $text = '</dd></dl>';
1992                         }
1993                 } else {
1994                         return '<!-- ERR 3 -->';
1995                 }
1996                 return $text."\n";
1997         }
1998         /**#@-*/
1999
2000         /**
2001          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2002          *
2003          * @param $linestart bool whether or not this is at the start of a line.
2004          * @private
2005          * @return string the lists rendered as HTML
2006          */
2007         function doBlockLevels( $text, $linestart ) {
2008                 wfProfileIn( __METHOD__ );
2009
2010                 # Parsing through the text line by line.  The main thing
2011                 # happening here is handling of block-level elements p, pre,
2012                 # and making lists from lines starting with * # : etc.
2013                 #
2014                 $textLines = StringUtils::explode( "\n", $text );
2015
2016                 $lastPrefix = $output = '';
2017                 $this->mDTopen = $inBlockElem = false;
2018                 $prefixLength = 0;
2019                 $paragraphStack = false;
2020
2021                 foreach ( $textLines as $oLine ) {
2022                         # Fix up $linestart
2023                         if ( !$linestart ) {
2024                                 $output .= $oLine;
2025                                 $linestart = true;
2026                                 continue;
2027                         }
2028                         # * = ul
2029                         # # = ol
2030                         # ; = dt
2031                         # : = dd
2032
2033                         $lastPrefixLength = strlen( $lastPrefix );
2034                         $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2035                         $preOpenMatch = preg_match( '/<pre/i', $oLine );
2036                         # If not in a <pre> element, scan for and figure out what prefixes are there.
2037                         if ( !$this->mInPre ) {
2038                                 # Multiple prefixes may abut each other for nested lists.
2039                                 $prefixLength = strspn( $oLine, '*#:;' );
2040                                 $prefix = substr( $oLine, 0, $prefixLength );
2041
2042                                 # eh?
2043                                 # ; and : are both from definition-lists, so they're equivalent
2044                                 #  for the purposes of determining whether or not we need to open/close
2045                                 #  elements.
2046                                 $prefix2 = str_replace( ';', ':', $prefix );
2047                                 $t = substr( $oLine, $prefixLength );
2048                                 $this->mInPre = (bool)$preOpenMatch;
2049                         } else {
2050                                 # Don't interpret any other prefixes in preformatted text
2051                                 $prefixLength = 0;
2052                                 $prefix = $prefix2 = '';
2053                                 $t = $oLine;
2054                         }
2055
2056                         # List generation
2057                         if ( $prefixLength && $lastPrefix === $prefix2 ) {
2058                                 # Same as the last item, so no need to deal with nesting or opening stuff
2059                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2060                                 $paragraphStack = false;
2061
2062                                 if ( substr( $prefix, -1 ) === ';') {
2063                                         # The one nasty exception: definition lists work like this:
2064                                         # ; title : definition text
2065                                         # So we check for : in the remainder text to split up the
2066                                         # title and definition, without b0rking links.
2067                                         $term = $t2 = '';
2068                                         if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2069                                                 $t = $t2;
2070                                                 $output .= $term . $this->nextItem( ':' );
2071                                         }
2072                                 }
2073                         } elseif ( $prefixLength || $lastPrefixLength ) {
2074                                 # We need to open or close prefixes, or both.
2075
2076                                 # Either open or close a level...
2077                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2078                                 $paragraphStack = false;
2079
2080                                 # Close all the prefixes which aren't shared.
2081                                 while ( $commonPrefixLength < $lastPrefixLength ) {
2082                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2083                                         --$lastPrefixLength;
2084                                 }
2085
2086                                 # Continue the current prefix if appropriate.
2087                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2088                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2089                                 }
2090
2091                                 # Open prefixes where appropriate.
2092                                 while ( $prefixLength > $commonPrefixLength ) {
2093                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2094                                         $output .= $this->openList( $char );
2095
2096                                         if ( ';' === $char ) {
2097                                                 # FIXME: This is dupe of code above
2098                                                 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2099                                                         $t = $t2;
2100                                                         $output .= $term . $this->nextItem( ':' );
2101                                                 }
2102                                         }
2103                                         ++$commonPrefixLength;
2104                                 }
2105                                 $lastPrefix = $prefix2;
2106                         }
2107
2108                         # If we have no prefixes, go to paragraph mode.
2109                         if ( 0 == $prefixLength ) {
2110                                 wfProfileIn( __METHOD__."-paragraph" );
2111                                 # No prefix (not in list)--go to paragraph mode
2112                                 # XXX: use a stack for nestable elements like span, table and div
2113                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2114                                 $closematch = preg_match(
2115                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2116                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2117                                 if ( $openmatch or $closematch ) {
2118                                         $paragraphStack = false;
2119                                         # TODO bug 5718: paragraph closed
2120                                         $output .= $this->closeParagraph();
2121                                         if ( $preOpenMatch and !$preCloseMatch ) {
2122                                                 $this->mInPre = true;
2123                                         }
2124                                         if ( $closematch ) {
2125                                                 $inBlockElem = false;
2126                                         } else {
2127                                                 $inBlockElem = true;
2128                                         }
2129                                 } elseif ( !$inBlockElem && !$this->mInPre ) {
2130                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) ) {
2131                                                 # pre
2132                                                 if ( $this->mLastSection !== 'pre' ) {
2133                                                         $paragraphStack = false;
2134                                                         $output .= $this->closeParagraph().'<pre>';
2135                                                         $this->mLastSection = 'pre';
2136                                                 }
2137                                                 $t = substr( $t, 1 );
2138                                         } else {
2139                                                 # paragraph
2140                                                 if ( trim( $t ) == '' ) {
2141                                                         if ( $paragraphStack ) {
2142                                                                 $output .= $paragraphStack.'<br />';
2143                                                                 $paragraphStack = false;
2144                                                                 $this->mLastSection = 'p';
2145                                                         } else {
2146                                                                 if ( $this->mLastSection !== 'p' ) {
2147                                                                         $output .= $this->closeParagraph();
2148                                                                         $this->mLastSection = '';
2149                                                                         $paragraphStack = '<p>';
2150                                                                 } else {
2151                                                                         $paragraphStack = '</p><p>';
2152                                                                 }
2153                                                         }
2154                                                 } else {
2155                                                         if ( $paragraphStack ) {
2156                                                                 $output .= $paragraphStack;
2157                                                                 $paragraphStack = false;
2158                                                                 $this->mLastSection = 'p';
2159                                                         } elseif ( $this->mLastSection !== 'p' ) {
2160                                                                 $output .= $this->closeParagraph().'<p>';
2161                                                                 $this->mLastSection = 'p';
2162                                                         }
2163                                                 }
2164                                         }
2165                                 }
2166                                 wfProfileOut( __METHOD__."-paragraph" );
2167                         }
2168                         # somewhere above we forget to get out of pre block (bug 785)
2169                         if ( $preCloseMatch && $this->mInPre ) {
2170                                 $this->mInPre = false;
2171                         }
2172                         if ( $paragraphStack === false ) {
2173                                 $output .= $t."\n";
2174                         }
2175                 }
2176                 while ( $prefixLength ) {
2177                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2178                         --$prefixLength;
2179                 }
2180                 if ( $this->mLastSection != '' ) {
2181                         $output .= '</' . $this->mLastSection . '>';
2182                         $this->mLastSection = '';
2183                 }
2184
2185                 wfProfileOut( __METHOD__ );
2186                 return $output;
2187         }
2188
2189         /**
2190          * Split up a string on ':', ignoring any occurences inside tags
2191          * to prevent illegal overlapping.
2192          * @param string $str the string to split
2193          * @param string &$before set to everything before the ':'
2194          * @param string &$after set to everything after the ':'
2195          * return string the position of the ':', or false if none found
2196          */
2197         function findColonNoLinks( $str, &$before, &$after ) {
2198                 wfProfileIn( __METHOD__ );
2199
2200                 $pos = strpos( $str, ':' );
2201                 if ( $pos === false ) {
2202                         # Nothing to find!
2203                         wfProfileOut( __METHOD__ );
2204                         return false;
2205                 }
2206
2207                 $lt = strpos( $str, '<' );
2208                 if ( $lt === false || $lt > $pos ) {
2209                         # Easy; no tag nesting to worry about
2210                         $before = substr( $str, 0, $pos );
2211                         $after = substr( $str, $pos+1 );
2212                         wfProfileOut( __METHOD__ );
2213                         return $pos;
2214                 }
2215
2216                 # Ugly state machine to walk through avoiding tags.
2217                 $state = self::COLON_STATE_TEXT;
2218                 $stack = 0;
2219                 $len = strlen( $str );
2220                 for( $i = 0; $i < $len; $i++ ) {
2221                         $c = $str{$i};
2222
2223                         switch( $state ) {
2224                         # (Using the number is a performance hack for common cases)
2225                         case 0: # self::COLON_STATE_TEXT:
2226                                 switch( $c ) {
2227                                 case "<":
2228                                         # Could be either a <start> tag or an </end> tag
2229                                         $state = self::COLON_STATE_TAGSTART;
2230                                         break;
2231                                 case ":":
2232                                         if ( $stack == 0 ) {
2233                                                 # We found it!
2234                                                 $before = substr( $str, 0, $i );
2235                                                 $after = substr( $str, $i + 1 );
2236                                                 wfProfileOut( __METHOD__ );
2237                                                 return $i;
2238                                         }
2239                                         # Embedded in a tag; don't break it.
2240                                         break;
2241                                 default:
2242                                         # Skip ahead looking for something interesting
2243                                         $colon = strpos( $str, ':', $i );
2244                                         if ( $colon === false ) {
2245                                                 # Nothing else interesting
2246                                                 wfProfileOut( __METHOD__ );
2247                                                 return false;
2248                                         }
2249                                         $lt = strpos( $str, '<', $i );
2250                                         if ( $stack === 0 ) {
2251                                                 if ( $lt === false || $colon < $lt ) {
2252                                                         # We found it!
2253                                                         $before = substr( $str, 0, $colon );
2254                                                         $after = substr( $str, $colon + 1 );
2255                                                         wfProfileOut( __METHOD__ );
2256                                                         return $i;
2257                                                 }
2258                                         }
2259                                         if ( $lt === false ) {
2260                                                 # Nothing else interesting to find; abort!
2261                                                 # We're nested, but there's no close tags left. Abort!
2262                                                 break 2;
2263                                         }
2264                                         # Skip ahead to next tag start
2265                                         $i = $lt;
2266                                         $state = self::COLON_STATE_TAGSTART;
2267                                 }
2268                                 break;
2269                         case 1: # self::COLON_STATE_TAG:
2270                                 # In a <tag>
2271                                 switch( $c ) {
2272                                 case ">":
2273                                         $stack++;
2274                                         $state = self::COLON_STATE_TEXT;
2275                                         break;
2276                                 case "/":
2277                                         # Slash may be followed by >?
2278                                         $state = self::COLON_STATE_TAGSLASH;
2279                                         break;
2280                                 default:
2281                                         # ignore
2282                                 }
2283                                 break;
2284                         case 2: # self::COLON_STATE_TAGSTART:
2285                                 switch( $c ) {
2286                                 case "/":
2287                                         $state = self::COLON_STATE_CLOSETAG;
2288                                         break;
2289                                 case "!":
2290                                         $state = self::COLON_STATE_COMMENT;
2291                                         break;
2292                                 case ">":
2293                                         # Illegal early close? This shouldn't happen D:
2294                                         $state = self::COLON_STATE_TEXT;
2295                                         break;
2296                                 default:
2297                                         $state = self::COLON_STATE_TAG;
2298                                 }
2299                                 break;
2300                         case 3: # self::COLON_STATE_CLOSETAG:
2301                                 # In a </tag>
2302                                 if ( $c === ">" ) {
2303                                         $stack--;
2304                                         if ( $stack < 0 ) {
2305                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2306                                                 wfProfileOut( __METHOD__ );
2307                                                 return false;
2308                                         }
2309                                         $state = self::COLON_STATE_TEXT;
2310                                 }
2311                                 break;
2312                         case self::COLON_STATE_TAGSLASH:
2313                                 if ( $c === ">" ) {
2314                                         # Yes, a self-closed tag <blah/>
2315                                         $state = self::COLON_STATE_TEXT;
2316                                 } else {
2317                                         # Probably we're jumping the gun, and this is an attribute
2318                                         $state = self::COLON_STATE_TAG;
2319                                 }
2320                                 break;
2321                         case 5: # self::COLON_STATE_COMMENT:
2322                                 if ( $c === "-" ) {
2323                                         $state = self::COLON_STATE_COMMENTDASH;
2324                                 }
2325                                 break;
2326                         case self::COLON_STATE_COMMENTDASH:
2327                                 if ( $c === "-" ) {
2328                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2329                                 } else {
2330                                         $state = self::COLON_STATE_COMMENT;
2331                                 }
2332                                 break;
2333                         case self::COLON_STATE_COMMENTDASHDASH:
2334                                 if ( $c === ">" ) {
2335                                         $state = self::COLON_STATE_TEXT;
2336                                 } else {
2337                                         $state = self::COLON_STATE_COMMENT;
2338                                 }
2339                                 break;
2340                         default:
2341                                 throw new MWException( "State machine error in " . __METHOD__ );
2342                         }
2343                 }
2344                 if ( $stack > 0 ) {
2345                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2346                         return false;
2347                 }
2348                 wfProfileOut( __METHOD__ );
2349                 return false;
2350         }
2351
2352         /**
2353          * Return value of a magic variable (like PAGENAME)
2354          *
2355          * @private
2356          */
2357         function getVariableValue( $index, $frame=false ) {
2358                 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2359                 global $wgScriptPath, $wgStylePath;
2360
2361                 /**
2362                  * Some of these require message or data lookups and can be
2363                  * expensive to check many times.
2364                  */
2365                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2366                         if ( isset( $this->mVarCache[$index] ) ) {
2367                                 return $this->mVarCache[$index];
2368                         }
2369                 }
2370
2371                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2372                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2373
2374                 # Use the time zone
2375                 global $wgLocaltimezone;
2376                 if ( isset( $wgLocaltimezone ) ) {
2377                         $oldtz = date_default_timezone_get();
2378                         date_default_timezone_set( $wgLocaltimezone );
2379                 }
2380
2381                 $localTimestamp = date( 'YmdHis', $ts );
2382                 $localMonth = date( 'm', $ts );
2383                 $localMonth1 = date( 'n', $ts );
2384                 $localMonthName = date( 'n', $ts );
2385                 $localDay = date( 'j', $ts );
2386                 $localDay2 = date( 'd', $ts );
2387                 $localDayOfWeek = date( 'w', $ts );
2388                 $localWeek = date( 'W', $ts );
2389                 $localYear = date( 'Y', $ts );
2390                 $localHour = date( 'H', $ts );
2391                 if ( isset( $wgLocaltimezone ) ) {
2392                         date_default_timezone_set( $oldtz );
2393                 }
2394
2395                 switch ( $index ) {
2396                         case 'currentmonth':
2397                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2398                                 break;
2399                         case 'currentmonth1':
2400                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2401                                 break;
2402                         case 'currentmonthname':
2403                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2404                                 break;
2405                         case 'currentmonthnamegen':
2406                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2407                                 break;
2408                         case 'currentmonthabbrev':
2409                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2410                                 break;
2411                         case 'currentday':
2412                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2413                                 break;
2414                         case 'currentday2':
2415                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2416                                 break;
2417                         case 'localmonth':
2418                                 $value = $wgContLang->formatNum( $localMonth );
2419                                 break;
2420                         case 'localmonth1':
2421                                 $value = $wgContLang->formatNum( $localMonth1 );
2422                                 break;
2423                         case 'localmonthname':
2424                                 $value = $wgContLang->getMonthName( $localMonthName );
2425                                 break;
2426                         case 'localmonthnamegen':
2427                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2428                                 break;
2429                         case 'localmonthabbrev':
2430                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2431                                 break;
2432                         case 'localday':
2433                                 $value = $wgContLang->formatNum( $localDay );
2434                                 break;
2435                         case 'localday2':
2436                                 $value = $wgContLang->formatNum( $localDay2 );
2437                                 break;
2438                         case 'pagename':
2439                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2440                                 break;
2441                         case 'pagenamee':
2442                                 $value = $this->mTitle->getPartialURL();
2443                                 break;
2444                         case 'fullpagename':
2445                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2446                                 break;
2447                         case 'fullpagenamee':
2448                                 $value = $this->mTitle->getPrefixedURL();
2449                                 break;
2450                         case 'subpagename':
2451                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2452                                 break;
2453                         case 'subpagenamee':
2454                                 $value = $this->mTitle->getSubpageUrlForm();
2455                                 break;
2456                         case 'basepagename':
2457                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2458                                 break;
2459                         case 'basepagenamee':
2460                                 $value = wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2461                                 break;
2462                         case 'talkpagename':
2463                                 if ( $this->mTitle->canTalk() ) {
2464                                         $talkPage = $this->mTitle->getTalkPage();
2465                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2466                                 } else {
2467                                         $value = '';
2468                                 }
2469                                 break;
2470                         case 'talkpagenamee':
2471                                 if ( $this->mTitle->canTalk() ) {
2472                                         $talkPage = $this->mTitle->getTalkPage();
2473                                         $value = $talkPage->getPrefixedUrl();
2474                                 } else {
2475                                         $value = '';
2476                                 }
2477                                 break;
2478                         case 'subjectpagename':
2479                                 $subjPage = $this->mTitle->getSubjectPage();
2480                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2481                                 break;
2482                         case 'subjectpagenamee':
2483                                 $subjPage = $this->mTitle->getSubjectPage();
2484                                 $value = $subjPage->getPrefixedUrl();
2485                                 break;
2486                         case 'revisionid':
2487                                 # Let the edit saving system know we should parse the page
2488                                 # *after* a revision ID has been assigned.
2489                                 $this->mOutput->setFlag( 'vary-revision' );
2490                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2491                                 $value = $this->mRevisionId;
2492                                 break;
2493                         case 'revisionday':
2494                                 # Let the edit saving system know we should parse the page
2495                                 # *after* a revision ID has been assigned. This is for null edits.
2496                                 $this->mOutput->setFlag( 'vary-revision' );
2497                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2498                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2499                                 break;
2500                         case 'revisionday2':
2501                                 # Let the edit saving system know we should parse the page
2502                                 # *after* a revision ID has been assigned. This is for null edits.
2503                                 $this->mOutput->setFlag( 'vary-revision' );
2504                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2505                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2506                                 break;
2507                         case 'revisionmonth':
2508                                 # Let the edit saving system know we should parse the page
2509                                 # *after* a revision ID has been assigned. This is for null edits.
2510                                 $this->mOutput->setFlag( 'vary-revision' );
2511                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2512                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2513                                 break;
2514                         case 'revisionyear':
2515                                 # Let the edit saving system know we should parse the page
2516                                 # *after* a revision ID has been assigned. This is for null edits.
2517                                 $this->mOutput->setFlag( 'vary-revision' );
2518                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2519                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2520                                 break;
2521                         case 'revisiontimestamp':
2522                                 # Let the edit saving system know we should parse the page
2523                                 # *after* a revision ID has been assigned. This is for null edits.
2524                                 $this->mOutput->setFlag( 'vary-revision' );
2525                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2526                                 $value = $this->getRevisionTimestamp();
2527                                 break;
2528                         case 'revisionuser':
2529                                 # Let the edit saving system know we should parse the page
2530                                 # *after* a revision ID has been assigned. This is for null edits.
2531                                 $this->mOutput->setFlag( 'vary-revision' );
2532                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2533                                 $value = $this->getRevisionUser();
2534                                 break;
2535                         case 'namespace':
2536                                 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2537                                 break;
2538                         case 'namespacee':
2539                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2540                                 break;
2541                         case 'talkspace':
2542                                 $value = $this->mTitle->canTalk() ? str_replace( '_',' ',$this->mTitle->getTalkNsText() ) : '';
2543                                 break;
2544                         case 'talkspacee':
2545                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2546                                 break;
2547                         case 'subjectspace':
2548                                 $value = $this->mTitle->getSubjectNsText();
2549                                 break;
2550                         case 'subjectspacee':
2551                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2552                                 break;
2553                         case 'currentdayname':
2554                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2555                                 break;
2556                         case 'currentyear':
2557                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2558                                 break;
2559                         case 'currenttime':
2560                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2561                                 break;
2562                         case 'currenthour':
2563                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2564                                 break;
2565                         case 'currentweek':
2566                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2567                                 # int to remove the padding
2568                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2569                                 break;
2570                         case 'currentdow':
2571                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2572                                 break;
2573                         case 'localdayname':
2574                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2575                                 break;
2576                         case 'localyear':
2577                                 $value = $wgContLang->formatNum( $localYear, true );
2578                                 break;
2579                         case 'localtime':
2580                                 $value = $wgContLang->time( $localTimestamp, false, false );
2581                                 break;
2582                         case 'localhour':
2583                                 $value = $wgContLang->formatNum( $localHour, true );
2584                                 break;
2585                         case 'localweek':
2586                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2587                                 # int to remove the padding
2588                                 $value = $wgContLang->formatNum( (int)$localWeek );
2589                                 break;
2590                         case 'localdow':
2591                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2592                                 break;
2593                         case 'numberofarticles':
2594                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2595                                 break;
2596                         case 'numberoffiles':
2597                                 $value = $wgContLang->formatNum( SiteStats::images() );
2598                                 break;
2599                         case 'numberofusers':
2600                                 $value = $wgContLang->formatNum( SiteStats::users() );
2601                                 break;
2602                         case 'numberofactiveusers':
2603                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2604                                 break;
2605                         case 'numberofpages':
2606                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2607                                 break;
2608                         case 'numberofadmins':
2609                                 $value = $wgContLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2610                                 break;
2611                         case 'numberofedits':
2612                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2613                                 break;
2614                         case 'numberofviews':
2615                                 $value = $wgContLang->formatNum( SiteStats::views() );
2616                                 break;
2617                         case 'currenttimestamp':
2618                                 $value = wfTimestamp( TS_MW, $ts );
2619                                 break;
2620                         case 'localtimestamp':
2621                                 $value = $localTimestamp;
2622                                 break;
2623                         case 'currentversion':
2624                                 $value = SpecialVersion::getVersion();
2625                                 break;
2626                         case 'sitename':
2627                                 return $wgSitename;
2628                         case 'server':
2629                                 return $wgServer;
2630                         case 'servername':
2631                                 return $wgServerName;
2632                         case 'scriptpath':
2633                                 return $wgScriptPath;
2634                         case 'stylepath':
2635                                 return $wgStylePath;
2636                         case 'directionmark':
2637                                 return $wgContLang->getDirMark();
2638                         case 'contentlanguage':
2639                                 global $wgContLanguageCode;
2640                                 return $wgContLanguageCode;
2641                         default:
2642                                 $ret = null;
2643                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
2644                                         return $ret;
2645                                 } else {
2646                                         return null;
2647                                 }
2648                 }
2649
2650                 if ( $index )
2651                         $this->mVarCache[$index] = $value;
2652
2653                 return $value;
2654         }
2655
2656         /**
2657          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2658          *
2659          * @private
2660          */
2661         function initialiseVariables() {
2662                 wfProfileIn( __METHOD__ );
2663                 $variableIDs = MagicWord::getVariableIDs();
2664                 $substIDs = MagicWord::getSubstIDs();
2665
2666                 $this->mVariables = new MagicWordArray( $variableIDs );
2667                 $this->mSubstWords = new MagicWordArray( $substIDs );
2668                 wfProfileOut( __METHOD__ );
2669         }
2670
2671         /**
2672          * Preprocess some wikitext and return the document tree.
2673          * This is the ghost of replace_variables().
2674          *
2675          * @param string $text The text to parse
2676          * @param integer flags Bitwise combination of:
2677          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2678          *                                     included. Default is to assume a direct page view.
2679          *
2680          * The generated DOM tree must depend only on the input text and the flags.
2681          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2682          *
2683          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2684          * change in the DOM tree for a given text, must be passed through the section identifier
2685          * in the section edit link and thus back to extractSections().
2686          *
2687          * The output of this function is currently only cached in process memory, but a persistent
2688          * cache may be implemented at a later date which takes further advantage of these strict
2689          * dependency requirements.
2690          *
2691          * @private
2692          */
2693         function preprocessToDom( $text, $flags = 0 ) {
2694                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2695                 return $dom;
2696         }
2697
2698         /**
2699          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2700          */
2701         public static function splitWhitespace( $s ) {
2702                 $ltrimmed = ltrim( $s );
2703                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2704                 $trimmed = rtrim( $ltrimmed );
2705                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2706                 if ( $diff > 0 ) {
2707                         $w2 = substr( $ltrimmed, -$diff );
2708                 } else {
2709                         $w2 = '';
2710                 }
2711                 return array( $w1, $trimmed, $w2 );
2712         }
2713
2714         /**
2715          * Replace magic variables, templates, and template arguments
2716          * with the appropriate text. Templates are substituted recursively,
2717          * taking care to avoid infinite loops.
2718          *
2719          * Note that the substitution depends on value of $mOutputType:
2720          *  self::OT_WIKI: only {{subst:}} templates
2721          *  self::OT_PREPROCESS: templates but not extension tags
2722          *  self::OT_HTML: all templates and extension tags
2723          *
2724          * @param string $tex The text to transform
2725          * @param PPFrame $frame Object describing the arguments passed to the template.
2726          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2727          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2728          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2729          * @private
2730          */
2731         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2732                 # Is there any text? Also, Prevent too big inclusions!
2733                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2734                         return $text;
2735                 }
2736                 wfProfileIn( __METHOD__ );
2737
2738                 if ( $frame === false ) {
2739                         $frame = $this->getPreprocessor()->newFrame();
2740                 } elseif ( !( $frame instanceof PPFrame ) ) {
2741                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2742                         $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2743                 }
2744
2745                 $dom = $this->preprocessToDom( $text );
2746                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2747                 $text = $frame->expand( $dom, $flags );
2748
2749                 wfProfileOut( __METHOD__ );
2750                 return $text;
2751         }
2752
2753         # Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2754         static function createAssocArgs( $args ) {
2755                 $assocArgs = array();
2756                 $index = 1;
2757                 foreach ( $args as $arg ) {
2758                         $eqpos = strpos( $arg, '=' );
2759                         if ( $eqpos === false ) {
2760                                 $assocArgs[$index++] = $arg;
2761                         } else {
2762                                 $name = trim( substr( $arg, 0, $eqpos ) );
2763                                 $value = trim( substr( $arg, $eqpos+1 ) );
2764                                 if ( $value === false ) {
2765                                         $value = '';
2766                                 }
2767                                 if ( $name !== false ) {
2768                                         $assocArgs[$name] = $value;
2769                                 }
2770                         }
2771                 }
2772
2773                 return $assocArgs;
2774         }
2775
2776         /**
2777          * Warn the user when a parser limitation is reached
2778          * Will warn at most once the user per limitation type
2779          *
2780          * @param string $limitationType, should be one of:
2781          *   'expensive-parserfunction' (corresponding messages:
2782          *       'expensive-parserfunction-warning',
2783          *       'expensive-parserfunction-category')
2784          *   'post-expand-template-argument' (corresponding messages:
2785          *       'post-expand-template-argument-warning',
2786          *       'post-expand-template-argument-category')
2787          *   'post-expand-template-inclusion' (corresponding messages:
2788          *       'post-expand-template-inclusion-warning',
2789          *       'post-expand-template-inclusion-category')
2790          * @params int $current, $max When an explicit limit has been
2791          *       exceeded, provide the values (optional)
2792          */
2793         function limitationWarn( $limitationType, $current=null, $max=null) {
2794                 # does no harm if $current and $max are present but are unnecessary for the message
2795                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
2796                 $this->mOutput->addWarning( $warning );
2797                 $this->addTrackingCategory( "$limitationType-category" );
2798         }
2799
2800         /**
2801          * Return the text of a template, after recursively
2802          * replacing any variables or templates within the template.
2803          *
2804          * @param array $piece The parts of the template
2805          *  $piece['title']: the title, i.e. the part before the |
2806          *  $piece['parts']: the parameter array
2807          *  $piece['lineStart']: whether the brace was at the start of a line
2808          * @param PPFrame The current frame, contains template arguments
2809          * @return string the text of the template
2810          * @private
2811          */
2812         function braceSubstitution( $piece, $frame ) {
2813                 global $wgContLang, $wgNonincludableNamespaces;
2814                 wfProfileIn( __METHOD__ );
2815                 wfProfileIn( __METHOD__.'-setup' );
2816
2817                 # Flags
2818                 $found = false;             # $text has been filled
2819                 $nowiki = false;            # wiki markup in $text should be escaped
2820                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2821                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2822                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2823                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2824
2825                 # Title object, where $text came from
2826                 $title = null;
2827
2828                 # $part1 is the bit before the first |, and must contain only title characters.
2829                 # Various prefixes will be stripped from it later.
2830                 $titleWithSpaces = $frame->expand( $piece['title'] );
2831                 $part1 = trim( $titleWithSpaces );
2832                 $titleText = false;
2833
2834                 # Original title text preserved for various purposes
2835                 $originalTitle = $part1;
2836
2837                 # $args is a list of argument nodes, starting from index 0, not including $part1
2838                 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
2839                 wfProfileOut( __METHOD__.'-setup' );
2840
2841                 # SUBST
2842                 wfProfileIn( __METHOD__.'-modifiers' );
2843                 if ( !$found ) {
2844
2845                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
2846
2847                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
2848                         # Decide whether to expand template or keep wikitext as-is.
2849                         if ( $this->ot['wiki'] ) {
2850                                 if ( $substMatch === false ) {
2851                                         $literal = true;  # literal when in PST with no prefix
2852                                 } else {
2853                                         $literal = false; # expand when in PST with subst: or safesubst:
2854                                 }
2855                         } else {
2856                                 if ( $substMatch == 'subst' ) {
2857                                         $literal = true;  # literal when not in PST with plain subst:
2858                                 } else {
2859                                         $literal = false; # expand when not in PST with safesubst: or no prefix
2860                                 }
2861                         }
2862                         if ( $literal ) {
2863                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2864                                 $isLocalObj = true;
2865                                 $found = true;
2866                         }
2867                 }
2868
2869                 # Variables
2870                 if ( !$found && $args->getLength() == 0 ) {
2871                         $id = $this->mVariables->matchStartToEnd( $part1 );
2872                         if ( $id !== false ) {
2873                                 $text = $this->getVariableValue( $id, $frame );
2874                                 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
2875                                         $this->mOutput->mContainsOldMagic = true;
2876                                 }
2877                                 $found = true;
2878                         }
2879                 }
2880
2881                 # MSG, MSGNW and RAW
2882                 if ( !$found ) {
2883                         # Check for MSGNW:
2884                         $mwMsgnw = MagicWord::get( 'msgnw' );
2885                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2886                                 $nowiki = true;
2887                         } else {
2888                                 # Remove obsolete MSG:
2889                                 $mwMsg = MagicWord::get( 'msg' );
2890                                 $mwMsg->matchStartAndRemove( $part1 );
2891                         }
2892
2893                         # Check for RAW:
2894                         $mwRaw = MagicWord::get( 'raw' );
2895                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2896                                 $forceRawInterwiki = true;
2897                         }
2898                 }
2899                 wfProfileOut( __METHOD__.'-modifiers' );
2900
2901                 # Parser functions
2902                 if ( !$found ) {
2903                         wfProfileIn( __METHOD__ . '-pfunc' );
2904
2905                         $colonPos = strpos( $part1, ':' );
2906                         if ( $colonPos !== false ) {
2907                                 # Case sensitive functions
2908                                 $function = substr( $part1, 0, $colonPos );
2909                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2910                                         $function = $this->mFunctionSynonyms[1][$function];
2911                                 } else {
2912                                         # Case insensitive functions
2913                                         $function = $wgContLang->lc( $function );
2914                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2915                                                 $function = $this->mFunctionSynonyms[0][$function];
2916                                         } else {
2917                                                 $function = false;
2918                                         }
2919                                 }
2920                                 if ( $function ) {
2921                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2922                                         $initialArgs = array( &$this );
2923                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2924                                         if ( $flags & SFH_OBJECT_ARGS ) {
2925                                                 # Add a frame parameter, and pass the arguments as an array
2926                                                 $allArgs = $initialArgs;
2927                                                 $allArgs[] = $frame;
2928                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2929                                                         $funcArgs[] = $args->item( $i );
2930                                                 }
2931                                                 $allArgs[] = $funcArgs;
2932                                         } else {
2933                                                 # Convert arguments to plain text
2934                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2935                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2936                                                 }
2937                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2938                                         }
2939
2940                                         # Workaround for PHP bug 35229 and similar
2941                                         if ( !is_callable( $callback ) ) {
2942                                                 wfProfileOut( __METHOD__ . '-pfunc' );
2943                                                 wfProfileOut( __METHOD__ );
2944                                                 throw new MWException( "Tag hook for $function is not callable\n" );
2945                                         }
2946                                         $result = call_user_func_array( $callback, $allArgs );
2947                                         $found = true;
2948                                         $noparse = true;
2949                                         $preprocessFlags = 0;
2950
2951                                         if ( is_array( $result ) ) {
2952                                                 if ( isset( $result[0] ) ) {
2953                                                         $text = $result[0];
2954                                                         unset( $result[0] );
2955                                                 }
2956
2957                                                 # Extract flags into the local scope
2958                                                 # This allows callers to set flags such as nowiki, found, etc.
2959                                                 extract( $result );
2960                                         } else {
2961                                                 $text = $result;
2962                                         }
2963                                         if ( !$noparse ) {
2964                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
2965                                                 $isChildObj = true;
2966                                         }
2967                                 }
2968                         }
2969                         wfProfileOut( __METHOD__ . '-pfunc' );
2970                 }
2971
2972                 # Finish mangling title and then check for loops.
2973                 # Set $title to a Title object and $titleText to the PDBK
2974                 if ( !$found ) {
2975                         $ns = NS_TEMPLATE;
2976                         # Split the title into page and subpage
2977                         $subpage = '';
2978                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2979                         if ( $subpage !== '' ) {
2980                                 $ns = $this->mTitle->getNamespace();
2981                         }
2982                         $title = Title::newFromText( $part1, $ns );
2983                         if ( $title ) {
2984                                 $titleText = $title->getPrefixedText();
2985                                 # Check for language variants if the template is not found
2986                                 if ( $wgContLang->hasVariants() && $title->getArticleID() == 0 ) {
2987                                         $wgContLang->findVariantLink( $part1, $title, true );
2988                                 }
2989                                 # Do recursion depth check
2990                                 $limit = $this->mOptions->getMaxTemplateDepth();
2991                                 if ( $frame->depth >= $limit ) {
2992                                         $found = true;
2993                                         $text = '<span class="error">'
2994                                                 . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit )
2995                                                 . '</span>';
2996                                 }
2997                         }
2998                 }
2999
3000                 # Load from database
3001                 if ( !$found && $title ) {
3002                         wfProfileIn( __METHOD__ . '-loadtpl' );
3003                         if ( !$title->isExternal() ) {
3004                                 if ( $title->getNamespace() == NS_SPECIAL
3005                                         && $this->mOptions->getAllowSpecialInclusion()
3006                                         && $this->ot['html'] )
3007                                 {
3008                                         $text = SpecialPage::capturePath( $title );
3009                                         if ( is_string( $text ) ) {
3010                                                 $found = true;
3011                                                 $isHTML = true;
3012                                                 $this->disableCache();
3013                                         }
3014                                 } elseif ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3015                                         $found = false; # access denied
3016                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3017                                 } else {
3018                                         list( $text, $title ) = $this->getTemplateDom( $title );
3019                                         if ( $text !== false ) {
3020                                                 $found = true;
3021                                                 $isChildObj = true;
3022                                         }
3023                                 }
3024
3025                                 # If the title is valid but undisplayable, make a link to it
3026                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3027                                         $text = "[[:$titleText]]";
3028                                         $found = true;
3029                                 }
3030                         } elseif ( $title->isTrans() ) {
3031                                 # Interwiki transclusion
3032                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3033                                         $text = $this->interwikiTransclude( $title, 'render' );
3034                                         $isHTML = true;
3035                                 } else {
3036                                         $text = $this->interwikiTransclude( $title, 'raw' );
3037                                         # Preprocess it like a template
3038                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3039                                         $isChildObj = true;
3040                                 }
3041                                 $found = true;
3042                         }
3043
3044                         # Do infinite loop check
3045                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3046                         if ( !$frame->loopCheck( $title ) ) {
3047                                 $found = true;
3048                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3049                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3050                         }
3051                         wfProfileOut( __METHOD__ . '-loadtpl' );
3052                 }
3053
3054                 # If we haven't found text to substitute by now, we're done
3055                 # Recover the source wikitext and return it
3056                 if ( !$found ) {
3057                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3058                         wfProfileOut( __METHOD__ );
3059                         return array( 'object' => $text );
3060                 }
3061
3062                 # Expand DOM-style return values in a child frame
3063                 if ( $isChildObj ) {
3064                         # Clean up argument array
3065                         $newFrame = $frame->newChild( $args, $title );
3066
3067                         if ( $nowiki ) {
3068                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3069                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3070                                 # Expansion is eligible for the empty-frame cache
3071                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3072                                         $text = $this->mTplExpandCache[$titleText];
3073                                 } else {
3074                                         $text = $newFrame->expand( $text );
3075                                         $this->mTplExpandCache[$titleText] = $text;
3076                                 }
3077                         } else {
3078                                 # Uncached expansion
3079                                 $text = $newFrame->expand( $text );
3080                         }
3081                 }
3082                 if ( $isLocalObj && $nowiki ) {
3083                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3084                         $isLocalObj = false;
3085                 }
3086
3087                 # Replace raw HTML by a placeholder
3088                 # Add a blank line preceding, to prevent it from mucking up
3089                 # immediately preceding headings
3090                 if ( $isHTML ) {
3091                         $text = "\n\n" . $this->insertStripItem( $text );
3092                 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3093                         # Escape nowiki-style return values
3094                         $text = wfEscapeWikiText( $text );
3095                 } elseif ( is_string( $text )
3096                         && !$piece['lineStart']
3097                         && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3098                 {
3099                         # Bug 529: if the template begins with a table or block-level
3100                         # element, it should be treated as beginning a new line.
3101                         # This behaviour is somewhat controversial.
3102                         $text = "\n" . $text;
3103                 }
3104
3105                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3106                         # Error, oversize inclusion
3107                         $text = "[[$originalTitle]]" .
3108                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3109                         $this->limitationWarn( 'post-expand-template-inclusion' );
3110                 }
3111
3112                 if ( $isLocalObj ) {
3113                         $ret = array( 'object' => $text );
3114                 } else {
3115                         $ret = array( 'text' => $text );
3116                 }
3117
3118                 wfProfileOut( __METHOD__ );
3119                 return $ret;
3120         }
3121
3122         /**
3123          * Get the semi-parsed DOM representation of a template with a given title,
3124          * and its redirect destination title. Cached.
3125          */
3126         function getTemplateDom( $title ) {
3127                 $cacheTitle = $title;
3128                 $titleText = $title->getPrefixedDBkey();
3129
3130                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3131                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3132                         $title = Title::makeTitle( $ns, $dbk );
3133                         $titleText = $title->getPrefixedDBkey();
3134                 }
3135                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3136                         return array( $this->mTplDomCache[$titleText], $title );
3137                 }
3138
3139                 # Cache miss, go to the database
3140                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3141
3142                 if ( $text === false ) {
3143                         $this->mTplDomCache[$titleText] = false;
3144                         return array( false, $title );
3145                 }
3146
3147                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3148                 $this->mTplDomCache[ $titleText ] = $dom;
3149
3150                 if ( !$title->equals( $cacheTitle ) ) {
3151                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3152                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3153                 }
3154
3155                 return array( $dom, $title );
3156         }
3157
3158         /**
3159          * Fetch the unparsed text of a template and register a reference to it.
3160          */
3161         function fetchTemplateAndTitle( $title ) {
3162                 $templateCb = $this->mOptions->getTemplateCallback();
3163                 $stuff = call_user_func( $templateCb, $title, $this );
3164                 $text = $stuff['text'];
3165                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3166                 if ( isset( $stuff['deps'] ) ) {
3167                         foreach ( $stuff['deps'] as $dep ) {
3168                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3169                         }
3170                 }
3171                 return array( $text, $finalTitle );
3172         }
3173
3174         function fetchTemplate( $title ) {
3175                 $rv = $this->fetchTemplateAndTitle( $title );
3176                 return $rv[0];
3177         }
3178
3179         /**
3180          * Static function to get a template
3181          * Can be overridden via ParserOptions::setTemplateCallback().
3182          */
3183         static function statelessFetchTemplate( $title, $parser=false ) {
3184                 $text = $skip = false;
3185                 $finalTitle = $title;
3186                 $deps = array();
3187
3188                 # Loop to fetch the article, with up to 1 redirect
3189                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3190                         # Give extensions a chance to select the revision instead
3191                         $id = false; # Assume current
3192                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3193
3194                         if ( $skip ) {
3195                                 $text = false;
3196                                 $deps[] = array(
3197                                         'title' => $title,
3198                                         'page_id' => $title->getArticleID(),
3199                                         'rev_id' => null );
3200                                 break;
3201                         }
3202                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3203                         $rev_id = $rev ? $rev->getId() : 0;
3204                         # If there is no current revision, there is no page
3205                         if ( $id === false && !$rev ) {
3206                                 $linkCache = LinkCache::singleton();
3207                                 $linkCache->addBadLinkObj( $title );
3208                         }
3209
3210                         $deps[] = array(
3211                                 'title' => $title,
3212                                 'page_id' => $title->getArticleID(),
3213                                 'rev_id' => $rev_id );
3214
3215                         if ( $rev ) {
3216                                 $text = $rev->getText();
3217                         } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3218                                 global $wgContLang;
3219                                 $message = $wgContLang->lcfirst( $title->getText() );
3220                                 $text = wfMsgForContentNoTrans( $message );
3221                                 if ( wfEmptyMsg( $message, $text ) ) {
3222                                         $text = false;
3223                                         break;
3224                                 }
3225                         } else {
3226                                 break;
3227                         }
3228                         if ( $text === false ) {
3229                                 break;
3230                         }
3231                         # Redirect?
3232                         $finalTitle = $title;
3233                         $title = Title::newFromRedirect( $text );
3234                 }
3235                 return array(
3236                         'text' => $text,
3237                         'finalTitle' => $finalTitle,
3238                         'deps' => $deps );
3239         }
3240
3241         /**
3242          * Transclude an interwiki link.
3243          */
3244         function interwikiTransclude( $title, $action ) {
3245                 global $wgEnableScaryTranscluding;
3246
3247                 if ( !$wgEnableScaryTranscluding ) {
3248                         return wfMsg('scarytranscludedisabled');
3249                 }
3250
3251                 $url = $title->getFullUrl( "action=$action" );
3252
3253                 if ( strlen( $url ) > 255 ) {
3254                         return wfMsg( 'scarytranscludetoolong' );
3255                 }
3256                 return $this->fetchScaryTemplateMaybeFromCache( $url );
3257         }
3258
3259         function fetchScaryTemplateMaybeFromCache( $url ) {
3260                 global $wgTranscludeCacheExpiry;
3261                 $dbr = wfGetDB( DB_SLAVE );
3262                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3263                 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3264                                 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3265                 if ( $obj ) {
3266                         return $obj->tc_contents;
3267                 }
3268
3269                 $text = Http::get( $url );
3270                 if ( !$text ) {
3271                         return wfMsg( 'scarytranscludefailed', $url );
3272                 }
3273
3274                 $dbw = wfGetDB( DB_MASTER );
3275                 $dbw->replace( 'transcache', array('tc_url'), array(
3276                         'tc_url' => $url,
3277                         'tc_time' => $dbw->timestamp( time() ),
3278                         'tc_contents' => $text)
3279                 );
3280                 return $text;
3281         }
3282
3283
3284         /**
3285          * Triple brace replacement -- used for template arguments
3286          * @private
3287          */
3288         function argSubstitution( $piece, $frame ) {
3289                 wfProfileIn( __METHOD__ );
3290
3291                 $error = false;
3292                 $parts = $piece['parts'];
3293                 $nameWithSpaces = $frame->expand( $piece['title'] );
3294                 $argName = trim( $nameWithSpaces );
3295                 $object = false;
3296                 $text = $frame->getArgument( $argName );
3297                 if (  $text === false && $parts->getLength() > 0
3298                   && (
3299                     $this->ot['html']
3300                     || $this->ot['pre']
3301                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3302                   )
3303                 ) {
3304                         # No match in frame, use the supplied default
3305                         $object = $parts->item( 0 )->getChildren();
3306                 }
3307                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3308                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3309                         $this->limitationWarn( 'post-expand-template-argument' );
3310                 }
3311
3312                 if ( $text === false && $object === false ) {
3313                         # No match anywhere
3314                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3315                 }
3316                 if ( $error !== false ) {
3317                         $text .= $error;
3318                 }
3319                 if ( $object !== false ) {
3320                         $ret = array( 'object' => $object );
3321                 } else {
3322                         $ret = array( 'text' => $text );
3323                 }
3324
3325                 wfProfileOut( __METHOD__ );
3326                 return $ret;
3327         }
3328
3329         /**
3330          * Return the text to be used for a given extension tag.
3331          * This is the ghost of strip().
3332          *
3333          * @param array $params Associative array of parameters:
3334          *     name       PPNode for the tag name
3335          *     attr       PPNode for unparsed text where tag attributes are thought to be
3336          *     attributes Optional associative array of parsed attributes
3337          *     inner      Contents of extension element
3338          *     noClose    Original text did not have a close tag
3339          * @param PPFrame $frame
3340          */
3341         function extensionSubstitution( $params, $frame ) {
3342                 global $wgRawHtml, $wgContLang;
3343
3344                 $name = $frame->expand( $params['name'] );
3345                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3346                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3347                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3348
3349                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3350                         ( $this->ot['html'] || $this->ot['pre'] );
3351                 if ( $isFunctionTag ) {
3352                         $markerType = 'none';
3353                 } else {
3354                         $markerType = 'general';
3355                 }
3356                 if ( $this->ot['html'] || $isFunctionTag ) {
3357                         $name = strtolower( $name );
3358                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3359                         if ( isset( $params['attributes'] ) ) {
3360                                 $attributes = $attributes + $params['attributes'];
3361                         }
3362
3363                         if ( isset( $this->mTagHooks[$name] ) ) {
3364                                 # Workaround for PHP bug 35229 and similar
3365                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3366                                         throw new MWException( "Tag hook for $name is not callable\n" );
3367                                 }
3368                                 $output = call_user_func_array( $this->mTagHooks[$name],
3369                                         array( $content, $attributes, $this, $frame ) );
3370                         } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3371                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3372                                 if ( !is_callable( $callback ) ) {
3373                                         throw new MWException( "Tag hook for $name is not callable\n" );
3374                                 }
3375
3376                                 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3377                         } else {
3378                                 $output = '<span class="error">Invalid tag extension name: ' .
3379                                         htmlspecialchars( $name ) . '</span>';
3380                         }
3381
3382                         if ( is_array( $output ) ) {
3383                                 # Extract flags to local scope (to override $markerType)
3384                                 $flags = $output;
3385                                 $output = $flags[0];
3386                                 unset( $flags[0] );
3387                                 extract( $flags );
3388                         }
3389                 } else {
3390                         if ( is_null( $attrText ) ) {
3391                                 $attrText = '';
3392                         }
3393                         if ( isset( $params['attributes'] ) ) {
3394                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3395                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3396                                                 htmlspecialchars( $attrValue ) . '"';
3397                                 }
3398                         }
3399                         if ( $content === null ) {
3400                                 $output = "<$name$attrText/>";
3401                         } else {
3402                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3403                                 $output = "<$name$attrText>$content$close";
3404                         }
3405                 }
3406
3407                 if ( $markerType === 'none' ) {
3408                         return $output;
3409                 } elseif ( $markerType === 'nowiki' ) {
3410                         $this->mStripState->nowiki->setPair( $marker, $output );
3411                 } elseif ( $markerType === 'general' ) {
3412                         $this->mStripState->general->setPair( $marker, $output );
3413                 } else {
3414                         throw new MWException( __METHOD__.': invalid marker type' );
3415                 }
3416                 return $marker;
3417         }
3418
3419         /**
3420          * Increment an include size counter
3421          *
3422          * @param string $type The type of expansion
3423          * @param integer $size The size of the text
3424          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3425          */
3426         function incrementIncludeSize( $type, $size ) {
3427                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3428                         return false;
3429                 } else {
3430                         $this->mIncludeSizes[$type] += $size;
3431                         return true;
3432                 }
3433         }
3434
3435         /**
3436          * Increment the expensive function count
3437          *
3438          * @return boolean False if the limit has been exceeded
3439          */
3440         function incrementExpensiveFunctionCount() {
3441                 global $wgExpensiveParserFunctionLimit;
3442                 $this->mExpensiveFunctionCount++;
3443                 if ( $this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit ) {
3444                         return true;
3445                 }
3446                 return false;
3447         }
3448
3449         /**
3450          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3451          * Fills $this->mDoubleUnderscores, returns the modified text
3452          */
3453         function doDoubleUnderscore( $text ) {
3454                 wfProfileIn( __METHOD__ );
3455
3456                 # The position of __TOC__ needs to be recorded
3457                 $mw = MagicWord::get( 'toc' );
3458                 if ( $mw->match( $text ) ) {
3459                         $this->mShowToc = true;
3460                         $this->mForceTocPosition = true;
3461
3462                         # Set a placeholder. At the end we'll fill it in with the TOC.
3463                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3464
3465                         # Only keep the first one.
3466                         $text = $mw->replace( '', $text );
3467                 }
3468
3469                 # Now match and remove the rest of them
3470                 $mwa = MagicWord::getDoubleUnderscoreArray();
3471                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3472
3473                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3474                         $this->mOutput->mNoGallery = true;
3475                 }
3476                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3477                         $this->mShowToc = false;
3478                 }
3479                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3480                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3481                         $this->addTrackingCategory( 'hidden-category-category' );
3482                 }
3483                 # (bug 8068) Allow control over whether robots index a page.
3484                 #
3485                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3486                 # is not desirable, the last one on the page should win.
3487                 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3488                         $this->mOutput->setIndexPolicy( 'noindex' );
3489                         $this->addTrackingCategory( 'noindex-category' );
3490                 }
3491                 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3492                         $this->mOutput->setIndexPolicy( 'index' );
3493                         $this->addTrackingCategory( 'index-category' );
3494                 }
3495
3496                 wfProfileOut( __METHOD__ );
3497                 return $text;
3498         }
3499
3500         /**
3501          * Add a tracking category, getting the title from a system message,
3502          * or print a debug message if the title is invalid.
3503          * @param $msg String message key
3504          * @return Bool whether the addition was successful
3505          */
3506         protected function addTrackingCategory( $msg ) {
3507                 $cat = wfMsgForContent( $msg );
3508
3509                 # Allow tracking categories to be disabled by setting them to "-"
3510                 if ( $cat === '-' ) {
3511                         return false;
3512                 }
3513
3514                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3515                 if ( $containerCategory ) {
3516                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3517                         return true;
3518                 } else {
3519                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3520                         return false;
3521                 }
3522         }
3523
3524         /**
3525          * This function accomplishes several tasks:
3526          * 1) Auto-number headings if that option is enabled
3527          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3528          * 3) Add a Table of contents on the top for users who have enabled the option
3529          * 4) Auto-anchor headings
3530          *
3531          * It loops through all headlines, collects the necessary data, then splits up the
3532          * string and re-inserts the newly formatted headlines.
3533          *
3534          * @param string $text
3535          * @param string $origText Original, untouched wikitext
3536          * @param boolean $isMain
3537          * @private
3538          */
3539         function formatHeadings( $text, $origText, $isMain=true ) {
3540                 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3541
3542                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3543                 $showEditLink = $this->mOptions->getEditSection();
3544
3545                 # Do not call quickUserCan unless necessary
3546                 if ( $showEditLink && !$this->mTitle->quickUserCan( 'edit' ) ) {
3547                         $showEditLink = 0;
3548                 }
3549
3550                 # Inhibit editsection links if requested in the page
3551                 if ( isset( $this->mDoubleUnderscores['noeditsection'] )  || $this->mOptions->getIsPrintable() ) {
3552                         $showEditLink = 0;
3553                 }
3554
3555                 # Get all headlines for numbering them and adding funky stuff like [edit]
3556                 # links - this is for later, but we need the number of headlines right now
3557                 $matches = array();
3558                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3559
3560                 # if there are fewer than 4 headlines in the article, do not show TOC
3561                 # unless it's been explicitly enabled.
3562                 $enoughToc = $this->mShowToc &&
3563                         ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3564
3565                 # Allow user to stipulate that a page should have a "new section"
3566                 # link added via __NEWSECTIONLINK__
3567                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3568                         $this->mOutput->setNewSection( true );
3569                 }
3570
3571                 # Allow user to remove the "new section"
3572                 # link via __NONEWSECTIONLINK__
3573                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3574                         $this->mOutput->hideNewSection( true );
3575                 }
3576
3577                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3578                 # override above conditions and always show TOC above first header
3579                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3580                         $this->mShowToc = true;
3581                         $enoughToc = true;
3582                 }
3583
3584                 # We need this to perform operations on the HTML
3585                 $sk = $this->mOptions->getSkin();
3586
3587                 # headline counter
3588                 $headlineCount = 0;
3589                 $numVisible = 0;
3590
3591                 # Ugh .. the TOC should have neat indentation levels which can be
3592                 # passed to the skin functions. These are determined here
3593                 $toc = '';
3594                 $full = '';
3595                 $head = array();
3596                 $sublevelCount = array();
3597                 $levelCount = array();
3598                 $toclevel = 0;
3599                 $level = 0;
3600                 $prevlevel = 0;
3601                 $toclevel = 0;
3602                 $prevtoclevel = 0;
3603                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3604                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3605                 $oldType = $this->mOutputType;
3606                 $this->setOutputType( self::OT_WIKI );
3607                 $frame = $this->getPreprocessor()->newFrame();
3608                 $root = $this->preprocessToDom( $origText );
3609                 $node = $root->getFirstChild();
3610                 $byteOffset = 0;
3611                 $tocraw = array();
3612
3613                 foreach ( $matches[3] as $headline ) {
3614                         $isTemplate = false;
3615                         $titleText = false;
3616                         $sectionIndex = false;
3617                         $numbering = '';
3618                         $markerMatches = array();
3619                         if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
3620                                 $serial = $markerMatches[1];
3621                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3622                                 $isTemplate = ( $titleText != $baseTitleText );
3623                                 $headline = preg_replace( "/^$markerRegex/", "", $headline );
3624                         }
3625
3626                         if ( $toclevel ) {
3627                                 $prevlevel = $level;
3628                                 $prevtoclevel = $toclevel;
3629                         }
3630                         $level = $matches[1][$headlineCount];
3631
3632                         if ( $level > $prevlevel ) {
3633                                 # Increase TOC level
3634                                 $toclevel++;
3635                                 $sublevelCount[$toclevel] = 0;
3636                                 if ( $toclevel<$wgMaxTocLevel ) {
3637                                         $prevtoclevel = $toclevel;
3638                                         $toc .= $sk->tocIndent();
3639                                         $numVisible++;
3640                                 }
3641                         } elseif ( $level < $prevlevel && $toclevel > 1 ) {
3642                                 # Decrease TOC level, find level to jump to
3643
3644                                 for ( $i = $toclevel; $i > 0; $i-- ) {
3645                                         if ( $levelCount[$i] == $level ) {
3646                                                 # Found last matching level
3647                                                 $toclevel = $i;
3648                                                 break;
3649                                         } elseif ( $levelCount[$i] < $level ) {
3650                                                 # Found first matching level below current level
3651                                                 $toclevel = $i + 1;
3652                                                 break;
3653                                         }
3654                                 }
3655                                 if ( $i == 0 ) {
3656                                         $toclevel = 1;
3657                                 }
3658                                 if ( $toclevel<$wgMaxTocLevel ) {
3659                                         if ( $prevtoclevel < $wgMaxTocLevel ) {
3660                                                 # Unindent only if the previous toc level was shown :p
3661                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3662                                                 $prevtoclevel = $toclevel;
3663                                         } else {
3664                                                 $toc .= $sk->tocLineEnd();
3665                                         }
3666                                 }
3667                         } else {
3668                                 # No change in level, end TOC line
3669                                 if ( $toclevel<$wgMaxTocLevel ) {
3670                                         $toc .= $sk->tocLineEnd();
3671                                 }
3672                         }
3673
3674                         $levelCount[$toclevel] = $level;
3675
3676                         # count number of headlines for each level
3677                         @$sublevelCount[$toclevel]++;
3678                         $dot = 0;
3679                         for( $i = 1; $i <= $toclevel; $i++ ) {
3680                                 if ( !empty( $sublevelCount[$i] ) ) {
3681                                         if ( $dot ) {
3682                                                 $numbering .= '.';
3683                                         }
3684                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3685                                         $dot = 1;
3686                                 }
3687                         }
3688
3689                         # The safe header is a version of the header text safe to use for links
3690                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3691                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3692
3693                         # Remove link placeholders by the link text.
3694                         #     <!--LINK number-->
3695                         # turns into
3696                         #     link text with suffix
3697                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3698
3699                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3700                         $tocline = preg_replace(
3701                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3702                                 array( '',                          '<$1>' ),
3703                                 $safeHeadline
3704                         );
3705                         $tocline = trim( $tocline );
3706
3707                         # For the anchor, strip out HTML-y stuff period
3708                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3709                         $safeHeadline = preg_replace( '/[ _]+/', ' ', $safeHeadline );
3710                         $safeHeadline = trim( $safeHeadline );
3711
3712                         # Save headline for section edit hint before it's escaped
3713                         $headlineHint = $safeHeadline;
3714
3715                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
3716                                 # For reverse compatibility, provide an id that's
3717                                 # HTML4-compatible, like we used to.
3718                                 #
3719                                 # It may be worth noting, academically, that it's possible for
3720                                 # the legacy anchor to conflict with a non-legacy headline
3721                                 # anchor on the page.  In this case likely the "correct" thing
3722                                 # would be to either drop the legacy anchors or make sure
3723                                 # they're numbered first.  However, this would require people
3724                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3725                                 # manually, so let's not bother worrying about it.
3726                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3727                                         array( 'noninitial', 'legacy' ) );
3728                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3729
3730                                 if ( $legacyHeadline == $safeHeadline ) {
3731                                         # No reason to have both (in fact, we can't)
3732                                         $legacyHeadline = false;
3733                                 }
3734                         } else {
3735                                 $legacyHeadline = false;
3736                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3737                                         'noninitial' );
3738                         }
3739
3740                         # HTML names must be case-insensitively unique (bug 10721).  FIXME:
3741                         # Does this apply to Unicode characters?  Because we aren't
3742                         # handling those here.
3743                         $arrayKey = strtolower( $safeHeadline );
3744                         if ( $legacyHeadline === false ) {
3745                                 $legacyArrayKey = false;
3746                         } else {
3747                                 $legacyArrayKey = strtolower( $legacyHeadline );
3748                         }
3749
3750                         # count how many in assoc. array so we can track dupes in anchors
3751                         if ( isset( $refers[$arrayKey] ) ) {
3752                                 $refers[$arrayKey]++;
3753                         } else {
3754                                 $refers[$arrayKey] = 1;
3755                         }
3756                         if ( isset( $refers[$legacyArrayKey] ) ) {
3757                                 $refers[$legacyArrayKey]++;
3758                         } else {
3759                                 $refers[$legacyArrayKey] = 1;
3760                         }
3761
3762                         # Don't number the heading if it is the only one (looks silly)
3763                         if ( $doNumberHeadings && count( $matches[3] ) > 1) {
3764                                 # the two are different if the line contains a link
3765                                 $headline = $numbering . ' ' . $headline;
3766                         }
3767
3768                         # Create the anchor for linking from the TOC to the section
3769                         $anchor = $safeHeadline;
3770                         $legacyAnchor = $legacyHeadline;
3771                         if ( $refers[$arrayKey] > 1 ) {
3772                                 $anchor .= '_' . $refers[$arrayKey];
3773                         }
3774                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3775                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3776                         }
3777                         if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
3778                                 $toc .= $sk->tocLine( $anchor, $tocline,
3779                                         $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
3780                         }
3781
3782                         # Add the section to the section tree
3783                         # Find the DOM node for this header
3784                         while ( $node && !$isTemplate ) {
3785                                 if ( $node->getName() === 'h' ) {
3786                                         $bits = $node->splitHeading();
3787                                         if ( $bits['i'] == $sectionIndex )
3788                                                 break;
3789                                 }
3790                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
3791                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
3792                                 $node = $node->getNextSibling();
3793                         }
3794                         $tocraw[] = array(
3795                                 'toclevel' => $toclevel,
3796                                 'level' => $level,
3797                                 'line' => $tocline,
3798                                 'number' => $numbering,
3799                                 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
3800                                 'fromtitle' => $titleText,
3801                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
3802                                 'anchor' => $anchor,
3803                         );
3804
3805                         # give headline the correct <h#> tag
3806                         if ( $showEditLink && $sectionIndex !== false ) {
3807                                 if ( $isTemplate ) {
3808                                         # Put a T flag in the section identifier, to indicate to extractSections()
3809                                         # that sections inside <includeonly> should be counted.
3810                                         $editlink = $sk->doEditSectionLink( Title::newFromText( $titleText ), "T-$sectionIndex" );
3811                                 } else {
3812                                         $editlink = $sk->doEditSectionLink( $this->mTitle, $sectionIndex, $headlineHint );
3813                                 }
3814                         } else {
3815                                 $editlink = '';
3816                         }
3817                         $head[$headlineCount] = $sk->makeHeadline( $level,
3818                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3819                                 $editlink, $legacyAnchor );
3820
3821                         $headlineCount++;
3822                 }
3823
3824                 $this->setOutputType( $oldType );
3825
3826                 # Never ever show TOC if no headers
3827                 if ( $numVisible < 1 ) {
3828                         $enoughToc = false;
3829                 }
3830
3831                 if ( $enoughToc ) {
3832                         if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3833                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3834                         }
3835                         $toc = $sk->tocList( $toc );
3836                         $this->mOutput->setTOCHTML( $toc );
3837                 }
3838
3839                 if ( $isMain ) {
3840                         $this->mOutput->setSections( $tocraw );
3841                 }
3842
3843                 # split up and insert constructed headlines
3844
3845                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3846                 $i = 0;
3847
3848                 foreach ( $blocks as $block ) {
3849                         if ( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3850                                 # This is the [edit] link that appears for the top block of text when
3851                                 # section editing is enabled
3852
3853                                 # Disabled because it broke block formatting
3854                                 # For example, a bullet point in the top line
3855                                 # $full .= $sk->editSectionLink(0);
3856                         }
3857                         $full .= $block;
3858                         if ( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3859                                 # Top anchor now in skin
3860                                 $full = $full.$toc;
3861                         }
3862
3863                         if ( !empty( $head[$i] ) ) {
3864                                 $full .= $head[$i];
3865                         }
3866                         $i++;
3867                 }
3868                 if ( $this->mForceTocPosition ) {
3869                         return str_replace( '<!--MWTOC-->', $toc, $full );
3870                 } else {
3871                         return $full;
3872                 }
3873         }
3874
3875         /**
3876          * Merge $tree2 into $tree1 by replacing the section with index
3877          * $section in $tree1 and its descendants with the sections in $tree2.
3878          * Note that in the returned section tree, only the 'index' and
3879          * 'byteoffset' fields are guaranteed to be correct.
3880          * @param $tree1 array Section tree from ParserOutput::getSectons()
3881         *  @param $tree2 array Section tree
3882          * @param $section int Section index
3883          * @param $title Title Title both section trees come from
3884          * @param $len2 int Length of the original wikitext for $tree2
3885          * @return array Merged section tree
3886          */
3887         public static function mergeSectionTrees( $tree1, $tree2, $section, $title, $len2 ) {
3888                 global $wgContLang;
3889                 $newTree = array();
3890                 $targetLevel = false;
3891                 $merged = false;
3892                 $lastLevel = 1;
3893                 $nextIndex = 1;
3894                 $numbering = array( 0 );
3895                 $titletext = $title->getPrefixedDBkey();
3896                 foreach ( $tree1 as $s ) {
3897                         if ( $targetLevel !== false ) {
3898                                 if ( $s['level'] <= $targetLevel ) {
3899                                         # We've skipped enough
3900                                         $targetLevel = false;
3901                                 } else {
3902                                         continue;
3903                                 }
3904                         }
3905                         if ( $s['index'] != $section ||
3906                                         $s['fromtitle'] != $titletext ) {
3907                                 self::incrementNumbering( $numbering,
3908                                         $s['toclevel'], $lastLevel );
3909
3910                                 # Rewrite index, byteoffset and number
3911                                 if ( $s['fromtitle'] == $titletext ) {
3912                                         $s['index'] = $nextIndex++;
3913                                         if ( $merged ) {
3914                                                 $s['byteoffset'] += $len2;
3915                                         }
3916                                 }
3917                                 $s['number']  = implode( '.', array_map(
3918                                         array( $wgContLang, 'formatnum' ),
3919                                         $numbering ) );
3920                                 $lastLevel = $s['toclevel'];
3921                                 $newTree[] = $s;
3922                         } else {
3923                                 # We're at $section
3924                                 # Insert sections from $tree2 here
3925                                 foreach ( $tree2 as $s2 ) {
3926                                         # Rewrite the fields in $s2
3927                                         # before inserting it
3928                                         $s2['toclevel'] += $s['toclevel'] - 1;
3929                                         $s2['level'] += $s['level'] - 1;
3930                                         $s2['index'] = $nextIndex++;
3931                                         $s2['byteoffset'] += $s['byteoffset'];
3932
3933                                         self::incrementNumbering( $numbering,
3934                                                 $s2['toclevel'], $lastLevel );
3935                                         $s2['number']  = implode( '.', array_map(
3936                                                 array( $wgContLang, 'formatnum' ),
3937                                                 $numbering ) );
3938                                         $lastLevel = $s2['toclevel'];
3939                                         $newTree[] = $s2;
3940                                 }
3941                                 # Skip all descendants of $section in $tree1
3942                                 $targetLevel = $s['level'];
3943                                 $merged = true;
3944                         }
3945                 }
3946                 return $newTree;
3947         }
3948
3949         /**
3950          * Increment a section number. Helper function for mergeSectionTrees()
3951          * @param $number array Array representing a section number
3952          * @param $level int Current TOC level (depth)
3953          * @param $lastLevel int Level of previous TOC entry
3954          */
3955         private static function incrementNumbering( &$number, $level, $lastLevel ) {
3956                 if ( $level > $lastLevel ) {
3957                         $number[$level - 1] = 1;
3958                 } elseif ( $level < $lastLevel ) {
3959                         foreach ( $number as $key => $unused )
3960                                 if ( $key >= $level ) {
3961                                         unset( $number[$key] );
3962                                 }
3963                         $number[$level - 1]++;
3964                 } else {
3965                         $number[$level - 1]++;
3966                 }
3967         }
3968
3969         /**
3970          * Transform wiki markup when saving a page by doing \r\n -> \n
3971          * conversion, substitting signatures, {{subst:}} templates, etc.
3972          *
3973          * @param string $text the text to transform
3974          * @param Title &$title the Title object for the current article
3975          * @param User $user the User object describing the current user
3976          * @param ParserOptions $options parsing options
3977          * @param bool $clearState whether to clear the parser state first
3978          * @return string the altered wiki markup
3979          * @public
3980          */
3981         function preSaveTransform( $text, Title $title, $user, $options, $clearState = true ) {
3982                 $this->mOptions = $options;
3983                 $this->setTitle( $title );
3984                 $this->setOutputType( self::OT_WIKI );
3985
3986                 if ( $clearState ) {
3987                         $this->clearState();
3988                 }
3989
3990                 $pairs = array(
3991                         "\r\n" => "\n",
3992                 );
3993                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3994                 $text = $this->pstPass2( $text, $user );
3995                 $text = $this->mStripState->unstripBoth( $text );
3996                 return $text;
3997         }
3998
3999         /**
4000          * Pre-save transform helper function
4001          * @private
4002          */
4003         function pstPass2( $text, $user ) {
4004                 global $wgContLang, $wgLocaltimezone;
4005
4006                 # Note: This is the timestamp saved as hardcoded wikitext to
4007                 # the database, we use $wgContLang here in order to give
4008                 # everyone the same signature and use the default one rather
4009                 # than the one selected in each user's preferences.
4010                 # (see also bug 12815)
4011                 $ts = $this->mOptions->getTimestamp();
4012                 if ( isset( $wgLocaltimezone ) ) {
4013                         $tz = $wgLocaltimezone;
4014                 } else {
4015                         $tz = date_default_timezone_get();
4016                 }
4017
4018                 $unixts = wfTimestamp( TS_UNIX, $ts );
4019                 $oldtz = date_default_timezone_get();
4020                 date_default_timezone_set( $tz );
4021                 $ts = date( 'YmdHis', $unixts );
4022                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4023
4024                 # Allow translation of timezones trough wiki. date() can return
4025                 # whatever crap the system uses, localised or not, so we cannot
4026                 # ship premade translations.
4027                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4028                 $value = wfMsgForContent( $key );
4029                 if ( !wfEmptyMsg( $key, $value ) ) {
4030                         $tzMsg = $value;
4031                 }
4032
4033                 date_default_timezone_set( $oldtz );
4034
4035                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4036
4037                 # Variable replacement
4038                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4039                 $text = $this->replaceVariables( $text );
4040
4041                 # Signatures
4042                 $sigText = $this->getUserSig( $user );
4043                 $text = strtr( $text, array(
4044                         '~~~~~' => $d,
4045                         '~~~~' => "$sigText $d",
4046                         '~~~' => $sigText
4047                 ) );
4048
4049                 # Context links: [[|name]] and [[name (context)|]]
4050                 global $wgLegalTitleChars;
4051                 $tc = "[$wgLegalTitleChars]";
4052                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4053
4054                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4055                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
4056                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4057                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4058
4059                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4060                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4061                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4062                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4063
4064                 $t = $this->mTitle->getText();
4065                 $m = array();
4066                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4067                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4068                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4069                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4070                 } else {
4071                         # if there's no context, don't bother duplicating the title
4072                         $text = preg_replace( $p2, '[[\\1]]', $text );
4073                 }
4074
4075                 # Trim trailing whitespace
4076                 $text = rtrim( $text );
4077
4078                 return $text;
4079         }
4080
4081         /**
4082          * Fetch the user's signature text, if any, and normalize to
4083          * validated, ready-to-insert wikitext.
4084          * If you have pre-fetched the nickname or the fancySig option, you can
4085          * specify them here to save a database query.
4086          *
4087          * @param User $user
4088          * @return string
4089          */
4090         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4091                 global $wgMaxSigChars;
4092
4093                 $username = $user->getName();
4094
4095                 # If not given, retrieve from the user object.
4096                 if ( $nickname === false )
4097                         $nickname = $user->getOption( 'nickname' );
4098
4099                 if ( is_null( $fancySig ) ) {
4100                         $fancySig = $user->getBoolOption( 'fancysig' );
4101                 }
4102
4103                 $nickname = $nickname == null ? $username : $nickname;
4104
4105                 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4106                         $nickname = $username;
4107                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4108                 } elseif ( $fancySig !== false ) {
4109                         # Sig. might contain markup; validate this
4110                         if ( $this->validateSig( $nickname ) !== false ) {
4111                                 # Validated; clean up (if needed) and return it
4112                                 return $this->cleanSig( $nickname, true );
4113                         } else {
4114                                 # Failed to validate; fall back to the default
4115                                 $nickname = $username;
4116                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4117                         }
4118                 }
4119
4120                 # Make sure nickname doesnt get a sig in a sig
4121                 $nickname = $this->cleanSigInSig( $nickname );
4122
4123                 # If we're still here, make it a link to the user page
4124                 $userText = wfEscapeWikiText( $username );
4125                 $nickText = wfEscapeWikiText( $nickname );
4126                 if ( $user->isAnon() )  {
4127                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4128                 } else {
4129                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4130                 }
4131         }
4132
4133         /**
4134          * Check that the user's signature contains no bad XML
4135          *
4136          * @param string $text
4137          * @return mixed An expanded string, or false if invalid.
4138          */
4139         function validateSig( $text ) {
4140                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4141         }
4142
4143         /**
4144          * Clean up signature text
4145          *
4146          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4147          * 2) Substitute all transclusions
4148          *
4149          * @param string $text
4150          * @param $parsing Whether we're cleaning (preferences save) or parsing
4151          * @return string Signature text
4152          */
4153         function cleanSig( $text, $parsing = false ) {
4154                 if ( !$parsing ) {
4155                         global $wgTitle;
4156                         $this->clearState();
4157                         $this->setTitle( $wgTitle );
4158                         $this->mOptions = new ParserOptions;
4159                         $this->setOutputType = self::OT_PREPROCESS;
4160                 }
4161
4162                 # Option to disable this feature
4163                 if ( !$this->mOptions->getCleanSignatures() ) {
4164                         return $text;
4165                 }
4166
4167                 # FIXME: regex doesn't respect extension tags or nowiki
4168                 #  => Move this logic to braceSubstitution()
4169                 $substWord = MagicWord::get( 'subst' );
4170                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4171                 $substText = '{{' . $substWord->getSynonym( 0 );
4172
4173                 $text = preg_replace( $substRegex, $substText, $text );
4174                 $text = $this->cleanSigInSig( $text );
4175                 $dom = $this->preprocessToDom( $text );
4176                 $frame = $this->getPreprocessor()->newFrame();
4177                 $text = $frame->expand( $dom );
4178
4179                 if ( !$parsing ) {
4180                         $text = $this->mStripState->unstripBoth( $text );
4181                 }
4182
4183                 return $text;
4184         }
4185
4186         /**
4187          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4188          * @param string $text
4189          * @return string Signature text with /~{3,5}/ removed
4190          */
4191         function cleanSigInSig( $text ) {
4192                 $text = preg_replace( '/~{3,5}/', '', $text );
4193                 return $text;
4194         }
4195
4196         /**
4197          * Set up some variables which are usually set up in parse()
4198          * so that an external function can call some class members with confidence
4199          * @public
4200          */
4201         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4202                 $this->setTitle( $title );
4203                 $this->mOptions = $options;
4204                 $this->setOutputType( $outputType );
4205                 if ( $clearState ) {
4206                         $this->clearState();
4207                 }
4208         }
4209
4210         /**
4211          * Wrapper for preprocess()
4212          *
4213          * @param string $text the text to preprocess
4214          * @param ParserOptions $options  options
4215          * @return string
4216          * @public
4217          */
4218         function transformMsg( $text, $options ) {
4219                 global $wgTitle;
4220                 static $executing = false;
4221
4222                 # Guard against infinite recursion
4223                 if ( $executing ) {
4224                         return $text;
4225                 }
4226                 $executing = true;
4227
4228                 wfProfileIn( __METHOD__ );
4229                 $text = $this->preprocess( $text, $wgTitle, $options );
4230
4231                 $executing = false;
4232                 wfProfileOut( __METHOD__ );
4233                 return $text;
4234         }
4235
4236         /**
4237          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4238          * The callback should have the following form:
4239          *    function myParserHook( $text, $params, $parser ) { ... }
4240          *
4241          * Transform and return $text. Use $parser for any required context, e.g. use
4242          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4243          *
4244          * @public
4245          *
4246          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4247          * @param mixed $callback The callback function (and object) to use for the tag
4248          *
4249          * @return The old value of the mTagHooks array associated with the hook
4250          */
4251         function setHook( $tag, $callback ) {
4252                 $tag = strtolower( $tag );
4253                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4254                 $this->mTagHooks[$tag] = $callback;
4255                 if ( !in_array( $tag, $this->mStripList ) ) {
4256                         $this->mStripList[] = $tag;
4257                 }
4258
4259                 return $oldVal;
4260         }
4261
4262         function setTransparentTagHook( $tag, $callback ) {
4263                 $tag = strtolower( $tag );
4264                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4265                 $this->mTransparentTagHooks[$tag] = $callback;
4266
4267                 return $oldVal;
4268         }
4269
4270         /**
4271          * Remove all tag hooks
4272          */
4273         function clearTagHooks() {
4274                 $this->mTagHooks = array();
4275                 $this->mStripList = $this->mDefaultStripList;
4276         }
4277
4278         /**
4279          * Create a function, e.g. {{sum:1|2|3}}
4280          * The callback function should have the form:
4281          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4282          *
4283          * Or with SFH_OBJECT_ARGS:
4284          *    function myParserFunction( $parser, $frame, $args ) { ... }
4285          *
4286          * The callback may either return the text result of the function, or an array with the text
4287          * in element 0, and a number of flags in the other elements. The names of the flags are
4288          * specified in the keys. Valid flags are:
4289          *   found                     The text returned is valid, stop processing the template. This
4290          *                             is on by default.
4291          *   nowiki                    Wiki markup in the return value should be escaped
4292          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4293          *
4294          * @public
4295          *
4296          * @param string $id The magic word ID
4297          * @param mixed $callback The callback function (and object) to use
4298          * @param integer $flags a combination of the following flags:
4299          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4300          *
4301          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4302          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4303          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4304          *     the arguments, and to control the way they are expanded.
4305          *
4306          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4307          *     arguments, for instance:
4308          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4309          *
4310          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4311          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4312          *     working if/when this is changed.
4313          *
4314          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4315          *     expansion.
4316          *
4317          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4318          *     about the methods available in PPFrame and PPNode.
4319          *
4320          * @return The old callback function for this name, if any
4321          */
4322         function setFunctionHook( $id, $callback, $flags = 0 ) {
4323                 global $wgContLang;
4324
4325                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4326                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4327
4328                 # Add to function cache
4329                 $mw = MagicWord::get( $id );
4330                 if ( !$mw )
4331                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4332
4333                 $synonyms = $mw->getSynonyms();
4334                 $sensitive = intval( $mw->isCaseSensitive() );
4335
4336                 foreach ( $synonyms as $syn ) {
4337                         # Case
4338                         if ( !$sensitive ) {
4339                                 $syn = $wgContLang->lc( $syn );
4340                         }
4341                         # Add leading hash
4342                         if ( !( $flags & SFH_NO_HASH ) ) {
4343                                 $syn = '#' . $syn;
4344                         }
4345                         # Remove trailing colon
4346                         if ( substr( $syn, -1, 1 ) === ':' ) {
4347                                 $syn = substr( $syn, 0, -1 );
4348                         }
4349                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4350                 }
4351                 return $oldVal;
4352         }
4353
4354         /**
4355          * Get all registered function hook identifiers
4356          *
4357          * @return array
4358          */
4359         function getFunctionHooks() {
4360                 return array_keys( $this->mFunctionHooks );
4361         }
4362
4363         /**
4364          * Create a tag function, e.g. <test>some stuff</test>.
4365          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4366          * Unlike parser functions, their content is not preprocessed.
4367          */
4368         function setFunctionTagHook( $tag, $callback, $flags ) {
4369                 $tag = strtolower( $tag );
4370                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4371                         $this->mFunctionTagHooks[$tag] : null;
4372                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4373
4374                 if ( !in_array( $tag, $this->mStripList ) ) {
4375                         $this->mStripList[] = $tag;
4376                 }
4377
4378                 return $old;
4379         }
4380
4381         /**
4382          * FIXME: update documentation. makeLinkObj() is deprecated.
4383          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4384          * Placeholders created in Skin::makeLinkObj()
4385          * Returns an array of link CSS classes, indexed by PDBK.
4386          */
4387         function replaceLinkHolders( &$text, $options = 0 ) {
4388                 return $this->mLinkHolders->replace( $text );
4389         }
4390
4391         /**
4392          * Replace <!--LINK--> link placeholders with plain text of links
4393          * (not HTML-formatted).
4394          * @param string $text
4395          * @return string
4396          */
4397         function replaceLinkHoldersText( $text ) {
4398                 return $this->mLinkHolders->replaceText( $text );
4399         }
4400
4401         /**
4402          * Renders an image gallery from a text with one line per image.
4403          * text labels may be given by using |-style alternative text. E.g.
4404          *   Image:one.jpg|The number "1"
4405          *   Image:tree.jpg|A tree
4406          * given as text will return the HTML of a gallery with two images,
4407          * labeled 'The number "1"' and
4408          * 'A tree'.
4409          */
4410         function renderImageGallery( $text, $params ) {
4411                 $ig = new ImageGallery();
4412                 $ig->setContextTitle( $this->mTitle );
4413                 $ig->setShowBytes( false );
4414                 $ig->setShowFilename( false );
4415                 $ig->setParser( $this );
4416                 $ig->setHideBadImages();
4417                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4418                 $ig->useSkin( $this->mOptions->getSkin() );
4419                 $ig->mRevisionId = $this->mRevisionId;
4420
4421                 if ( isset( $params['showfilename'] ) ) {
4422                         $ig->setShowFilename( true );
4423                 } else {
4424                         $ig->setShowFilename( false );
4425                 }
4426                 if ( isset( $params['caption'] ) ) {
4427                         $caption = $params['caption'];
4428                         $caption = htmlspecialchars( $caption );
4429                         $caption = $this->replaceInternalLinks( $caption );
4430                         $ig->setCaptionHtml( $caption );
4431                 }
4432                 if ( isset( $params['perrow'] ) ) {
4433                         $ig->setPerRow( $params['perrow'] );
4434                 }
4435                 if ( isset( $params['widths'] ) ) {
4436                         $ig->setWidths( $params['widths'] );
4437                 }
4438                 if ( isset( $params['heights'] ) ) {
4439                         $ig->setHeights( $params['heights'] );
4440                 }
4441
4442                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4443
4444                 $lines = StringUtils::explode( "\n", $text );
4445                 foreach ( $lines as $line ) {
4446                         # match lines like these:
4447                         # Image:someimage.jpg|This is some image
4448                         $matches = array();
4449                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4450                         # Skip empty lines
4451                         if ( count( $matches ) == 0 ) {
4452                                 continue;
4453                         }
4454
4455                         if ( strpos( $matches[0], '%' ) !== false ) {
4456                                 $matches[1] = urldecode( $matches[1] );
4457                         }
4458                         $tp = Title::newFromText( $matches[1] );
4459                         $nt =& $tp;
4460                         if ( is_null( $nt ) ) {
4461                                 # Bogus title. Ignore these so we don't bomb out later.
4462                                 continue;
4463                         }
4464                         if ( isset( $matches[3] ) ) {
4465                                 $label = $matches[3];
4466                         } else {
4467                                 $label = '';
4468                         }
4469
4470                         $html = $this->recursiveTagParse( trim( $label ) );
4471
4472                         $ig->add( $nt, $html );
4473
4474                         # Only add real images (bug #5586)
4475                         if ( $nt->getNamespace() == NS_FILE ) {
4476                                 $this->mOutput->addImage( $nt->getDBkey() );
4477                         }
4478                 }
4479                 return $ig->toHTML();
4480         }
4481
4482         function getImageParams( $handler ) {
4483                 if ( $handler ) {
4484                         $handlerClass = get_class( $handler );
4485                 } else {
4486                         $handlerClass = '';
4487                 }
4488                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4489                         # Initialise static lists
4490                         static $internalParamNames = array(
4491                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4492                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4493                                         'bottom', 'text-bottom' ),
4494                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4495                                         'upright', 'border', 'link', 'alt' ),
4496                         );
4497                         static $internalParamMap;
4498                         if ( !$internalParamMap ) {
4499                                 $internalParamMap = array();
4500                                 foreach ( $internalParamNames as $type => $names ) {
4501                                         foreach ( $names as $name ) {
4502                                                 $magicName = str_replace( '-', '_', "img_$name" );
4503                                                 $internalParamMap[$magicName] = array( $type, $name );
4504                                         }
4505                                 }
4506                         }
4507
4508                         # Add handler params
4509                         $paramMap = $internalParamMap;
4510                         if ( $handler ) {
4511                                 $handlerParamMap = $handler->getParamMap();
4512                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4513                                         $paramMap[$magic] = array( 'handler', $paramName );
4514                                 }
4515                         }
4516                         $this->mImageParams[$handlerClass] = $paramMap;
4517                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4518                 }
4519                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4520         }
4521
4522         /**
4523          * Parse image options text and use it to make an image
4524          * @param Title $title
4525          * @param string $options
4526          * @param LinkHolderArray $holders
4527          */
4528         function makeImage( $title, $options, $holders = false ) {
4529                 # Check if the options text is of the form "options|alt text"
4530                 # Options are:
4531                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4532                 #  * left       no resizing, just left align. label is used for alt= only
4533                 #  * right      same, but right aligned
4534                 #  * none       same, but not aligned
4535                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4536                 #  * center     center the image
4537                 #  * frame      Keep original image size, no magnify-button.
4538                 #  * framed     Same as "frame"
4539                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4540                 #  * upright    reduce width for upright images, rounded to full __0 px
4541                 #  * border     draw a 1px border around the image
4542                 #  * alt        Text for HTML alt attribute (defaults to empty)
4543                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4544                 # vertical-align values (no % or length right now):
4545                 #  * baseline
4546                 #  * sub
4547                 #  * super
4548                 #  * top
4549                 #  * text-top
4550                 #  * middle
4551                 #  * bottom
4552                 #  * text-bottom
4553
4554                 $parts = StringUtils::explode( "|", $options );
4555                 $sk = $this->mOptions->getSkin();
4556
4557                 # Give extensions a chance to select the file revision for us
4558                 $skip = $time = $descQuery = false;
4559                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4560
4561                 if ( $skip ) {
4562                         return $sk->link( $title );
4563                 }
4564
4565                 # Get the file
4566                 $imagename = $title->getDBkey();
4567                 $file = wfFindFile( $title, array( 'time' => $time ) );
4568                 # Get parameter map
4569                 $handler = $file ? $file->getHandler() : false;
4570
4571                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4572
4573                 # Process the input parameters
4574                 $caption = '';
4575                 $params = array( 'frame' => array(), 'handler' => array(),
4576                         'horizAlign' => array(), 'vertAlign' => array() );
4577                 foreach ( $parts as $part ) {
4578                         $part = trim( $part );
4579                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4580                         $validated = false;
4581                         if ( isset( $paramMap[$magicName] ) ) {
4582                                 list( $type, $paramName ) = $paramMap[$magicName];
4583
4584                                 # Special case; width and height come in one variable together
4585                                 if ( $type === 'handler' && $paramName === 'width' ) {
4586                                         $m = array();
4587                                         # (bug 13500) In both cases (width/height and width only),
4588                                         # permit trailing "px" for backward compatibility.
4589                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4590                                                 $width = intval( $m[1] );
4591                                                 $height = intval( $m[2] );
4592                                                 if ( $handler->validateParam( 'width', $width ) ) {
4593                                                         $params[$type]['width'] = $width;
4594                                                         $validated = true;
4595                                                 }
4596                                                 if ( $handler->validateParam( 'height', $height ) ) {
4597                                                         $params[$type]['height'] = $height;
4598                                                         $validated = true;
4599                                                 }
4600                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4601                                                 $width = intval( $value );
4602                                                 if ( $handler->validateParam( 'width', $width ) ) {
4603                                                         $params[$type]['width'] = $width;
4604                                                         $validated = true;
4605                                                 }
4606                                         } # else no validation -- bug 13436
4607                                 } else {
4608                                         if ( $type === 'handler' ) {
4609                                                 # Validate handler parameter
4610                                                 $validated = $handler->validateParam( $paramName, $value );
4611                                         } else {
4612                                                 # Validate internal parameters
4613                                                 switch( $paramName ) {
4614                                                 case 'manualthumb':
4615                                                 case 'alt':
4616                                                         # @todo Fixme: possibly check validity here for
4617                                                         # manualthumb? downstream behavior seems odd with
4618                                                         # missing manual thumbs.
4619                                                         $validated = true;
4620                                                         $value = $this->stripAltText( $value, $holders );
4621                                                         break;
4622                                                 case 'link':
4623                                                         $chars = self::EXT_LINK_URL_CLASS;
4624                                                         $prots = $this->mUrlProtocols;
4625                                                         if ( $value === '' ) {
4626                                                                 $paramName = 'no-link';
4627                                                                 $value = true;
4628                                                                 $validated = true;
4629                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4630                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4631                                                                         $paramName = 'link-url';
4632                                                                         $this->mOutput->addExternalLink( $value );
4633                                                                         $validated = true;
4634                                                                 }
4635                                                         } else {
4636                                                                 $linkTitle = Title::newFromText( $value );
4637                                                                 if ( $linkTitle ) {
4638                                                                         $paramName = 'link-title';
4639                                                                         $value = $linkTitle;
4640                                                                         $this->mOutput->addLink( $linkTitle );
4641                                                                         $validated = true;
4642                                                                 }
4643                                                         }
4644                                                         break;
4645                                                 default:
4646                                                         # Most other things appear to be empty or numeric...
4647                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4648                                                 }
4649                                         }
4650
4651                                         if ( $validated ) {
4652                                                 $params[$type][$paramName] = $value;
4653                                         }
4654                                 }
4655                         }
4656                         if ( !$validated ) {
4657                                 $caption = $part;
4658                         }
4659                 }
4660
4661                 # Process alignment parameters
4662                 if ( $params['horizAlign'] ) {
4663                         $params['frame']['align'] = key( $params['horizAlign'] );
4664                 }
4665                 if ( $params['vertAlign'] ) {
4666                         $params['frame']['valign'] = key( $params['vertAlign'] );
4667                 }
4668
4669                 $params['frame']['caption'] = $caption;
4670
4671                 # Will the image be presented in a frame, with the caption below?
4672                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4673                                  isset( $params['frame']['framed'] ) ||
4674                                  isset( $params['frame']['thumbnail'] ) ||
4675                                  isset( $params['frame']['manualthumb'] );
4676
4677                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4678                 # came to also set the caption, ordinary text after the image -- which
4679                 # makes no sense, because that just repeats the text multiple times in
4680                 # screen readers.  It *also* came to set the title attribute.
4681                 #
4682                 # Now that we have an alt attribute, we should not set the alt text to
4683                 # equal the caption: that's worse than useless, it just repeats the
4684                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4685                 # use the unnamed parameter for alt text as well, just for the time be-
4686                 # ing, if the unnamed param is set and the alt param is not.
4687                 #
4688                 # For the future, we need to figure out if we want to tweak this more,
4689                 # e.g., introducing a title= parameter for the title; ignoring the un-
4690                 # named parameter entirely for images without a caption; adding an ex-
4691                 # plicit caption= parameter and preserving the old magic unnamed para-
4692                 # meter for BC; ...
4693                 if ( $imageIsFramed ) { # Framed image
4694                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4695                                 # No caption or alt text, add the filename as the alt text so
4696                                 # that screen readers at least get some description of the image
4697                                 $params['frame']['alt'] = $title->getText();
4698                         }
4699                         # Do not set $params['frame']['title'] because tooltips don't make sense
4700                         # for framed images
4701                 } else { # Inline image
4702                         if ( !isset( $params['frame']['alt'] ) ) {
4703                                 # No alt text, use the "caption" for the alt text
4704                                 if ( $caption !== '') {
4705                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4706                                 } else {
4707                                         # No caption, fall back to using the filename for the
4708                                         # alt text
4709                                         $params['frame']['alt'] = $title->getText();
4710                                 }
4711                         }
4712                         # Use the "caption" for the tooltip text
4713                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4714                 }
4715
4716                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4717
4718                 # Linker does the rest
4719                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4720
4721                 # Give the handler a chance to modify the parser object
4722                 if ( $handler ) {
4723                         $handler->parserTransformHook( $this, $file );
4724                 }
4725
4726                 return $ret;
4727         }
4728
4729         protected function stripAltText( $caption, $holders ) {
4730                 # Strip bad stuff out of the title (tooltip).  We can't just use
4731                 # replaceLinkHoldersText() here, because if this function is called
4732                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4733                 if ( $holders ) {
4734                         $tooltip = $holders->replaceText( $caption );
4735                 } else {
4736                         $tooltip = $this->replaceLinkHoldersText( $caption );
4737                 }
4738
4739                 # make sure there are no placeholders in thumbnail attributes
4740                 # that are later expanded to html- so expand them now and
4741                 # remove the tags
4742                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4743                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4744
4745                 return $tooltip;
4746         }
4747
4748         /**
4749          * Set a flag in the output object indicating that the content is dynamic and
4750          * shouldn't be cached.
4751          */
4752         function disableCache() {
4753                 wfDebug( "Parser output marked as uncacheable.\n" );
4754                 $this->mOutput->mCacheTime = -1;
4755         }
4756
4757         /**#@+
4758          * Callback from the Sanitizer for expanding items found in HTML attribute
4759          * values, so they can be safely tested and escaped.
4760          * @param string $text
4761          * @param PPFrame $frame
4762          * @return string
4763          * @private
4764          */
4765         function attributeStripCallback( &$text, $frame = false ) {
4766                 $text = $this->replaceVariables( $text, $frame );
4767                 $text = $this->mStripState->unstripBoth( $text );
4768                 return $text;
4769         }
4770
4771         /**#@-*/
4772
4773         /**#@+
4774          * Accessor/mutator
4775          */
4776         function Title( $x = null ) { return wfSetVar( $this->mTitle, $x ); }
4777         function Options( $x = null ) { return wfSetVar( $this->mOptions, $x ); }
4778         function OutputType( $x = null ) { return wfSetVar( $this->mOutputType, $x ); }
4779         /**#@-*/
4780
4781         /**#@+
4782          * Accessor
4783          */
4784         function getTags() {
4785                 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ) );
4786         }
4787         /**#@-*/
4788
4789
4790         /**
4791          * Break wikitext input into sections, and either pull or replace
4792          * some particular section's text.
4793          *
4794          * External callers should use the getSection and replaceSection methods.
4795          *
4796          * @param string $text Page wikitext
4797          * @param string $section A section identifier string of the form:
4798          *   <flag1> - <flag2> - ... - <section number>
4799          *
4800          * Currently the only recognised flag is "T", which means the target section number
4801          * was derived during a template inclusion parse, in other words this is a template
4802          * section edit link. If no flags are given, it was an ordinary section edit link.
4803          * This flag is required to avoid a section numbering mismatch when a section is
4804          * enclosed by <includeonly> (bug 6563).
4805          *
4806          * The section number 0 pulls the text before the first heading; other numbers will
4807          * pull the given section along with its lower-level subsections. If the section is
4808          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4809          *
4810          * @param string $mode One of "get" or "replace"
4811          * @param string $newText Replacement text for section data.
4812          * @return string for "get", the extracted section text.
4813          *                for "replace", the whole page with the section replaced.
4814          */
4815         private function extractSections( $text, $section, $mode, $newText='' ) {
4816                 global $wgTitle;
4817                 $this->clearState();
4818                 $this->setTitle( $wgTitle ); # not generally used but removes an ugly failure mode
4819                 $this->mOptions = new ParserOptions;
4820                 $this->setOutputType( self::OT_PLAIN );
4821                 $outText = '';
4822                 $frame = $this->getPreprocessor()->newFrame();
4823
4824                 # Process section extraction flags
4825                 $flags = 0;
4826                 $sectionParts = explode( '-', $section );
4827                 $sectionIndex = array_pop( $sectionParts );
4828                 foreach ( $sectionParts as $part ) {
4829                         if ( $part === 'T' ) {
4830                                 $flags |= self::PTD_FOR_INCLUSION;
4831                         }
4832                 }
4833                 # Preprocess the text
4834                 $root = $this->preprocessToDom( $text, $flags );
4835
4836                 # <h> nodes indicate section breaks
4837                 # They can only occur at the top level, so we can find them by iterating the root's children
4838                 $node = $root->getFirstChild();
4839
4840                 # Find the target section
4841                 if ( $sectionIndex == 0 ) {
4842                         # Section zero doesn't nest, level=big
4843                         $targetLevel = 1000;
4844                 } else {
4845             while ( $node ) {
4846                 if ( $node->getName() === 'h' ) {
4847                     $bits = $node->splitHeading();
4848                                         if ( $bits['i'] == $sectionIndex ) {
4849                                         $targetLevel = $bits['level'];
4850                                                 break;
4851                                         }
4852                                 }
4853                                 if ( $mode === 'replace' ) {
4854                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4855                                 }
4856                                 $node = $node->getNextSibling();
4857                         }
4858                 }
4859
4860                 if ( !$node ) {
4861                         # Not found
4862                         if ( $mode === 'get' ) {
4863                                 return $newText;
4864                         } else {
4865                                 return $text;
4866                         }
4867                 }
4868
4869                 # Find the end of the section, including nested sections
4870                 do {
4871                         if ( $node->getName() === 'h' ) {
4872                                 $bits = $node->splitHeading();
4873                                 $curLevel = $bits['level'];
4874                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4875                                         break;
4876                                 }
4877                         }
4878                         if ( $mode === 'get' ) {
4879                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4880                         }
4881                         $node = $node->getNextSibling();
4882                 } while ( $node );
4883
4884                 # Write out the remainder (in replace mode only)
4885                 if ( $mode === 'replace' ) {
4886                         # Output the replacement text
4887                         # Add two newlines on -- trailing whitespace in $newText is conventionally
4888                         # stripped by the editor, so we need both newlines to restore the paragraph gap
4889                         # Only add trailing whitespace if there is newText
4890                         if ( $newText != "" ) {
4891                                 $outText .= $newText . "\n\n";
4892                         }
4893
4894                         while ( $node ) {
4895                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4896                                 $node = $node->getNextSibling();
4897                         }
4898                 }
4899
4900                 if ( is_string( $outText ) ) {
4901                         # Re-insert stripped tags
4902                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4903                 }
4904
4905                 return $outText;
4906         }
4907
4908         /**
4909          * This function returns the text of a section, specified by a number ($section).
4910          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4911          * the first section before any such heading (section 0).
4912          *
4913          * If a section contains subsections, these are also returned.
4914          *
4915          * @param string $text text to look in
4916          * @param string $section section identifier
4917          * @param string $deftext default to return if section is not found
4918          * @return string text of the requested section
4919          */
4920         public function getSection( $text, $section, $deftext='' ) {
4921                 return $this->extractSections( $text, $section, "get", $deftext );
4922         }
4923
4924         public function replaceSection( $oldtext, $section, $text ) {
4925                 return $this->extractSections( $oldtext, $section, "replace", $text );
4926         }
4927
4928         /**
4929          * Get the timestamp associated with the current revision, adjusted for
4930          * the default server-local timestamp
4931          */
4932         function getRevisionTimestamp() {
4933                 if ( is_null( $this->mRevisionTimestamp ) ) {
4934                         wfProfileIn( __METHOD__ );
4935                         global $wgContLang;
4936                         $dbr = wfGetDB( DB_SLAVE );
4937                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4938                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4939
4940                         # Normalize timestamp to internal MW format for timezone processing.
4941                         # This has the added side-effect of replacing a null value with
4942                         # the current time, which gives us more sensible behavior for
4943                         # previews.
4944                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4945
4946                         # The cryptic '' timezone parameter tells to use the site-default
4947                         # timezone offset instead of the user settings.
4948                         #
4949                         # Since this value will be saved into the parser cache, served
4950                         # to other users, and potentially even used inside links and such,
4951                         # it needs to be consistent for all visitors.
4952                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4953
4954                         wfProfileOut( __METHOD__ );
4955                 }
4956                 return $this->mRevisionTimestamp;
4957         }
4958
4959         /**
4960          * Get the name of the user that edited the last revision
4961          */
4962         function getRevisionUser() {
4963                 # if this template is subst: the revision id will be blank,
4964                 # so just use the current user's name
4965                 if ( $this->mRevisionId ) {
4966                         $revision = Revision::newFromId( $this->mRevisionId );
4967                         $revuser = $revision->getUserText();
4968                 } else {
4969                         global $wgUser;
4970                         $revuser = $wgUser->getName();
4971                 }
4972                 return $revuser;
4973         }
4974
4975         /**
4976          * Mutator for $mDefaultSort
4977          *
4978          * @param $sort New value
4979          */
4980         public function setDefaultSort( $sort ) {
4981                 $this->mDefaultSort = $sort;
4982         }
4983
4984         /**
4985          * Accessor for $mDefaultSort
4986          * Will use the title/prefixed title if none is set
4987          *
4988          * @return string
4989          */
4990         public function getDefaultSort() {
4991                 global $wgCategoryPrefixedDefaultSortkey;
4992                 if ( $this->mDefaultSort !== false ) {
4993                         return $this->mDefaultSort;
4994                 } elseif ( $this->mTitle->getNamespace() == NS_CATEGORY ||
4995                         !$wgCategoryPrefixedDefaultSortkey )
4996                 {
4997                         return $this->mTitle->getText();
4998                 } else {
4999                         return $this->mTitle->getPrefixedText();
5000                 }
5001         }
5002
5003         /**
5004          * Accessor for $mDefaultSort
5005          * Unlike getDefaultSort(), will return false if none is set
5006          *
5007          * @return string or false
5008          */
5009         public function getCustomDefaultSort() {
5010                 return $this->mDefaultSort;
5011         }
5012
5013         /**
5014          * Try to guess the section anchor name based on a wikitext fragment
5015          * presumably extracted from a heading, for example "Header" from
5016          * "== Header ==".
5017          */
5018         public function guessSectionNameFromWikiText( $text ) {
5019                 # Strip out wikitext links(they break the anchor)
5020                 $text = $this->stripSectionName( $text );
5021                 $headline = Sanitizer::decodeCharReferences( $text );
5022                 # strip out HTML
5023                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
5024                 $headline = trim( $headline );
5025                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
5026                 $replacearray = array(
5027                         '%3A' => ':',
5028                         '%' => '.'
5029                 );
5030                 return str_replace(
5031                         array_keys( $replacearray ),
5032                         array_values( $replacearray ),
5033                         $sectionanchor );
5034         }
5035
5036         /**
5037          * Strips a text string of wikitext for use in a section anchor
5038          *
5039          * Accepts a text string and then removes all wikitext from the
5040          * string and leaves only the resultant text (i.e. the result of
5041          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5042          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5043          * to create valid section anchors by mimicing the output of the
5044          * parser when headings are parsed.
5045          *
5046          * @param $text string Text string to be stripped of wikitext
5047          * for use in a Section anchor
5048          * @return Filtered text string
5049          */
5050         public function stripSectionName( $text ) {
5051                 # Strip internal link markup
5052                 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5053                 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5054
5055                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5056                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5057                 # on how many empty links there are on the page - need to figure that out.
5058                 $text = preg_replace( '/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5059
5060                 # Parse wikitext quotes (italics & bold)
5061                 $text = $this->doQuotes( $text );
5062
5063                 # Strip HTML tags
5064                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5065                 return $text;
5066         }
5067
5068         function srvus( $text ) {
5069                 return $this->testSrvus( $text, $this->mOutputType );
5070         }
5071
5072         /**
5073          * strip/replaceVariables/unstrip for preprocessor regression testing
5074          */
5075         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
5076                 $this->clearState();
5077                 if ( !$title instanceof Title ) {
5078                         $title = Title::newFromText( $title );
5079                 }
5080                 $this->mTitle = $title;
5081                 $this->mOptions = $options;
5082                 $this->setOutputType( $outputType );
5083                 $text = $this->replaceVariables( $text );
5084                 $text = $this->mStripState->unstripBoth( $text );
5085                 $text = Sanitizer::removeHTMLtags( $text );
5086                 return $text;
5087         }
5088
5089         function testPst( $text, $title, $options ) {
5090                 global $wgUser;
5091                 if ( !$title instanceof Title ) {
5092                         $title = Title::newFromText( $title );
5093                 }
5094                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5095         }
5096
5097         function testPreprocess( $text, $title, $options ) {
5098                 if ( !$title instanceof Title ) {
5099                         $title = Title::newFromText( $title );
5100                 }
5101                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5102         }
5103
5104         function markerSkipCallback( $s, $callback ) {
5105                 $i = 0;
5106                 $out = '';
5107                 while ( $i < strlen( $s ) ) {
5108                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5109                         if ( $markerStart === false ) {
5110                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5111                                 break;
5112                         } else {
5113                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5114                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5115                                 if ( $markerEnd === false ) {
5116                                         $out .= substr( $s, $markerStart );
5117                                         break;
5118                                 } else {
5119                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5120                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5121                                         $i = $markerEnd;
5122                                 }
5123                         }
5124                 }
5125                 return $out;
5126         }
5127
5128         function serialiseHalfParsedText( $text ) {
5129                 $data = array();
5130                 $data['text'] = $text;
5131
5132                 # First, find all strip markers, and store their
5133                 #  data in an array.
5134                 $stripState = new StripState;
5135                 $pos = 0;
5136                 while ( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) )
5137                         && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) )
5138                 {
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         /**
5192          * TODO: document
5193          * @param $data Array
5194          * @param $intPrefix String unique identifying prefix
5195          * @return String
5196          */
5197         function unserialiseHalfParsedText( $data, $intPrefix = null ) {
5198                 if ( !$intPrefix ) {
5199                         $intPrefix = $this->getRandomString();
5200                 }
5201
5202                 # First, extract the strip state.
5203                 $stripState = $data['stripstate'];
5204                 $this->mStripState->general->merge( $stripState->general );
5205                 $this->mStripState->nowiki->merge( $stripState->nowiki );
5206
5207                 # Now, extract the text, and renumber links
5208                 $text = $data['text'];
5209                 $links = $data['linkholder'];
5210
5211                 # Internal...
5212                 foreach ( $links['internal'] as $ns => $nsLinks ) {
5213                         foreach ( $nsLinks as $key => $entry ) {
5214                                 $newKey = $intPrefix . '-' . $key;
5215                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
5216
5217                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
5218                         }
5219                 }
5220
5221                 # Interwiki...
5222                 foreach ( $links['interwiki'] as $key => $entry ) {
5223                         $newKey = "$intPrefix-$key";
5224                         $this->mLinkHolders->interwikis[$newKey] = $entry;
5225
5226                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
5227                 }
5228
5229                 # Should be good to go.
5230                 return $text;
5231         }
5232 }
5233
5234 /**
5235  * @todo document, briefly.
5236  * @ingroup Parser
5237  */
5238 class StripState {
5239         var $general, $nowiki;
5240
5241         function __construct() {
5242                 $this->general = new ReplacementArray;
5243                 $this->nowiki = new ReplacementArray;
5244         }
5245
5246         function unstripGeneral( $text ) {
5247                 wfProfileIn( __METHOD__ );
5248                 do {
5249                         $oldText = $text;
5250                         $text = $this->general->replace( $text );
5251                 } while ( $text !== $oldText );
5252                 wfProfileOut( __METHOD__ );
5253                 return $text;
5254         }
5255
5256         function unstripNoWiki( $text ) {
5257                 wfProfileIn( __METHOD__ );
5258                 do {
5259                         $oldText = $text;
5260                         $text = $this->nowiki->replace( $text );
5261                 } while ( $text !== $oldText );
5262                 wfProfileOut( __METHOD__ );
5263                 return $text;
5264         }
5265
5266         function unstripBoth( $text ) {
5267                 wfProfileIn( __METHOD__ );
5268                 do {
5269                         $oldText = $text;
5270                         $text = $this->general->replace( $text );
5271                         $text = $this->nowiki->replace( $text );
5272                 } while ( $text !== $oldText );
5273                 wfProfileOut( __METHOD__ );
5274                 return $text;
5275         }
5276 }
5277
5278 /**
5279  * @todo document, briefly.
5280  * @ingroup Parser
5281  */
5282 class OnlyIncludeReplacer {
5283         var $output = '';
5284
5285         function replace( $matches ) {
5286                 if ( substr( $matches[1], -1 ) === "\n" ) {
5287                         $this->output .= substr( $matches[1], 0, -1 );
5288                 } else {
5289                         $this->output .= $matches[1];
5290                 }
5291         }
5292 }