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