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() / cleanSigInSig()
  26  *     Cleans a signature before saving it to preferences
  27  * getSection()
  28  *     Return the content of a section from an article for section editing
  29  * replaceSection()
  30  *     Replaces a section by number inside an article
  31  * getPreloadText()
  32  *     Removes <noinclude> sections, and <includeonly> tags.
  33  *
  34  * Globals used:
  35  *    objects:   $wgLang, $wgContLang
  36  *
  37  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  38  *
  39  * settings:
  40  *  $wgUseDynamicDates*, $wgInterwikiMagic*,
  41  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  42  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  43  *  $wgMaxArticleSize*
  44  *
  45  *  * only within ParserOptions
  46  * </pre>
  47  *
  48  * @ingroup Parser
  49  */
  50 class Parser {
  51         /**
  52          * Update this version number when the ParserOutput format
  53          * changes in an incompatible way, so the parser cache
  54          * can automatically discard old data.
  55          */
  56         const VERSION = '1.6.5';
  57
  58         /**
  59          * Update this version number when the output of serialiseHalfParsedText()
  60          * changes in an incompatible way
  61          */
  62         const HALF_PARSED_VERSION = 2;
  63
  64         # Flags for Parser::setFunctionHook
  65         # Also available as global constants from Defines.php
  66         const SFH_NO_HASH = 1;
  67         const SFH_OBJECT_ARGS = 2;
  68
  69         # Constants needed for external link processing
  70         # Everything except bracket, space, or control characters
  71         const EXT_LINK_URL_CLASS = '(?:[^\]\[<>"\\x00-\\x20\\x7F]|(?:\[\]))';
  72         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  73                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  74
  75         # State constants for the definition list colon extraction
  76         const COLON_STATE_TEXT = 0;
  77         const COLON_STATE_TAG = 1;
  78         const COLON_STATE_TAGSTART = 2;
  79         const COLON_STATE_CLOSETAG = 3;
  80         const COLON_STATE_TAGSLASH = 4;
  81         const COLON_STATE_COMMENT = 5;
  82         const COLON_STATE_COMMENTDASH = 6;
  83         const COLON_STATE_COMMENTDASHDASH = 7;
  84
  85         # Flags for preprocessToDom
  86         const PTD_FOR_INCLUSION = 1;
  87
  88         # Allowed values for $this->mOutputType
  89         # Parameter to startExternalParse().
  90         const OT_HTML = 1; # like parse()
  91         const OT_WIKI = 2; # like preSaveTransform()
  92         const OT_PREPROCESS = 3; # like preprocess()
  93         const OT_MSG = 3;
  94         const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
  95
  96         # Marker Suffix needs to be accessible staticly.
  97         const MARKER_SUFFIX = "-QINU\x7f";
  98
  99         # Persistent:
 100         var $mTagHooks = array();
 101         var $mTransparentTagHooks = array();
 102         var $mFunctionHooks = array();
 103         var $mFunctionSynonyms = array( 0 => array(), 1 => array() );
 104         var $mFunctionTagHooks = array();
 105         var $mStripList  = array();
 106         var $mDefaultStripList  = array();
 107         var $mVarCache = array();
 108         var $mImageParams = array();
 109         var $mImageParamsMagicArray = array();
 110         var $mMarkerIndex = 0;
 111         var $mFirstCall = true;
 112         var $mVariables, $mSubstWords; # Initialised by initialiseVariables()
 113         var $mConf, $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols; # Initialised in constructor
 114
 115         # Cleared with clearState():
 116         /**
 117          * @var ParserOutput
 118          */
 119         var $mOutput;
 120         var $mAutonumber, $mDTopen;
 121
 122         /**
 123          * @var StripState
 124          */
 125         var $mStripState;
 126
 127         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 128         var $mLinkHolders, $mLinkID;
 129         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 130         var $mTplExpandCache; # empty-frame expansion cache
 131         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 132         var $mExpensiveFunctionCount; # number of expensive parser function calls
 133
 134         /**
 135          * @var User
 136          */
 137         var $mUser; # User object; only used when doing pre-save transform
 138
 139         # Temporary
 140         # These are variables reset at least once per parse regardless of $clearState
 141
 142         /**
 143          * @var ParserOptions
 144          */
 145         var $mOptions;
 146
 147         /**
 148          * @var Title
 149          */
 150         var $mTitle;        # Title context, used for self-link rendering and similar things
 151         var $mOutputType;   # Output type, one of the OT_xxx constants
 152         var $ot;            # Shortcut alias, see setOutputType()
 153         var $mRevisionObject; # The revision object of the specified revision ID
 154         var $mRevisionId;   # ID to display in {{REVISIONID}} tags
 155         var $mRevisionTimestamp; # The timestamp of the specified revision ID
 156         var $mRevisionUser; # Userto display in {{REVISIONUSER}} tag
 157         var $mRevIdForTs;   # The revision ID which was used to fetch the timestamp
 158
 159         /**
 160          * Constructor
 161          */
 162         public function __construct( $conf = array() ) {
 163                 $this->mConf = $conf;
 164                 $this->mUrlProtocols = wfUrlProtocols();
 165                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 166                         '(?:[^\]\[<>"\x00-\x20\x7F]|\[\])+) *([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/S';
 167                 if ( isset( $conf['preprocessorClass'] ) ) {
 168                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 169                 } elseif ( extension_loaded( 'domxml' ) ) {
 170                         # PECL extension that conflicts with the core DOM extension (bug 13770)
 171                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 172                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 173                 } elseif ( extension_loaded( 'dom' ) ) {
 174                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 175                 } else {
 176                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 177                 }
 178         }
 179
 180         /**
 181          * Reduce memory usage to reduce the impact of circular references
 182          */
 183         function __destruct() {
 184                 if ( isset( $this->mLinkHolders ) ) {
 185                         unset( $this->mLinkHolders );
 186                 }
 187                 foreach ( $this as $name => $value ) {
 188                         unset( $this->$name );
 189                 }
 190         }
 191
 192         /**
 193          * Do various kinds of initialisation on the first call of the parser
 194          */
 195         function firstCallInit() {
 196                 if ( !$this->mFirstCall ) {
 197                         return;
 198                 }
 199                 $this->mFirstCall = false;
 200
 201                 wfProfileIn( __METHOD__ );
 202
 203                 CoreParserFunctions::register( $this );
 204                 CoreTagHooks::register( $this );
 205                 $this->initialiseVariables();
 206
 207                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 208                 wfProfileOut( __METHOD__ );
 209         }
 210
 211         /**
 212          * Clear Parser state
 213          *
 214          * @private
 215          */
 216         function clearState() {
 217                 wfProfileIn( __METHOD__ );
 218                 if ( $this->mFirstCall ) {
 219                         $this->firstCallInit();
 220                 }
 221                 $this->mOutput = new ParserOutput;
 222                 $this->mOptions->registerWatcher( array( $this->mOutput, 'recordOption' ) );
 223                 $this->mAutonumber = 0;
 224                 $this->mLastSection = '';
 225                 $this->mDTopen = false;
 226                 $this->mIncludeCount = array();
 227                 $this->mArgStack = false;
 228                 $this->mInPre = false;
 229                 $this->mLinkHolders = new LinkHolderArray( $this );
 230                 $this->mLinkID = 0;
 231                 $this->mRevisionObject = $this->mRevisionTimestamp =
 232                         $this->mRevisionId = $this->mRevisionUser = null;
 233                 $this->mVarCache = array();
 234                 $this->mUser = null;
 235
 236                 /**
 237                  * Prefix for temporary replacement strings for the multipass parser.
 238                  * \x07 should never appear in input as it's disallowed in XML.
 239                  * Using it at the front also gives us a little extra robustness
 240                  * since it shouldn't match when butted up against identifier-like
 241                  * string constructs.
 242                  *
 243                  * Must not consist of all title characters, or else it will change
 244                  * the behaviour of <nowiki> in a link.
 245                  */
 246                 # $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 247                 # Changed to \x7f to allow XML double-parsing -- TS
 248                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 249                 $this->mStripState = new StripState( $this->mUniqPrefix );
 250
 251
 252                 # Clear these on every parse, bug 4549
 253                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 254
 255                 $this->mShowToc = true;
 256                 $this->mForceTocPosition = false;
 257                 $this->mIncludeSizes = array(
 258                         'post-expand' => 0,
 259                         'arg' => 0,
 260                 );
 261                 $this->mPPNodeCount = 0;
 262                 $this->mDefaultSort = false;
 263                 $this->mHeadings = array();
 264                 $this->mDoubleUnderscores = array();
 265                 $this->mExpensiveFunctionCount = 0;
 266
 267                 # Fix cloning
 268                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 269                         $this->mPreprocessor = null;
 270                 }
 271
 272                 wfRunHooks( 'ParserClearState', array( &$this ) );
 273                 wfProfileOut( __METHOD__ );
 274         }
 275
 276         /**
 277          * Convert wikitext to HTML
 278          * Do not call this function recursively.
 279          *
 280          * @param $text String: text we want to parse
 281          * @param $title Title object
 282          * @param $options ParserOptions
 283          * @param $linestart boolean
 284          * @param $clearState boolean
 285          * @param $revid Int: number to pass in {{REVISIONID}}
 286          * @return ParserOutput a ParserOutput
 287          */
 288         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 289                 /**
 290                  * First pass--just handle <nowiki> sections, pass the rest off
 291                  * to internalParse() which does all the real work.
 292                  */
 293
 294                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang, $wgDisableLangConversion, $wgDisableTitleConversion;
 295                 $fname = __METHOD__.'-' . wfGetCaller();
 296                 wfProfileIn( __METHOD__ );
 297                 wfProfileIn( $fname );
 298
 299                 $this->startParse( $title, $options, self::OT_HTML, $clearState );
 300
 301                 $oldRevisionId = $this->mRevisionId;
 302                 $oldRevisionObject = $this->mRevisionObject;
 303                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 304                 $oldRevisionUser = $this->mRevisionUser;
 305                 if ( $revid !== null ) {
 306                         $this->mRevisionId = $revid;
 307                         $this->mRevisionObject = null;
 308                         $this->mRevisionTimestamp = null;
 309                         $this->mRevisionUser = null;
 310                 }
 311
 312                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 313                 # No more strip!
 314                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 315                 $text = $this->internalParse( $text );
 316
 317                 $text = $this->mStripState->unstripGeneral( $text );
 318
 319                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 320                 $fixtags = array(
 321                         # french spaces, last one Guillemet-left
 322                         # only if there is something before the space
 323                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;\\2',
 324                         # french spaces, Guillemet-right
 325                         '/(\\302\\253) /' => '\\1&#160;',
 326                         '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
 327                 );
 328                 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
 329
 330                 $text = $this->doBlockLevels( $text, $linestart );
 331
 332                 $this->replaceLinkHolders( $text );
 333
 334                 /**
 335                  * The page doesn't get language converted if
 336                  * a) It's disabled
 337                  * b) Content isn't converted
 338                  * c) It's a conversion table
 339                  */
 340                 if ( !( $wgDisableLangConversion
 341                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 342                                 || $this->mTitle->isConversionTable() ) ) {
 343
 344                         # The position of the convert() call should not be changed. it
 345                         # assumes that the links are all replaced and the only thing left
 346                         # is the <nowiki> mark.
 347
 348                         $text = $wgContLang->convert( $text );
 349                 }
 350
 351                 /**
 352                  * A converted title will be provided in the output object if title and
 353                  * content conversion are enabled, the article text does not contain
 354                  * a conversion-suppressing double-underscore tag, and no
 355                  * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
 356                  * automatic link conversion.
 357                  */
 358                 if ( !( $wgDisableLangConversion
 359                                 || $wgDisableTitleConversion
 360                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 361                                 || isset( $this->mDoubleUnderscores['notitleconvert'] )
 362                                 || $this->mOutput->getDisplayTitle() !== false ) )
 363                 {
 364                         $convruletitle = $wgContLang->getConvRuleTitle();
 365                         if ( $convruletitle ) {
 366                                 $this->mOutput->setTitleText( $convruletitle );
 367                         } else {
 368                                 $titleText = $wgContLang->convertTitle( $title );
 369                                 $this->mOutput->setTitleText( $titleText );
 370                         }
 371                 }
 372
 373                 $text = $this->mStripState->unstripNoWiki( $text );
 374
 375                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 376
 377                 $text = $this->replaceTransparentTags( $text );
 378                 $text = $this->mStripState->unstripGeneral( $text );
 379
 380                 $text = Sanitizer::normalizeCharReferences( $text );
 381
 382                 if ( ( $wgUseTidy && $this->mOptions->getTidy() ) || $wgAlwaysUseTidy ) {
 383                         $text = MWTidy::tidy( $text );
 384                 } else {
 385                         # attempt to sanitize at least some nesting problems
 386                         # (bug #2702 and quite a few others)
 387                         $tidyregs = array(
 388                                 # ''Something [http://www.cool.com cool''] -->
 389                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 390                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 391                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 392                                 # fix up an anchor inside another anchor, only
 393                                 # at least for a single single nested link (bug 3695)
 394                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 395                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 396                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 397                                 # contains a div, so fix it up here; replace
 398                                 # div with escaped text
 399                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 400                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 401                                 # remove empty italic or bold tag pairs, some
 402                                 # introduced by rules above
 403                                 '/<([bi])><\/\\1>/' => '',
 404                         );
 405
 406                         $text = preg_replace(
 407                                 array_keys( $tidyregs ),
 408                                 array_values( $tidyregs ),
 409                                 $text );
 410                 }
 411                 global $wgExpensiveParserFunctionLimit;
 412                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 413                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 414                 }
 415
 416                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 417
 418                 # Information on include size limits, for the benefit of users who try to skirt them
 419                 if ( $this->mOptions->getEnableLimitReport() ) {
 420                         $max = $this->mOptions->getMaxIncludeSize();
 421                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 422                         $limitReport =
 423                                 "NewPP limit report\n" .
 424                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->getMaxPPNodeCount()}\n" .
 425                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 426                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 427                                 $PFreport;
 428                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 429                         $text .= "\n<!-- \n$limitReport-->\n";
 430                 }
 431                 $this->mOutput->setText( $text );
 432
 433                 $this->mRevisionId = $oldRevisionId;
 434                 $this->mRevisionObject = $oldRevisionObject;
 435                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 436                 $this->mRevisionUser = $oldRevisionUser;
 437                 wfProfileOut( $fname );
 438                 wfProfileOut( __METHOD__ );
 439
 440                 return $this->mOutput;
 441         }
 442
 443         /**
 444          * Recursive parser entry point that can be called from an extension tag
 445          * hook.
 446          *
 447          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 448          *
 449          * @param $text String: text extension wants to have parsed
 450          * @param $frame PPFrame: The frame to use for expanding any template variables
 451          */
 452         function recursiveTagParse( $text, $frame=false ) {
 453                 wfProfileIn( __METHOD__ );
 454                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 455                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 456                 $text = $this->internalParse( $text, false, $frame );
 457                 wfProfileOut( __METHOD__ );
 458                 return $text;
 459         }
 460
 461         /**
 462          * Expand templates and variables in the text, producing valid, static wikitext.
 463          * Also removes comments.
 464          */
 465         function preprocess( $text, Title $title, ParserOptions $options, $revid = null ) {
 466                 wfProfileIn( __METHOD__ );
 467                 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
 468                 if ( $revid !== null ) {
 469                         $this->mRevisionId = $revid;
 470                 }
 471                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 472                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 473                 $text = $this->replaceVariables( $text );
 474                 $text = $this->mStripState->unstripBoth( $text );
 475                 wfProfileOut( __METHOD__ );
 476                 return $text;
 477         }
 478
 479         /**
 480          * Process the wikitext for the ?preload= feature. (bug 5210)
 481          *
 482          * <noinclude>, <includeonly> etc. are parsed as for template transclusion,
 483          * comments, templates, arguments, tags hooks and parser functions are untouched.
 484          */
 485         public function getPreloadText( $text, Title $title, ParserOptions $options ) {
 486                 # Parser (re)initialisation
 487                 $this->startParse( $title, $options, self::OT_PLAIN, true );
 488
 489                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 490                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
 491                 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
 492                 $text = $this->mStripState->unstripBoth( $text );
 493                 return $text;
 494         }
 495
 496         /**
 497          * Get a random string
 498          */
 499         static public function getRandomString() {
 500                 return dechex( mt_rand( 0, 0x7fffffff ) ) . dechex( mt_rand( 0, 0x7fffffff ) );
 501         }
 502
 503         /**
 504          * Set the current user.
 505          * Should only be used when doing pre-save transform.
 506          *
 507          * @param $user Mixed: User object or null (to reset)
 508          */
 509         function setUser( $user ) {
 510                 $this->mUser = $user;
 511         }
 512
 513         /**
 514          * Accessor for mUniqPrefix.
 515          *
 516          * @return String
 517          */
 518         public function uniqPrefix() {
 519                 if ( !isset( $this->mUniqPrefix ) ) {
 520                         # @todo Fixme: this is probably *horribly wrong*
 521                         # LanguageConverter seems to want $wgParser's uniqPrefix, however
 522                         # if this is called for a parser cache hit, the parser may not
 523                         # have ever been initialized in the first place.
 524                         # Not really sure what the heck is supposed to be going on here.
 525                         return '';
 526                         # throw new MWException( "Accessing uninitialized mUniqPrefix" );
 527                 }
 528                 return $this->mUniqPrefix;
 529         }
 530
 531         /**
 532          * Set the context title
 533          */
 534         function setTitle( $t ) {
 535                 if ( !$t || $t instanceof FakeTitle ) {
 536                         $t = Title::newFromText( 'NO TITLE' );
 537                 }
 538
 539                 if ( strval( $t->getFragment() ) !== '' ) {
 540                         # Strip the fragment to avoid various odd effects
 541                         $this->mTitle = clone $t;
 542                         $this->mTitle->setFragment( '' );
 543                 } else {
 544                         $this->mTitle = $t;
 545                 }
 546         }
 547
 548         /**
 549          * Accessor for the Title object
 550          *
 551          * @return Title object
 552          */
 553         function getTitle() {
 554                 return $this->mTitle;
 555         }
 556
 557         /**
 558          * Accessor/mutator for the Title object
 559          *
 560          * @param $x New Title object or null to just get the current one
 561          * @return Title object
 562          */
 563         function Title( $x = null ) {
 564                 return wfSetVar( $this->mTitle, $x );
 565         }
 566
 567         /**
 568          * Set the output type
 569          *
 570          * @param $ot Integer: new value
 571          */
 572         function setOutputType( $ot ) {
 573                 $this->mOutputType = $ot;
 574                 # Shortcut alias
 575                 $this->ot = array(
 576                         'html' => $ot == self::OT_HTML,
 577                         'wiki' => $ot == self::OT_WIKI,
 578                         'pre' => $ot == self::OT_PREPROCESS,
 579                         'plain' => $ot == self::OT_PLAIN,
 580                 );
 581         }
 582
 583         /**
 584          * Accessor/mutator for the output type
 585          *
 586          * @param $x New value or null to just get the current one
 587          * @return Integer
 588          */
 589         function OutputType( $x = null ) {
 590                 return wfSetVar( $this->mOutputType, $x );
 591         }
 592
 593         /**
 594          * Get the ParserOutput object
 595          *
 596          * @return ParserOutput object
 597          */
 598         function getOutput() {
 599                 return $this->mOutput;
 600         }
 601
 602         /**
 603          * Get the ParserOptions object
 604          *
 605          * @return ParserOptions object
 606          */
 607         function getOptions() {
 608                 return $this->mOptions;
 609         }
 610
 611         /**
 612          * Accessor/mutator for the ParserOptions object
 613          *
 614          * @param $x New value or null to just get the current one
 615          * @return Current ParserOptions object
 616          */
 617         function Options( $x = null ) {
 618                 return wfSetVar( $this->mOptions, $x );
 619         }
 620
 621         function nextLinkID() {
 622                 return $this->mLinkID++;
 623         }
 624
 625         function setLinkID( $id ) {
 626                 $this->mLinkID = $id;
 627         }
 628
 629         /**
 630          * @return Language
 631          */
 632         function getFunctionLang() {
 633                 global $wgLang, $wgContLang;
 634
 635                 $target = $this->mOptions->getTargetLanguage();
 636                 if ( $target !== null ) {
 637                         return $target;
 638                 } else {
 639                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 640                 }
 641         }
 642
 643         /**
 644          * Get a User object either from $this->mUser, if set, or from the
 645          * ParserOptions object otherwise
 646          *
 647          * @return User object
 648          */
 649         function getUser() {
 650                 if ( !is_null( $this->mUser ) ) {
 651                         return $this->mUser;
 652                 }
 653                 return $this->mOptions->getUser();
 654         }
 655
 656         /**
 657          * Get a preprocessor object
 658          *
 659          * @return Preprocessor instance
 660          */
 661         function getPreprocessor() {
 662                 if ( !isset( $this->mPreprocessor ) ) {
 663                         $class = $this->mPreprocessorClass;
 664                         $this->mPreprocessor = new $class( $this );
 665                 }
 666                 return $this->mPreprocessor;
 667         }
 668
 669         /**
 670          * Replaces all occurrences of HTML-style comments and the given tags
 671          * in the text with a random marker and returns the next text. The output
 672          * parameter $matches will be an associative array filled with data in
 673          * the form:
 674          *   'UNIQ-xxxxx' => array(
 675          *     'element',
 676          *     'tag content',
 677          *     array( 'param' => 'x' ),
 678          *     '<element param="x">tag content</element>' ) )
 679          *
 680          * @param $elements list of element names. Comments are always extracted.
 681          * @param $text Source text string.
 682          * @param $matches Out parameter, Array: extracted tags
 683          * @param $uniq_prefix
 684          * @return String: stripped text
 685          */
 686         public static function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
 687                 static $n = 1;
 688                 $stripped = '';
 689                 $matches = array();
 690
 691                 $taglist = implode( '|', $elements );
 692                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 693
 694                 while ( $text != '' ) {
 695                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 696                         $stripped .= $p[0];
 697                         if ( count( $p ) < 5 ) {
 698                                 break;
 699                         }
 700                         if ( count( $p ) > 5 ) {
 701                                 # comment
 702                                 $element    = $p[4];
 703                                 $attributes = '';
 704                                 $close      = '';
 705                                 $inside     = $p[5];
 706                         } else {
 707                                 # tag
 708                                 $element    = $p[1];
 709                                 $attributes = $p[2];
 710                                 $close      = $p[3];
 711                                 $inside     = $p[4];
 712                         }
 713
 714                         $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
 715                         $stripped .= $marker;
 716
 717                         if ( $close === '/>' ) {
 718                                 # Empty element tag, <tag />
 719                                 $content = null;
 720                                 $text = $inside;
 721                                 $tail = null;
 722                         } else {
 723                                 if ( $element === '!--' ) {
 724                                         $end = '/(-->)/';
 725                                 } else {
 726                                         $end = "/(<\\/$element\\s*>)/i";
 727                                 }
 728                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 729                                 $content = $q[0];
 730                                 if ( count( $q ) < 3 ) {
 731                                         # No end tag -- let it run out to the end of the text.
 732                                         $tail = '';
 733                                         $text = '';
 734                                 } else {
 735                                         $tail = $q[1];
 736                                         $text = $q[2];
 737                                 }
 738                         }
 739
 740                         $matches[$marker] = array( $element,
 741                                 $content,
 742                                 Sanitizer::decodeTagAttributes( $attributes ),
 743                                 "<$element$attributes$close$content$tail" );
 744                 }
 745                 return $stripped;
 746         }
 747
 748         /**
 749          * Get a list of strippable XML-like elements
 750          */
 751         function getStripList() {
 752                 return $this->mStripList;
 753         }
 754
 755         /**
 756          * Add an item to the strip state
 757          * Returns the unique tag which must be inserted into the stripped text
 758          * The tag will be replaced with the original text in unstrip()
 759          *
 760          * @private
 761          */
 762         function insertStripItem( $text ) {
 763                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 764                 $this->mMarkerIndex++;
 765                 $this->mStripState->addGeneral( $rnd, $text );
 766                 return $rnd;
 767         }
 768
 769         /**
 770          * parse the wiki syntax used to render tables
 771          *
 772          * @private
 773          */
 774         function doTableStuff( $text ) {
 775                 wfProfileIn( __METHOD__ );
 776
 777                 $lines = StringUtils::explode( "\n", $text );
 778                 $out = '';
 779                 $output =& $out;
 780
 781                 foreach ( $lines as $outLine ) {
 782                         $line = trim( $outLine );
 783
 784                         # empty line, go to next line,
 785                         # but only append \n if outside of table
 786                         if ( $line === '') {
 787                                 $output .= $outLine . "\n";
 788                                 continue;
 789                         }
 790                         $firstChars = $line[0];
 791                         if ( strlen( $line ) > 1 ) {
 792                                 $firstChars .= in_array( $line[1], array( '}', '+', '-' ) ) ? $line[1] : '';
 793                         }
 794                         $matches = array();
 795
 796                         if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line , $matches ) ) {
 797                                 $tables[] = array();
 798                                 $table =& $this->last( $tables );
 799                                 $table[0] = array(); // first row
 800                                 $currentRow =& $table[0];
 801                                 $table['indent'] = strlen( $matches[1] );
 802
 803                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 804                                 $attributes = Sanitizer::fixTagAttributes( $attributes , 'table' );
 805
 806                                 if ( $attributes !== '' ) {
 807                                         $table['attributes'] = $attributes;
 808                                 }
 809                         } else if ( !isset( $tables[0] ) ) {
 810                                 // we're outside the table
 811
 812                                 $out .= $outLine . "\n";
 813                         } else if ( $firstChars === '|}' ) {
 814                                 // trim the |} code from the line
 815                                 $line = substr ( $line , 2 );
 816
 817                                 // Shorthand for last row
 818                                 $lastRow =& $this->last( $table );
 819
 820                                 // a thead at the end becomes a tfoot, unless there is only one row
 821                                 // Do this before deleting empty last lines to allow headers at the bottom of tables
 822                                 if ( isset( $lastRow['type'] ) && $lastRow['type'] == 'thead' && isset( $table[1] ) ) {
 823                                         $lastRow['type'] = 'tfoot';
 824                                         for ( $i = 0; isset( $lastRow[$i] ); $i++ ) {
 825                                                 $lastRow[$i]['type'] = 'th';
 826                                         }
 827                                 }
 828
 829                                 // Delete empty last lines
 830                                 if ( empty( $lastRow ) ) {
 831                                         $lastRow = NULL;
 832                                 }
 833                                 $o = '';
 834                                 $curtable = array_pop( $tables );
 835
 836                                 #Add a line-ending before the table, but only if there isn't one already
 837                                 if ( substr( $out, -1 ) !== "\n" ) {
 838                                         $o .= "\n";
 839                                 }
 840                                 $o .= $this->generateTableHTML( $curtable ) . $line . "\n";
 841
 842                                 if ( count( $tables ) > 0 ) {
 843                                         $table =& $this->last( $tables );
 844                                         $currentRow =& $this->last( $table );
 845                                         $currentElement =& $this->last( $currentRow );
 846
 847                                         $output =& $currentElement['content'];
 848                                 } else {
 849                                         $output =& $out;
 850                                 }
 851
 852                                 $output .= $o;
 853
 854                         } else if ( $firstChars === '|-' ) {
 855                                 // start a new row element
 856                                 // but only when we haven't started one already
 857                                 if ( count( $currentRow ) != 0 ) {
 858                                         $table[] = array();
 859                                         $currentRow =& $this->last( $table );
 860                                 }
 861                                 // Get the attributes, there's nothing else useful in $line now
 862                                 $line = substr ( $line , 2 );
 863                                 $attributes = $this->mStripState->unstripBoth( $line );
 864                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 865                                 if ( $attributes !== '' ) {
 866                                         $currentRow['attributes'] = $attributes;
 867                                 }
 868
 869                         } else if ( $firstChars  === '|+' ) {
 870                                 // a table caption, but only proceed if there isn't one already
 871                                 if ( !isset ( $table['caption'] ) ) {
 872                                         $line = substr ( $line , 2 );
 873
 874                                         $c = $this->getCellAttr( $line , 'caption' );
 875                                         $table['caption'] = array();
 876                                         $table['caption']['content'] = $c[0];
 877                                         if ( isset( $c[1] ) ) $table['caption']['attributes'] = $c[1];
 878                                         unset( $c );
 879                                         $output =& $table['caption']['content'];
 880                                 }
 881                         } else if ( $firstChars === '|' || $firstChars === '!' || $firstChars === '!+' ) {
 882                                 // Which kind of cells are we dealing with
 883                                 $currentTag = 'td';
 884                                 $line = substr ( $line , 1 );
 885
 886                                 if ( $firstChars === '!'  || $firstChars === '!+' ) {
 887                                         $line = str_replace ( '!!' , '||' , $line );
 888                                         $currentTag = 'th';
 889                                 }
 890
 891                                 // Split up multiple cells on the same line.
 892                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 893                                 $line = ''; // save memory
 894
 895                                 // decide whether thead to tbody
 896                                 if ( !array_key_exists( 'type', $currentRow ) ) {
 897                                         $currentRow['type'] = ( $firstChars === '!' ) ? 'thead' : 'tbody' ;
 898                                 } else if ( $firstChars === '|' ) {
 899                                         $currentRow['type'] = 'tbody';
 900                                 }
 901
 902                                 // Loop through each table cell
 903                                 foreach ( $cells as $cell ) {
 904                                         // a new cell
 905                                         $currentRow[] = array();
 906                                         $currentElement =& $this->last( $currentRow );
 907
 908                                         $currentElement['type'] = $currentTag;
 909
 910                                         $c = $this->getCellAttr( $cell , $currentTag );
 911                                         $currentElement['content'] = $c[0];
 912                                         if ( isset( $c[1] ) ) $currentElement['attributes'] = $c[1];
 913                                         unset( $c );
 914                                 }
 915                                 $output =& $currentElement['content'];
 916
 917                         } else {
 918                                 $output .= "\n$outLine";
 919                         }
 920                 }
 921
 922                 # Remove trailing line-ending (b/c)
 923                 if ( substr( $out, -1 ) === "\n" ) {
 924                         $out = substr( $out, 0, -1 );
 925                 }
 926
 927                 # Close any unclosed tables
 928                 if ( isset( $tables ) && count( $tables ) > 0 ) {
 929                         for ( $i = 0; $i < count( $tables ); $i++ ) {
 930                                 $curtable = array_pop( $tables );
 931                                 $curtable = $this->generateTableHTML( $curtable );
 932                                 #Add a line-ending before the table, but only if there isn't one already
 933                                 if ( substr( $out, -1 ) !== "\n"  && $curtable !== "" ) {
 934                                         $out .= "\n";
 935                                 }
 936                                 $out .= $curtable;
 937                         }
 938                 }
 939
 940                 wfProfileOut( __METHOD__ );
 941
 942                 return $out;
 943         }
 944
 945         /**
 946          * Helper function for doTableStuff() separating the contents of cells from
 947          * attributes. Particularly useful as there's a possible bug and this action
 948          * is repeated twice.
 949          *
 950          * @private
 951          */
 952         function getCellAttr ( $cell, $tagName ) {
 953                 $content = null;
 954                 $attributes = null;
 955
 956                 $cell = trim ( $cell );
 957
 958                 // A cell could contain both parameters and data
 959                 $cellData = explode ( '|' , $cell , 2 );
 960
 961                 // Bug 553: Note that a '|' inside an invalid link should not
 962                 // be mistaken as delimiting cell parameters
 963                 if ( strpos( $cellData[0], '[[' ) !== false ) {
 964                         $content = trim ( $cell );
 965                 }
 966                 else if ( count ( $cellData ) == 1 ) {
 967                         $content = trim ( $cellData[0] );
 968                 }
 969                 else {
 970                         $attributes = $this->mStripState->unstripBoth( $cellData[0] );
 971                         $attributes = Sanitizer::fixTagAttributes( $attributes , $tagName );
 972
 973                         $content = trim ( $cellData[1] );
 974                 }
 975                 return array( $content, $attributes );
 976         }
 977
 978
 979         /**
 980          * Helper function for doTableStuff(). This converts the structured array into html.
 981          *
 982          * @private
 983          */
 984         function generateTableHTML ( &$table ) {
 985                 $return = "";
 986                 $return .= str_repeat( '<dl><dd>' , $table['indent'] );
 987                 $return .= '<table';
 988                 $return .= isset( $table['attributes'] ) ? $table['attributes'] : '';
 989                 $return .= '>';
 990                 unset( $table['attributes'] );
 991
 992                 if ( isset( $table['caption'] ) ) {
 993                         $return .= "\n<caption";
 994                         $return .= isset( $table['caption']['attributes'] ) ? $table['caption']['attributes'] : '';
 995                         $return .= '>';
 996                         $return .= $table['caption']['content'];
 997                         $return .= "\n</caption>";
 998                 }
 999                 $lastSection = '';
1000                 $empty = true;
1001                 $simple = true;
1002
1003                 // If we only have tbodies, mark table as simple
1004                 for ( $i = 0; isset( $table[$i] ); $i++ ) {
1005                         if ( !count( $table[$i] ) ) continue;
1006                         if ( !isset( $table[$i]['type'] ) ) $table[$i]['type'] = 'tbody';
1007                         if ( !$lastSection ) {
1008                                 $lastSection = $table[$i]['type'];
1009                         } else if ( $lastSection != $table[$i]['type'] ) {
1010                                 $simple = false;
1011                                 break;
1012                         }
1013                 }
1014                 $lastSection = '';
1015                 for ( $i = 0; isset( $table[$i] ); $i++ ) {
1016                         // Check for empty tables
1017                         if ( count( $table[$i] ) ) {
1018                                 $empty = false;
1019                         } else {
1020                                 continue;
1021                         }
1022                         if ( $table[$i]['type'] != $lastSection && !$simple ) {
1023                                 $return .= "\n<" . $table[$i]['type'] . '>';
1024                         }
1025
1026                         $return .= "\n<tr";
1027                         $return .= isset( $table[$i]['attributes'] ) ? $table[$i]['attributes'] : '';
1028                         $return .= '>';
1029                         for ( $j = 0; isset( $table[$i][$j] ); $j++ ) {
1030                                 if ( !isset( $table[$i][$j]['type'] ) ) $table[$i][$j]['type'] = 'td';
1031                                 $return .= "\n<" . $table[$i][$j]['type'];
1032                                 $return .= isset( $table[$i][$j]['attributes'] ) ? $table[$i][$j]['attributes'] : '';
1033                                 $return .= '>';
1034
1035                                 $return .= $table[$i][$j]['content'];
1036                                 if ( $table[$i][$j]['content'] != '' )
1037                                         $return .= "\n";
1038
1039                                 $return .= '</' . $table[$i][$j]['type'] . '>';
1040                                 unset( $table[$i][$j] );
1041                         }
1042                         $return .= "\n</tr>";
1043
1044                         if ( ( !isset( $table[$i + 1] ) && !$simple ) || ( isset( $table[$i + 1] ) && isset( $table[$i + 1]['type'] ) && $table[$i]['type'] != $table[$i + 1]['type'] ) ) {
1045                                 $return .= '</' . $table[$i]['type'] . '>';
1046                         }
1047                         $lastSection = $table[$i]['type'];
1048                         unset( $table[$i] );
1049                 }
1050                 if ( $empty ) {
1051                         if ( isset( $table['caption'] ) ) {
1052                                 $return .= "\n<tr><td></td></tr>";
1053                         } else {
1054                                 return '';
1055                         }
1056                 }
1057                 $return .= "\n</table>";
1058                 $return .= str_repeat( '</dd></dl>' , $table['indent'] );
1059
1060                 return $return;
1061         }
1062
1063         /**
1064          * like end() but only works on the numeric array index and php's internal pointers
1065          * returns a reference to the last element of an array much like "\$arr[-1]" in perl
1066          * ignores associative elements and will create a 0 key will a NULL value if there were
1067          * no numric elements and an array itself if not previously defined.
1068          *
1069          * @private
1070          */
1071         function &last ( &$arr ) {
1072                 for ( $i = count( $arr ); ( !isset( $arr[$i] ) && $i > 0 ); $i-- ) {  }
1073                 return $arr[$i];
1074         }
1075
1076         /**
1077          * Helper function for parse() that transforms wiki markup into
1078          * HTML. Only called for $mOutputType == self::OT_HTML.
1079          *
1080          * @private
1081          */
1082         function internalParse( $text, $isMain = true, $frame=false ) {
1083                 wfProfileIn( __METHOD__ );
1084
1085                 $origText = $text;
1086
1087                 # Hook to suspend the parser in this state
1088                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1089                         wfProfileOut( __METHOD__ );
1090                         return $text ;
1091                 }
1092
1093                 # if $frame is provided, then use $frame for replacing any variables
1094                 if ( $frame ) {
1095                         # use frame depth to infer how include/noinclude tags should be handled
1096                         # depth=0 means this is the top-level document; otherwise it's an included document
1097                         if ( !$frame->depth ) {
1098                                 $flag = 0;
1099                         } else {
1100                                 $flag = Parser::PTD_FOR_INCLUSION;
1101                         }
1102                         $dom = $this->preprocessToDom( $text, $flag );
1103                         $text = $frame->expand( $dom );
1104                 } else {
1105                         # if $frame is not provided, then use old-style replaceVariables
1106                         $text = $this->replaceVariables( $text );
1107                 }
1108
1109                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
1110                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1111
1112                 # Tables need to come after variable replacement for things to work
1113                 # properly; putting them before other transformations should keep
1114                 # exciting things like link expansions from showing up in surprising
1115                 # places.
1116                 $text = $this->doTableStuff( $text );
1117
1118                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1119
1120                 $text = $this->doDoubleUnderscore( $text );
1121
1122                 $text = $this->doHeadings( $text );
1123                 if ( $this->mOptions->getUseDynamicDates() ) {
1124                         $df = DateFormatter::getInstance();
1125                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1126                 }
1127                 $text = $this->replaceInternalLinks( $text );
1128                 $text = $this->doAllQuotes( $text );
1129                 $text = $this->replaceExternalLinks( $text );
1130
1131                 # replaceInternalLinks may sometimes leave behind
1132                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1133                 $text = str_replace( $this->mUniqPrefix.'NOPARSE', '', $text );
1134
1135                 $text = $this->doMagicLinks( $text );
1136                 $text = $this->formatHeadings( $text, $origText, $isMain );
1137
1138                 wfProfileOut( __METHOD__ );
1139                 return $text;
1140         }
1141
1142         /**
1143          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1144          * magic external links.
1145          *
1146          * DML
1147          * @private
1148          */
1149         function doMagicLinks( $text ) {
1150                 wfProfileIn( __METHOD__ );
1151                 $prots = $this->mUrlProtocols;
1152                 $urlChar = self::EXT_LINK_URL_CLASS;
1153                 $text = preg_replace_callback(
1154                         '!(?:                           # Start cases
1155                                 (<a[ \t\r\n>].*?</a>) |     # m[1]: Skip link text
1156                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1157                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
1158                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1159                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1160                                         (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1161                                         (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1162                                         [0-9Xx]                 # check digit
1163                                         \b)
1164                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1165                 wfProfileOut( __METHOD__ );
1166                 return $text;
1167         }
1168
1169         function magicLinkCallback( $m ) {
1170                 if ( isset( $m[1] ) && $m[1] !== '' ) {
1171                         # Skip anchor
1172                         return $m[0];
1173                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1174                         # Skip HTML element
1175                         return $m[0];
1176                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1177                         # Free external link
1178                         return $this->makeFreeExternalLink( $m[0] );
1179                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1180                         # RFC or PMID
1181                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1182                                 $keyword = 'RFC';
1183                                 $urlmsg = 'rfcurl';
1184                                 $CssClass = 'mw-magiclink-rfc';
1185                                 $id = $m[4];
1186                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1187                                 $keyword = 'PMID';
1188                                 $urlmsg = 'pubmedurl';
1189                                 $CssClass = 'mw-magiclink-pmid';
1190                                 $id = $m[4];
1191                         } else {
1192                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1193                                         substr( $m[0], 0, 20 ) . '"' );
1194                         }
1195                         $url = wfMsgForContent( $urlmsg, $id );
1196                         return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $CssClass );
1197                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1198                         # ISBN
1199                         $isbn = $m[5];
1200                         $num = strtr( $isbn, array(
1201                                 '-' => '',
1202                                 ' ' => '',
1203                                 'x' => 'X',
1204                         ));
1205                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1206                         return'<a href="' .
1207                                 $titleObj->escapeLocalUrl() .
1208                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1209                 } else {
1210                         return $m[0];
1211                 }
1212         }
1213
1214         /**
1215          * Make a free external link, given a user-supplied URL
1216          * @return HTML
1217          * @private
1218          */
1219         function makeFreeExternalLink( $url ) {
1220                 global $wgContLang;
1221                 wfProfileIn( __METHOD__ );
1222
1223                 $trail = '';
1224
1225                 # The characters '<' and '>' (which were escaped by
1226                 # removeHTMLtags()) should not be included in
1227                 # URLs, per RFC 2396.
1228                 $m2 = array();
1229                 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1230                         $trail = substr( $url, $m2[0][1] ) . $trail;
1231                         $url = substr( $url, 0, $m2[0][1] );
1232                 }
1233
1234                 # Move trailing punctuation to $trail
1235                 $sep = ',;\.:!?';
1236                 # If there is no left bracket, then consider right brackets fair game too
1237                 if ( strpos( $url, '(' ) === false ) {
1238                         $sep .= ')';
1239                 }
1240
1241                 $numSepChars = strspn( strrev( $url ), $sep );
1242                 if ( $numSepChars ) {
1243                         $trail = substr( $url, -$numSepChars ) . $trail;
1244                         $url = substr( $url, 0, -$numSepChars );
1245                 }
1246
1247                 $url = Sanitizer::cleanUrl( $url );
1248
1249                 # Is this an external image?
1250                 $text = $this->maybeMakeExternalImage( $url );
1251                 if ( $text === false ) {
1252                         # Not an image, make a link
1253                         $text = Linker::makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1254                                 $this->getExternalLinkAttribs( $url ) );
1255                         # Register it in the output object...
1256                         # Replace unnecessary URL escape codes with their equivalent characters
1257                         $pasteurized = self::replaceUnusualEscapes( $url );
1258                         $this->mOutput->addExternalLink( $pasteurized );
1259                 }
1260                 wfProfileOut( __METHOD__ );
1261                 return $text . $trail;
1262         }
1263
1264
1265         /**
1266          * Parse headers and return html
1267          *
1268          * @private
1269          */
1270         function doHeadings( $text ) {
1271                 wfProfileIn( __METHOD__ );
1272                 for ( $i = 6; $i >= 1; --$i ) {
1273                         $h = str_repeat( '=', $i );
1274                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1275                           "<h$i>\\1</h$i>", $text );
1276                 }
1277                 wfProfileOut( __METHOD__ );
1278                 return $text;
1279         }
1280
1281         /**
1282          * Replace single quotes with HTML markup
1283          * @private
1284          * @return string the altered text
1285          */
1286         function doAllQuotes( $text ) {
1287                 wfProfileIn( __METHOD__ );
1288                 $outtext = '';
1289                 $lines = StringUtils::explode( "\n", $text );
1290                 foreach ( $lines as $line ) {
1291                         $outtext .= $this->doQuotes( $line ) . "\n";
1292                 }
1293                 $outtext = substr( $outtext, 0,-1 );
1294                 wfProfileOut( __METHOD__ );
1295                 return $outtext;
1296         }
1297
1298         /**
1299          * Helper function for doAllQuotes()
1300          */
1301         public function doQuotes( $text ) {
1302                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1303                 if ( count( $arr ) == 1 ) {
1304                         return $text;
1305                 } else {
1306                         # First, do some preliminary work. This may shift some apostrophes from
1307                         # being mark-up to being text. It also counts the number of occurrences
1308                         # of bold and italics mark-ups.
1309                         $numbold = 0;
1310                         $numitalics = 0;
1311                         for ( $i = 0; $i < count( $arr ); $i++ ) {
1312                                 if ( ( $i % 2 ) == 1 ) {
1313                                         # If there are ever four apostrophes, assume the first is supposed to
1314                                         # be text, and the remaining three constitute mark-up for bold text.
1315                                         if ( strlen( $arr[$i] ) == 4 ) {
1316                                                 $arr[$i-1] .= "'";
1317                                                 $arr[$i] = "'''";
1318                                         } elseif ( strlen( $arr[$i] ) > 5 ) {
1319                                                 # If there are more than 5 apostrophes in a row, assume they're all
1320                                                 # text except for the last 5.
1321                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1322                                                 $arr[$i] = "'''''";
1323                                         }
1324                                         # Count the number of occurrences of bold and italics mark-ups.
1325                                         # We are not counting sequences of five apostrophes.
1326                                         if ( strlen( $arr[$i] ) == 2 ) {
1327                                                 $numitalics++;
1328                                         } elseif ( strlen( $arr[$i] ) == 3 ) {
1329                                                 $numbold++;
1330                                         } elseif ( strlen( $arr[$i] ) == 5 ) {
1331                                                 $numitalics++;
1332                                                 $numbold++;
1333                                         }
1334                                 }
1335                         }
1336
1337                         # If there is an odd number of both bold and italics, it is likely
1338                         # that one of the bold ones was meant to be an apostrophe followed
1339                         # by italics. Which one we cannot know for certain, but it is more
1340                         # likely to be one that has a single-letter word before it.
1341                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1342                                 $i = 0;
1343                                 $firstsingleletterword = -1;
1344                                 $firstmultiletterword = -1;
1345                                 $firstspace = -1;
1346                                 foreach ( $arr as $r ) {
1347                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1348                                                 $x1 = substr( $arr[$i-1], -1 );
1349                                                 $x2 = substr( $arr[$i-1], -2, 1 );
1350                                                 if ( $x1 === ' ' ) {
1351                                                         if ( $firstspace == -1 ) {
1352                                                                 $firstspace = $i;
1353                                                         }
1354                                                 } elseif ( $x2 === ' ') {
1355                                                         if ( $firstsingleletterword == -1 ) {
1356                                                                 $firstsingleletterword = $i;
1357                                                         }
1358                                                 } else {
1359                                                         if ( $firstmultiletterword == -1 ) {
1360                                                                 $firstmultiletterword = $i;
1361                                                         }
1362                                                 }
1363                                         }
1364                                         $i++;
1365                                 }
1366
1367                                 # If there is a single-letter word, use it!
1368                                 if ( $firstsingleletterword > -1 ) {
1369                                         $arr[$firstsingleletterword] = "''";
1370                                         $arr[$firstsingleletterword-1] .= "'";
1371                                 } elseif ( $firstmultiletterword > -1 ) {
1372                                         # If not, but there's a multi-letter word, use that one.
1373                                         $arr[$firstmultiletterword] = "''";
1374                                         $arr[$firstmultiletterword-1] .= "'";
1375                                 } elseif ( $firstspace > -1 ) {
1376                                         # ... otherwise use the first one that has neither.
1377                                         # (notice that it is possible for all three to be -1 if, for example,
1378                                         # there is only one pentuple-apostrophe in the line)
1379                                         $arr[$firstspace] = "''";
1380                                         $arr[$firstspace-1] .= "'";
1381                                 }
1382                         }
1383
1384                         # Now let's actually convert our apostrophic mush to HTML!
1385                         $output = '';
1386                         $buffer = '';
1387                         $state = '';
1388                         $i = 0;
1389                         foreach ( $arr as $r ) {
1390                                 if ( ( $i % 2 ) == 0 ) {
1391                                         if ( $state === 'both' ) {
1392                                                 $buffer .= $r;
1393                                         } else {
1394                                                 $output .= $r;
1395                                         }
1396                                 } else {
1397                                         if ( strlen( $r ) == 2 ) {
1398                                                 if ( $state === 'i' ) {
1399                                                         $output .= '</i>'; $state = '';
1400                                                 } elseif ( $state === 'bi' ) {
1401                                                         $output .= '</i>'; $state = 'b';
1402                                                 } elseif ( $state === 'ib' ) {
1403                                                         $output .= '</b></i><b>'; $state = 'b';
1404                                                 } elseif ( $state === 'both' ) {
1405                                                         $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1406                                                 } else { # $state can be 'b' or ''
1407                                                         $output .= '<i>'; $state .= 'i';
1408                                                 }
1409                                         } elseif ( strlen( $r ) == 3 ) {
1410                                                 if ( $state === 'b' ) {
1411                                                         $output .= '</b>'; $state = '';
1412                                                 } elseif ( $state === 'bi' ) {
1413                                                         $output .= '</i></b><i>'; $state = 'i';
1414                                                 } elseif ( $state === 'ib' ) {
1415                                                         $output .= '</b>'; $state = 'i';
1416                                                 } elseif ( $state === 'both' ) {
1417                                                         $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1418                                                 } else { # $state can be 'i' or ''
1419                                                         $output .= '<b>'; $state .= 'b';
1420                                                 }
1421                                         } elseif ( strlen( $r ) == 5 ) {
1422                                                 if ( $state === 'b' ) {
1423                                                         $output .= '</b><i>'; $state = 'i';
1424                                                 } elseif ( $state === 'i' ) {
1425                                                         $output .= '</i><b>'; $state = 'b';
1426                                                 } elseif ( $state === 'bi' ) {
1427                                                         $output .= '</i></b>'; $state = '';
1428                                                 } elseif ( $state === 'ib' ) {
1429                                                         $output .= '</b></i>'; $state = '';
1430                                                 } elseif ( $state === 'both' ) {
1431                                                         $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1432                                                 } else { # ($state == '')
1433                                                         $buffer = ''; $state = 'both';
1434                                                 }
1435                                         }
1436                                 }
1437                                 $i++;
1438                         }
1439                         # Now close all remaining tags.  Notice that the order is important.
1440                         if ( $state === 'b' || $state === 'ib' ) {
1441                                 $output .= '</b>';
1442                         }
1443                         if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1444                                 $output .= '</i>';
1445                         }
1446                         if ( $state === 'bi' ) {
1447                                 $output .= '</b>';
1448                         }
1449                         # There might be lonely ''''', so make sure we have a buffer
1450                         if ( $state === 'both' && $buffer ) {
1451                                 $output .= '<b><i>'.$buffer.'</i></b>';
1452                         }
1453                         return $output;
1454                 }
1455         }
1456
1457         /**
1458          * Replace external links (REL)
1459          *
1460          * Note: this is all very hackish and the order of execution matters a lot.
1461          * Make sure to run maintenance/parserTests.php if you change this code.
1462          *
1463          * @private
1464          */
1465         function replaceExternalLinks( $text ) {
1466                 global $wgContLang;
1467                 wfProfileIn( __METHOD__ );
1468
1469                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1470                 $s = array_shift( $bits );
1471
1472                 $i = 0;
1473                 while ( $i<count( $bits ) ) {
1474                         $url = $bits[$i++];
1475                         $protocol = $bits[$i++];
1476                         $text = $bits[$i++];
1477                         $trail = $bits[$i++];
1478
1479                         # The characters '<' and '>' (which were escaped by
1480                         # removeHTMLtags()) should not be included in
1481                         # URLs, per RFC 2396.
1482                         $m2 = array();
1483                         if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1484                                 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1485                                 $url = substr( $url, 0, $m2[0][1] );
1486                         }
1487
1488                         # If the link text is an image URL, replace it with an <img> tag
1489                         # This happened by accident in the original parser, but some people used it extensively
1490                         $img = $this->maybeMakeExternalImage( $text );
1491                         if ( $img !== false ) {
1492                                 $text = $img;
1493                         }
1494
1495                         $dtrail = '';
1496
1497                         # Set linktype for CSS - if URL==text, link is essentially free
1498                         $linktype = ( $text === $url ) ? 'free' : 'text';
1499
1500                         # No link text, e.g. [http://domain.tld/some.link]
1501                         if ( $text == '' ) {
1502                                 # Autonumber if allowed. See bug #5918
1503                                 if ( strpos( wfUrlProtocols(), substr( $protocol, 0, strpos( $protocol, ':' ) ) ) !== false ) {
1504                                         $langObj = $this->getFunctionLang();
1505                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1506                                         $linktype = 'autonumber';
1507                                 } else {
1508                                         # Otherwise just use the URL
1509                                         $text = htmlspecialchars( $url );
1510                                         $linktype = 'free';
1511                                 }
1512                         } else {
1513                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1514                                 # Check for trail
1515                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1516                         }
1517
1518                         $text = $wgContLang->markNoConversion( $text );
1519
1520                         $url = Sanitizer::cleanUrl( $url );
1521
1522                         # Use the encoded URL
1523                         # This means that users can paste URLs directly into the text
1524                         # Funny characters like ö aren't valid in URLs anyway
1525                         # This was changed in August 2004
1526                         $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1527                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1528
1529                         # Register link in the output object.
1530                         # Replace unnecessary URL escape codes with the referenced character
1531                         # This prevents spammers from hiding links from the filters
1532                         $pasteurized = self::replaceUnusualEscapes( $url );
1533                         $this->mOutput->addExternalLink( $pasteurized );
1534                 }
1535
1536                 wfProfileOut( __METHOD__ );
1537                 return $s;
1538         }
1539
1540         /**
1541          * Get an associative array of additional HTML attributes appropriate for a
1542          * particular external link.  This currently may include rel => nofollow
1543          * (depending on configuration, namespace, and the URL's domain) and/or a
1544          * target attribute (depending on configuration).
1545          *
1546          * @param $url String: optional URL, to extract the domain from for rel =>
1547          *   nofollow if appropriate
1548          * @return Array: associative array of HTML attributes
1549          */
1550         function getExternalLinkAttribs( $url = false ) {
1551                 $attribs = array();
1552                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1553                 $ns = $this->mTitle->getNamespace();
1554                 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) ) {
1555                         $attribs['rel'] = 'nofollow';
1556
1557                         global $wgNoFollowDomainExceptions;
1558                         if ( $wgNoFollowDomainExceptions ) {
1559                                 $bits = wfParseUrl( $url );
1560                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1561                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1562                                                 if ( substr( $bits['host'], -strlen( $domain ) ) == $domain ) {
1563                                                         unset( $attribs['rel'] );
1564                                                         break;
1565                                                 }
1566                                         }
1567                                 }
1568                         }
1569                 }
1570                 if ( $this->mOptions->getExternalLinkTarget() ) {
1571                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1572                 }
1573                 return $attribs;
1574         }
1575
1576         /**
1577          * Replace unusual URL escape codes with their equivalent characters
1578          *
1579          * @param $url String
1580          * @return String
1581          *
1582          * @todo  This can merge genuinely required bits in the path or query string,
1583          *        breaking legit URLs. A proper fix would treat the various parts of
1584          *        the URL differently; as a workaround, just use the output for
1585          *        statistical records, not for actual linking/output.
1586          */
1587         static function replaceUnusualEscapes( $url ) {
1588                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1589                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1590         }
1591
1592         /**
1593          * Callback function used in replaceUnusualEscapes().
1594          * Replaces unusual URL escape codes with their equivalent character
1595          */
1596         private static function replaceUnusualEscapesCallback( $matches ) {
1597                 $char = urldecode( $matches[0] );
1598                 $ord = ord( $char );
1599                 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1600                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1601                         # No, shouldn't be escaped
1602                         return $char;
1603                 } else {
1604                         # Yes, leave it escaped
1605                         return $matches[0];
1606                 }
1607         }
1608
1609         /**
1610          * make an image if it's allowed, either through the global
1611          * option, through the exception, or through the on-wiki whitelist
1612          * @private
1613          */
1614         function maybeMakeExternalImage( $url ) {
1615                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1616                 $imagesexception = !empty( $imagesfrom );
1617                 $text = false;
1618                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1619                 if ( $imagesexception && is_array( $imagesfrom ) ) {
1620                         $imagematch = false;
1621                         foreach ( $imagesfrom as $match ) {
1622                                 if ( strpos( $url, $match ) === 0 ) {
1623                                         $imagematch = true;
1624                                         break;
1625                                 }
1626                         }
1627                 } elseif ( $imagesexception ) {
1628                         $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1629                 } else {
1630                         $imagematch = false;
1631                 }
1632                 if ( $this->mOptions->getAllowExternalImages()
1633                          || ( $imagesexception && $imagematch ) ) {
1634                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1635                                 # Image found
1636                                 $text = Linker::makeExternalImage( $url );
1637                         }
1638                 }
1639                 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1640                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1641                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1642                         foreach ( $whitelist as $entry ) {
1643                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1644                                 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1645                                         continue;
1646                                 }
1647                                 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1648                                         # Image matches a whitelist entry
1649                                         $text = Linker::makeExternalImage( $url );
1650                                         break;
1651                                 }
1652                         }
1653                 }
1654                 return $text;
1655         }
1656
1657         /**
1658          * Process [[ ]] wikilinks
1659          * @return String: processed text
1660          *
1661          * @private
1662          */
1663         function replaceInternalLinks( $s ) {
1664                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1665                 return $s;
1666         }
1667
1668         /**
1669          * Process [[ ]] wikilinks (RIL)
1670          * @return LinkHolderArray
1671          *
1672          * @private
1673          */
1674         function replaceInternalLinks2( &$s ) {
1675                 global $wgContLang;
1676
1677                 wfProfileIn( __METHOD__ );
1678
1679                 wfProfileIn( __METHOD__.'-setup' );
1680                 static $tc = FALSE, $e1, $e1_img;
1681                 # the % is needed to support urlencoded titles as well
1682                 if ( !$tc ) {
1683                         $tc = Title::legalChars() . '#%';
1684                         # Match a link having the form [[namespace:link|alternate]]trail
1685                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1686                         # Match cases where there is no "]]", which might still be images
1687                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1688                 }
1689
1690                 $holders = new LinkHolderArray( $this );
1691
1692                 # split the entire text string on occurences of [[
1693                 $a = StringUtils::explode( '[[', ' ' . $s );
1694                 # get the first element (all text up to first [[), and remove the space we added
1695                 $s = $a->current();
1696                 $a->next();
1697                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1698                 $s = substr( $s, 1 );
1699
1700                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1701                 $e2 = null;
1702                 if ( $useLinkPrefixExtension ) {
1703                         # Match the end of a line for a word that's not followed by whitespace,
1704                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1705                         $e2 = wfMsgForContent( 'linkprefix' );
1706                 }
1707
1708                 if ( is_null( $this->mTitle ) ) {
1709                         wfProfileOut( __METHOD__.'-setup' );
1710                         wfProfileOut( __METHOD__ );
1711                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1712                 }
1713                 $nottalk = !$this->mTitle->isTalkPage();
1714
1715                 if ( $useLinkPrefixExtension ) {
1716                         $m = array();
1717                         if ( preg_match( $e2, $s, $m ) ) {
1718                                 $first_prefix = $m[2];
1719                         } else {
1720                                 $first_prefix = false;
1721                         }
1722                 } else {
1723                         $prefix = '';
1724                 }
1725
1726                 if ( $wgContLang->hasVariants() ) {
1727                         $selflink = $wgContLang->autoConvertToAllVariants( $this->mTitle->getPrefixedText() );
1728                 } else {
1729                         $selflink = array( $this->mTitle->getPrefixedText() );
1730                 }
1731                 $useSubpages = $this->areSubpagesAllowed();
1732                 wfProfileOut( __METHOD__.'-setup' );
1733
1734                 # Loop for each link
1735                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1736                         # Check for excessive memory usage
1737                         if ( $holders->isBig() ) {
1738                                 # Too big
1739                                 # Do the existence check, replace the link holders and clear the array
1740                                 $holders->replace( $s );
1741                                 $holders->clear();
1742                         }
1743
1744                         if ( $useLinkPrefixExtension ) {
1745                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1746                                 if ( preg_match( $e2, $s, $m ) ) {
1747                                         $prefix = $m[2];
1748                                         $s = $m[1];
1749                                 } else {
1750                                         $prefix='';
1751                                 }
1752                                 # first link
1753                                 if ( $first_prefix ) {
1754                                         $prefix = $first_prefix;
1755                                         $first_prefix = false;
1756                                 }
1757                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1758                         }
1759
1760                         $might_be_img = false;
1761
1762                         wfProfileIn( __METHOD__."-e1" );
1763                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1764                                 $text = $m[2];
1765                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1766                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1767                                 # the real problem is with the $e1 regex
1768                                 # See bug 1300.
1769                                 #
1770                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1771                                 # and no image is in sight. See bug 2095.
1772                                 #
1773                                 if ( $text !== '' &&
1774                                         substr( $m[3], 0, 1 ) === ']' &&
1775                                         strpos( $text, '[' ) !== false
1776                                 )
1777                                 {
1778                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1779                                         $m[3] = substr( $m[3], 1 );
1780                                 }
1781                                 # fix up urlencoded title texts
1782                                 if ( strpos( $m[1], '%' ) !== false ) {
1783                                         # Should anchors '#' also be rejected?
1784                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), rawurldecode( $m[1] ) );
1785                                 }
1786                                 $trail = $m[3];
1787                         } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1788                                 $might_be_img = true;
1789                                 $text = $m[2];
1790                                 if ( strpos( $m[1], '%' ) !== false ) {
1791                                         $m[1] = rawurldecode( $m[1] );
1792                                 }
1793                                 $trail = "";
1794                         } else { # Invalid form; output directly
1795                                 $s .= $prefix . '[[' . $line ;
1796                                 wfProfileOut( __METHOD__."-e1" );
1797                                 continue;
1798                         }
1799                         wfProfileOut( __METHOD__."-e1" );
1800                         wfProfileIn( __METHOD__."-misc" );
1801
1802                         # Don't allow internal links to pages containing
1803                         # PROTO: where PROTO is a valid URL protocol; these
1804                         # should be external links.
1805                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1806                                 $s .= $prefix . '[[' . $line ;
1807                                 wfProfileOut( __METHOD__."-misc" );
1808                                 continue;
1809                         }
1810
1811                         # Make subpage if necessary
1812                         if ( $useSubpages ) {
1813                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1814                         } else {
1815                                 $link = $m[1];
1816                         }
1817
1818                         $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1819                         if ( !$noforce ) {
1820                                 # Strip off leading ':'
1821                                 $link = substr( $link, 1 );
1822                         }
1823
1824                         wfProfileOut( __METHOD__."-misc" );
1825                         wfProfileIn( __METHOD__."-title" );
1826                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1827                         if ( $nt === null ) {
1828                                 $s .= $prefix . '[[' . $line;
1829                                 wfProfileOut( __METHOD__."-title" );
1830                                 continue;
1831                         }
1832
1833                         $ns = $nt->getNamespace();
1834                         $iw = $nt->getInterWiki();
1835                         wfProfileOut( __METHOD__."-title" );
1836
1837                         if ( $might_be_img ) { # if this is actually an invalid link
1838                                 wfProfileIn( __METHOD__."-might_be_img" );
1839                                 if ( $ns == NS_FILE && $noforce ) { # but might be an image
1840                                         $found = false;
1841                                         while ( true ) {
1842                                                 # look at the next 'line' to see if we can close it there
1843                                                 $a->next();
1844                                                 $next_line = $a->current();
1845                                                 if ( $next_line === false || $next_line === null ) {
1846                                                         break;
1847                                                 }
1848                                                 $m = explode( ']]', $next_line, 3 );
1849                                                 if ( count( $m ) == 3 ) {
1850                                                         # the first ]] closes the inner link, the second the image
1851                                                         $found = true;
1852                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1853                                                         $trail = $m[2];
1854                                                         break;
1855                                                 } elseif ( count( $m ) == 2 ) {
1856                                                         # if there's exactly one ]] that's fine, we'll keep looking
1857                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1858                                                 } else {
1859                                                         # if $next_line is invalid too, we need look no further
1860                                                         $text .= '[[' . $next_line;
1861                                                         break;
1862                                                 }
1863                                         }
1864                                         if ( !$found ) {
1865                                                 # we couldn't find the end of this imageLink, so output it raw
1866                                                 # but don't ignore what might be perfectly normal links in the text we've examined
1867                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1868                                                 $s .= "{$prefix}[[$link|$text";
1869                                                 # note: no $trail, because without an end, there *is* no trail
1870                                                 wfProfileOut( __METHOD__."-might_be_img" );
1871                                                 continue;
1872                                         }
1873                                 } else { # it's not an image, so output it raw
1874                                         $s .= "{$prefix}[[$link|$text";
1875                                         # note: no $trail, because without an end, there *is* no trail
1876                                         wfProfileOut( __METHOD__."-might_be_img" );
1877                                         continue;
1878                                 }
1879                                 wfProfileOut( __METHOD__."-might_be_img" );
1880                         }
1881
1882                         $wasblank = ( $text  == '' );
1883                         if ( $wasblank ) {
1884                                 $text = $link;
1885                         } else {
1886                                 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
1887                                 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
1888                                 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
1889                                 #    -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
1890                                 $text = $this->doQuotes( $text );
1891                         }
1892
1893                         # Link not escaped by : , create the various objects
1894                         if ( $noforce ) {
1895
1896                                 # Interwikis
1897                                 wfProfileIn( __METHOD__."-interwiki" );
1898                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1899                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1900                                         $s = rtrim( $s . $prefix );
1901                                         $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
1902                                         wfProfileOut( __METHOD__."-interwiki" );
1903                                         continue;
1904                                 }
1905                                 wfProfileOut( __METHOD__."-interwiki" );
1906
1907                                 if ( $ns == NS_FILE ) {
1908                                         wfProfileIn( __METHOD__."-image" );
1909                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1910                                                 if ( $wasblank ) {
1911                                                         # if no parameters were passed, $text
1912                                                         # becomes something like "File:Foo.png",
1913                                                         # which we don't want to pass on to the
1914                                                         # image generator
1915                                                         $text = '';
1916                                                 } else {
1917                                                         # recursively parse links inside the image caption
1918                                                         # actually, this will parse them in any other parameters, too,
1919                                                         # but it might be hard to fix that, and it doesn't matter ATM
1920                                                         $text = $this->replaceExternalLinks( $text );
1921                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1922                                                 }
1923                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1924                                                 $s .= $prefix . $this->armorLinks(
1925                                                         $this->makeImage( $nt, $text, $holders ) ) . $trail;
1926                                         } else {
1927                                                 $s .= $prefix . $trail;
1928                                         }
1929                                         wfProfileOut( __METHOD__."-image" );
1930                                         continue;
1931                                 }
1932
1933                                 if ( $ns == NS_CATEGORY ) {
1934                                         wfProfileIn( __METHOD__."-category" );
1935                                         $s = rtrim( $s . "\n" ); # bug 87
1936
1937                                         if ( $wasblank ) {
1938                                                 $sortkey = $this->getDefaultSort();
1939                                         } else {
1940                                                 $sortkey = $text;
1941                                         }
1942                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1943                                         $sortkey = str_replace( "\n", '', $sortkey );
1944                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1945                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1946
1947                                         /**
1948                                          * Strip the whitespace Category links produce, see bug 87
1949                                          * @todo We might want to use trim($tmp, "\n") here.
1950                                          */
1951                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
1952
1953                                         wfProfileOut( __METHOD__."-category" );
1954                                         continue;
1955                                 }
1956                         }
1957
1958                         # Self-link checking
1959                         if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1960                                 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1961                                         $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
1962                                         continue;
1963                                 }
1964                         }
1965
1966                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1967                         # FIXME: Should do batch file existence checks, see comment below
1968                         if ( $ns == NS_MEDIA ) {
1969                                 wfProfileIn( __METHOD__."-media" );
1970                                 # Give extensions a chance to select the file revision for us
1971                                 $time = $sha1 = $descQuery = false;
1972                                 wfRunHooks( 'BeforeParserFetchFileAndTitle',
1973                                         array( $this, $nt, &$time, &$sha1, &$descQuery ) );
1974                                 # Fetch and register the file (file title may be different via hooks)
1975                                 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $time, $sha1 );
1976                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1977                                 $s .= $prefix . $this->armorLinks(
1978                                         Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
1979                                 wfProfileOut( __METHOD__."-media" );
1980                                 continue;
1981                         }
1982
1983                         wfProfileIn( __METHOD__."-always_known" );
1984                         # Some titles, such as valid special pages or files in foreign repos, should
1985                         # be shown as bluelinks even though they're not included in the page table
1986                         #
1987                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1988                         # batch file existence checks for NS_FILE and NS_MEDIA
1989                         if ( $iw == '' && $nt->isAlwaysKnown() ) {
1990                                 $this->mOutput->addLink( $nt );
1991                                 $s .= $this->makeKnownLinkHolder( $nt, $text, array(), $trail, $prefix );
1992                         } else {
1993                                 # Links will be added to the output link list after checking
1994                                 $s .= $holders->makeHolder( $nt, $text, array(), $trail, $prefix );
1995                         }
1996                         wfProfileOut( __METHOD__."-always_known" );
1997                 }
1998                 wfProfileOut( __METHOD__ );
1999                 return $holders;
2000         }
2001
2002         /**
2003          * Render a forced-blue link inline; protect against double expansion of
2004          * URLs if we're in a mode that prepends full URL prefixes to internal links.
2005          * Since this little disaster has to split off the trail text to avoid
2006          * breaking URLs in the following text without breaking trails on the
2007          * wiki links, it's been made into a horrible function.
2008          *
2009          * @param $nt Title
2010          * @param $text String
2011          * @param $query Array or String
2012          * @param $trail String
2013          * @param $prefix String
2014          * @return String: HTML-wikitext mix oh yuck
2015          */
2016         function makeKnownLinkHolder( $nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
2017                 list( $inside, $trail ) = Linker::splitTrail( $trail );
2018
2019                 if ( is_string( $query ) ) {
2020                         $query = wfCgiToArray( $query );
2021                 }
2022                 if ( $text == '' ) {
2023                         $text = htmlspecialchars( $nt->getPrefixedText() );
2024                 }
2025
2026                 $link = Linker::linkKnown( $nt, "$prefix$text$inside", array(), $query );
2027
2028                 return $this->armorLinks( $link ) . $trail;
2029         }
2030
2031         /**
2032          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2033          * going to go through further parsing steps before inline URL expansion.
2034          *
2035          * Not needed quite as much as it used to be since free links are a bit
2036          * more sensible these days. But bracketed links are still an issue.
2037          *
2038          * @param $text String: more-or-less HTML
2039          * @return String: less-or-more HTML with NOPARSE bits
2040          */
2041         function armorLinks( $text ) {
2042                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
2043                         "{$this->mUniqPrefix}NOPARSE$1", $text );
2044         }
2045
2046         /**
2047          * Return true if subpage links should be expanded on this page.
2048          * @return Boolean
2049          */
2050         function areSubpagesAllowed() {
2051                 # Some namespaces don't allow subpages
2052                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2053         }
2054
2055         /**
2056          * Handle link to subpage if necessary
2057          *
2058          * @param $target String: the source of the link
2059          * @param &$text String: the link text, modified as necessary
2060          * @return string the full name of the link
2061          * @private
2062          */
2063         function maybeDoSubpageLink( $target, &$text ) {
2064                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2065         }
2066
2067         /**#@+
2068          * Used by doBlockLevels()
2069          * @private
2070          *
2071          * @return string
2072          */
2073         function closeParagraph() {
2074                 $result = '';
2075                 if ( $this->mLastSection != '' ) {
2076                         $result = '</' . $this->mLastSection  . ">\n";
2077                 }
2078                 $this->mInPre = false;
2079                 $this->mLastSection = '';
2080                 return $result;
2081         }
2082
2083         /**
2084          * getCommon() returns the length of the longest common substring
2085          * of both arguments, starting at the beginning of both.
2086          * @private
2087          */
2088         function getCommon( $st1, $st2 ) {
2089                 $fl = strlen( $st1 );
2090                 $shorter = strlen( $st2 );
2091                 if ( $fl < $shorter ) {
2092                         $shorter = $fl;
2093                 }
2094
2095                 for ( $i = 0; $i < $shorter; ++$i ) {
2096                         if ( $st1[$i] != $st2[$i] ) {
2097                                 break;
2098                         }
2099                 }
2100                 return $i;
2101         }
2102
2103         /**
2104          * These next three functions open, continue, and close the list
2105          * element appropriate to the prefix character passed into them.
2106          * @private
2107          *
2108          * @return string
2109          */
2110         function openList( $char ) {
2111                 $result = $this->closeParagraph();
2112
2113                 if ( '*' === $char ) {
2114                         $result .= '<ul><li>';
2115                 } elseif ( '#' === $char ) {
2116                         $result .= '<ol><li>';
2117                 } elseif ( ':' === $char ) {
2118                         $result .= '<dl><dd>';
2119                 } elseif ( ';' === $char ) {
2120                         $result .= '<dl><dt>';
2121                         $this->mDTopen = true;
2122                 } else {
2123                         $result = '<!-- ERR 1 -->';
2124                 }
2125
2126                 return $result;
2127         }
2128
2129         /**
2130          * TODO: document
2131          * @param $char String
2132          * @private
2133          *
2134          * @return string
2135          */
2136         function nextItem( $char ) {
2137                 if ( '*' === $char || '#' === $char ) {
2138                         return '</li><li>';
2139                 } elseif ( ':' === $char || ';' === $char ) {
2140                         $close = '</dd>';
2141                         if ( $this->mDTopen ) {
2142                                 $close = '</dt>';
2143                         }
2144                         if ( ';' === $char ) {
2145                                 $this->mDTopen = true;
2146                                 return $close . '<dt>';
2147                         } else {
2148                                 $this->mDTopen = false;
2149                                 return $close . '<dd>';
2150                         }
2151                 }
2152                 return '<!-- ERR 2 -->';
2153         }
2154
2155         /**
2156          * TODO: document
2157          * @param $char String
2158          * @private
2159          *
2160          * @return string
2161          */
2162         function closeList( $char ) {
2163                 if ( '*' === $char ) {
2164                         $text = '</li></ul>';
2165                 } elseif ( '#' === $char ) {
2166                         $text = '</li></ol>';
2167                 } elseif ( ':' === $char ) {
2168                         if ( $this->mDTopen ) {
2169                                 $this->mDTopen = false;
2170                                 $text = '</dt></dl>';
2171                         } else {
2172                                 $text = '</dd></dl>';
2173                         }
2174                 } else {
2175                         return '<!-- ERR 3 -->';
2176                 }
2177                 return $text."\n";
2178         }
2179         /**#@-*/
2180
2181         /**
2182          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2183          *
2184          * @param $text String
2185          * @param $linestart Boolean: whether or not this is at the start of a line.
2186          * @private
2187          * @return string the lists rendered as HTML
2188          */
2189         function doBlockLevels( $text, $linestart ) {
2190                 wfProfileIn( __METHOD__ );
2191
2192                 # Parsing through the text line by line.  The main thing
2193                 # happening here is handling of block-level elements p, pre,
2194                 # and making lists from lines starting with * # : etc.
2195                 #
2196                 $textLines = StringUtils::explode( "\n", $text );
2197
2198                 $lastPrefix = $output = '';
2199                 $this->mDTopen = $inBlockElem = false;
2200                 $prefixLength = 0;
2201                 $paragraphStack = false;
2202
2203                 foreach ( $textLines as $oLine ) {
2204                         # Fix up $linestart
2205                         if ( !$linestart ) {
2206                                 $output .= $oLine;
2207                                 $linestart = true;
2208                                 continue;
2209                         }
2210                         # * = ul
2211                         # # = ol
2212                         # ; = dt
2213                         # : = dd
2214
2215                         $lastPrefixLength = strlen( $lastPrefix );
2216                         $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2217                         $preOpenMatch = preg_match( '/<pre/i', $oLine );
2218                         # If not in a <pre> element, scan for and figure out what prefixes are there.
2219                         if ( !$this->mInPre ) {
2220                                 # Multiple prefixes may abut each other for nested lists.
2221                                 $prefixLength = strspn( $oLine, '*#:;' );
2222                                 $prefix = substr( $oLine, 0, $prefixLength );
2223
2224                                 # eh?
2225                                 # ; and : are both from definition-lists, so they're equivalent
2226                                 #  for the purposes of determining whether or not we need to open/close
2227                                 #  elements.
2228                                 $prefix2 = str_replace( ';', ':', $prefix );
2229                                 $t = substr( $oLine, $prefixLength );
2230                                 $this->mInPre = (bool)$preOpenMatch;
2231                         } else {
2232                                 # Don't interpret any other prefixes in preformatted text
2233                                 $prefixLength = 0;
2234                                 $prefix = $prefix2 = '';
2235                                 $t = $oLine;
2236                         }
2237
2238                         # List generation
2239                         if ( $prefixLength && $lastPrefix === $prefix2 ) {
2240                                 # Same as the last item, so no need to deal with nesting or opening stuff
2241                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2242                                 $paragraphStack = false;
2243
2244                                 if ( substr( $prefix, -1 ) === ';') {
2245                                         # The one nasty exception: definition lists work like this:
2246                                         # ; title : definition text
2247                                         # So we check for : in the remainder text to split up the
2248                                         # title and definition, without b0rking links.
2249                                         $term = $t2 = '';
2250                                         if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2251                                                 $t = $t2;
2252                                                 $output .= $term . $this->nextItem( ':' );
2253                                         }
2254                                 }
2255                         } elseif ( $prefixLength || $lastPrefixLength ) {
2256                                 # We need to open or close prefixes, or both.
2257
2258                                 # Either open or close a level...
2259                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2260                                 $paragraphStack = false;
2261
2262                                 # Close all the prefixes which aren't shared.
2263                                 while ( $commonPrefixLength < $lastPrefixLength ) {
2264                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2265                                         --$lastPrefixLength;
2266                                 }
2267
2268                                 # Continue the current prefix if appropriate.
2269                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2270                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2271                                 }
2272
2273                                 # Open prefixes where appropriate.
2274                                 while ( $prefixLength > $commonPrefixLength ) {
2275                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2276                                         $output .= $this->openList( $char );
2277
2278                                         if ( ';' === $char ) {
2279                                                 # FIXME: This is dupe of code above
2280                                                 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2281                                                         $t = $t2;
2282                                                         $output .= $term . $this->nextItem( ':' );
2283                                                 }
2284                                         }
2285                                         ++$commonPrefixLength;
2286                                 }
2287                                 $lastPrefix = $prefix2;
2288                         }
2289
2290                         # If we have no prefixes, go to paragraph mode.
2291                         if ( 0 == $prefixLength ) {
2292                                 wfProfileIn( __METHOD__."-paragraph" );
2293                                 # No prefix (not in list)--go to paragraph mode
2294                                 # XXX: use a stack for nestable elements like span, table and div
2295                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2296                                 $closematch = preg_match(
2297                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2298                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2299                                 if ( $openmatch or $closematch ) {
2300                                         $paragraphStack = false;
2301                                         # TODO bug 5718: paragraph closed
2302                                         $output .= $this->closeParagraph();
2303                                         if ( $preOpenMatch and !$preCloseMatch ) {
2304                                                 $this->mInPre = true;
2305                                         }
2306                                         $inBlockElem = !$closematch;
2307                                 } elseif ( !$inBlockElem && !$this->mInPre ) {
2308                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) ) {
2309                                                 # pre
2310                                                 if ( $this->mLastSection !== 'pre' ) {
2311                                                         $paragraphStack = false;
2312                                                         $output .= $this->closeParagraph().'<pre>';
2313                                                         $this->mLastSection = 'pre';
2314                                                 }
2315                                                 $t = substr( $t, 1 );
2316                                         } else {
2317                                                 # paragraph
2318                                                 if ( trim( $t ) === '' ) {
2319                                                         if ( $paragraphStack ) {
2320                                                                 $output .= $paragraphStack.'<br />';
2321                                                                 $paragraphStack = false;
2322                                                                 $this->mLastSection = 'p';
2323                                                         } else {
2324                                                                 if ( $this->mLastSection !== 'p' ) {
2325                                                                         $output .= $this->closeParagraph();
2326                                                                         $this->mLastSection = '';
2327                                                                         $paragraphStack = '<p>';
2328                                                                 } else {
2329                                                                         $paragraphStack = '</p><p>';
2330                                                                 }
2331                                                         }
2332                                                 } else {
2333                                                         if ( $paragraphStack ) {
2334                                                                 $output .= $paragraphStack;
2335                                                                 $paragraphStack = false;
2336                                                                 $this->mLastSection = 'p';
2337                                                         } elseif ( $this->mLastSection !== 'p' ) {
2338                                                                 $output .= $this->closeParagraph().'<p>';
2339                                                                 $this->mLastSection = 'p';
2340                                                         }
2341                                                 }
2342                                         }
2343                                 }
2344                                 wfProfileOut( __METHOD__."-paragraph" );
2345                         }
2346                         # somewhere above we forget to get out of pre block (bug 785)
2347                         if ( $preCloseMatch && $this->mInPre ) {
2348                                 $this->mInPre = false;
2349                         }
2350                         if ( $paragraphStack === false ) {
2351                                 $output .= $t."\n";
2352                         }
2353                 }
2354                 while ( $prefixLength ) {
2355                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2356                         --$prefixLength;
2357                 }
2358                 if ( $this->mLastSection != '' ) {
2359                         $output .= '</' . $this->mLastSection . '>';
2360                         $this->mLastSection = '';
2361                 }
2362
2363                 wfProfileOut( __METHOD__ );
2364                 return $output;
2365         }
2366
2367         /**
2368          * Split up a string on ':', ignoring any occurences inside tags
2369          * to prevent illegal overlapping.
2370          *
2371          * @param $str String: the string to split
2372          * @param &$before String: set to everything before the ':'
2373          * @param &$after String: set to everything after the ':'
2374          * return String: the position of the ':', or false if none found
2375          */
2376         function findColonNoLinks( $str, &$before, &$after ) {
2377                 wfProfileIn( __METHOD__ );
2378
2379                 $pos = strpos( $str, ':' );
2380                 if ( $pos === false ) {
2381                         # Nothing to find!
2382                         wfProfileOut( __METHOD__ );
2383                         return false;
2384                 }
2385
2386                 $lt = strpos( $str, '<' );
2387                 if ( $lt === false || $lt > $pos ) {
2388                         # Easy; no tag nesting to worry about
2389                         $before = substr( $str, 0, $pos );
2390                         $after = substr( $str, $pos+1 );
2391                         wfProfileOut( __METHOD__ );
2392                         return $pos;
2393                 }
2394
2395                 # Ugly state machine to walk through avoiding tags.
2396                 $state = self::COLON_STATE_TEXT;
2397                 $stack = 0;
2398                 $len = strlen( $str );
2399                 for( $i = 0; $i < $len; $i++ ) {
2400                         $c = $str[$i];
2401
2402                         switch( $state ) {
2403                         # (Using the number is a performance hack for common cases)
2404                         case 0: # self::COLON_STATE_TEXT:
2405                                 switch( $c ) {
2406                                 case "<":
2407                                         # Could be either a <start> tag or an </end> tag
2408                                         $state = self::COLON_STATE_TAGSTART;
2409                                         break;
2410                                 case ":":
2411                                         if ( $stack == 0 ) {
2412                                                 # We found it!
2413                                                 $before = substr( $str, 0, $i );
2414                                                 $after = substr( $str, $i + 1 );
2415                                                 wfProfileOut( __METHOD__ );
2416                                                 return $i;
2417                                         }
2418                                         # Embedded in a tag; don't break it.
2419                                         break;
2420                                 default:
2421                                         # Skip ahead looking for something interesting
2422                                         $colon = strpos( $str, ':', $i );
2423                                         if ( $colon === false ) {
2424                                                 # Nothing else interesting
2425                                                 wfProfileOut( __METHOD__ );
2426                                                 return false;
2427                                         }
2428                                         $lt = strpos( $str, '<', $i );
2429                                         if ( $stack === 0 ) {
2430                                                 if ( $lt === false || $colon < $lt ) {
2431                                                         # We found it!
2432                                                         $before = substr( $str, 0, $colon );
2433                                                         $after = substr( $str, $colon + 1 );
2434                                                         wfProfileOut( __METHOD__ );
2435                                                         return $i;
2436                                                 }
2437                                         }
2438                                         if ( $lt === false ) {
2439                                                 # Nothing else interesting to find; abort!
2440                                                 # We're nested, but there's no close tags left. Abort!
2441                                                 break 2;
2442                                         }
2443                                         # Skip ahead to next tag start
2444                                         $i = $lt;
2445                                         $state = self::COLON_STATE_TAGSTART;
2446                                 }
2447                                 break;
2448                         case 1: # self::COLON_STATE_TAG:
2449                                 # In a <tag>
2450                                 switch( $c ) {
2451                                 case ">":
2452                                         $stack++;
2453                                         $state = self::COLON_STATE_TEXT;
2454                                         break;
2455                                 case "/":
2456                                         # Slash may be followed by >?
2457                                         $state = self::COLON_STATE_TAGSLASH;
2458                                         break;
2459                                 default:
2460                                         # ignore
2461                                 }
2462                                 break;
2463                         case 2: # self::COLON_STATE_TAGSTART:
2464                                 switch( $c ) {
2465                                 case "/":
2466                                         $state = self::COLON_STATE_CLOSETAG;
2467                                         break;
2468                                 case "!":
2469                                         $state = self::COLON_STATE_COMMENT;
2470                                         break;
2471                                 case ">":
2472                                         # Illegal early close? This shouldn't happen D:
2473                                         $state = self::COLON_STATE_TEXT;
2474                                         break;
2475                                 default:
2476                                         $state = self::COLON_STATE_TAG;
2477                                 }
2478                                 break;
2479                         case 3: # self::COLON_STATE_CLOSETAG:
2480                                 # In a </tag>
2481                                 if ( $c === ">" ) {
2482                                         $stack--;
2483                                         if ( $stack < 0 ) {
2484                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2485                                                 wfProfileOut( __METHOD__ );
2486                                                 return false;
2487                                         }
2488                                         $state = self::COLON_STATE_TEXT;
2489                                 }
2490                                 break;
2491                         case self::COLON_STATE_TAGSLASH:
2492                                 if ( $c === ">" ) {
2493                                         # Yes, a self-closed tag <blah/>
2494                                         $state = self::COLON_STATE_TEXT;
2495                                 } else {
2496                                         # Probably we're jumping the gun, and this is an attribute
2497                                         $state = self::COLON_STATE_TAG;
2498                                 }
2499                                 break;
2500                         case 5: # self::COLON_STATE_COMMENT:
2501                                 if ( $c === "-" ) {
2502                                         $state = self::COLON_STATE_COMMENTDASH;
2503                                 }
2504                                 break;
2505                         case self::COLON_STATE_COMMENTDASH:
2506                                 if ( $c === "-" ) {
2507                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2508                                 } else {
2509                                         $state = self::COLON_STATE_COMMENT;
2510                                 }
2511                                 break;
2512                         case self::COLON_STATE_COMMENTDASHDASH:
2513                                 if ( $c === ">" ) {
2514                                         $state = self::COLON_STATE_TEXT;
2515                                 } else {
2516                                         $state = self::COLON_STATE_COMMENT;
2517                                 }
2518                                 break;
2519                         default:
2520                                 throw new MWException( "State machine error in " . __METHOD__ );
2521                         }
2522                 }
2523                 if ( $stack > 0 ) {
2524                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2525                         wfProfileOut( __METHOD__ );
2526                         return false;
2527                 }
2528                 wfProfileOut( __METHOD__ );
2529                 return false;
2530         }
2531
2532         /**
2533          * Return value of a magic variable (like PAGENAME)
2534          *
2535          * @private
2536          *
2537          * @param $index integer
2538          * @param $frame PPFrame
2539          */
2540         function getVariableValue( $index, $frame=false ) {
2541                 global $wgContLang, $wgSitename, $wgServer;
2542                 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2543
2544                 /**
2545                  * Some of these require message or data lookups and can be
2546                  * expensive to check many times.
2547                  */
2548                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2549                         if ( isset( $this->mVarCache[$index] ) ) {
2550                                 return $this->mVarCache[$index];
2551                         }
2552                 }
2553
2554                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2555                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2556
2557                 # Use the time zone
2558                 global $wgLocaltimezone;
2559                 if ( isset( $wgLocaltimezone ) ) {
2560                         $oldtz = date_default_timezone_get();
2561                         date_default_timezone_set( $wgLocaltimezone );
2562                 }
2563
2564                 $localTimestamp = date( 'YmdHis', $ts );
2565                 $localMonth = date( 'm', $ts );
2566                 $localMonth1 = date( 'n', $ts );
2567                 $localMonthName = date( 'n', $ts );
2568                 $localDay = date( 'j', $ts );
2569                 $localDay2 = date( 'd', $ts );
2570                 $localDayOfWeek = date( 'w', $ts );
2571                 $localWeek = date( 'W', $ts );
2572                 $localYear = date( 'Y', $ts );
2573                 $localHour = date( 'H', $ts );
2574                 if ( isset( $wgLocaltimezone ) ) {
2575                         date_default_timezone_set( $oldtz );
2576                 }
2577
2578                 switch ( $index ) {
2579                         case 'currentmonth':
2580                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2581                                 break;
2582                         case 'currentmonth1':
2583                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2584                                 break;
2585                         case 'currentmonthname':
2586                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2587                                 break;
2588                         case 'currentmonthnamegen':
2589                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2590                                 break;
2591                         case 'currentmonthabbrev':
2592                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2593                                 break;
2594                         case 'currentday':
2595                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2596                                 break;
2597                         case 'currentday2':
2598                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2599                                 break;
2600                         case 'localmonth':
2601                                 $value = $wgContLang->formatNum( $localMonth );
2602                                 break;
2603                         case 'localmonth1':
2604                                 $value = $wgContLang->formatNum( $localMonth1 );
2605                                 break;
2606                         case 'localmonthname':
2607                                 $value = $wgContLang->getMonthName( $localMonthName );
2608                                 break;
2609                         case 'localmonthnamegen':
2610                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2611                                 break;
2612                         case 'localmonthabbrev':
2613                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2614                                 break;
2615                         case 'localday':
2616                                 $value = $wgContLang->formatNum( $localDay );
2617                                 break;
2618                         case 'localday2':
2619                                 $value = $wgContLang->formatNum( $localDay2 );
2620                                 break;
2621                         case 'pagename':
2622                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2623                                 break;
2624                         case 'pagenamee':
2625                                 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2626                                 break;
2627                         case 'fullpagename':
2628                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2629                                 break;
2630                         case 'fullpagenamee':
2631                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2632                                 break;
2633                         case 'subpagename':
2634                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2635                                 break;
2636                         case 'subpagenamee':
2637                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2638                                 break;
2639                         case 'basepagename':
2640                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2641                                 break;
2642                         case 'basepagenamee':
2643                                 $value = wfEscapeWikiText( wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) ) );
2644                                 break;
2645                         case 'talkpagename':
2646                                 if ( $this->mTitle->canTalk() ) {
2647                                         $talkPage = $this->mTitle->getTalkPage();
2648                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2649                                 } else {
2650                                         $value = '';
2651                                 }
2652                                 break;
2653                         case 'talkpagenamee':
2654                                 if ( $this->mTitle->canTalk() ) {
2655                                         $talkPage = $this->mTitle->getTalkPage();
2656                                         $value = wfEscapeWikiText( $talkPage->getPrefixedUrl() );
2657                                 } else {
2658                                         $value = '';
2659                                 }
2660                                 break;
2661                         case 'subjectpagename':
2662                                 $subjPage = $this->mTitle->getSubjectPage();
2663                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2664                                 break;
2665                         case 'subjectpagenamee':
2666                                 $subjPage = $this->mTitle->getSubjectPage();
2667                                 $value = wfEscapeWikiText( $subjPage->getPrefixedUrl() );
2668                                 break;
2669                         case 'revisionid':
2670                                 # Let the edit saving system know we should parse the page
2671                                 # *after* a revision ID has been assigned.
2672                                 $this->mOutput->setFlag( 'vary-revision' );
2673                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2674                                 $value = $this->mRevisionId;
2675                                 break;
2676                         case 'revisionday':
2677                                 # Let the edit saving system know we should parse the page
2678                                 # *after* a revision ID has been assigned. This is for null edits.
2679                                 $this->mOutput->setFlag( 'vary-revision' );
2680                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2681                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2682                                 break;
2683                         case 'revisionday2':
2684                                 # Let the edit saving system know we should parse the page
2685                                 # *after* a revision ID has been assigned. This is for null edits.
2686                                 $this->mOutput->setFlag( 'vary-revision' );
2687                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2688                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2689                                 break;
2690                         case 'revisionmonth':
2691                                 # Let the edit saving system know we should parse the page
2692                                 # *after* a revision ID has been assigned. This is for null edits.
2693                                 $this->mOutput->setFlag( 'vary-revision' );
2694                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2695                                 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2696                                 break;
2697                         case 'revisionmonth1':
2698                                 # Let the edit saving system know we should parse the page
2699                                 # *after* a revision ID has been assigned. This is for null edits.
2700                                 $this->mOutput->setFlag( 'vary-revision' );
2701                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2702                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2703                                 break;
2704                         case 'revisionyear':
2705                                 # Let the edit saving system know we should parse the page
2706                                 # *after* a revision ID has been assigned. This is for null edits.
2707                                 $this->mOutput->setFlag( 'vary-revision' );
2708                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2709                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2710                                 break;
2711                         case 'revisiontimestamp':
2712                                 # Let the edit saving system know we should parse the page
2713                                 # *after* a revision ID has been assigned. This is for null edits.
2714                                 $this->mOutput->setFlag( 'vary-revision' );
2715                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2716                                 $value = $this->getRevisionTimestamp();
2717                                 break;
2718                         case 'revisionuser':
2719                                 # Let the edit saving system know we should parse the page
2720                                 # *after* a revision ID has been assigned. This is for null edits.
2721                                 $this->mOutput->setFlag( 'vary-revision' );
2722                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2723                                 $value = $this->getRevisionUser();
2724                                 break;
2725                         case 'namespace':
2726                                 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2727                                 break;
2728                         case 'namespacee':
2729                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2730                                 break;
2731                         case 'talkspace':
2732                                 $value = $this->mTitle->canTalk() ? str_replace( '_',' ',$this->mTitle->getTalkNsText() ) : '';
2733                                 break;
2734                         case 'talkspacee':
2735                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2736                                 break;
2737                         case 'subjectspace':
2738                                 $value = $this->mTitle->getSubjectNsText();
2739                                 break;
2740                         case 'subjectspacee':
2741                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2742                                 break;
2743                         case 'currentdayname':
2744                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2745                                 break;
2746                         case 'currentyear':
2747                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2748                                 break;
2749                         case 'currenttime':
2750                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2751                                 break;
2752                         case 'currenthour':
2753                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2754                                 break;
2755                         case 'currentweek':
2756                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2757                                 # int to remove the padding
2758                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2759                                 break;
2760                         case 'currentdow':
2761                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2762                                 break;
2763                         case 'localdayname':
2764                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2765                                 break;
2766                         case 'localyear':
2767                                 $value = $wgContLang->formatNum( $localYear, true );
2768                                 break;
2769                         case 'localtime':
2770                                 $value = $wgContLang->time( $localTimestamp, false, false );
2771                                 break;
2772                         case 'localhour':
2773                                 $value = $wgContLang->formatNum( $localHour, true );
2774                                 break;
2775                         case 'localweek':
2776                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2777                                 # int to remove the padding
2778                                 $value = $wgContLang->formatNum( (int)$localWeek );
2779                                 break;
2780                         case 'localdow':
2781                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2782                                 break;
2783                         case 'numberofarticles':
2784                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2785                                 break;
2786                         case 'numberoffiles':
2787                                 $value = $wgContLang->formatNum( SiteStats::images() );
2788                                 break;
2789                         case 'numberofusers':
2790                                 $value = $wgContLang->formatNum( SiteStats::users() );
2791                                 break;
2792                         case 'numberofactiveusers':
2793                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2794                                 break;
2795                         case 'numberofpages':
2796                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2797                                 break;
2798                         case 'numberofadmins':
2799                                 $value = $wgContLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2800                                 break;
2801                         case 'numberofedits':
2802                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2803                                 break;
2804                         case 'numberofviews':
2805                                 $value = $wgContLang->formatNum( SiteStats::views() );
2806                                 break;
2807                         case 'currenttimestamp':
2808                                 $value = wfTimestamp( TS_MW, $ts );
2809                                 break;
2810                         case 'localtimestamp':
2811                                 $value = $localTimestamp;
2812                                 break;
2813                         case 'currentversion':
2814                                 $value = SpecialVersion::getVersion();
2815                                 break;
2816                         case 'articlepath':
2817                                 return $wgArticlePath;
2818                         case 'sitename':
2819                                 return $wgSitename;
2820                         case 'server':
2821                                 return $wgServer;
2822                         case 'servername':
2823                                 wfSuppressWarnings(); # May give an E_WARNING in PHP < 5.3.3
2824                                 $serverName = parse_url( $wgServer, PHP_URL_HOST );
2825                                 wfRestoreWarnings();
2826                                 return $serverName ? $serverName : $wgServer;
2827                         case 'scriptpath':
2828                                 return $wgScriptPath;
2829                         case 'stylepath':
2830                                 return $wgStylePath;
2831                         case 'directionmark':
2832                                 return $wgContLang->getDirMark();
2833                         case 'contentlanguage':
2834                                 global $wgLanguageCode;
2835                                 return $wgLanguageCode;
2836                         default:
2837                                 $ret = null;
2838                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
2839                                         return $ret;
2840                                 } else {
2841                                         return null;
2842                                 }
2843                 }
2844
2845                 if ( $index )
2846                         $this->mVarCache[$index] = $value;
2847
2848                 return $value;
2849         }
2850
2851         /**
2852          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2853          *
2854          * @private
2855          */
2856         function initialiseVariables() {
2857                 wfProfileIn( __METHOD__ );
2858                 $variableIDs = MagicWord::getVariableIDs();
2859                 $substIDs = MagicWord::getSubstIDs();
2860
2861                 $this->mVariables = new MagicWordArray( $variableIDs );
2862                 $this->mSubstWords = new MagicWordArray( $substIDs );
2863                 wfProfileOut( __METHOD__ );
2864         }
2865
2866         /**
2867          * Preprocess some wikitext and return the document tree.
2868          * This is the ghost of replace_variables().
2869          *
2870          * @param $text String: The text to parse
2871          * @param $flags Integer: bitwise combination of:
2872          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2873          *                                     included. Default is to assume a direct page view.
2874          *
2875          * The generated DOM tree must depend only on the input text and the flags.
2876          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2877          *
2878          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2879          * change in the DOM tree for a given text, must be passed through the section identifier
2880          * in the section edit link and thus back to extractSections().
2881          *
2882          * The output of this function is currently only cached in process memory, but a persistent
2883          * cache may be implemented at a later date which takes further advantage of these strict
2884          * dependency requirements.
2885          *
2886          * @private
2887          *
2888          * @return PPNode
2889          */
2890         function preprocessToDom( $text, $flags = 0 ) {
2891                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2892                 return $dom;
2893         }
2894
2895         /**
2896          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2897          *
2898          * @return array
2899          */
2900         public static function splitWhitespace( $s ) {
2901                 $ltrimmed = ltrim( $s );
2902                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2903                 $trimmed = rtrim( $ltrimmed );
2904                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2905                 if ( $diff > 0 ) {
2906                         $w2 = substr( $ltrimmed, -$diff );
2907                 } else {
2908                         $w2 = '';
2909                 }
2910                 return array( $w1, $trimmed, $w2 );
2911         }
2912
2913         /**
2914          * Replace magic variables, templates, and template arguments
2915          * with the appropriate text. Templates are substituted recursively,
2916          * taking care to avoid infinite loops.
2917          *
2918          * Note that the substitution depends on value of $mOutputType:
2919          *  self::OT_WIKI: only {{subst:}} templates
2920          *  self::OT_PREPROCESS: templates but not extension tags
2921          *  self::OT_HTML: all templates and extension tags
2922          *
2923          * @param $text String: the text to transform
2924          * @param $frame PPFrame Object describing the arguments passed to the template.
2925          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2926          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2927          * @param $argsOnly Boolean: only do argument (triple-brace) expansion, not double-brace expansion
2928          * @private
2929          *
2930          * @return string
2931          */
2932         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2933                 # Is there any text? Also, Prevent too big inclusions!
2934                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2935                         return $text;
2936                 }
2937                 wfProfileIn( __METHOD__ );
2938
2939                 if ( $frame === false ) {
2940                         $frame = $this->getPreprocessor()->newFrame();
2941                 } elseif ( !( $frame instanceof PPFrame ) ) {
2942                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2943                         $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2944                 }
2945
2946                 $dom = $this->preprocessToDom( $text );
2947                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2948                 $text = $frame->expand( $dom, $flags );
2949
2950                 wfProfileOut( __METHOD__ );
2951                 return $text;
2952         }
2953
2954         /**
2955          * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2956          *
2957          * @return array
2958          */
2959         static function createAssocArgs( $args ) {
2960                 $assocArgs = array();
2961                 $index = 1;
2962                 foreach ( $args as $arg ) {
2963                         $eqpos = strpos( $arg, '=' );
2964                         if ( $eqpos === false ) {
2965                                 $assocArgs[$index++] = $arg;
2966                         } else {
2967                                 $name = trim( substr( $arg, 0, $eqpos ) );
2968                                 $value = trim( substr( $arg, $eqpos+1 ) );
2969                                 if ( $value === false ) {
2970                                         $value = '';
2971                                 }
2972                                 if ( $name !== false ) {
2973                                         $assocArgs[$name] = $value;
2974                                 }
2975                         }
2976                 }
2977
2978                 return $assocArgs;
2979         }
2980
2981         /**
2982          * Warn the user when a parser limitation is reached
2983          * Will warn at most once the user per limitation type
2984          *
2985          * @param $limitationType String: should be one of:
2986          *   'expensive-parserfunction' (corresponding messages:
2987          *       'expensive-parserfunction-warning',
2988          *       'expensive-parserfunction-category')
2989          *   'post-expand-template-argument' (corresponding messages:
2990          *       'post-expand-template-argument-warning',
2991          *       'post-expand-template-argument-category')
2992          *   'post-expand-template-inclusion' (corresponding messages:
2993          *       'post-expand-template-inclusion-warning',
2994          *       'post-expand-template-inclusion-category')
2995          * @param $current Current value
2996          * @param $max Maximum allowed, when an explicit limit has been
2997          *       exceeded, provide the values (optional)
2998          */
2999         function limitationWarn( $limitationType, $current=null, $max=null) {
3000                 # does no harm if $current and $max are present but are unnecessary for the message
3001                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
3002                 $this->mOutput->addWarning( $warning );
3003                 $this->addTrackingCategory( "$limitationType-category" );
3004         }
3005
3006         /**
3007          * Return the text of a template, after recursively
3008          * replacing any variables or templates within the template.
3009          *
3010          * @param $piece Array: the parts of the template
3011          *  $piece['title']: the title, i.e. the part before the |
3012          *  $piece['parts']: the parameter array
3013          *  $piece['lineStart']: whether the brace was at the start of a line
3014          * @param $frame PPFrame The current frame, contains template arguments
3015          * @return String: the text of the template
3016          * @private
3017          */
3018         function braceSubstitution( $piece, $frame ) {
3019                 global $wgContLang, $wgNonincludableNamespaces;
3020                 wfProfileIn( __METHOD__ );
3021                 wfProfileIn( __METHOD__.'-setup' );
3022
3023                 # Flags
3024                 $found = false;             # $text has been filled
3025                 $nowiki = false;            # wiki markup in $text should be escaped
3026                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
3027                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
3028                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
3029                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
3030
3031                 # Title object, where $text came from
3032                 $title = null;
3033
3034                 # $part1 is the bit before the first |, and must contain only title characters.
3035                 # Various prefixes will be stripped from it later.
3036                 $titleWithSpaces = $frame->expand( $piece['title'] );
3037                 $part1 = trim( $titleWithSpaces );
3038                 $titleText = false;
3039
3040                 # Original title text preserved for various purposes
3041                 $originalTitle = $part1;
3042
3043                 # $args is a list of argument nodes, starting from index 0, not including $part1
3044                 # *** FIXME if piece['parts'] is null then the call to getLength() below won't work b/c this $args isn't an object
3045                 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
3046                 wfProfileOut( __METHOD__.'-setup' );
3047
3048                 # SUBST
3049                 wfProfileIn( __METHOD__.'-modifiers' );
3050                 if ( !$found ) {
3051
3052                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3053
3054                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
3055                         # Decide whether to expand template or keep wikitext as-is.
3056                         if ( $this->ot['wiki'] ) {
3057                                 if ( $substMatch === false ) {
3058                                         $literal = true;  # literal when in PST with no prefix
3059                                 } else {
3060                                         $literal = false; # expand when in PST with subst: or safesubst:
3061                                 }
3062                         } else {
3063                                 if ( $substMatch == 'subst' ) {
3064                                         $literal = true;  # literal when not in PST with plain subst:
3065                                 } else {
3066                                         $literal = false; # expand when not in PST with safesubst: or no prefix
3067                                 }
3068                         }
3069                         if ( $literal ) {
3070                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3071                                 $isLocalObj = true;
3072                                 $found = true;
3073                         }
3074                 }
3075
3076                 # Variables
3077                 if ( !$found && $args->getLength() == 0 ) {
3078                         $id = $this->mVariables->matchStartToEnd( $part1 );
3079                         if ( $id !== false ) {
3080                                 $text = $this->getVariableValue( $id, $frame );
3081                                 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3082                                         $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3083                                 }
3084                                 $found = true;
3085                         }
3086                 }
3087
3088                 # MSG, MSGNW and RAW
3089                 if ( !$found ) {
3090                         # Check for MSGNW:
3091                         $mwMsgnw = MagicWord::get( 'msgnw' );
3092                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3093                                 $nowiki = true;
3094                         } else {
3095                                 # Remove obsolete MSG:
3096                                 $mwMsg = MagicWord::get( 'msg' );
3097                                 $mwMsg->matchStartAndRemove( $part1 );
3098                         }
3099
3100                         # Check for RAW:
3101                         $mwRaw = MagicWord::get( 'raw' );
3102                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3103                                 $forceRawInterwiki = true;
3104                         }
3105                 }
3106                 wfProfileOut( __METHOD__.'-modifiers' );
3107
3108                 # Parser functions
3109                 if ( !$found ) {
3110                         wfProfileIn( __METHOD__ . '-pfunc' );
3111
3112                         $colonPos = strpos( $part1, ':' );
3113                         if ( $colonPos !== false ) {
3114                                 # Case sensitive functions
3115                                 $function = substr( $part1, 0, $colonPos );
3116                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3117                                         $function = $this->mFunctionSynonyms[1][$function];
3118                                 } else {
3119                                         # Case insensitive functions
3120                                         $function = $wgContLang->lc( $function );
3121                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3122                                                 $function = $this->mFunctionSynonyms[0][$function];
3123                                         } else {
3124                                                 $function = false;
3125                                         }
3126                                 }
3127                                 if ( $function ) {
3128                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3129                                         $initialArgs = array( &$this );
3130                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3131                                         if ( $flags & SFH_OBJECT_ARGS ) {
3132                                                 # Add a frame parameter, and pass the arguments as an array
3133                                                 $allArgs = $initialArgs;
3134                                                 $allArgs[] = $frame;
3135                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3136                                                         $funcArgs[] = $args->item( $i );
3137                                                 }
3138                                                 $allArgs[] = $funcArgs;
3139                                         } else {
3140                                                 # Convert arguments to plain text
3141                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3142                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
3143                                                 }
3144                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3145                                         }
3146
3147                                         # Workaround for PHP bug 35229 and similar
3148                                         if ( !is_callable( $callback ) ) {
3149                                                 wfProfileOut( __METHOD__ . '-pfunc' );
3150                                                 wfProfileOut( __METHOD__ );
3151                                                 throw new MWException( "Tag hook for $function is not callable\n" );
3152                                         }
3153                                         $result = call_user_func_array( $callback, $allArgs );
3154                                         $found = true;
3155                                         $noparse = true;
3156                                         $preprocessFlags = 0;
3157
3158                                         if ( is_array( $result ) ) {
3159                                                 if ( isset( $result[0] ) ) {
3160                                                         $text = $result[0];
3161                                                         unset( $result[0] );
3162                                                 }
3163
3164                                                 # Extract flags into the local scope
3165                                                 # This allows callers to set flags such as nowiki, found, etc.
3166                                                 extract( $result );
3167                                         } else {
3168                                                 $text = $result;
3169                                         }
3170                                         if ( !$noparse ) {
3171                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
3172                                                 $isChildObj = true;
3173                                         }
3174                                 }
3175                         }
3176                         wfProfileOut( __METHOD__ . '-pfunc' );
3177                 }
3178
3179                 # Finish mangling title and then check for loops.
3180                 # Set $title to a Title object and $titleText to the PDBK
3181                 if ( !$found ) {
3182                         $ns = NS_TEMPLATE;
3183                         # Split the title into page and subpage
3184                         $subpage = '';
3185                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3186                         if ( $subpage !== '' ) {
3187                                 $ns = $this->mTitle->getNamespace();
3188                         }
3189                         $title = Title::newFromText( $part1, $ns );
3190                         if ( $title ) {
3191                                 $titleText = $title->getPrefixedText();
3192                                 # Check for language variants if the template is not found
3193                                 if ( $wgContLang->hasVariants() && $title->getArticleID() == 0 ) {
3194                                         $wgContLang->findVariantLink( $part1, $title, true );
3195                                 }
3196                                 # Do recursion depth check
3197                                 $limit = $this->mOptions->getMaxTemplateDepth();
3198                                 if ( $frame->depth >= $limit ) {
3199                                         $found = true;
3200                                         $text = '<span class="error">'
3201                                                 . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit )
3202                                                 . '</span>';
3203                                 }
3204                         }
3205                 }
3206
3207                 # Load from database
3208                 if ( !$found && $title ) {
3209                         wfProfileIn( __METHOD__ . '-loadtpl' );
3210                         if ( !$title->isExternal() ) {
3211                                 if ( $title->getNamespace() == NS_SPECIAL
3212                                         && $this->mOptions->getAllowSpecialInclusion()
3213                                         && $this->ot['html'] )
3214                                 {
3215                                         $text = SpecialPageFactory::capturePath( $title );
3216                                         if ( is_string( $text ) ) {
3217                                                 $found = true;
3218                                                 $isHTML = true;
3219                                                 $this->disableCache();
3220                                         }
3221                                 } elseif ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3222                                         $found = false; # access denied
3223                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3224                                 } else {
3225                                         list( $text, $title ) = $this->getTemplateDom( $title );
3226                                         if ( $text !== false ) {
3227                                                 $found = true;
3228                                                 $isChildObj = true;
3229                                         }
3230                                 }
3231
3232                                 # If the title is valid but undisplayable, make a link to it
3233                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3234                                         $text = "[[:$titleText]]";
3235                                         $found = true;
3236                                 }
3237                         } elseif ( $title->isTrans() ) {
3238                                 # Interwiki transclusion
3239                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3240                                         $text = $this->interwikiTransclude( $title, 'render' );
3241                                         $isHTML = true;
3242                                 } else {
3243                                         $text = $this->interwikiTransclude( $title, 'raw' );
3244                                         # Preprocess it like a template
3245                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3246                                         $isChildObj = true;
3247                                 }
3248                                 $found = true;
3249                         }
3250
3251                         # Do infinite loop check
3252                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3253                         if ( !$frame->loopCheck( $title ) ) {
3254                                 $found = true;
3255                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3256                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3257                         }
3258                         wfProfileOut( __METHOD__ . '-loadtpl' );
3259                 }
3260
3261                 # If we haven't found text to substitute by now, we're done
3262                 # Recover the source wikitext and return it
3263                 if ( !$found ) {
3264                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3265                         wfProfileOut( __METHOD__ );
3266                         return array( 'object' => $text );
3267                 }
3268
3269                 # Expand DOM-style return values in a child frame
3270                 if ( $isChildObj ) {
3271                         # Clean up argument array
3272                         $newFrame = $frame->newChild( $args, $title );
3273
3274                         if ( $nowiki ) {
3275                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3276                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3277                                 # Expansion is eligible for the empty-frame cache
3278                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3279                                         $text = $this->mTplExpandCache[$titleText];
3280                                 } else {
3281                                         $text = $newFrame->expand( $text );
3282                                         $this->mTplExpandCache[$titleText] = $text;
3283                                 }
3284                         } else {
3285                                 # Uncached expansion
3286                                 $text = $newFrame->expand( $text );
3287                         }
3288                 }
3289                 if ( $isLocalObj && $nowiki ) {
3290                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3291                         $isLocalObj = false;
3292                 }
3293
3294                 # Replace raw HTML by a placeholder
3295                 # Add a blank line preceding, to prevent it from mucking up
3296                 # immediately preceding headings
3297                 if ( $isHTML ) {
3298                         $text = "\n\n" . $this->insertStripItem( $text );
3299                 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3300                         # Escape nowiki-style return values
3301                         $text = wfEscapeWikiText( $text );
3302                 } elseif ( is_string( $text )
3303                         && !$piece['lineStart']
3304                         && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3305                 {
3306                         # Bug 529: if the template begins with a table or block-level
3307                         # element, it should be treated as beginning a new line.
3308                         # This behaviour is somewhat controversial.
3309                         $text = "\n" . $text;
3310                 }
3311
3312                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3313                         # Error, oversize inclusion
3314                         if ( $titleText !== false ) {
3315                                 # Make a working, properly escaped link if possible (bug 23588)
3316                                 $text = "[[:$titleText]]";
3317                         } else {
3318                                 # This will probably not be a working link, but at least it may
3319                                 # provide some hint of where the problem is
3320                                 preg_replace( '/^:/', '', $originalTitle );
3321                                 $text = "[[:$originalTitle]]";
3322                         }
3323                         $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3324                         $this->limitationWarn( 'post-expand-template-inclusion' );
3325                 }
3326
3327                 if ( $isLocalObj ) {
3328                         $ret = array( 'object' => $text );
3329                 } else {
3330                         $ret = array( 'text' => $text );
3331                 }
3332
3333                 wfProfileOut( __METHOD__ );
3334                 return $ret;
3335         }
3336
3337         /**
3338          * Get the semi-parsed DOM representation of a template with a given title,
3339          * and its redirect destination title. Cached.
3340          *
3341          * @return array
3342          */
3343         function getTemplateDom( $title ) {
3344                 $cacheTitle = $title;
3345                 $titleText = $title->getPrefixedDBkey();
3346
3347                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3348                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3349                         $title = Title::makeTitle( $ns, $dbk );
3350                         $titleText = $title->getPrefixedDBkey();
3351                 }
3352                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3353                         return array( $this->mTplDomCache[$titleText], $title );
3354                 }
3355
3356                 # Cache miss, go to the database
3357                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3358
3359                 if ( $text === false ) {
3360                         $this->mTplDomCache[$titleText] = false;
3361                         return array( false, $title );
3362                 }
3363
3364                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3365                 $this->mTplDomCache[ $titleText ] = $dom;
3366
3367                 if ( !$title->equals( $cacheTitle ) ) {
3368                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3369                                 array( $title->getNamespace(), $cdb = $title->getDBkey() );
3370                 }
3371
3372                 return array( $dom, $title );
3373         }
3374
3375         /**
3376          * Fetch the unparsed text of a template and register a reference to it.
3377          * @param Title $title
3378          * @return Array ( string or false, Title )
3379          */
3380         function fetchTemplateAndTitle( $title ) {
3381                 $templateCb = $this->mOptions->getTemplateCallback(); # Defaults to Parser::statelessFetchTemplate()
3382                 $stuff = call_user_func( $templateCb, $title, $this );
3383                 $text = $stuff['text'];
3384                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3385                 if ( isset( $stuff['deps'] ) ) {
3386                         foreach ( $stuff['deps'] as $dep ) {
3387                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3388                         }
3389                 }
3390                 return array( $text, $finalTitle );
3391         }
3392
3393         /**
3394          * Fetch the unparsed text of a template and register a reference to it.
3395          * @param Title $title
3396          * @return mixed string or false
3397          */
3398         function fetchTemplate( $title ) {
3399                 $rv = $this->fetchTemplateAndTitle( $title );
3400                 return $rv[0];
3401         }
3402
3403         /**
3404          * Static function to get a template
3405          * Can be overridden via ParserOptions::setTemplateCallback().
3406          *
3407          * @return array
3408          */
3409         static function statelessFetchTemplate( $title, $parser = false ) {
3410                 $text = $skip = false;
3411                 $finalTitle = $title;
3412                 $deps = array();
3413
3414                 # Loop to fetch the article, with up to 1 redirect
3415                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3416                         # Give extensions a chance to select the revision instead
3417                         $id = false; # Assume current
3418                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle',
3419                                 array( $parser, $title, &$skip, &$id ) );
3420
3421                         if ( $skip ) {
3422                                 $text = false;
3423                                 $deps[] = array(
3424                                         'title'         => $title,
3425                                         'page_id'       => $title->getArticleID(),
3426                                         'rev_id'        => null
3427                                 );
3428                                 break;
3429                         }
3430                         # Get the revision
3431                         $rev = $id
3432                                 ? Revision::newFromId( $id )
3433                                 : Revision::newFromTitle( $title );
3434                         $rev_id = $rev ? $rev->getId() : 0;
3435                         # If there is no current revision, there is no page
3436                         if ( $id === false && !$rev ) {
3437                                 $linkCache = LinkCache::singleton();
3438                                 $linkCache->addBadLinkObj( $title );
3439                         }
3440
3441                         $deps[] = array(
3442                                 'title'         => $title,
3443                                 'page_id'       => $title->getArticleID(),
3444                                 'rev_id'        => $rev_id );
3445                         if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3446                                 # We fetched a rev from a different title; register it too...
3447                                 $deps[] = array(
3448                                         'title'         => $rev->getTitle(),
3449                                         'page_id'       => $rev->getPage(),
3450                                         'rev_id'        => $rev_id );
3451                         }
3452
3453                         if ( $rev ) {
3454                                 $text = $rev->getText();
3455                         } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3456                                 global $wgContLang;
3457                                 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3458                                 if ( !$message->exists() ) {
3459                                         $text = false;
3460                                         break;
3461                                 }
3462                                 $text = $message->plain();
3463                         } else {
3464                                 break;
3465                         }
3466                         if ( $text === false ) {
3467                                 break;
3468                         }
3469                         # Redirect?
3470                         $finalTitle = $title;
3471                         $title = Title::newFromRedirect( $text );
3472                 }
3473                 return array(
3474                         'text' => $text,
3475                         'finalTitle' => $finalTitle,
3476                         'deps' => $deps );
3477         }
3478
3479         /**
3480          * Fetch a file and its title and register a reference to it.
3481          * @param Title $title
3482          * @param string $time MW timestamp
3483          * @param string $sha1 base 36 SHA-1
3484          * @return mixed File or false
3485          */
3486         function fetchFile( $title, $time = false, $sha1 = false ) {
3487                 $res = $this->fetchFileAndTitle( $title, $time, $sha1 );
3488                 return $res[0];
3489         }
3490
3491         /**
3492          * Fetch a file and its title and register a reference to it.
3493          * @param Title $title
3494          * @param string $time MW timestamp
3495          * @param string $sha1 base 36 SHA-1
3496          * @return Array ( File or false, Title of file )
3497          */
3498         function fetchFileAndTitle( $title, $time = false, $sha1 = false ) {
3499                 if ( $time === '0' ) {
3500                         $file = false; // broken thumbnail forced by hook
3501                 } elseif ( $sha1 ) { // get by (sha1,timestamp)
3502                         $file = RepoGroup::singleton()->findFileFromKey( $sha1, array( 'time' => $time ) );
3503                 } else { // get by (name,timestamp)
3504                         $file = wfFindFile( $title, array( 'time' => $time ) );
3505                 }
3506                 $time = $file ? $file->getTimestamp() : false;
3507                 $sha1 = $file ? $file->getSha1() : false;
3508                 # Register the file as a dependency...
3509                 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3510                 if ( $file && !$title->equals( $file->getTitle() ) ) {
3511                         # Update fetched file title
3512                         $title = $file->getTitle();
3513                 }
3514                 return array( $file, $title );
3515         }
3516
3517         /**
3518          * Transclude an interwiki link.
3519          *
3520          * @return string
3521          */
3522         function interwikiTransclude( $title, $action ) {
3523                 global $wgEnableScaryTranscluding;
3524
3525                 if ( !$wgEnableScaryTranscluding ) {
3526                         return wfMsgForContent('scarytranscludedisabled');
3527                 }
3528
3529                 $url = $title->getFullUrl( "action=$action" );
3530
3531                 if ( strlen( $url ) > 255 ) {
3532                         return wfMsgForContent( 'scarytranscludetoolong' );
3533                 }
3534                 return $this->fetchScaryTemplateMaybeFromCache( $url );
3535         }
3536
3537         /**
3538          * @param $url string
3539          * @return Mixed|String
3540          */
3541         function fetchScaryTemplateMaybeFromCache( $url ) {
3542                 global $wgTranscludeCacheExpiry;
3543                 $dbr = wfGetDB( DB_SLAVE );
3544                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3545                 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3546                                 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3547                 if ( $obj ) {
3548                         return $obj->tc_contents;
3549                 }
3550
3551                 $text = Http::get( $url );
3552                 if ( !$text ) {
3553                         return wfMsgForContent( 'scarytranscludefailed', $url );
3554                 }
3555
3556                 $dbw = wfGetDB( DB_MASTER );
3557                 $dbw->replace( 'transcache', array('tc_url'), array(
3558                         'tc_url' => $url,
3559                         'tc_time' => $dbw->timestamp( time() ),
3560                         'tc_contents' => $text)
3561                 );
3562                 return $text;
3563         }
3564
3565         /**
3566          * Triple brace replacement -- used for template arguments
3567          * @private
3568          *
3569          * @param $peice array
3570          * @param $frame PPFrame
3571          *
3572          * @return array
3573          */
3574         function argSubstitution( $piece, $frame ) {
3575                 wfProfileIn( __METHOD__ );
3576
3577                 $error = false;
3578                 $parts = $piece['parts'];
3579                 $nameWithSpaces = $frame->expand( $piece['title'] );
3580                 $argName = trim( $nameWithSpaces );
3581                 $object = false;
3582                 $text = $frame->getArgument( $argName );
3583                 if (  $text === false && $parts->getLength() > 0
3584                   && (
3585                         $this->ot['html']
3586                         || $this->ot['pre']
3587                         || ( $this->ot['wiki'] && $frame->isTemplate() )
3588                   )
3589                 ) {
3590                         # No match in frame, use the supplied default
3591                         $object = $parts->item( 0 )->getChildren();
3592                 }
3593                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3594                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3595                         $this->limitationWarn( 'post-expand-template-argument' );
3596                 }
3597
3598                 if ( $text === false && $object === false ) {
3599                         # No match anywhere
3600                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3601                 }
3602                 if ( $error !== false ) {
3603                         $text .= $error;
3604                 }
3605                 if ( $object !== false ) {
3606                         $ret = array( 'object' => $object );
3607                 } else {
3608                         $ret = array( 'text' => $text );
3609                 }
3610
3611                 wfProfileOut( __METHOD__ );
3612                 return $ret;
3613         }
3614
3615         /**
3616          * Return the text to be used for a given extension tag.
3617          * This is the ghost of strip().
3618          *
3619          * @param $params Associative array of parameters:
3620          *     name       PPNode for the tag name
3621          *     attr       PPNode for unparsed text where tag attributes are thought to be
3622          *     attributes Optional associative array of parsed attributes
3623          *     inner      Contents of extension element
3624          *     noClose    Original text did not have a close tag
3625          * @param $frame PPFrame
3626          *
3627          * @return string
3628          */
3629         function extensionSubstitution( $params, $frame ) {
3630                 $name = $frame->expand( $params['name'] );
3631                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3632                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3633                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3634
3635                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3636                         ( $this->ot['html'] || $this->ot['pre'] );
3637                 if ( $isFunctionTag ) {
3638                         $markerType = 'none';
3639                 } else {
3640                         $markerType = 'general';
3641                 }
3642                 if ( $this->ot['html'] || $isFunctionTag ) {
3643                         $name = strtolower( $name );
3644                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3645                         if ( isset( $params['attributes'] ) ) {
3646                                 $attributes = $attributes + $params['attributes'];
3647                         }
3648
3649                         if ( isset( $this->mTagHooks[$name] ) ) {
3650                                 # Workaround for PHP bug 35229 and similar
3651                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3652                                         throw new MWException( "Tag hook for $name is not callable\n" );
3653                                 }
3654                                 $output = call_user_func_array( $this->mTagHooks[$name],
3655                                         array( $content, $attributes, $this, $frame ) );
3656                         } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3657                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3658                                 if ( !is_callable( $callback ) ) {
3659                                         throw new MWException( "Tag hook for $name is not callable\n" );
3660                                 }
3661
3662                                 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3663                         } else {
3664                                 $output = '<span class="error">Invalid tag extension name: ' .
3665                                         htmlspecialchars( $name ) . '</span>';
3666                         }
3667
3668                         if ( is_array( $output ) ) {
3669                                 # Extract flags to local scope (to override $markerType)
3670                                 $flags = $output;
3671                                 $output = $flags[0];
3672                                 unset( $flags[0] );
3673                                 extract( $flags );
3674                         }
3675                 } else {
3676                         if ( is_null( $attrText ) ) {
3677                                 $attrText = '';
3678                         }
3679                         if ( isset( $params['attributes'] ) ) {
3680                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3681                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3682                                                 htmlspecialchars( $attrValue ) . '"';
3683                                 }
3684                         }
3685                         if ( $content === null ) {
3686                                 $output = "<$name$attrText/>";
3687                         } else {
3688                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3689                                 $output = "<$name$attrText>$content$close";
3690                         }
3691                 }
3692
3693                 if ( $markerType === 'none' ) {
3694                         return $output;
3695                 } elseif ( $markerType === 'nowiki' ) {
3696                         $this->mStripState->addNoWiki( $marker, $output );
3697                 } elseif ( $markerType === 'general' ) {
3698                         $this->mStripState->addGeneral( $marker, $output );
3699                 } else {
3700                         throw new MWException( __METHOD__.': invalid marker type' );
3701                 }
3702                 return $marker;
3703         }
3704
3705         /**
3706          * Increment an include size counter
3707          *
3708          * @param $type String: the type of expansion
3709          * @param $size Integer: the size of the text
3710          * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
3711          */
3712         function incrementIncludeSize( $type, $size ) {
3713                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3714                         return false;
3715                 } else {
3716                         $this->mIncludeSizes[$type] += $size;
3717                         return true;
3718                 }
3719         }
3720
3721         /**
3722          * Increment the expensive function count
3723          *
3724          * @return Boolean: false if the limit has been exceeded
3725          */
3726         function incrementExpensiveFunctionCount() {
3727                 global $wgExpensiveParserFunctionLimit;
3728                 $this->mExpensiveFunctionCount++;
3729                 if ( $this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit ) {
3730                         return true;
3731                 }
3732                 return false;
3733         }
3734
3735         /**
3736          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3737          * Fills $this->mDoubleUnderscores, returns the modified text
3738          */
3739         function doDoubleUnderscore( $text ) {
3740                 wfProfileIn( __METHOD__ );
3741
3742                 # The position of __TOC__ needs to be recorded
3743                 $mw = MagicWord::get( 'toc' );
3744                 if ( $mw->match( $text ) ) {
3745                         $this->mShowToc = true;
3746                         $this->mForceTocPosition = true;
3747
3748                         # Set a placeholder. At the end we'll fill it in with the TOC.
3749                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3750
3751                         # Only keep the first one.
3752                         $text = $mw->replace( '', $text );
3753                 }
3754
3755                 # Now match and remove the rest of them
3756                 $mwa = MagicWord::getDoubleUnderscoreArray();
3757                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3758
3759                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3760                         $this->mOutput->mNoGallery = true;
3761                 }
3762                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3763                         $this->mShowToc = false;
3764                 }
3765                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3766                         $this->addTrackingCategory( 'hidden-category-category' );
3767                 }
3768                 # (bug 8068) Allow control over whether robots index a page.
3769                 #
3770                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3771                 # is not desirable, the last one on the page should win.
3772                 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3773                         $this->mOutput->setIndexPolicy( 'noindex' );
3774                         $this->addTrackingCategory( 'noindex-category' );
3775                 }
3776                 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3777                         $this->mOutput->setIndexPolicy( 'index' );
3778                         $this->addTrackingCategory( 'index-category' );
3779                 }
3780
3781                 # Cache all double underscores in the database
3782                 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3783                         $this->mOutput->setProperty( $key, '' );
3784                 }
3785
3786                 wfProfileOut( __METHOD__ );
3787                 return $text;
3788         }
3789
3790         /**
3791          * Add a tracking category, getting the title from a system message,
3792          * or print a debug message if the title is invalid.
3793          *
3794          * @param $msg String: message key
3795          * @return Boolean: whether the addition was successful
3796          */
3797         protected function addTrackingCategory( $msg ) {
3798                 $cat = wfMsgForContent( $msg );
3799
3800                 # Allow tracking categories to be disabled by setting them to "-"
3801                 if ( $cat === '-' ) {
3802                         return false;
3803                 }
3804
3805                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3806                 if ( $containerCategory ) {
3807                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3808                         return true;
3809                 } else {
3810                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3811                         return false;
3812                 }
3813         }
3814
3815         /**
3816          * This function accomplishes several tasks:
3817          * 1) Auto-number headings if that option is enabled
3818          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3819          * 3) Add a Table of contents on the top for users who have enabled the option
3820          * 4) Auto-anchor headings
3821          *
3822          * It loops through all headlines, collects the necessary data, then splits up the
3823          * string and re-inserts the newly formatted headlines.
3824          *
3825          * @param $text String
3826          * @param $origText String: original, untouched wikitext
3827          * @param $isMain Boolean
3828          * @private
3829          */
3830         function formatHeadings( $text, $origText, $isMain=true ) {
3831                 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3832
3833                 # Inhibit editsection links if requested in the page
3834                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3835                         $showEditLink = 0;
3836                 } else {
3837                         $showEditLink = $this->mOptions->getEditSection();
3838                 }
3839                 if ( $showEditLink ) {
3840                         $this->mOutput->setEditSectionTokens( true );
3841                 }
3842
3843                 # Get all headlines for numbering them and adding funky stuff like [edit]
3844                 # links - this is for later, but we need the number of headlines right now
3845                 $matches = array();
3846                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3847
3848                 # if there are fewer than 4 headlines in the article, do not show TOC
3849                 # unless it's been explicitly enabled.
3850                 $enoughToc = $this->mShowToc &&
3851                         ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3852
3853                 # Allow user to stipulate that a page should have a "new section"
3854                 # link added via __NEWSECTIONLINK__
3855                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3856                         $this->mOutput->setNewSection( true );
3857                 }
3858
3859                 # Allow user to remove the "new section"
3860                 # link via __NONEWSECTIONLINK__
3861                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3862                         $this->mOutput->hideNewSection( true );
3863                 }
3864
3865                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3866                 # override above conditions and always show TOC above first header
3867                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3868                         $this->mShowToc = true;
3869                         $enoughToc = true;
3870                 }
3871
3872                 # headline counter
3873                 $headlineCount = 0;
3874                 $numVisible = 0;
3875
3876                 # Ugh .. the TOC should have neat indentation levels which can be
3877                 # passed to the skin functions. These are determined here
3878                 $toc = '';
3879                 $full = '';
3880                 $head = array();
3881                 $sublevelCount = array();
3882                 $levelCount = array();
3883                 $level = 0;
3884                 $prevlevel = 0;
3885                 $toclevel = 0;
3886                 $prevtoclevel = 0;
3887                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3888                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3889                 $oldType = $this->mOutputType;
3890                 $this->setOutputType( self::OT_WIKI );
3891                 $frame = $this->getPreprocessor()->newFrame();
3892                 $root = $this->preprocessToDom( $origText );
3893                 $node = $root->getFirstChild();
3894                 $byteOffset = 0;
3895                 $tocraw = array();
3896                 $refers = array();
3897
3898                 foreach ( $matches[3] as $headline ) {
3899                         $isTemplate = false;
3900                         $titleText = false;
3901                         $sectionIndex = false;
3902                         $numbering = '';
3903                         $markerMatches = array();
3904                         if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
3905                                 $serial = $markerMatches[1];
3906                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3907                                 $isTemplate = ( $titleText != $baseTitleText );
3908                                 $headline = preg_replace( "/^$markerRegex/", "", $headline );
3909                         }
3910
3911                         if ( $toclevel ) {
3912                                 $prevlevel = $level;
3913                         }
3914                         $level = $matches[1][$headlineCount];
3915
3916                         if ( $level > $prevlevel ) {
3917                                 # Increase TOC level
3918                                 $toclevel++;
3919                                 $sublevelCount[$toclevel] = 0;
3920                                 if ( $toclevel<$wgMaxTocLevel ) {
3921                                         $prevtoclevel = $toclevel;
3922                                         $toc .= Linker::tocIndent();
3923                                         $numVisible++;
3924                                 }
3925                         } elseif ( $level < $prevlevel && $toclevel > 1 ) {
3926                                 # Decrease TOC level, find level to jump to
3927
3928                                 for ( $i = $toclevel; $i > 0; $i-- ) {
3929                                         if ( $levelCount[$i] == $level ) {
3930                                                 # Found last matching level
3931                                                 $toclevel = $i;
3932                                                 break;
3933                                         } elseif ( $levelCount[$i] < $level ) {
3934                                                 # Found first matching level below current level
3935                                                 $toclevel = $i + 1;
3936                                                 break;
3937                                         }
3938                                 }
3939                                 if ( $i == 0 ) {
3940                                         $toclevel = 1;
3941                                 }
3942                                 if ( $toclevel<$wgMaxTocLevel ) {
3943                                         if ( $prevtoclevel < $wgMaxTocLevel ) {
3944                                                 # Unindent only if the previous toc level was shown :p
3945                                                 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
3946                                                 $prevtoclevel = $toclevel;
3947                                         } else {
3948                                                 $toc .= Linker::tocLineEnd();
3949                                         }
3950                                 }
3951                         } else {
3952                                 # No change in level, end TOC line
3953                                 if ( $toclevel<$wgMaxTocLevel ) {
3954                                         $toc .= Linker::tocLineEnd();
3955                                 }
3956                         }
3957
3958                         $levelCount[$toclevel] = $level;
3959
3960                         # count number of headlines for each level
3961                         @$sublevelCount[$toclevel]++;
3962                         $dot = 0;
3963                         for( $i = 1; $i <= $toclevel; $i++ ) {
3964                                 if ( !empty( $sublevelCount[$i] ) ) {
3965                                         if ( $dot ) {
3966                                                 $numbering .= '.';
3967                                         }
3968                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3969                                         $dot = 1;
3970                                 }
3971                         }
3972
3973                         # The safe header is a version of the header text safe to use for links
3974                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3975                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3976
3977                         # Remove link placeholders by the link text.
3978                         #     <!--LINK number-->
3979                         # turns into
3980                         #     link text with suffix
3981                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3982
3983                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3984                         $tocline = preg_replace(
3985                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3986                                 array( '',                          '<$1>' ),
3987                                 $safeHeadline
3988                         );
3989                         $tocline = trim( $tocline );
3990
3991                         # For the anchor, strip out HTML-y stuff period
3992                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3993                         $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
3994
3995                         # Save headline for section edit hint before it's escaped
3996                         $headlineHint = $safeHeadline;
3997
3998                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
3999                                 # For reverse compatibility, provide an id that's
4000                                 # HTML4-compatible, like we used to.
4001                                 #
4002                                 # It may be worth noting, academically, that it's possible for
4003                                 # the legacy anchor to conflict with a non-legacy headline
4004                                 # anchor on the page.  In this case likely the "correct" thing
4005                                 # would be to either drop the legacy anchors or make sure
4006                                 # they're numbered first.  However, this would require people
4007                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4008                                 # manually, so let's not bother worrying about it.
4009                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
4010                                         array( 'noninitial', 'legacy' ) );
4011                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4012
4013                                 if ( $legacyHeadline == $safeHeadline ) {
4014                                         # No reason to have both (in fact, we can't)
4015                                         $legacyHeadline = false;
4016                                 }
4017                         } else {
4018                                 $legacyHeadline = false;
4019                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
4020                                         'noninitial' );
4021                         }
4022
4023                         # HTML names must be case-insensitively unique (bug 10721).
4024                         # This does not apply to Unicode characters per
4025                         # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
4026                         # FIXME: We may be changing them depending on the current locale.
4027                         $arrayKey = strtolower( $safeHeadline );
4028                         if ( $legacyHeadline === false ) {
4029                                 $legacyArrayKey = false;
4030                         } else {
4031                                 $legacyArrayKey = strtolower( $legacyHeadline );
4032                         }
4033
4034                         # count how many in assoc. array so we can track dupes in anchors
4035                         if ( isset( $refers[$arrayKey] ) ) {
4036                                 $refers[$arrayKey]++;
4037                         } else {
4038                                 $refers[$arrayKey] = 1;
4039                         }
4040                         if ( isset( $refers[$legacyArrayKey] ) ) {
4041                                 $refers[$legacyArrayKey]++;
4042                         } else {
4043                                 $refers[$legacyArrayKey] = 1;
4044                         }
4045
4046                         # Don't number the heading if it is the only one (looks silly)
4047                         if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4048                                 # the two are different if the line contains a link
4049                                 $headline = $numbering . ' ' . $headline;
4050                         }
4051
4052                         # Create the anchor for linking from the TOC to the section
4053                         $anchor = $safeHeadline;
4054                         $legacyAnchor = $legacyHeadline;
4055                         if ( $refers[$arrayKey] > 1 ) {
4056                                 $anchor .= '_' . $refers[$arrayKey];
4057                         }
4058                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
4059                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
4060                         }
4061                         if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4062                                 $toc .= Linker::tocLine( $anchor, $tocline,
4063                                         $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4064                         }
4065
4066                         # Add the section to the section tree
4067                         # Find the DOM node for this header
4068                         while ( $node && !$isTemplate ) {
4069                                 if ( $node->getName() === 'h' ) {
4070                                         $bits = $node->splitHeading();
4071                                         if ( $bits['i'] == $sectionIndex ) {
4072                                                 break;
4073                                         }
4074                                 }
4075                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4076                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4077                                 $node = $node->getNextSibling();
4078                         }
4079                         $tocraw[] = array(
4080                                 'toclevel' => $toclevel,
4081                                 'level' => $level,
4082                                 'line' => $tocline,
4083                                 'number' => $numbering,
4084                                 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4085                                 'fromtitle' => $titleText,
4086                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
4087                                 'anchor' => $anchor,
4088                         );
4089
4090                         # give headline the correct <h#> tag
4091                         if ( $showEditLink && $sectionIndex !== false ) {
4092                                 // Output edit section links as markers with styles that can be customized by skins
4093                                 if ( $isTemplate ) {
4094                                         # Put a T flag in the section identifier, to indicate to extractSections()
4095                                         # that sections inside <includeonly> should be counted.
4096                                         $editlinkArgs = array( $titleText, "T-$sectionIndex"/*, null */ );
4097                                 } else {
4098                                         $editlinkArgs = array( $this->mTitle->getPrefixedText(), $sectionIndex, $headlineHint );
4099                                 }
4100                                 // We use a bit of pesudo-xml for editsection markers. The language converter is run later on
4101                                 // Using a UNIQ style marker leads to the converter screwing up the tokens when it converts stuff
4102                                 // And trying to insert strip tags fails too. At this point all real inputted tags have already been escaped
4103                                 // so we don't have to worry about a user trying to input one of these markers directly.
4104                                 // We use a page and section attribute to stop the language converter from converting these important bits
4105                                 // of data, but put the headline hint inside a content block because the language converter is supposed to
4106                                 // be able to convert that piece of data.
4107                                 $editlink = '<mw:editsection page="' . htmlspecialchars($editlinkArgs[0]);
4108                                 $editlink .= '" section="' . htmlspecialchars($editlinkArgs[1]) .'"';
4109                                 if ( isset($editlinkArgs[2]) ) {
4110                                         $editlink .= '>' . $editlinkArgs[2] . '</mw:editsection>';
4111                                 } else {
4112                                         $editlink .= '/>';
4113                                 }
4114                         } else {
4115                                 $editlink = '';
4116                         }
4117                         $head[$headlineCount] = Linker::makeHeadline( $level,
4118                                 $matches['attrib'][$headlineCount], $anchor, $headline,
4119                                 $editlink, $legacyAnchor );
4120
4121                         $headlineCount++;
4122                 }
4123
4124                 $this->setOutputType( $oldType );
4125
4126                 # Never ever show TOC if no headers
4127                 if ( $numVisible < 1 ) {
4128                         $enoughToc = false;
4129                 }
4130
4131                 if ( $enoughToc ) {
4132                         if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4133                                 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4134                         }
4135                         $toc = Linker::tocList( $toc, $this->mOptions->getUserLang() );
4136                         $this->mOutput->setTOCHTML( $toc );
4137                 }
4138
4139                 if ( $isMain ) {
4140                         $this->mOutput->setSections( $tocraw );
4141                 }
4142
4143                 # split up and insert constructed headlines
4144
4145                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
4146                 $i = 0;
4147
4148                 foreach ( $blocks as $block ) {
4149                         if ( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
4150                                 # This is the [edit] link that appears for the top block of text when
4151                                 # section editing is enabled
4152
4153                                 # Disabled because it broke block formatting
4154                                 # For example, a bullet point in the top line
4155                                 # $full .= $sk->editSectionLink(0);
4156                         }
4157                         $full .= $block;
4158                         if ( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
4159                                 # Top anchor now in skin
4160                                 $full = $full.$toc;
4161                         }
4162
4163                         if ( !empty( $head[$i] ) ) {
4164                                 $full .= $head[$i];
4165                         }
4166                         $i++;
4167                 }
4168                 if ( $this->mForceTocPosition ) {
4169                         return str_replace( '<!--MWTOC-->', $toc, $full );
4170                 } else {
4171                         return $full;
4172                 }
4173         }
4174
4175         /**
4176          * Transform wiki markup when saving a page by doing \r\n -> \n
4177          * conversion, substitting signatures, {{subst:}} templates, etc.
4178          *
4179          * @param $text String: the text to transform
4180          * @param $title Title: the Title object for the current article
4181          * @param $user User: the User object describing the current user
4182          * @param $options ParserOptions: parsing options
4183          * @param $clearState Boolean: whether to clear the parser state first
4184          * @return String: the altered wiki markup
4185          */
4186         public function preSaveTransform( $text, Title $title, User $user, ParserOptions $options, $clearState = true ) {
4187                 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4188                 $this->setUser( $user );
4189
4190                 $pairs = array(
4191                         "\r\n" => "\n",
4192                 );
4193                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4194                 if( $options->getPreSaveTransform() ) {
4195                         $text = $this->pstPass2( $text, $user );
4196                 }
4197                 $text = $this->mStripState->unstripBoth( $text );
4198
4199                 $this->setUser( null ); #Reset
4200
4201                 return $text;
4202         }
4203
4204         /**
4205          * Pre-save transform helper function
4206          * @private
4207          */
4208         function pstPass2( $text, $user ) {
4209                 global $wgContLang, $wgLocaltimezone;
4210
4211                 # Note: This is the timestamp saved as hardcoded wikitext to
4212                 # the database, we use $wgContLang here in order to give
4213                 # everyone the same signature and use the default one rather
4214                 # than the one selected in each user's preferences.
4215                 # (see also bug 12815)
4216                 $ts = $this->mOptions->getTimestamp();
4217                 if ( isset( $wgLocaltimezone ) ) {
4218                         $tz = $wgLocaltimezone;
4219                 } else {
4220                         $tz = date_default_timezone_get();
4221                 }
4222
4223                 $unixts = wfTimestamp( TS_UNIX, $ts );
4224                 $oldtz = date_default_timezone_get();
4225                 date_default_timezone_set( $tz );
4226                 $ts = date( 'YmdHis', $unixts );
4227                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4228
4229                 # Allow translation of timezones through wiki. date() can return
4230                 # whatever crap the system uses, localised or not, so we cannot
4231                 # ship premade translations.
4232                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4233                 $msg = wfMessage( $key )->inContentLanguage();
4234                 if ( $msg->exists() ) {
4235                         $tzMsg = $msg->text();
4236                 }
4237
4238                 date_default_timezone_set( $oldtz );
4239
4240                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4241
4242                 # Variable replacement
4243                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4244                 $text = $this->replaceVariables( $text );
4245
4246                 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4247                 # which may corrupt this parser instance via its wfMsgExt( parsemag ) call-
4248
4249                 # Signatures
4250                 $sigText = $this->getUserSig( $user );
4251                 $text = strtr( $text, array(
4252                         '~~~~~' => $d,
4253                         '~~~~' => "$sigText $d",
4254                         '~~~' => $sigText
4255                 ) );
4256
4257                 # Context links: [[|name]] and [[name (context)|]]
4258                 global $wgLegalTitleChars;
4259                 $tc = "[$wgLegalTitleChars]";
4260                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4261
4262                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4263                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
4264                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4265                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4266
4267                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4268                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4269                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4270                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4271
4272                 $t = $this->mTitle->getText();
4273                 $m = array();
4274                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4275                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4276                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4277                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4278                 } else {
4279                         # if there's no context, don't bother duplicating the title
4280                         $text = preg_replace( $p2, '[[\\1]]', $text );
4281                 }
4282
4283                 # Trim trailing whitespace
4284                 $text = rtrim( $text );
4285
4286                 return $text;
4287         }
4288
4289         /**
4290          * Fetch the user's signature text, if any, and normalize to
4291          * validated, ready-to-insert wikitext.
4292          * If you have pre-fetched the nickname or the fancySig option, you can
4293          * specify them here to save a database query.
4294          * Do not reuse this parser instance after calling getUserSig(),
4295          * as it may have changed if it's the $wgParser.
4296          *
4297          * @param $user User
4298          * @param $nickname String: nickname to use or false to use user's default nickname
4299          * @param $fancySig Boolean: whether the nicknname is the complete signature
4300          *                  or null to use default value
4301          * @return string
4302          */
4303         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4304                 global $wgMaxSigChars;
4305
4306                 $username = $user->getName();
4307
4308                 # If not given, retrieve from the user object.
4309                 if ( $nickname === false )
4310                         $nickname = $user->getOption( 'nickname' );
4311
4312                 if ( is_null( $fancySig ) ) {
4313                         $fancySig = $user->getBoolOption( 'fancysig' );
4314                 }
4315
4316                 $nickname = $nickname == null ? $username : $nickname;
4317
4318                 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4319                         $nickname = $username;
4320                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4321                 } elseif ( $fancySig !== false ) {
4322                         # Sig. might contain markup; validate this
4323                         if ( $this->validateSig( $nickname ) !== false ) {
4324                                 # Validated; clean up (if needed) and return it
4325                                 return $this->cleanSig( $nickname, true );
4326                         } else {
4327                                 # Failed to validate; fall back to the default
4328                                 $nickname = $username;
4329                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4330                         }
4331                 }
4332
4333                 # Make sure nickname doesnt get a sig in a sig
4334                 $nickname = $this->cleanSigInSig( $nickname );
4335
4336                 # If we're still here, make it a link to the user page
4337                 $userText = wfEscapeWikiText( $username );
4338                 $nickText = wfEscapeWikiText( $nickname );
4339                 if ( $user->isAnon() )  {
4340                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4341                 } else {
4342                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4343                 }
4344         }
4345
4346         /**
4347          * Check that the user's signature contains no bad XML
4348          *
4349          * @param $text String
4350          * @return mixed An expanded string, or false if invalid.
4351          */
4352         function validateSig( $text ) {
4353                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4354         }
4355
4356         /**
4357          * Clean up signature text
4358          *
4359          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4360          * 2) Substitute all transclusions
4361          *
4362          * @param $text String
4363          * @param $parsing Whether we're cleaning (preferences save) or parsing
4364          * @return String: signature text
4365          */
4366         function cleanSig( $text, $parsing = false ) {
4367                 if ( !$parsing ) {
4368                         global $wgTitle;
4369                         $this->mOptions = new ParserOptions;
4370                         $this->clearState();
4371                         $this->setTitle( $wgTitle );
4372                         $this->setOutputType = self::OT_PREPROCESS;
4373                 }
4374
4375                 # Option to disable this feature
4376                 if ( !$this->mOptions->getCleanSignatures() ) {
4377                         return $text;
4378                 }
4379
4380                 # FIXME: regex doesn't respect extension tags or nowiki
4381                 #  => Move this logic to braceSubstitution()
4382                 $substWord = MagicWord::get( 'subst' );
4383                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4384                 $substText = '{{' . $substWord->getSynonym( 0 );
4385
4386                 $text = preg_replace( $substRegex, $substText, $text );
4387                 $text = $this->cleanSigInSig( $text );
4388                 $dom = $this->preprocessToDom( $text );
4389                 $frame = $this->getPreprocessor()->newFrame();
4390                 $text = $frame->expand( $dom );
4391
4392                 if ( !$parsing ) {
4393                         $text = $this->mStripState->unstripBoth( $text );
4394                 }
4395
4396                 return $text;
4397         }
4398
4399         /**
4400          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4401          *
4402          * @param $text String
4403          * @return String: signature text with /~{3,5}/ removed
4404          */
4405         function cleanSigInSig( $text ) {
4406                 $text = preg_replace( '/~{3,5}/', '', $text );
4407                 return $text;
4408         }
4409
4410         /**
4411          * Set up some variables which are usually set up in parse()
4412          * so that an external function can call some class members with confidence
4413          */
4414         public function startExternalParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4415                 $this->startParse( $title, $options, $outputType, $clearState );
4416         }
4417
4418         private function startParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4419                 $this->setTitle( $title );
4420                 $this->mOptions = $options;
4421                 $this->setOutputType( $outputType );
4422                 if ( $clearState ) {
4423                         $this->clearState();
4424                 }
4425         }
4426
4427         /**
4428          * Wrapper for preprocess()
4429          *
4430          * @param $text String: the text to preprocess
4431          * @param $options ParserOptions: options
4432          * @param $title Title object or null to use $wgTitle
4433          * @return String
4434          */
4435         public function transformMsg( $text, $options, $title = null ) {
4436                 static $executing = false;
4437
4438                 # Guard against infinite recursion
4439                 if ( $executing ) {
4440                         return $text;
4441                 }
4442                 $executing = true;
4443
4444                 wfProfileIn( __METHOD__ );
4445                 if ( !$title ) {
4446                         global $wgTitle;
4447                         $title = $wgTitle;
4448                 }
4449                 if ( !$title ) {
4450                         # It's not uncommon having a null $wgTitle in scripts. See r80898
4451                         # Create a ghost title in such case
4452                         $title = Title::newFromText( 'Dwimmerlaik' );
4453                 }
4454                 $text = $this->preprocess( $text, $title, $options );
4455
4456                 $executing = false;
4457                 wfProfileOut( __METHOD__ );
4458                 return $text;
4459         }
4460
4461         /**
4462          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4463          * The callback should have the following form:
4464          *    function myParserHook( $text, $params, $parser, $frame ) { ... }
4465          *
4466          * Transform and return $text. Use $parser for any required context, e.g. use
4467          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4468          *
4469          * Hooks may return extended information by returning an array, of which the
4470          * first numbered element (index 0) must be the return string, and all other
4471          * entries are extracted into local variables within an internal function
4472          * in the Parser class.
4473          *
4474          * This interface (introduced r61913) appears to be undocumented, but
4475          * 'markerName' is used by some core tag hooks to override which strip
4476          * array their results are placed in. **Use great caution if attempting
4477          * this interface, as it is not documented and injudicious use could smash
4478          * private variables.**
4479          *
4480          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4481          * @param $callback Mixed: the callback function (and object) to use for the tag
4482          * @return The old value of the mTagHooks array associated with the hook
4483          */
4484         public function setHook( $tag, $callback ) {
4485                 $tag = strtolower( $tag );
4486                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4487                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4488                 $this->mTagHooks[$tag] = $callback;
4489                 if ( !in_array( $tag, $this->mStripList ) ) {
4490                         $this->mStripList[] = $tag;
4491                 }
4492
4493                 return $oldVal;
4494         }
4495
4496         /**
4497          * As setHook(), but letting the contents be parsed.
4498          *
4499          * Transparent tag hooks are like regular XML-style tag hooks, except they
4500          * operate late in the transformation sequence, on HTML instead of wikitext.
4501          *
4502          * This is probably obsoleted by things dealing with parser frames?
4503          * The only extension currently using it is geoserver.
4504          *
4505          * @since 1.10
4506          * @todo better document or deprecate this
4507          *
4508          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4509          * @param $callback Mixed: the callback function (and object) to use for the tag
4510          * @return The old value of the mTagHooks array associated with the hook
4511          */
4512         function setTransparentTagHook( $tag, $callback ) {
4513                 $tag = strtolower( $tag );
4514                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4515                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4516                 $this->mTransparentTagHooks[$tag] = $callback;
4517
4518                 return $oldVal;
4519         }
4520
4521         /**
4522          * Remove all tag hooks
4523          */
4524         function clearTagHooks() {
4525                 $this->mTagHooks = array();
4526                 $this->mStripList = $this->mDefaultStripList;
4527         }
4528
4529         /**
4530          * Create a function, e.g. {{sum:1|2|3}}
4531          * The callback function should have the form:
4532          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4533          *
4534          * Or with SFH_OBJECT_ARGS:
4535          *    function myParserFunction( $parser, $frame, $args ) { ... }
4536          *
4537          * The callback may either return the text result of the function, or an array with the text
4538          * in element 0, and a number of flags in the other elements. The names of the flags are
4539          * specified in the keys. Valid flags are:
4540          *   found                     The text returned is valid, stop processing the template. This
4541          *                             is on by default.
4542          *   nowiki                    Wiki markup in the return value should be escaped
4543          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4544          *
4545          * @param $id String: The magic word ID
4546          * @param $callback Mixed: the callback function (and object) to use
4547          * @param $flags Integer: a combination of the following flags:
4548          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4549          *
4550          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4551          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4552          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4553          *     the arguments, and to control the way they are expanded.
4554          *
4555          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4556          *     arguments, for instance:
4557          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4558          *
4559          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4560          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4561          *     working if/when this is changed.
4562          *
4563          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4564          *     expansion.
4565          *
4566          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4567          *     about the methods available in PPFrame and PPNode.
4568          *
4569          * @return The old callback function for this name, if any
4570          */
4571         public function setFunctionHook( $id, $callback, $flags = 0 ) {
4572                 global $wgContLang;
4573
4574                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4575                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4576
4577                 # Add to function cache
4578                 $mw = MagicWord::get( $id );
4579                 if ( !$mw )
4580                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4581
4582                 $synonyms = $mw->getSynonyms();
4583                 $sensitive = intval( $mw->isCaseSensitive() );
4584
4585                 foreach ( $synonyms as $syn ) {
4586                         # Case
4587                         if ( !$sensitive ) {
4588                                 $syn = $wgContLang->lc( $syn );
4589                         }
4590                         # Add leading hash
4591                         if ( !( $flags & SFH_NO_HASH ) ) {
4592                                 $syn = '#' . $syn;
4593                         }
4594                         # Remove trailing colon
4595                         if ( substr( $syn, -1, 1 ) === ':' ) {
4596                                 $syn = substr( $syn, 0, -1 );
4597                         }
4598                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4599                 }
4600                 return $oldVal;
4601         }
4602
4603         /**
4604          * Get all registered function hook identifiers
4605          *
4606          * @return Array
4607          */
4608         function getFunctionHooks() {
4609                 return array_keys( $this->mFunctionHooks );
4610         }
4611
4612         /**
4613          * Create a tag function, e.g. <test>some stuff</test>.
4614          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4615          * Unlike parser functions, their content is not preprocessed.
4616          */
4617         function setFunctionTagHook( $tag, $callback, $flags ) {
4618                 $tag = strtolower( $tag );
4619                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4620                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4621                         $this->mFunctionTagHooks[$tag] : null;
4622                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4623
4624                 if ( !in_array( $tag, $this->mStripList ) ) {
4625                         $this->mStripList[] = $tag;
4626                 }
4627
4628                 return $old;
4629         }
4630
4631         /**
4632          * FIXME: update documentation. makeLinkObj() is deprecated.
4633          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4634          * Placeholders created in Skin::makeLinkObj()
4635          * Returns an array of link CSS classes, indexed by PDBK.
4636          */
4637         function replaceLinkHolders( &$text, $options = 0 ) {
4638                 return $this->mLinkHolders->replace( $text );
4639         }
4640
4641         /**
4642          * Replace <!--LINK--> link placeholders with plain text of links
4643          * (not HTML-formatted).
4644          *
4645          * @param $text String
4646          * @return String
4647          */
4648         function replaceLinkHoldersText( $text ) {
4649                 return $this->mLinkHolders->replaceText( $text );
4650         }
4651
4652         /**
4653          * Renders an image gallery from a text with one line per image.
4654          * text labels may be given by using |-style alternative text. E.g.
4655          *   Image:one.jpg|The number "1"
4656          *   Image:tree.jpg|A tree
4657          * given as text will return the HTML of a gallery with two images,
4658          * labeled 'The number "1"' and
4659          * 'A tree'.
4660          *
4661          * @param string $text
4662          * @param array $param
4663          * @return string HTML
4664          */
4665         function renderImageGallery( $text, $params ) {
4666                 $ig = new ImageGallery();
4667                 $ig->setContextTitle( $this->mTitle );
4668                 $ig->setShowBytes( false );
4669                 $ig->setShowFilename( false );
4670                 $ig->setParser( $this );
4671                 $ig->setHideBadImages();
4672                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4673
4674                 if ( isset( $params['showfilename'] ) ) {
4675                         $ig->setShowFilename( true );
4676                 } else {
4677                         $ig->setShowFilename( false );
4678                 }
4679                 if ( isset( $params['caption'] ) ) {
4680                         $caption = $params['caption'];
4681                         $caption = htmlspecialchars( $caption );
4682                         $caption = $this->replaceInternalLinks( $caption );
4683                         $ig->setCaptionHtml( $caption );
4684                 }
4685                 if ( isset( $params['perrow'] ) ) {
4686                         $ig->setPerRow( $params['perrow'] );
4687                 }
4688                 if ( isset( $params['widths'] ) ) {
4689                         $ig->setWidths( $params['widths'] );
4690                 }
4691                 if ( isset( $params['heights'] ) ) {
4692                         $ig->setHeights( $params['heights'] );
4693                 }
4694
4695                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4696
4697                 $lines = StringUtils::explode( "\n", $text );
4698                 foreach ( $lines as $line ) {
4699                         # match lines like these:
4700                         # Image:someimage.jpg|This is some image
4701                         $matches = array();
4702                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4703                         # Skip empty lines
4704                         if ( count( $matches ) == 0 ) {
4705                                 continue;
4706                         }
4707
4708                         if ( strpos( $matches[0], '%' ) !== false ) {
4709                                 $matches[1] = rawurldecode( $matches[1] );
4710                         }
4711                         $title = Title::newFromText( $matches[1], NS_FILE );
4712                         if ( is_null( $title ) ) {
4713                                 # Bogus title. Ignore these so we don't bomb out later.
4714                                 continue;
4715                         }
4716
4717                         $label = '';
4718                         $alt = '';
4719                         if ( isset( $matches[3] ) ) {
4720                                 // look for an |alt= definition while trying not to break existing
4721                                 // captions with multiple pipes (|) in it, until a more sensible grammar
4722                                 // is defined for images in galleries
4723
4724                                 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
4725                                 $altmatches = StringUtils::explode('|', $matches[3]);
4726                                 $magicWordAlt = MagicWord::get( 'img_alt' );
4727
4728                                 foreach ( $altmatches as $altmatch ) {
4729                                         $match = $magicWordAlt->matchVariableStartToEnd( $altmatch );
4730                                         if ( $match ) {
4731                                                 $alt = $this->stripAltText( $match, false );
4732                                         }
4733                                         else {
4734                                                 // concatenate all other pipes
4735                                                 $label .= '|' . $altmatch;
4736                                         }
4737                                 }
4738                                 // remove the first pipe
4739                                 $label = substr( $label, 1 );
4740                         }
4741
4742                         $ig->add( $title, $label, $alt );
4743                 }
4744                 return $ig->toHTML();
4745         }
4746
4747         function getImageParams( $handler ) {
4748                 if ( $handler ) {
4749                         $handlerClass = get_class( $handler );
4750                 } else {
4751                         $handlerClass = '';
4752                 }
4753                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4754                         # Initialise static lists
4755                         static $internalParamNames = array(
4756                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4757                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4758                                         'bottom', 'text-bottom' ),
4759                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4760                                         'upright', 'border', 'link', 'alt' ),
4761                         );
4762                         static $internalParamMap;
4763                         if ( !$internalParamMap ) {
4764                                 $internalParamMap = array();
4765                                 foreach ( $internalParamNames as $type => $names ) {
4766                                         foreach ( $names as $name ) {
4767                                                 $magicName = str_replace( '-', '_', "img_$name" );
4768                                                 $internalParamMap[$magicName] = array( $type, $name );
4769                                         }
4770                                 }
4771                         }
4772
4773                         # Add handler params
4774                         $paramMap = $internalParamMap;
4775                         if ( $handler ) {
4776                                 $handlerParamMap = $handler->getParamMap();
4777                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4778                                         $paramMap[$magic] = array( 'handler', $paramName );
4779                                 }
4780                         }
4781                         $this->mImageParams[$handlerClass] = $paramMap;
4782                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4783                 }
4784                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4785         }
4786
4787         /**
4788          * Parse image options text and use it to make an image
4789          *
4790          * @param $title Title
4791          * @param $options String
4792          * @param $holders LinkHolderArray
4793          * @return string HTML
4794          */
4795         function makeImage( $title, $options, $holders = false ) {
4796                 # Check if the options text is of the form "options|alt text"
4797                 # Options are:
4798                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4799                 #  * left       no resizing, just left align. label is used for alt= only
4800                 #  * right      same, but right aligned
4801                 #  * none       same, but not aligned
4802                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4803                 #  * center     center the image
4804                 #  * frame      Keep original image size, no magnify-button.
4805                 #  * framed     Same as "frame"
4806                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4807                 #  * upright    reduce width for upright images, rounded to full __0 px
4808                 #  * border     draw a 1px border around the image
4809                 #  * alt        Text for HTML alt attribute (defaults to empty)
4810                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4811                 # vertical-align values (no % or length right now):
4812                 #  * baseline
4813                 #  * sub
4814                 #  * super
4815                 #  * top
4816                 #  * text-top
4817                 #  * middle
4818                 #  * bottom
4819                 #  * text-bottom
4820
4821                 $parts = StringUtils::explode( "|", $options );
4822
4823                 # Give extensions a chance to select the file revision for us
4824                 $time = $sha1 = $descQuery = false;
4825                 wfRunHooks( 'BeforeParserFetchFileAndTitle',
4826                         array( $this, $title, &$time, &$sha1, &$descQuery ) );
4827                 # Fetch and register the file (file title may be different via hooks)
4828                 list( $file, $title ) = $this->fetchFileAndTitle( $title, $time, $sha1 );
4829
4830                 # Get parameter map
4831                 $handler = $file ? $file->getHandler() : false;
4832
4833                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4834
4835                 if ( !$file ) {
4836                         $this->addTrackingCategory( 'broken-file-category' );
4837                 }
4838
4839                 # Process the input parameters
4840                 $caption = '';
4841                 $params = array( 'frame' => array(), 'handler' => array(),
4842                         'horizAlign' => array(), 'vertAlign' => array() );
4843                 foreach ( $parts as $part ) {
4844                         $part = trim( $part );
4845                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4846                         $validated = false;
4847                         if ( isset( $paramMap[$magicName] ) ) {
4848                                 list( $type, $paramName ) = $paramMap[$magicName];
4849
4850                                 # Special case; width and height come in one variable together
4851                                 if ( $type === 'handler' && $paramName === 'width' ) {
4852                                         $m = array();
4853                                         # (bug 13500) In both cases (width/height and width only),
4854                                         # permit trailing "px" for backward compatibility.
4855                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4856                                                 $width = intval( $m[1] );
4857                                                 $height = intval( $m[2] );
4858                                                 if ( $handler->validateParam( 'width', $width ) ) {
4859                                                         $params[$type]['width'] = $width;
4860                                                         $validated = true;
4861                                                 }
4862                                                 if ( $handler->validateParam( 'height', $height ) ) {
4863                                                         $params[$type]['height'] = $height;
4864                                                         $validated = true;
4865                                                 }
4866                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4867                                                 $width = intval( $value );
4868                                                 if ( $handler->validateParam( 'width', $width ) ) {
4869                                                         $params[$type]['width'] = $width;
4870                                                         $validated = true;
4871                                                 }
4872                                         } # else no validation -- bug 13436
4873                                 } else {
4874                                         if ( $type === 'handler' ) {
4875                                                 # Validate handler parameter
4876                                                 $validated = $handler->validateParam( $paramName, $value );
4877                                         } else {
4878                                                 # Validate internal parameters
4879                                                 switch( $paramName ) {
4880                                                 case 'manualthumb':
4881                                                 case 'alt':
4882                                                         # @todo Fixme: possibly check validity here for
4883                                                         # manualthumb? downstream behavior seems odd with
4884                                                         # missing manual thumbs.
4885                                                         $validated = true;
4886                                                         $value = $this->stripAltText( $value, $holders );
4887                                                         break;
4888                                                 case 'link':
4889                                                         $chars = self::EXT_LINK_URL_CLASS;
4890                                                         $prots = $this->mUrlProtocols;
4891                                                         if ( $value === '' ) {
4892                                                                 $paramName = 'no-link';
4893                                                                 $value = true;
4894                                                                 $validated = true;
4895                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4896                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4897                                                                         $paramName = 'link-url';
4898                                                                         $this->mOutput->addExternalLink( $value );
4899                                                                         if ( $this->mOptions->getExternalLinkTarget() ) {
4900                                                                                 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
4901                                                                         }
4902                                                                         $validated = true;
4903                                                                 }
4904                                                         } else {
4905                                                                 $linkTitle = Title::newFromText( $value );
4906                                                                 if ( $linkTitle ) {
4907                                                                         $paramName = 'link-title';
4908                                                                         $value = $linkTitle;
4909                                                                         $this->mOutput->addLink( $linkTitle );
4910                                                                         $validated = true;
4911                                                                 }
4912                                                         }
4913                                                         break;
4914                                                 default:
4915                                                         # Most other things appear to be empty or numeric...
4916                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4917                                                 }
4918                                         }
4919
4920                                         if ( $validated ) {
4921                                                 $params[$type][$paramName] = $value;
4922                                         }
4923                                 }
4924                         }
4925                         if ( !$validated ) {
4926                                 $caption = $part;
4927                         }
4928                 }
4929
4930                 # Process alignment parameters
4931                 if ( $params['horizAlign'] ) {
4932                         $params['frame']['align'] = key( $params['horizAlign'] );
4933                 }
4934                 if ( $params['vertAlign'] ) {
4935                         $params['frame']['valign'] = key( $params['vertAlign'] );
4936                 }
4937
4938                 $params['frame']['caption'] = $caption;
4939
4940                 # Will the image be presented in a frame, with the caption below?
4941                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4942                                                  isset( $params['frame']['framed'] ) ||
4943                                                  isset( $params['frame']['thumbnail'] ) ||
4944                                                  isset( $params['frame']['manualthumb'] );
4945
4946                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4947                 # came to also set the caption, ordinary text after the image -- which
4948                 # makes no sense, because that just repeats the text multiple times in
4949                 # screen readers.  It *also* came to set the title attribute.
4950                 #
4951                 # Now that we have an alt attribute, we should not set the alt text to
4952                 # equal the caption: that's worse than useless, it just repeats the
4953                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4954                 # use the unnamed parameter for alt text as well, just for the time be-
4955                 # ing, if the unnamed param is set and the alt param is not.
4956                 #
4957                 # For the future, we need to figure out if we want to tweak this more,
4958                 # e.g., introducing a title= parameter for the title; ignoring the un-
4959                 # named parameter entirely for images without a caption; adding an ex-
4960                 # plicit caption= parameter and preserving the old magic unnamed para-
4961                 # meter for BC; ...
4962                 if ( $imageIsFramed ) { # Framed image
4963                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4964                                 # No caption or alt text, add the filename as the alt text so
4965                                 # that screen readers at least get some description of the image
4966                                 $params['frame']['alt'] = $title->getText();
4967                         }
4968                         # Do not set $params['frame']['title'] because tooltips don't make sense
4969                         # for framed images
4970                 } else { # Inline image
4971                         if ( !isset( $params['frame']['alt'] ) ) {
4972                                 # No alt text, use the "caption" for the alt text
4973                                 if ( $caption !== '') {
4974                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4975                                 } else {
4976                                         # No caption, fall back to using the filename for the
4977                                         # alt text
4978                                         $params['frame']['alt'] = $title->getText();
4979                                 }
4980                         }
4981                         # Use the "caption" for the tooltip text
4982                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4983                 }
4984
4985                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4986
4987                 # Linker does the rest
4988                 $ret = Linker::makeImageLink2( $title, $file, $params['frame'], $params['handler'],
4989                         $time, $descQuery, $this->mOptions->getThumbSize() );
4990
4991                 # Give the handler a chance to modify the parser object
4992                 if ( $handler ) {
4993                         $handler->parserTransformHook( $this, $file );
4994                 }
4995
4996                 return $ret;
4997         }
4998
4999         protected function stripAltText( $caption, $holders ) {
5000                 # Strip bad stuff out of the title (tooltip).  We can't just use
5001                 # replaceLinkHoldersText() here, because if this function is called
5002                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5003                 if ( $holders ) {
5004                         $tooltip = $holders->replaceText( $caption );
5005                 } else {
5006                         $tooltip = $this->replaceLinkHoldersText( $caption );
5007                 }
5008
5009                 # make sure there are no placeholders in thumbnail attributes
5010                 # that are later expanded to html- so expand them now and
5011                 # remove the tags
5012                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5013                 $tooltip = Sanitizer::stripAllTags( $tooltip );
5014
5015                 return $tooltip;
5016         }
5017
5018         /**
5019          * Set a flag in the output object indicating that the content is dynamic and
5020          * shouldn't be cached.
5021          */
5022         function disableCache() {
5023                 wfDebug( "Parser output marked as uncacheable.\n" );
5024                 $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
5025                 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5026         }
5027
5028         /**
5029          * Callback from the Sanitizer for expanding items found in HTML attribute
5030          * values, so they can be safely tested and escaped.
5031          *
5032          * @param $text String
5033          * @param $frame PPFrame
5034          * @return String
5035          */
5036         function attributeStripCallback( &$text, $frame = false ) {
5037                 $text = $this->replaceVariables( $text, $frame );
5038                 $text = $this->mStripState->unstripBoth( $text );
5039                 return $text;
5040         }
5041
5042         /**
5043          * Accessor
5044          *
5045          * @return array
5046          */
5047         function getTags() {
5048                 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ) );
5049         }
5050
5051         /**
5052          * Replace transparent tags in $text with the values given by the callbacks.
5053          *
5054          * Transparent tag hooks are like regular XML-style tag hooks, except they
5055          * operate late in the transformation sequence, on HTML instead of wikitext.
5056          */
5057         function replaceTransparentTags( $text ) {
5058                 $matches = array();
5059                 $elements = array_keys( $this->mTransparentTagHooks );
5060                 $text = self::extractTagsAndParams( $elements, $text, $matches, $this->mUniqPrefix );
5061
5062                 foreach ( $matches as $marker => $data ) {
5063                         list( $element, $content, $params, $tag ) = $data;
5064                         $tagName = strtolower( $element );
5065                         if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5066                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
5067                         } else {
5068                                 $output = $tag;
5069                         }
5070                         $this->mStripState->addGeneral( $marker, $output );
5071                 }
5072                 return $text;
5073         }
5074
5075         /**
5076          * Break wikitext input into sections, and either pull or replace
5077          * some particular section's text.
5078          *
5079          * External callers should use the getSection and replaceSection methods.
5080          *
5081          * @param $text String: Page wikitext
5082          * @param $section String: a section identifier string of the form:
5083          *   <flag1> - <flag2> - ... - <section number>
5084          *
5085          * Currently the only recognised flag is "T", which means the target section number
5086          * was derived during a template inclusion parse, in other words this is a template
5087          * section edit link. If no flags are given, it was an ordinary section edit link.
5088          * This flag is required to avoid a section numbering mismatch when a section is
5089          * enclosed by <includeonly> (bug 6563).
5090          *
5091          * The section number 0 pulls the text before the first heading; other numbers will
5092          * pull the given section along with its lower-level subsections. If the section is
5093          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5094          *
5095          * @param $mode String: one of "get" or "replace"
5096          * @param $newText String: replacement text for section data.
5097          * @return String: for "get", the extracted section text.
5098          *                 for "replace", the whole page with the section replaced.
5099          */
5100         private function extractSections( $text, $section, $mode, $newText='' ) {
5101                 global $wgTitle; # not generally used but removes an ugly failure mode
5102                 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5103                 $outText = '';
5104                 $frame = $this->getPreprocessor()->newFrame();
5105
5106                 # Process section extraction flags
5107                 $flags = 0;
5108                 $sectionParts = explode( '-', $section );
5109                 $sectionIndex = array_pop( $sectionParts );
5110                 foreach ( $sectionParts as $part ) {
5111                         if ( $part === 'T' ) {
5112                                 $flags |= self::PTD_FOR_INCLUSION;
5113                         }
5114                 }
5115                 # Preprocess the text
5116                 $root = $this->preprocessToDom( $text, $flags );
5117
5118                 # <h> nodes indicate section breaks
5119                 # They can only occur at the top level, so we can find them by iterating the root's children
5120                 $node = $root->getFirstChild();
5121
5122                 # Find the target section
5123                 if ( $sectionIndex == 0 ) {
5124                         # Section zero doesn't nest, level=big
5125                         $targetLevel = 1000;
5126                         if ( !$node ) {
5127                                 # The page definitely exists - we checked that earlier - so it must be blank: see bug #14005
5128                                 return $text;
5129                         }
5130                 } else {
5131                         while ( $node ) {
5132                                 if ( $node->getName() === 'h' ) {
5133                                         $bits = $node->splitHeading();
5134                                         if ( $bits['i'] == $sectionIndex ) {
5135                                                 $targetLevel = $bits['level'];
5136                                                 break;
5137                                         }
5138                                 }
5139                                 if ( $mode === 'replace' ) {
5140                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5141                                 }
5142                                 $node = $node->getNextSibling();
5143                         }
5144                 }
5145
5146                 if ( !$node ) {
5147                         # Not found
5148                         if ( $mode === 'get' ) {
5149                                 return $newText;
5150                         } else {
5151                                 return $text;
5152                         }
5153                 }
5154
5155                 # Find the end of the section, including nested sections
5156                 do {
5157                         if ( $node->getName() === 'h' ) {
5158                                 $bits = $node->splitHeading();
5159                                 $curLevel = $bits['level'];
5160                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5161                                         break;
5162                                 }
5163                         }
5164                         if ( $mode === 'get' ) {
5165                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5166                         }
5167                         $node = $node->getNextSibling();
5168                 } while ( $node );
5169
5170                 # Write out the remainder (in replace mode only)
5171                 if ( $mode === 'replace' ) {
5172                         # Output the replacement text
5173                         # Add two newlines on -- trailing whitespace in $newText is conventionally
5174                         # stripped by the editor, so we need both newlines to restore the paragraph gap
5175                         # Only add trailing whitespace if there is newText
5176                         if ( $newText != "" ) {
5177                                 $outText .= $newText . "\n\n";
5178                         }
5179
5180                         while ( $node ) {
5181                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5182                                 $node = $node->getNextSibling();
5183                         }
5184                 }
5185
5186                 if ( is_string( $outText ) ) {
5187                         # Re-insert stripped tags
5188                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5189                 }
5190
5191                 return $outText;
5192         }
5193
5194         /**
5195          * This function returns the text of a section, specified by a number ($section).
5196          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5197          * the first section before any such heading (section 0).
5198          *
5199          * If a section contains subsections, these are also returned.
5200          *
5201          * @param $text String: text to look in
5202          * @param $section String: section identifier
5203          * @param $deftext String: default to return if section is not found
5204          * @return string text of the requested section
5205          */
5206         public function getSection( $text, $section, $deftext='' ) {
5207                 return $this->extractSections( $text, $section, "get", $deftext );
5208         }
5209
5210         /**
5211          * This function returns $oldtext after the content of the section
5212          * specified by $section has been replaced with $text.
5213          *
5214          * @param $oldtext String: former text of the article
5215          * @param $section Numeric: section identifier
5216          * @param $text String: replacing text
5217          * @return String: modified text
5218          */
5219         public function replaceSection( $oldtext, $section, $text ) {
5220                 return $this->extractSections( $oldtext, $section, "replace", $text );
5221         }
5222
5223         /**
5224          * Get the ID of the revision we are parsing
5225          *
5226          * @return Mixed: integer or null
5227          */
5228         function getRevisionId() {
5229                 return $this->mRevisionId;
5230         }
5231
5232         /**
5233          * Get the revision object for $this->mRevisionId
5234          *
5235          * @return Revision|null either a Revision object or null
5236          */
5237         protected function getRevisionObject() {
5238                 if ( !is_null( $this->mRevisionObject ) ) {
5239                         return $this->mRevisionObject;
5240                 }
5241                 if ( is_null( $this->mRevisionId ) ) {
5242                         return null;
5243                 }
5244
5245                 $this->mRevisionObject = Revision::newFromId( $this->mRevisionId );
5246                 return $this->mRevisionObject;
5247         }
5248
5249         /**
5250          * Get the timestamp associated with the current revision, adjusted for
5251          * the default server-local timestamp
5252          */
5253         function getRevisionTimestamp() {
5254                 if ( is_null( $this->mRevisionTimestamp ) ) {
5255                         wfProfileIn( __METHOD__ );
5256
5257                         $revObject = $this->getRevisionObject();
5258                         $timestamp = $revObject ? $revObject->getTimestamp() : false;
5259
5260                         if( $timestamp !== false ) {
5261                                 global $wgContLang;
5262
5263                                 # The cryptic '' timezone parameter tells to use the site-default
5264                                 # timezone offset instead of the user settings.
5265                                 #
5266                                 # Since this value will be saved into the parser cache, served
5267                                 # to other users, and potentially even used inside links and such,
5268                                 # it needs to be consistent for all visitors.
5269                                 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5270                         }
5271
5272                         wfProfileOut( __METHOD__ );
5273                 }
5274                 return $this->mRevisionTimestamp;
5275         }
5276
5277         /**
5278          * Get the name of the user that edited the last revision
5279          *
5280          * @return String: user name
5281          */
5282         function getRevisionUser() {
5283                 if( is_null( $this->mRevisionUser ) ) {
5284                         $revObject = $this->getRevisionObject();
5285
5286                         # if this template is subst: the revision id will be blank,
5287                         # so just use the current user's name
5288                         if( $revObject ) {
5289                                 $this->mRevisionUser = $revObject->getUserText();
5290                         } elseif( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5291                                 $this->mRevisionUser = $this->getUser()->getName();
5292                         }
5293                 }
5294                 return $this->mRevisionUser;
5295         }
5296
5297         /**
5298          * Mutator for $mDefaultSort
5299          *
5300          * @param $sort New value
5301          */
5302         public function setDefaultSort( $sort ) {
5303                 $this->mDefaultSort = $sort;
5304                 $this->mOutput->setProperty( 'defaultsort', $sort );
5305         }
5306
5307         /**
5308          * Accessor for $mDefaultSort
5309          * Will use the empty string if none is set.
5310          *
5311          * This value is treated as a prefix, so the
5312          * empty string is equivalent to sorting by
5313          * page name.
5314          *
5315          * @return string
5316          */
5317         public function getDefaultSort() {
5318                 if ( $this->mDefaultSort !== false ) {
5319                         return $this->mDefaultSort;
5320                 } else {
5321                         return '';
5322                 }
5323         }
5324
5325         /**
5326          * Accessor for $mDefaultSort
5327          * Unlike getDefaultSort(), will return false if none is set
5328          *
5329          * @return string or false
5330          */
5331         public function getCustomDefaultSort() {
5332                 return $this->mDefaultSort;
5333         }
5334
5335         /**
5336          * Try to guess the section anchor name based on a wikitext fragment
5337          * presumably extracted from a heading, for example "Header" from
5338          * "== Header ==".
5339          */
5340         public function guessSectionNameFromWikiText( $text ) {
5341                 # Strip out wikitext links(they break the anchor)
5342                 $text = $this->stripSectionName( $text );
5343                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5344                 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5345         }
5346
5347         /**
5348          * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5349          * instead.  For use in redirects, since IE6 interprets Redirect: headers
5350          * as something other than UTF-8 (apparently?), resulting in breakage.
5351          *
5352          * @param $text String: The section name
5353          * @return string An anchor
5354          */
5355         public function guessLegacySectionNameFromWikiText( $text ) {
5356                 # Strip out wikitext links(they break the anchor)
5357                 $text = $this->stripSectionName( $text );
5358                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5359                 return '#' . Sanitizer::escapeId( $text, array( 'noninitial', 'legacy' ) );
5360         }
5361
5362         /**
5363          * Strips a text string of wikitext for use in a section anchor
5364          *
5365          * Accepts a text string and then removes all wikitext from the
5366          * string and leaves only the resultant text (i.e. the result of
5367          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5368          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5369          * to create valid section anchors by mimicing the output of the
5370          * parser when headings are parsed.
5371          *
5372          * @param $text String: text string to be stripped of wikitext
5373          * for use in a Section anchor
5374          * @return Filtered text string
5375          */
5376         public function stripSectionName( $text ) {
5377                 # Strip internal link markup
5378                 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5379                 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5380
5381                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5382                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5383                 # on how many empty links there are on the page - need to figure that out.
5384                 $text = preg_replace( '/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5385
5386                 # Parse wikitext quotes (italics & bold)
5387                 $text = $this->doQuotes( $text );
5388
5389                 # Strip HTML tags
5390                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5391                 return $text;
5392         }
5393
5394         /**
5395          * strip/replaceVariables/unstrip for preprocessor regression testing
5396          *
5397          * @return string
5398          */
5399         function testSrvus( $text, $title, ParserOptions $options, $outputType = self::OT_HTML ) {
5400                 if ( !$title instanceof Title ) {
5401                         $title = Title::newFromText( $title );
5402                 }
5403                 $this->startParse( $title, $options, $outputType, true );
5404
5405                 $text = $this->replaceVariables( $text );
5406                 $text = $this->mStripState->unstripBoth( $text );
5407                 $text = Sanitizer::removeHTMLtags( $text );
5408                 return $text;
5409         }
5410
5411         function testPst( $text, $title, $options ) {
5412                 global $wgUser;
5413                 if ( !$title instanceof Title ) {
5414                         $title = Title::newFromText( $title );
5415                 }
5416                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5417         }
5418
5419         function testPreprocess( $text, $title, $options ) {
5420                 if ( !$title instanceof Title ) {
5421                         $title = Title::newFromText( $title );
5422                 }
5423                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5424         }
5425
5426         /**
5427          * Call a callback function on all regions of the given text that are not
5428          * inside strip markers, and replace those regions with the return value
5429          * of the callback. For example, with input:
5430          *
5431          *  aaa<MARKER>bbb
5432          *
5433          * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5434          * two strings will be replaced with the value returned by the callback in
5435          * each case.
5436          *
5437          * @return string
5438          */
5439         function markerSkipCallback( $s, $callback ) {
5440                 $i = 0;
5441                 $out = '';
5442                 while ( $i < strlen( $s ) ) {
5443                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5444                         if ( $markerStart === false ) {
5445                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5446                                 break;
5447                         } else {
5448                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5449                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5450                                 if ( $markerEnd === false ) {
5451                                         $out .= substr( $s, $markerStart );
5452                                         break;
5453                                 } else {
5454                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5455                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5456                                         $i = $markerEnd;
5457                                 }
5458                         }
5459                 }
5460                 return $out;
5461         }
5462
5463         /**
5464          * Save the parser state required to convert the given half-parsed text to
5465          * HTML. "Half-parsed" in this context means the output of
5466          * recursiveTagParse() or internalParse(). This output has strip markers
5467          * from replaceVariables (extensionSubstitution() etc.), and link
5468          * placeholders from replaceLinkHolders().
5469          *
5470          * Returns an array which can be serialized and stored persistently. This
5471          * array can later be loaded into another parser instance with
5472          * unserializeHalfParsedText(). The text can then be safely incorporated into
5473          * the return value of a parser hook.
5474          *
5475          * @return array
5476          */
5477         function serializeHalfParsedText( $text ) {
5478                 wfProfileIn( __METHOD__ );
5479                 $data = array(
5480                         'text' => $text,
5481                         'version' => self::HALF_PARSED_VERSION,
5482                         'stripState' => $this->mStripState->getSubState( $text ),
5483                         'linkHolders' => $this->mLinkHolders->getSubArray( $text )
5484                 );
5485                 wfProfileOut( __METHOD__ );
5486                 return $data;
5487         }
5488
5489         /**
5490          * Load the parser state given in the $data array, which is assumed to
5491          * have been generated by serializeHalfParsedText(). The text contents is
5492          * extracted from the array, and its markers are transformed into markers
5493          * appropriate for the current Parser instance. This transformed text is
5494          * returned, and can be safely included in the return value of a parser
5495          * hook.
5496          *
5497          * If the $data array has been stored persistently, the caller should first
5498          * check whether it is still valid, by calling isValidHalfParsedText().
5499          *
5500          * @param $data Serialized data
5501          * @return String
5502          */
5503         function unserializeHalfParsedText( $data ) {
5504                 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
5505                         throw new MWException( __METHOD__.': invalid version' );
5506                 }
5507
5508                 # First, extract the strip state.
5509                 $texts = array( $data['text'] );
5510                 $texts = $this->mStripState->merge( $data['stripState'], $texts );
5511
5512                 # Now renumber links
5513                 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
5514
5515                 # Should be good to go.
5516                 return $texts[0];
5517         }
5518
5519         /**
5520          * Returns true if the given array, presumed to be generated by
5521          * serializeHalfParsedText(), is compatible with the current version of the
5522          * parser.
5523          *
5524          * @param $data Array
5525          *
5526          * @return bool
5527          */
5528         function isValidHalfParsedText( $data ) {
5529                 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
5530         }
5531 }