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