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                         unset( $this->mLinkHolders );
 186                 }
 187                 foreach ( $this as $name => $value ) {
 188                         unset( $this->$name );
 189                 }
 190         }
 191
 192         /**
 193          * Do various kinds of initialisation on the first call of the parser
 194          */
 195         function firstCallInit() {
 196                 if ( !$this->mFirstCall ) {
 197                         return;
 198                 }
 199                 $this->mFirstCall = false;
 200
 201                 wfProfileIn( __METHOD__ );
 202
 203                 CoreParserFunctions::register( $this );
 204                 CoreTagHooks::register( $this );
 205                 $this->initialiseVariables();
 206
 207                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 208                 wfProfileOut( __METHOD__ );
 209         }
 210
 211         /**
 212          * Clear Parser state
 213          *
 214          * @private
 215          */
 216         function clearState() {
 217                 wfProfileIn( __METHOD__ );
 218                 if ( $this->mFirstCall ) {
 219                         $this->firstCallInit();
 220                 }
 221                 $this->mOutput = new ParserOutput;
 222                 $this->mOptions->registerWatcher( array( $this->mOutput, 'recordOption' ) );
 223                 $this->mAutonumber = 0;
 224                 $this->mLastSection = '';
 225                 $this->mDTopen = false;
 226                 $this->mIncludeCount = array();
 227                 $this->mArgStack = false;
 228                 $this->mInPre = false;
 229                 $this->mLinkHolders = new LinkHolderArray( $this );
 230                 $this->mLinkID = 0;
 231                 $this->mRevisionObject = $this->mRevisionTimestamp =
 232                         $this->mRevisionId = $this->mRevisionUser = null;
 233                 $this->mVarCache = array();
 234                 $this->mUser = null;
 235
 236                 /**
 237                  * Prefix for temporary replacement strings for the multipass parser.
 238                  * \x07 should never appear in input as it's disallowed in XML.
 239                  * Using it at the front also gives us a little extra robustness
 240                  * since it shouldn't match when butted up against identifier-like
 241                  * string constructs.
 242                  *
 243                  * Must not consist of all title characters, or else it will change
 244                  * the behaviour of <nowiki> in a link.
 245                  */
 246                 # $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 247                 # Changed to \x7f to allow XML double-parsing -- TS
 248                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 249                 $this->mStripState = new StripState( $this->mUniqPrefix );
 250
 251
 252                 # Clear these on every parse, bug 4549
 253                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 254
 255                 $this->mShowToc = true;
 256                 $this->mForceTocPosition = false;
 257                 $this->mIncludeSizes = array(
 258                         'post-expand' => 0,
 259                         'arg' => 0,
 260                 );
 261                 $this->mPPNodeCount = 0;
 262                 $this->mDefaultSort = false;
 263                 $this->mHeadings = array();
 264                 $this->mDoubleUnderscores = array();
 265                 $this->mExpensiveFunctionCount = 0;
 266
 267                 # Fix cloning
 268                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 269                         $this->mPreprocessor = null;
 270                 }
 271
 272                 wfRunHooks( 'ParserClearState', array( &$this ) );
 273                 wfProfileOut( __METHOD__ );
 274         }
 275
 276         /**
 277          * Convert wikitext to HTML
 278          * Do not call this function recursively.
 279          *
 280          * @param $text String: text we want to parse
 281          * @param $title Title object
 282          * @param $options ParserOptions
 283          * @param $linestart boolean
 284          * @param $clearState boolean
 285          * @param $revid Int: number to pass in {{REVISIONID}}
 286          * @return ParserOutput a ParserOutput
 287          */
 288         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 289                 /**
 290                  * First pass--just handle <nowiki> sections, pass the rest off
 291                  * to internalParse() which does all the real work.
 292                  */
 293
 294                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang, $wgDisableLangConversion, $wgDisableTitleConversion;
 295                 $fname = __METHOD__.'-' . wfGetCaller();
 296                 wfProfileIn( __METHOD__ );
 297                 wfProfileIn( $fname );
 298
 299                 $this->startParse( $title, $options, self::OT_HTML, $clearState );
 300
 301                 $oldRevisionId = $this->mRevisionId;
 302                 $oldRevisionObject = $this->mRevisionObject;
 303                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 304                 $oldRevisionUser = $this->mRevisionUser;
 305                 if ( $revid !== null ) {
 306                         $this->mRevisionId = $revid;
 307                         $this->mRevisionObject = null;
 308                         $this->mRevisionTimestamp = null;
 309                         $this->mRevisionUser = null;
 310                 }
 311
 312                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 313                 # No more strip!
 314                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 315                 $text = $this->internalParse( $text );
 316
 317                 $text = $this->mStripState->unstripGeneral( $text );
 318
 319                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 320                 $fixtags = array(
 321                         # french spaces, last one Guillemet-left
 322                         # only if there is something before the space
 323                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;\\2',
 324                         # french spaces, Guillemet-right
 325                         '/(\\302\\253) /' => '\\1&#160;',
 326                         '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
 327                 );
 328                 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
 329
 330                 $text = $this->doBlockLevels( $text, $linestart );
 331
 332                 $this->replaceLinkHolders( $text );
 333
 334                 /**
 335                  * The page doesn't get language converted if
 336                  * a) It's disabled
 337                  * b) Content isn't converted
 338                  * c) It's a conversion table
 339                  */
 340                 if ( !( $wgDisableLangConversion
 341                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 342                                 || $this->mTitle->isConversionTable() ) ) {
 343
 344                         # The position of the convert() call should not be changed. it
 345                         # assumes that the links are all replaced and the only thing left
 346                         # is the <nowiki> mark.
 347
 348                         $text = $wgContLang->convert( $text );
 349                 }
 350
 351                 /**
 352                  * A converted title will be provided in the output object if title and
 353                  * content conversion are enabled, the article text does not contain
 354                  * a conversion-suppressing double-underscore tag, and no
 355                  * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
 356                  * automatic link conversion.
 357                  */
 358                 if ( !( $wgDisableLangConversion
 359                                 || $wgDisableTitleConversion
 360                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 361                                 || isset( $this->mDoubleUnderscores['notitleconvert'] )
 362                                 || $this->mOutput->getDisplayTitle() !== false ) )
 363                 {
 364                         $convruletitle = $wgContLang->getConvRuleTitle();
 365                         if ( $convruletitle ) {
 366                                 $this->mOutput->setTitleText( $convruletitle );
 367                         } else {
 368                                 $titleText = $wgContLang->convertTitle( $title );
 369                                 $this->mOutput->setTitleText( $titleText );
 370                         }
 371                 }
 372
 373                 $text = $this->mStripState->unstripNoWiki( $text );
 374
 375                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 376
 377                 $text = $this->replaceTransparentTags( $text );
 378                 $text = $this->mStripState->unstripGeneral( $text );
 379
 380                 $text = Sanitizer::normalizeCharReferences( $text );
 381
 382                 if ( ( $wgUseTidy && $this->mOptions->getTidy() ) || $wgAlwaysUseTidy ) {
 383                         $text = MWTidy::tidy( $text );
 384                 } else {
 385                         # attempt to sanitize at least some nesting problems
 386                         # (bug #2702 and quite a few others)
 387                         $tidyregs = array(
 388                                 # ''Something [http://www.cool.com cool''] -->
 389                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 390                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 391                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 392                                 # fix up an anchor inside another anchor, only
 393                                 # at least for a single single nested link (bug 3695)
 394                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 395                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 396                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 397                                 # contains a div, so fix it up here; replace
 398                                 # div with escaped text
 399                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 400                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 401                                 # remove empty italic or bold tag pairs, some
 402                                 # introduced by rules above
 403                                 '/<([bi])><\/\\1>/' => '',
 404                         );
 405
 406                         $text = preg_replace(
 407                                 array_keys( $tidyregs ),
 408                                 array_values( $tidyregs ),
 409                                 $text );
 410                 }
 411                 global $wgExpensiveParserFunctionLimit;
 412                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 413                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 414                 }
 415
 416                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 417
 418                 # Information on include size limits, for the benefit of users who try to skirt them
 419                 if ( $this->mOptions->getEnableLimitReport() ) {
 420                         $max = $this->mOptions->getMaxIncludeSize();
 421                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 422                         $limitReport =
 423                                 "NewPP limit report\n" .
 424                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->getMaxPPNodeCount()}\n" .
 425                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 426                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 427                                 $PFreport;
 428                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 429                         $text .= "\n<!-- \n$limitReport-->\n";
 430                 }
 431                 $this->mOutput->setText( $text );
 432
 433                 $this->mRevisionId = $oldRevisionId;
 434                 $this->mRevisionObject = $oldRevisionObject;
 435                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 436                 $this->mRevisionUser = $oldRevisionUser;
 437                 wfProfileOut( $fname );
 438                 wfProfileOut( __METHOD__ );
 439
 440                 return $this->mOutput;
 441         }
 442
 443         /**
 444          * Recursive parser entry point that can be called from an extension tag
 445          * hook.
 446          *
 447          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 448          *
 449          * @param $text String: text extension wants to have parsed
 450          * @param $frame PPFrame: The frame to use for expanding any template variables
 451          */
 452         function recursiveTagParse( $text, $frame=false ) {
 453                 wfProfileIn( __METHOD__ );
 454                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 455                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 456                 $text = $this->internalParse( $text, false, $frame );
 457                 wfProfileOut( __METHOD__ );
 458                 return $text;
 459         }
 460
 461         /**
 462          * Expand templates and variables in the text, producing valid, static wikitext.
 463          * Also removes comments.
 464          */
 465         function preprocess( $text, Title $title, ParserOptions $options, $revid = null ) {
 466                 wfProfileIn( __METHOD__ );
 467                 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
 468                 if ( $revid !== null ) {
 469                         $this->mRevisionId = $revid;
 470                 }
 471                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 472                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 473                 $text = $this->replaceVariables( $text );
 474                 $text = $this->mStripState->unstripBoth( $text );
 475                 wfProfileOut( __METHOD__ );
 476                 return $text;
 477         }
 478
 479         /**
 480          * Process the wikitext for the ?preload= feature. (bug 5210)
 481          *
 482          * <noinclude>, <includeonly> etc. are parsed as for template transclusion,
 483          * comments, templates, arguments, tags hooks and parser functions are untouched.
 484          */
 485         public function getPreloadText( $text, Title $title, ParserOptions $options ) {
 486                 # Parser (re)initialisation
 487                 $this->startParse( $title, $options, self::OT_PLAIN, true );
 488
 489                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 490                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
 491                 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
 492                 $text = $this->mStripState->unstripBoth( $text );
 493                 return $text;
 494         }
 495
 496         /**
 497          * Get a random string
 498          */
 499         static public function getRandomString() {
 500                 return dechex( mt_rand( 0, 0x7fffffff ) ) . dechex( mt_rand( 0, 0x7fffffff ) );
 501         }
 502
 503         /**
 504          * Set the current user.
 505          * Should only be used when doing pre-save transform.
 506          *
 507          * @param $user Mixed: User object or null (to reset)
 508          */
 509         function setUser( $user ) {
 510                 $this->mUser = $user;
 511         }
 512
 513         /**
 514          * Accessor for mUniqPrefix.
 515          *
 516          * @return String
 517          */
 518         public function uniqPrefix() {
 519                 if ( !isset( $this->mUniqPrefix ) ) {
 520                         # @todo Fixme: this is probably *horribly wrong*
 521                         # LanguageConverter seems to want $wgParser's uniqPrefix, however
 522                         # if this is called for a parser cache hit, the parser may not
 523                         # have ever been initialized in the first place.
 524                         # Not really sure what the heck is supposed to be going on here.
 525                         return '';
 526                         # throw new MWException( "Accessing uninitialized mUniqPrefix" );
 527                 }
 528                 return $this->mUniqPrefix;
 529         }
 530
 531         /**
 532          * Set the context title
 533          */
 534         function setTitle( $t ) {
 535                 if ( !$t || $t instanceof FakeTitle ) {
 536                         $t = Title::newFromText( 'NO TITLE' );
 537                 }
 538
 539                 if ( strval( $t->getFragment() ) !== '' ) {
 540                         # Strip the fragment to avoid various odd effects
 541                         $this->mTitle = clone $t;
 542                         $this->mTitle->setFragment( '' );
 543                 } else {
 544                         $this->mTitle = $t;
 545                 }
 546         }
 547
 548         /**
 549          * Accessor for the Title object
 550          *
 551          * @return Title object
 552          */
 553         function getTitle() {
 554                 return $this->mTitle;
 555         }
 556
 557         /**
 558          * Accessor/mutator for the Title object
 559          *
 560          * @param $x New Title object or null to just get the current one
 561          * @return Title object
 562          */
 563         function Title( $x = null ) {
 564                 return wfSetVar( $this->mTitle, $x );
 565         }
 566
 567         /**
 568          * Set the output type
 569          *
 570          * @param $ot Integer: new value
 571          */
 572         function setOutputType( $ot ) {
 573                 $this->mOutputType = $ot;
 574                 # Shortcut alias
 575                 $this->ot = array(
 576                         'html' => $ot == self::OT_HTML,
 577                         'wiki' => $ot == self::OT_WIKI,
 578                         'pre' => $ot == self::OT_PREPROCESS,
 579                         'plain' => $ot == self::OT_PLAIN,
 580                 );
 581         }
 582
 583         /**
 584          * Accessor/mutator for the output type
 585          *
 586          * @param $x New value or null to just get the current one
 587          * @return Integer
 588          */
 589         function OutputType( $x = null ) {
 590                 return wfSetVar( $this->mOutputType, $x );
 591         }
 592
 593         /**
 594          * Get the ParserOutput object
 595          *
 596          * @return ParserOutput object
 597          */
 598         function getOutput() {
 599                 return $this->mOutput;
 600         }
 601
 602         /**
 603          * Get the ParserOptions object
 604          *
 605          * @return ParserOptions object
 606          */
 607         function getOptions() {
 608                 return $this->mOptions;
 609         }
 610
 611         /**
 612          * Accessor/mutator for the ParserOptions object
 613          *
 614          * @param $x New value or null to just get the current one
 615          * @return Current ParserOptions object
 616          */
 617         function Options( $x = null ) {
 618                 return wfSetVar( $this->mOptions, $x );
 619         }
 620
 621         function nextLinkID() {
 622                 return $this->mLinkID++;
 623         }
 624
 625         function setLinkID( $id ) {
 626                 $this->mLinkID = $id;
 627         }
 628
 629         /**
 630          * @return Language
 631          */
 632         function getFunctionLang() {
 633                 global $wgLang, $wgContLang;
 634
 635                 $target = $this->mOptions->getTargetLanguage();
 636                 if ( $target !== null ) {
 637                         return $target;
 638                 } else {
 639                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 640                 }
 641         }
 642
 643         /**
 644          * Get a User object either from $this->mUser, if set, or from the
 645          * ParserOptions object otherwise
 646          *
 647          * @return User object
 648          */
 649         function getUser() {
 650                 if ( !is_null( $this->mUser ) ) {
 651                         return $this->mUser;
 652                 }
 653                 return $this->mOptions->getUser();
 654         }
 655
 656         /**
 657          * Get a preprocessor object
 658          *
 659          * @return Preprocessor instance
 660          */
 661         function getPreprocessor() {
 662                 if ( !isset( $this->mPreprocessor ) ) {
 663                         $class = $this->mPreprocessorClass;
 664                         $this->mPreprocessor = new $class( $this );
 665                 }
 666                 return $this->mPreprocessor;
 667         }
 668
 669         /**
 670          * Replaces all occurrences of HTML-style comments and the given tags
 671          * in the text with a random marker and returns the next text. The output
 672          * parameter $matches will be an associative array filled with data in
 673          * the form:
 674          *   'UNIQ-xxxxx' => array(
 675          *     'element',
 676          *     'tag content',
 677          *     array( 'param' => 'x' ),
 678          *     '<element param="x">tag content</element>' ) )
 679          *
 680          * @param $elements list of element names. Comments are always extracted.
 681          * @param $text Source text string.
 682          * @param $matches Out parameter, Array: extracted tags
 683          * @param $uniq_prefix
 684          * @return String: stripped text
 685          *
 686          * @static
 687          */
 688         public function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
 689                 static $n = 1;
 690                 $stripped = '';
 691                 $matches = array();
 692
 693                 $taglist = implode( '|', $elements );
 694                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 695
 696                 while ( $text != '' ) {
 697                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 698                         $stripped .= $p[0];
 699                         if ( count( $p ) < 5 ) {
 700                                 break;
 701                         }
 702                         if ( count( $p ) > 5 ) {
 703                                 # comment
 704                                 $element    = $p[4];
 705                                 $attributes = '';
 706                                 $close      = '';
 707                                 $inside     = $p[5];
 708                         } else {
 709                                 # tag
 710                                 $element    = $p[1];
 711                                 $attributes = $p[2];
 712                                 $close      = $p[3];
 713                                 $inside     = $p[4];
 714                         }
 715
 716                         $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
 717                         $stripped .= $marker;
 718
 719                         if ( $close === '/>' ) {
 720                                 # Empty element tag, <tag />
 721                                 $content = null;
 722                                 $text = $inside;
 723                                 $tail = null;
 724                         } else {
 725                                 if ( $element === '!--' ) {
 726                                         $end = '/(-->)/';
 727                                 } else {
 728                                         $end = "/(<\\/$element\\s*>)/i";
 729                                 }
 730                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 731                                 $content = $q[0];
 732                                 if ( count( $q ) < 3 ) {
 733                                         # No end tag -- let it run out to the end of the text.
 734                                         $tail = '';
 735                                         $text = '';
 736                                 } else {
 737                                         $tail = $q[1];
 738                                         $text = $q[2];
 739                                 }
 740                         }
 741
 742                         $matches[$marker] = array( $element,
 743                                 $content,
 744                                 Sanitizer::decodeTagAttributes( $attributes ),
 745                                 "<$element$attributes$close$content$tail" );
 746                 }
 747                 return $stripped;
 748         }
 749
 750         /**
 751          * Get a list of strippable XML-like elements
 752          */
 753         function getStripList() {
 754                 return $this->mStripList;
 755         }
 756
 757         /**
 758          * @deprecated use replaceVariables
 759          */
 760         function strip( $text, $state, $stripcomments = false , $dontstrip = array() ) {
 761                 return $text;
 762         }
 763
 764         /**
 765          * Restores pre, math, and other extensions removed by strip()
 766          *
 767          * always call unstripNoWiki() after this one
 768          * @private
 769          * @deprecated use $this->mStripState->unstrip()
 770          */
 771         function unstrip( $text, $state ) {
 772                 return $state->unstripGeneral( $text );
 773         }
 774
 775         /**
 776          * Always call this after unstrip() to preserve the order
 777          *
 778          * @private
 779          * @deprecated use $this->mStripState->unstrip()
 780          */
 781         function unstripNoWiki( $text, $state ) {
 782                 return $state->unstripNoWiki( $text );
 783         }
 784
 785         /**
 786          * @deprecated use $this->mStripState->unstripBoth()
 787          */
 788         function unstripForHTML( $text ) {
 789                 return $this->mStripState->unstripBoth( $text );
 790         }
 791
 792         /**
 793          * Add an item to the strip state
 794          * Returns the unique tag which must be inserted into the stripped text
 795          * The tag will be replaced with the original text in unstrip()
 796          *
 797          * @private
 798          */
 799         function insertStripItem( $text ) {
 800                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 801                 $this->mMarkerIndex++;
 802                 $this->mStripState->addGeneral( $rnd, $text );
 803                 return $rnd;
 804         }
 805
 806         /**
 807          * Interface with html tidy
 808          * @deprecated Use MWTidy::tidy()
 809          */
 810         public static function tidy( $text ) {
 811                 wfDeprecated( __METHOD__ );
 812                 return MWTidy::tidy( $text );
 813         }
 814
 815         /**
 816          * parse the wiki syntax used to render tables
 817          *
 818          * @private
 819          */
 820         function doTableStuff( $text ) {
 821                 wfProfileIn( __METHOD__ );
 822
 823                 $lines = StringUtils::explode( "\n", $text );
 824                 $out = '';
 825                 $output =& $out;
 826
 827                 foreach ( $lines as $outLine ) {
 828                         $line = trim( $outLine );
 829
 830                         # empty line, go to next line,
 831                         # but only append \n if outside of table
 832                         if ( $line === '') {
 833                                 $output .= $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                                 $table['indent'] = strlen( $matches[1] );
 848
 849                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 850                                 $attributes = Sanitizer::fixTagAttributes( $attributes , 'table' );
 851
 852                                 if ( $attributes !== '' ) {
 853                                         $table['attributes'] = $attributes;
 854                                 }
 855                         } else if ( !isset( $tables[0] ) ) {
 856                                 // we're outside the table
 857
 858                                 $out .= $outLine . "\n";
 859                         } else if ( $firstChars === '|}' ) {
 860                                 // trim the |} code from the line
 861                                 $line = substr ( $line , 2 );
 862
 863                                 // Shorthand for last row
 864                                 $lastRow =& $this->last( $table );
 865
 866                                 // a thead at the end becomes a tfoot, unless there is only one row
 867                                 // Do this before deleting empty last lines to allow headers at the bottom of tables
 868                                 if ( isset( $lastRow['type'] ) && $lastRow['type'] == 'thead' && isset( $table[1] ) ) {
 869                                         $lastRow['type'] = 'tfoot';
 870                                         for ( $i = 0; isset( $lastRow[$i] ); $i++ ) {
 871                                                 $lastRow[$i]['type'] = 'th';
 872                                         }
 873                                 }
 874
 875                                 // Delete empty last lines
 876                                 if ( empty( $lastRow ) ) {
 877                                         $lastRow = NULL;
 878                                 }
 879                                 $o = '';
 880                                 $curtable = array_pop( $tables );
 881
 882                                 #Add a line-ending before the table, but only if there isn't one already
 883                                 if ( substr( $out, -1 ) !== "\n" ) {
 884                                         $o .= "\n";
 885                                 }
 886                                 $o .= $this->generateTableHTML( $curtable ) . $line . "\n";
 887
 888                                 if ( count( $tables ) > 0 ) {
 889                                         $table =& $this->last( $tables );
 890                                         $currentRow =& $this->last( $table );
 891                                         $currentElement =& $this->last( $currentRow );
 892
 893                                         $output =& $currentElement['content'];
 894                                 } else {
 895                                         $output =& $out;
 896                                 }
 897
 898                                 $output .= $o;
 899
 900                         } else if ( $firstChars === '|-' ) {
 901                                 // start a new row element
 902                                 // but only when we haven't started one already
 903                                 if ( count( $currentRow ) != 0 ) {
 904                                         $table[] = array();
 905                                         $currentRow =& $this->last( $table );
 906                                 }
 907                                 // Get the attributes, there's nothing else useful in $line now
 908                                 $line = substr ( $line , 2 );
 909                                 $attributes = $this->mStripState->unstripBoth( $line );
 910                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 911                                 if ( $attributes !== '' ) {
 912                                         $currentRow['attributes'] = $attributes;
 913                                 }
 914
 915                         } else if ( $firstChars  === '|+' ) {
 916                                 // a table caption, but only proceed if there isn't one already
 917                                 if ( !isset ( $table['caption'] ) ) {
 918                                         $line = substr ( $line , 2 );
 919
 920                                         $c = $this->getCellAttr( $line , 'caption' );
 921                                         $table['caption'] = array();
 922                                         $table['caption']['content'] = $c[0];
 923                                         if ( isset( $c[1] ) ) $table['caption']['attributes'] = $c[1];
 924                                         unset( $c );
 925                                         $output =& $table['caption']['content'];
 926                                 }
 927                         } else if ( $firstChars === '|' || $firstChars === '!' || $firstChars === '!+' ) {
 928                                 // Which kind of cells are we dealing with
 929                                 $currentTag = 'td';
 930                                 $line = substr ( $line , 1 );
 931
 932                                 if ( $firstChars === '!'  || $firstChars === '!+' ) {
 933                                         $line = str_replace ( '!!' , '||' , $line );
 934                                         $currentTag = 'th';
 935                                 }
 936
 937                                 // Split up multiple cells on the same line.
 938                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 939                                 $line = ''; // save memory
 940
 941                                 // decide whether thead to tbody
 942                                 if ( !array_key_exists( 'type', $currentRow ) ) {
 943                                         $currentRow['type'] = ( $firstChars === '!' ) ? 'thead' : 'tbody' ;
 944                                 } else if ( $firstChars === '|' ) {
 945                                         $currentRow['type'] = 'tbody';
 946                                 }
 947
 948                                 // Loop through each table cell
 949                                 foreach ( $cells as $cell ) {
 950                                         // a new cell
 951                                         $currentRow[] = array();
 952                                         $currentElement =& $this->last( $currentRow );
 953
 954                                         $currentElement['type'] = $currentTag;
 955
 956                                         $c = $this->getCellAttr( $cell , $currentTag );
 957                                         $currentElement['content'] = $c[0];
 958                                         if ( isset( $c[1] ) ) $currentElement['attributes'] = $c[1];
 959                                         unset( $c );
 960                                 }
 961                                 $output =& $currentElement['content'];
 962
 963                         } else {
 964                                 $output .= "\n$outLine";
 965                         }
 966                 }
 967
 968                 # Remove trailing line-ending (b/c)
 969                 if ( substr( $out, -1 ) === "\n" ) {
 970                         $out = substr( $out, 0, -1 );
 971                 }
 972
 973                 # Close any unclosed tables
 974                 if ( isset( $tables ) && count( $tables ) > 0 ) {
 975                         for ( $i = 0; $i < count( $tables ); $i++ ) {
 976                                 $curtable = array_pop( $tables );
 977                                 $curtable = $this->generateTableHTML( $curtable );
 978                                 #Add a line-ending before the table, but only if there isn't one already
 979                                 if ( substr( $out, -1 ) !== "\n"  && $curtable !== "" ) {
 980                                         $out .= "\n";
 981                                 }
 982                                 $out .= $curtable;
 983                         }
 984                 }
 985
 986                 wfProfileOut( __METHOD__ );
 987
 988                 return $out;
 989         }
 990
 991
 992         /**
 993          * Helper function for doTableStuff() separating the contents of cells from
 994          * attributes. Particularly useful as there's a possible bug and this action
 995          * is repeated twice.
 996          *
 997          * @private
 998          */
 999         function getCellAttr ( $cell, $tagName ) {
1000                 $content = null;
1001                 $attributes = null;
1002
1003                 $cell = trim ( $cell );
1004
1005                 // A cell could contain both parameters and data
1006                 $cellData = explode ( '|' , $cell , 2 );
1007
1008                 // Bug 553: Note that a '|' inside an invalid link should not
1009                 // be mistaken as delimiting cell parameters
1010                 if ( strpos( $cellData[0], '[[' ) !== false ) {
1011                         $content = trim ( $cell );
1012                 }
1013                 else if ( count ( $cellData ) == 1 ) {
1014                         $content = trim ( $cellData[0] );
1015                 }
1016                 else {
1017                         $attributes = $this->mStripState->unstripBoth( $cellData[0] );
1018                         $attributes = Sanitizer::fixTagAttributes( $attributes , $tagName );
1019
1020                         $content = trim ( $cellData[1] );
1021                 }
1022                 return array( $content, $attributes );
1023         }
1024
1025
1026         /**
1027          * Helper function for doTableStuff(). This converts the structured array into html.
1028          *
1029          * @private
1030          */
1031         function generateTableHTML ( &$table ) {
1032                 $return = "";
1033                 $return .= str_repeat( '<dl><dd>' , $table['indent'] );
1034                 $return .= '<table';
1035                 $return .= isset( $table['attributes'] ) ? $table['attributes'] : '';
1036                 $return .= '>';
1037                 unset( $table['attributes'] );
1038
1039                 if ( isset( $table['caption'] ) ) {
1040                         $return .= "\n<caption";
1041                         $return .= isset( $table['caption']['attributes'] ) ? $table['caption']['attributes'] : '';
1042                         $return .= '>';
1043                         $return .= $table['caption']['content'];
1044                         $return .= "\n</caption>";
1045                 }
1046                 $lastSection = '';
1047                 $empty = true;
1048                 $simple = true;
1049
1050                 // If we only have tbodies, mark table as simple
1051                 for ( $i = 0; isset( $table[$i] ); $i++ ) {
1052                         if ( !count( $table[$i] ) ) continue;
1053                         if ( !isset( $table[$i]['type'] ) ) $table[$i]['type'] = 'tbody';
1054                         if ( !$lastSection ) {
1055                                 $lastSection = $table[$i]['type'];
1056                         } else if ( $lastSection != $table[$i]['type'] ) {
1057                                 $simple = false;
1058                                 break;
1059                         }
1060                 }
1061                 $lastSection = '';
1062                 for ( $i = 0; isset( $table[$i] ); $i++ ) {
1063                         // Check for empty tables
1064                         if ( count( $table[$i] ) ) {
1065                                 $empty = false;
1066                         } else {
1067                                 continue;
1068                         }
1069                         if ( $table[$i]['type'] != $lastSection && !$simple ) {
1070                                 $return .= "\n<" . $table[$i]['type'] . '>';
1071                         }
1072
1073                         $return .= "\n<tr";
1074                         $return .= isset( $table[$i]['attributes'] ) ? $table[$i]['attributes'] : '';
1075                         $return .= '>';
1076                         for ( $j = 0; isset( $table[$i][$j] ); $j++ ) {
1077                                 if ( !isset( $table[$i][$j]['type'] ) ) $table[$i][$j]['type'] = 'td';
1078                                 $return .= "\n<" . $table[$i][$j]['type'];
1079                                 $return .= isset( $table[$i][$j]['attributes'] ) ? $table[$i][$j]['attributes'] : '';
1080                                 $return .= '>';
1081
1082                                 $return .= $table[$i][$j]['content'];
1083                                 if ( $table[$i][$j]['content'] != '' )
1084                                         $return .= "\n";
1085
1086                                 $return .= '</' . $table[$i][$j]['type'] . '>';
1087                                 unset( $table[$i][$j] );
1088                         }
1089                         $return .= "\n</tr>";
1090
1091                         if ( ( !isset( $table[$i + 1] ) && !$simple ) || ( isset( $table[$i + 1] ) && ( $table[$i]['type'] != $table[$i + 1]['type'] ) ) ) {
1092                                 $return .= '</' . $table[$i]['type'] . '>';
1093                         }
1094                         $lastSection = $table[$i]['type'];
1095                         unset( $table[$i] );
1096                 }
1097                 if ( $empty ) {
1098                         if ( isset( $table['caption'] ) ) {
1099                                 $return .= "\n<tr><td></td></tr>";
1100                         } else {
1101                                 return '';
1102                         }
1103                 }
1104                 $return .= "\n</table>";
1105                 $return .= str_repeat( '</dd></dl>' , $table['indent'] );
1106
1107                 return $return;
1108         }
1109
1110         /**
1111          * like end() but only works on the numeric array index and php's internal pointers
1112          * returns a reference to the last element of an array much like "\$arr[-1]" in perl
1113          * ignores associative elements and will create a 0 key will a NULL value if there were
1114          * no numric elements and an array itself if not previously defined.
1115          *
1116          * @private
1117          */
1118         function &last ( &$arr ) {
1119                 for ( $i = count( $arr ); ( !isset( $arr[$i] ) && $i > 0 ); $i-- ) {  }
1120                 return $arr[$i];
1121         }
1122
1123         /**
1124          * Helper function for parse() that transforms wiki markup into
1125          * HTML. Only called for $mOutputType == self::OT_HTML.
1126          *
1127          * @private
1128          */
1129         function internalParse( $text, $isMain = true, $frame=false ) {
1130                 wfProfileIn( __METHOD__ );
1131
1132                 $origText = $text;
1133
1134                 # Hook to suspend the parser in this state
1135                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1136                         wfProfileOut( __METHOD__ );
1137                         return $text ;
1138                 }
1139
1140                 # if $frame is provided, then use $frame for replacing any variables
1141                 if ( $frame ) {
1142                         # use frame depth to infer how include/noinclude tags should be handled
1143                         # depth=0 means this is the top-level document; otherwise it's an included document
1144                         if ( !$frame->depth ) {
1145                                 $flag = 0;
1146                         } else {
1147                                 $flag = Parser::PTD_FOR_INCLUSION;
1148                         }
1149                         $dom = $this->preprocessToDom( $text, $flag );
1150                         $text = $frame->expand( $dom );
1151                 } else {
1152                         # if $frame is not provided, then use old-style replaceVariables
1153                         $text = $this->replaceVariables( $text );
1154                 }
1155
1156                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
1157                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1158
1159                 # Tables need to come after variable replacement for things to work
1160                 # properly; putting them before other transformations should keep
1161                 # exciting things like link expansions from showing up in surprising
1162                 # places.
1163                 $text = $this->doTableStuff( $text );
1164
1165                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1166
1167                 $text = $this->doDoubleUnderscore( $text );
1168
1169                 $text = $this->doHeadings( $text );
1170                 if ( $this->mOptions->getUseDynamicDates() ) {
1171                         $df = DateFormatter::getInstance();
1172                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1173                 }
1174                 $text = $this->replaceInternalLinks( $text );
1175                 $text = $this->doAllQuotes( $text );
1176                 $text = $this->replaceExternalLinks( $text );
1177
1178                 # replaceInternalLinks may sometimes leave behind
1179                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1180                 $text = str_replace( $this->mUniqPrefix.'NOPARSE', '', $text );
1181
1182                 $text = $this->doMagicLinks( $text );
1183                 $text = $this->formatHeadings( $text, $origText, $isMain );
1184
1185                 wfProfileOut( __METHOD__ );
1186                 return $text;
1187         }
1188
1189         /**
1190          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1191          * magic external links.
1192          *
1193          * DML
1194          * @private
1195          */
1196         function doMagicLinks( $text ) {
1197                 wfProfileIn( __METHOD__ );
1198                 $prots = $this->mUrlProtocols;
1199                 $urlChar = self::EXT_LINK_URL_CLASS;
1200                 $text = preg_replace_callback(
1201                         '!(?:                           # Start cases
1202                                 (<a[ \t\r\n>].*?</a>) |     # m[1]: Skip link text
1203                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1204                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
1205                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1206                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1207                                         (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1208                                         (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1209                                         [0-9Xx]                 # check digit
1210                                         \b)
1211                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1212                 wfProfileOut( __METHOD__ );
1213                 return $text;
1214         }
1215
1216         function magicLinkCallback( $m ) {
1217                 if ( isset( $m[1] ) && $m[1] !== '' ) {
1218                         # Skip anchor
1219                         return $m[0];
1220                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1221                         # Skip HTML element
1222                         return $m[0];
1223                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1224                         # Free external link
1225                         return $this->makeFreeExternalLink( $m[0] );
1226                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1227                         # RFC or PMID
1228                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1229                                 $keyword = 'RFC';
1230                                 $urlmsg = 'rfcurl';
1231                                 $CssClass = 'mw-magiclink-rfc';
1232                                 $id = $m[4];
1233                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1234                                 $keyword = 'PMID';
1235                                 $urlmsg = 'pubmedurl';
1236                                 $CssClass = 'mw-magiclink-pmid';
1237                                 $id = $m[4];
1238                         } else {
1239                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1240                                         substr( $m[0], 0, 20 ) . '"' );
1241                         }
1242                         $url = wfMsgForContent( $urlmsg, $id );
1243                         return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $CssClass );
1244                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1245                         # ISBN
1246                         $isbn = $m[5];
1247                         $num = strtr( $isbn, array(
1248                                 '-' => '',
1249                                 ' ' => '',
1250                                 'x' => 'X',
1251                         ));
1252                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1253                         return'<a href="' .
1254                                 $titleObj->escapeLocalUrl() .
1255                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1256                 } else {
1257                         return $m[0];
1258                 }
1259         }
1260
1261         /**
1262          * Make a free external link, given a user-supplied URL
1263          * @return HTML
1264          * @private
1265          */
1266         function makeFreeExternalLink( $url ) {
1267                 global $wgContLang;
1268                 wfProfileIn( __METHOD__ );
1269
1270                 $trail = '';
1271
1272                 # The characters '<' and '>' (which were escaped by
1273                 # removeHTMLtags()) should not be included in
1274                 # URLs, per RFC 2396.
1275                 $m2 = array();
1276                 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1277                         $trail = substr( $url, $m2[0][1] ) . $trail;
1278                         $url = substr( $url, 0, $m2[0][1] );
1279                 }
1280
1281                 # Move trailing punctuation to $trail
1282                 $sep = ',;\.:!?';
1283                 # If there is no left bracket, then consider right brackets fair game too
1284                 if ( strpos( $url, '(' ) === false ) {
1285                         $sep .= ')';
1286                 }
1287
1288                 $numSepChars = strspn( strrev( $url ), $sep );
1289                 if ( $numSepChars ) {
1290                         $trail = substr( $url, -$numSepChars ) . $trail;
1291                         $url = substr( $url, 0, -$numSepChars );
1292                 }
1293
1294                 $url = Sanitizer::cleanUrl( $url );
1295
1296                 # Is this an external image?
1297                 $text = $this->maybeMakeExternalImage( $url );
1298                 if ( $text === false ) {
1299                         # Not an image, make a link
1300                         $text = Linker::makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1301                                 $this->getExternalLinkAttribs( $url ) );
1302                         # Register it in the output object...
1303                         # Replace unnecessary URL escape codes with their equivalent characters
1304                         $pasteurized = self::replaceUnusualEscapes( $url );
1305                         $this->mOutput->addExternalLink( $pasteurized );
1306                 }
1307                 wfProfileOut( __METHOD__ );
1308                 return $text . $trail;
1309         }
1310
1311
1312         /**
1313          * Parse headers and return html
1314          *
1315          * @private
1316          */
1317         function doHeadings( $text ) {
1318                 wfProfileIn( __METHOD__ );
1319                 for ( $i = 6; $i >= 1; --$i ) {
1320                         $h = str_repeat( '=', $i );
1321                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1322                           "<h$i>\\1</h$i>", $text );
1323                 }
1324                 wfProfileOut( __METHOD__ );
1325                 return $text;
1326         }
1327
1328         /**
1329          * Replace single quotes with HTML markup
1330          * @private
1331          * @return string the altered text
1332          */
1333         function doAllQuotes( $text ) {
1334                 wfProfileIn( __METHOD__ );
1335                 $outtext = '';
1336                 $lines = StringUtils::explode( "\n", $text );
1337                 foreach ( $lines as $line ) {
1338                         $outtext .= $this->doQuotes( $line ) . "\n";
1339                 }
1340                 $outtext = substr( $outtext, 0,-1 );
1341                 wfProfileOut( __METHOD__ );
1342                 return $outtext;
1343         }
1344
1345         /**
1346          * Helper function for doAllQuotes()
1347          */
1348         public function doQuotes( $text ) {
1349                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1350                 if ( count( $arr ) == 1 ) {
1351                         return $text;
1352                 } else {
1353                         # First, do some preliminary work. This may shift some apostrophes from
1354                         # being mark-up to being text. It also counts the number of occurrences
1355                         # of bold and italics mark-ups.
1356                         $numbold = 0;
1357                         $numitalics = 0;
1358                         for ( $i = 0; $i < count( $arr ); $i++ ) {
1359                                 if ( ( $i % 2 ) == 1 ) {
1360                                         # If there are ever four apostrophes, assume the first is supposed to
1361                                         # be text, and the remaining three constitute mark-up for bold text.
1362                                         if ( strlen( $arr[$i] ) == 4 ) {
1363                                                 $arr[$i-1] .= "'";
1364                                                 $arr[$i] = "'''";
1365                                         } elseif ( strlen( $arr[$i] ) > 5 ) {
1366                                                 # If there are more than 5 apostrophes in a row, assume they're all
1367                                                 # text except for the last 5.
1368                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1369                                                 $arr[$i] = "'''''";
1370                                         }
1371                                         # Count the number of occurrences of bold and italics mark-ups.
1372                                         # We are not counting sequences of five apostrophes.
1373                                         if ( strlen( $arr[$i] ) == 2 ) {
1374                                                 $numitalics++;
1375                                         } elseif ( strlen( $arr[$i] ) == 3 ) {
1376                                                 $numbold++;
1377                                         } elseif ( strlen( $arr[$i] ) == 5 ) {
1378                                                 $numitalics++;
1379                                                 $numbold++;
1380                                         }
1381                                 }
1382                         }
1383
1384                         # If there is an odd number of both bold and italics, it is likely
1385                         # that one of the bold ones was meant to be an apostrophe followed
1386                         # by italics. Which one we cannot know for certain, but it is more
1387                         # likely to be one that has a single-letter word before it.
1388                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1389                                 $i = 0;
1390                                 $firstsingleletterword = -1;
1391                                 $firstmultiletterword = -1;
1392                                 $firstspace = -1;
1393                                 foreach ( $arr as $r ) {
1394                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1395                                                 $x1 = substr( $arr[$i-1], -1 );
1396                                                 $x2 = substr( $arr[$i-1], -2, 1 );
1397                                                 if ( $x1 === ' ' ) {
1398                                                         if ( $firstspace == -1 ) {
1399                                                                 $firstspace = $i;
1400                                                         }
1401                                                 } elseif ( $x2 === ' ') {
1402                                                         if ( $firstsingleletterword == -1 ) {
1403                                                                 $firstsingleletterword = $i;
1404                                                         }
1405                                                 } else {
1406                                                         if ( $firstmultiletterword == -1 ) {
1407                                                                 $firstmultiletterword = $i;
1408                                                         }
1409                                                 }
1410                                         }
1411                                         $i++;
1412                                 }
1413
1414                                 # If there is a single-letter word, use it!
1415                                 if ( $firstsingleletterword > -1 ) {
1416                                         $arr[$firstsingleletterword] = "''";
1417                                         $arr[$firstsingleletterword-1] .= "'";
1418                                 } elseif ( $firstmultiletterword > -1 ) {
1419                                         # If not, but there's a multi-letter word, use that one.
1420                                         $arr[$firstmultiletterword] = "''";
1421                                         $arr[$firstmultiletterword-1] .= "'";
1422                                 } elseif ( $firstspace > -1 ) {
1423                                         # ... otherwise use the first one that has neither.
1424                                         # (notice that it is possible for all three to be -1 if, for example,
1425                                         # there is only one pentuple-apostrophe in the line)
1426                                         $arr[$firstspace] = "''";
1427                                         $arr[$firstspace-1] .= "'";
1428                                 }
1429                         }
1430
1431                         # Now let's actually convert our apostrophic mush to HTML!
1432                         $output = '';
1433                         $buffer = '';
1434                         $state = '';
1435                         $i = 0;
1436                         foreach ( $arr as $r ) {
1437                                 if ( ( $i % 2 ) == 0 ) {
1438                                         if ( $state === 'both' ) {
1439                                                 $buffer .= $r;
1440                                         } else {
1441                                                 $output .= $r;
1442                                         }
1443                                 } else {
1444                                         if ( strlen( $r ) == 2 ) {
1445                                                 if ( $state === 'i' ) {
1446                                                         $output .= '</i>'; $state = '';
1447                                                 } elseif ( $state === 'bi' ) {
1448                                                         $output .= '</i>'; $state = 'b';
1449                                                 } elseif ( $state === 'ib' ) {
1450                                                         $output .= '</b></i><b>'; $state = 'b';
1451                                                 } elseif ( $state === 'both' ) {
1452                                                         $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1453                                                 } else { # $state can be 'b' or ''
1454                                                         $output .= '<i>'; $state .= 'i';
1455                                                 }
1456                                         } elseif ( strlen( $r ) == 3 ) {
1457                                                 if ( $state === 'b' ) {
1458                                                         $output .= '</b>'; $state = '';
1459                                                 } elseif ( $state === 'bi' ) {
1460                                                         $output .= '</i></b><i>'; $state = 'i';
1461                                                 } elseif ( $state === 'ib' ) {
1462                                                         $output .= '</b>'; $state = 'i';
1463                                                 } elseif ( $state === 'both' ) {
1464                                                         $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1465                                                 } else { # $state can be 'i' or ''
1466                                                         $output .= '<b>'; $state .= 'b';
1467                                                 }
1468                                         } elseif ( strlen( $r ) == 5 ) {
1469                                                 if ( $state === 'b' ) {
1470                                                         $output .= '</b><i>'; $state = 'i';
1471                                                 } elseif ( $state === 'i' ) {
1472                                                         $output .= '</i><b>'; $state = 'b';
1473                                                 } elseif ( $state === 'bi' ) {
1474                                                         $output .= '</i></b>'; $state = '';
1475                                                 } elseif ( $state === 'ib' ) {
1476                                                         $output .= '</b></i>'; $state = '';
1477                                                 } elseif ( $state === 'both' ) {
1478                                                         $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1479                                                 } else { # ($state == '')
1480                                                         $buffer = ''; $state = 'both';
1481                                                 }
1482                                         }
1483                                 }
1484                                 $i++;
1485                         }
1486                         # Now close all remaining tags.  Notice that the order is important.
1487                         if ( $state === 'b' || $state === 'ib' ) {
1488                                 $output .= '</b>';
1489                         }
1490                         if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1491                                 $output .= '</i>';
1492                         }
1493                         if ( $state === 'bi' ) {
1494                                 $output .= '</b>';
1495                         }
1496                         # There might be lonely ''''', so make sure we have a buffer
1497                         if ( $state === 'both' && $buffer ) {
1498                                 $output .= '<b><i>'.$buffer.'</i></b>';
1499                         }
1500                         return $output;
1501                 }
1502         }
1503
1504         /**
1505          * Replace external links (REL)
1506          *
1507          * Note: this is all very hackish and the order of execution matters a lot.
1508          * Make sure to run maintenance/parserTests.php if you change this code.
1509          *
1510          * @private
1511          */
1512         function replaceExternalLinks( $text ) {
1513                 global $wgContLang;
1514                 wfProfileIn( __METHOD__ );
1515
1516                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1517                 $s = array_shift( $bits );
1518
1519                 $i = 0;
1520                 while ( $i<count( $bits ) ) {
1521                         $url = $bits[$i++];
1522                         $protocol = $bits[$i++];
1523                         $text = $bits[$i++];
1524                         $trail = $bits[$i++];
1525
1526                         # The characters '<' and '>' (which were escaped by
1527                         # removeHTMLtags()) should not be included in
1528                         # URLs, per RFC 2396.
1529                         $m2 = array();
1530                         if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1531                                 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1532                                 $url = substr( $url, 0, $m2[0][1] );
1533                         }
1534
1535                         # If the link text is an image URL, replace it with an <img> tag
1536                         # This happened by accident in the original parser, but some people used it extensively
1537                         $img = $this->maybeMakeExternalImage( $text );
1538                         if ( $img !== false ) {
1539                                 $text = $img;
1540                         }
1541
1542                         $dtrail = '';
1543
1544                         # Set linktype for CSS - if URL==text, link is essentially free
1545                         $linktype = ( $text === $url ) ? 'free' : 'text';
1546
1547                         # No link text, e.g. [http://domain.tld/some.link]
1548                         if ( $text == '' ) {
1549                                 # Autonumber if allowed. See bug #5918
1550                                 if ( strpos( wfUrlProtocols(), substr( $protocol, 0, strpos( $protocol, ':' ) ) ) !== false ) {
1551                                         $langObj = $this->getFunctionLang();
1552                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1553                                         $linktype = 'autonumber';
1554                                 } else {
1555                                         # Otherwise just use the URL
1556                                         $text = htmlspecialchars( $url );
1557                                         $linktype = 'free';
1558                                 }
1559                         } else {
1560                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1561                                 # Check for trail
1562                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1563                         }
1564
1565                         $text = $wgContLang->markNoConversion( $text );
1566
1567                         $url = Sanitizer::cleanUrl( $url );
1568
1569                         # Use the encoded URL
1570                         # This means that users can paste URLs directly into the text
1571                         # Funny characters like ö aren't valid in URLs anyway
1572                         # This was changed in August 2004
1573                         $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1574                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1575
1576                         # Register link in the output object.
1577                         # Replace unnecessary URL escape codes with the referenced character
1578                         # This prevents spammers from hiding links from the filters
1579                         $pasteurized = self::replaceUnusualEscapes( $url );
1580                         $this->mOutput->addExternalLink( $pasteurized );
1581                 }
1582
1583                 wfProfileOut( __METHOD__ );
1584                 return $s;
1585         }
1586
1587         /**
1588          * Get an associative array of additional HTML attributes appropriate for a
1589          * particular external link.  This currently may include rel => nofollow
1590          * (depending on configuration, namespace, and the URL's domain) and/or a
1591          * target attribute (depending on configuration).
1592          *
1593          * @param $url String: optional URL, to extract the domain from for rel =>
1594          *   nofollow if appropriate
1595          * @return Array: associative array of HTML attributes
1596          */
1597         function getExternalLinkAttribs( $url = false ) {
1598                 $attribs = array();
1599                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1600                 $ns = $this->mTitle->getNamespace();
1601                 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) ) {
1602                         $attribs['rel'] = 'nofollow';
1603
1604                         global $wgNoFollowDomainExceptions;
1605                         if ( $wgNoFollowDomainExceptions ) {
1606                                 $bits = wfParseUrl( $url );
1607                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1608                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1609                                                 if ( substr( $bits['host'], -strlen( $domain ) ) == $domain ) {
1610                                                         unset( $attribs['rel'] );
1611                                                         break;
1612                                                 }
1613                                         }
1614                                 }
1615                         }
1616                 }
1617                 if ( $this->mOptions->getExternalLinkTarget() ) {
1618                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1619                 }
1620                 return $attribs;
1621         }
1622
1623
1624         /**
1625          * Replace unusual URL escape codes with their equivalent characters
1626          *
1627          * @param $url String
1628          * @return String
1629          *
1630          * @todo  This can merge genuinely required bits in the path or query string,
1631          *        breaking legit URLs. A proper fix would treat the various parts of
1632          *        the URL differently; as a workaround, just use the output for
1633          *        statistical records, not for actual linking/output.
1634          */
1635         static function replaceUnusualEscapes( $url ) {
1636                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1637                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1638         }
1639
1640         /**
1641          * Callback function used in replaceUnusualEscapes().
1642          * Replaces unusual URL escape codes with their equivalent character
1643          */
1644         private static function replaceUnusualEscapesCallback( $matches ) {
1645                 $char = urldecode( $matches[0] );
1646                 $ord = ord( $char );
1647                 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1648                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1649                         # No, shouldn't be escaped
1650                         return $char;
1651                 } else {
1652                         # Yes, leave it escaped
1653                         return $matches[0];
1654                 }
1655         }
1656
1657         /**
1658          * make an image if it's allowed, either through the global
1659          * option, through the exception, or through the on-wiki whitelist
1660          * @private
1661          */
1662         function maybeMakeExternalImage( $url ) {
1663                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1664                 $imagesexception = !empty( $imagesfrom );
1665                 $text = false;
1666                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1667                 if ( $imagesexception && is_array( $imagesfrom ) ) {
1668                         $imagematch = false;
1669                         foreach ( $imagesfrom as $match ) {
1670                                 if ( strpos( $url, $match ) === 0 ) {
1671                                         $imagematch = true;
1672                                         break;
1673                                 }
1674                         }
1675                 } elseif ( $imagesexception ) {
1676                         $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1677                 } else {
1678                         $imagematch = false;
1679                 }
1680                 if ( $this->mOptions->getAllowExternalImages()
1681                          || ( $imagesexception && $imagematch ) ) {
1682                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1683                                 # Image found
1684                                 $text = Linker::makeExternalImage( $url );
1685                         }
1686                 }
1687                 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1688                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1689                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1690                         foreach ( $whitelist as $entry ) {
1691                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1692                                 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1693                                         continue;
1694                                 }
1695                                 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1696                                         # Image matches a whitelist entry
1697                                         $text = Linker::makeExternalImage( $url );
1698                                         break;
1699                                 }
1700                         }
1701                 }
1702                 return $text;
1703         }
1704
1705         /**
1706          * Process [[ ]] wikilinks
1707          * @return String: processed text
1708          *
1709          * @private
1710          */
1711         function replaceInternalLinks( $s ) {
1712                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1713                 return $s;
1714         }
1715
1716         /**
1717          * Process [[ ]] wikilinks (RIL)
1718          * @return LinkHolderArray
1719          *
1720          * @private
1721          */
1722         function replaceInternalLinks2( &$s ) {
1723                 global $wgContLang;
1724
1725                 wfProfileIn( __METHOD__ );
1726
1727                 wfProfileIn( __METHOD__.'-setup' );
1728                 static $tc = FALSE, $e1, $e1_img;
1729                 # the % is needed to support urlencoded titles as well
1730                 if ( !$tc ) {
1731                         $tc = Title::legalChars() . '#%';
1732                         # Match a link having the form [[namespace:link|alternate]]trail
1733                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1734                         # Match cases where there is no "]]", which might still be images
1735                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1736                 }
1737
1738                 $holders = new LinkHolderArray( $this );
1739
1740                 # split the entire text string on occurences of [[
1741                 $a = StringUtils::explode( '[[', ' ' . $s );
1742                 # get the first element (all text up to first [[), and remove the space we added
1743                 $s = $a->current();
1744                 $a->next();
1745                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1746                 $s = substr( $s, 1 );
1747
1748                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1749                 $e2 = null;
1750                 if ( $useLinkPrefixExtension ) {
1751                         # Match the end of a line for a word that's not followed by whitespace,
1752                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1753                         $e2 = wfMsgForContent( 'linkprefix' );
1754                 }
1755
1756                 if ( is_null( $this->mTitle ) ) {
1757                         wfProfileOut( __METHOD__.'-setup' );
1758                         wfProfileOut( __METHOD__ );
1759                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1760                 }
1761                 $nottalk = !$this->mTitle->isTalkPage();
1762
1763                 if ( $useLinkPrefixExtension ) {
1764                         $m = array();
1765                         if ( preg_match( $e2, $s, $m ) ) {
1766                                 $first_prefix = $m[2];
1767                         } else {
1768                                 $first_prefix = false;
1769                         }
1770                 } else {
1771                         $prefix = '';
1772                 }
1773
1774                 if ( $wgContLang->hasVariants() ) {
1775                         $selflink = $wgContLang->autoConvertToAllVariants( $this->mTitle->getPrefixedText() );
1776                 } else {
1777                         $selflink = array( $this->mTitle->getPrefixedText() );
1778                 }
1779                 $useSubpages = $this->areSubpagesAllowed();
1780                 wfProfileOut( __METHOD__.'-setup' );
1781
1782                 # Loop for each link
1783                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1784                         # Check for excessive memory usage
1785                         if ( $holders->isBig() ) {
1786                                 # Too big
1787                                 # Do the existence check, replace the link holders and clear the array
1788                                 $holders->replace( $s );
1789                                 $holders->clear();
1790                         }
1791
1792                         if ( $useLinkPrefixExtension ) {
1793                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1794                                 if ( preg_match( $e2, $s, $m ) ) {
1795                                         $prefix = $m[2];
1796                                         $s = $m[1];
1797                                 } else {
1798                                         $prefix='';
1799                                 }
1800                                 # first link
1801                                 if ( $first_prefix ) {
1802                                         $prefix = $first_prefix;
1803                                         $first_prefix = false;
1804                                 }
1805                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1806                         }
1807
1808                         $might_be_img = false;
1809
1810                         wfProfileIn( __METHOD__."-e1" );
1811                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1812                                 $text = $m[2];
1813                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1814                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1815                                 # the real problem is with the $e1 regex
1816                                 # See bug 1300.
1817                                 #
1818                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1819                                 # and no image is in sight. See bug 2095.
1820                                 #
1821                                 if ( $text !== '' &&
1822                                         substr( $m[3], 0, 1 ) === ']' &&
1823                                         strpos( $text, '[' ) !== false
1824                                 )
1825                                 {
1826                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1827                                         $m[3] = substr( $m[3], 1 );
1828                                 }
1829                                 # fix up urlencoded title texts
1830                                 if ( strpos( $m[1], '%' ) !== false ) {
1831                                         # Should anchors '#' also be rejected?
1832                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), rawurldecode( $m[1] ) );
1833                                 }
1834                                 $trail = $m[3];
1835                         } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1836                                 $might_be_img = true;
1837                                 $text = $m[2];
1838                                 if ( strpos( $m[1], '%' ) !== false ) {
1839                                         $m[1] = rawurldecode( $m[1] );
1840                                 }
1841                                 $trail = "";
1842                         } else { # Invalid form; output directly
1843                                 $s .= $prefix . '[[' . $line ;
1844                                 wfProfileOut( __METHOD__."-e1" );
1845                                 continue;
1846                         }
1847                         wfProfileOut( __METHOD__."-e1" );
1848                         wfProfileIn( __METHOD__."-misc" );
1849
1850                         # Don't allow internal links to pages containing
1851                         # PROTO: where PROTO is a valid URL protocol; these
1852                         # should be external links.
1853                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1854                                 $s .= $prefix . '[[' . $line ;
1855                                 wfProfileOut( __METHOD__."-misc" );
1856                                 continue;
1857                         }
1858
1859                         # Make subpage if necessary
1860                         if ( $useSubpages ) {
1861                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1862                         } else {
1863                                 $link = $m[1];
1864                         }
1865
1866                         $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1867                         if ( !$noforce ) {
1868                                 # Strip off leading ':'
1869                                 $link = substr( $link, 1 );
1870                         }
1871
1872                         wfProfileOut( __METHOD__."-misc" );
1873                         wfProfileIn( __METHOD__."-title" );
1874                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1875                         if ( $nt === null ) {
1876                                 $s .= $prefix . '[[' . $line;
1877                                 wfProfileOut( __METHOD__."-title" );
1878                                 continue;
1879                         }
1880
1881                         $ns = $nt->getNamespace();
1882                         $iw = $nt->getInterWiki();
1883                         wfProfileOut( __METHOD__."-title" );
1884
1885                         if ( $might_be_img ) { # if this is actually an invalid link
1886                                 wfProfileIn( __METHOD__."-might_be_img" );
1887                                 if ( $ns == NS_FILE && $noforce ) { # but might be an image
1888                                         $found = false;
1889                                         while ( true ) {
1890                                                 # look at the next 'line' to see if we can close it there
1891                                                 $a->next();
1892                                                 $next_line = $a->current();
1893                                                 if ( $next_line === false || $next_line === null ) {
1894                                                         break;
1895                                                 }
1896                                                 $m = explode( ']]', $next_line, 3 );
1897                                                 if ( count( $m ) == 3 ) {
1898                                                         # the first ]] closes the inner link, the second the image
1899                                                         $found = true;
1900                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1901                                                         $trail = $m[2];
1902                                                         break;
1903                                                 } elseif ( count( $m ) == 2 ) {
1904                                                         # if there's exactly one ]] that's fine, we'll keep looking
1905                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1906                                                 } else {
1907                                                         # if $next_line is invalid too, we need look no further
1908                                                         $text .= '[[' . $next_line;
1909                                                         break;
1910                                                 }
1911                                         }
1912                                         if ( !$found ) {
1913                                                 # we couldn't find the end of this imageLink, so output it raw
1914                                                 # but don't ignore what might be perfectly normal links in the text we've examined
1915                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1916                                                 $s .= "{$prefix}[[$link|$text";
1917                                                 # note: no $trail, because without an end, there *is* no trail
1918                                                 wfProfileOut( __METHOD__."-might_be_img" );
1919                                                 continue;
1920                                         }
1921                                 } else { # it's not an image, so output it raw
1922                                         $s .= "{$prefix}[[$link|$text";
1923                                         # note: no $trail, because without an end, there *is* no trail
1924                                         wfProfileOut( __METHOD__."-might_be_img" );
1925                                         continue;
1926                                 }
1927                                 wfProfileOut( __METHOD__."-might_be_img" );
1928                         }
1929
1930                         $wasblank = ( $text  == '' );
1931                         if ( $wasblank ) {
1932                                 $text = $link;
1933                         } else {
1934                                 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
1935                                 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
1936                                 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
1937                                 #    -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
1938                                 $text = $this->doQuotes( $text );
1939                         }
1940
1941                         # Link not escaped by : , create the various objects
1942                         if ( $noforce ) {
1943
1944                                 # Interwikis
1945                                 wfProfileIn( __METHOD__."-interwiki" );
1946                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1947                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1948                                         $s = rtrim( $s . $prefix );
1949                                         $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
1950                                         wfProfileOut( __METHOD__."-interwiki" );
1951                                         continue;
1952                                 }
1953                                 wfProfileOut( __METHOD__."-interwiki" );
1954
1955                                 if ( $ns == NS_FILE ) {
1956                                         wfProfileIn( __METHOD__."-image" );
1957                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1958                                                 if ( $wasblank ) {
1959                                                         # if no parameters were passed, $text
1960                                                         # becomes something like "File:Foo.png",
1961                                                         # which we don't want to pass on to the
1962                                                         # image generator
1963                                                         $text = '';
1964                                                 } else {
1965                                                         # recursively parse links inside the image caption
1966                                                         # actually, this will parse them in any other parameters, too,
1967                                                         # but it might be hard to fix that, and it doesn't matter ATM
1968                                                         $text = $this->replaceExternalLinks( $text );
1969                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1970                                                 }
1971                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1972                                                 $s .= $prefix . $this->armorLinks(
1973                                                         $this->makeImage( $nt, $text, $holders ) ) . $trail;
1974                                         } else {
1975                                                 $s .= $prefix . $trail;
1976                                         }
1977                                         wfProfileOut( __METHOD__."-image" );
1978                                         continue;
1979                                 }
1980
1981                                 if ( $ns == NS_CATEGORY ) {
1982                                         wfProfileIn( __METHOD__."-category" );
1983                                         $s = rtrim( $s . "\n" ); # bug 87
1984
1985                                         if ( $wasblank ) {
1986                                                 $sortkey = $this->getDefaultSort();
1987                                         } else {
1988                                                 $sortkey = $text;
1989                                         }
1990                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1991                                         $sortkey = str_replace( "\n", '', $sortkey );
1992                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1993                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1994
1995                                         /**
1996                                          * Strip the whitespace Category links produce, see bug 87
1997                                          * @todo We might want to use trim($tmp, "\n") here.
1998                                          */
1999                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
2000
2001                                         wfProfileOut( __METHOD__."-category" );
2002                                         continue;
2003                                 }
2004                         }
2005
2006                         # Self-link checking
2007                         if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
2008                                 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
2009                                         $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2010                                         continue;
2011                                 }
2012                         }
2013
2014                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
2015                         # FIXME: Should do batch file existence checks, see comment below
2016                         if ( $ns == NS_MEDIA ) {
2017                                 wfProfileIn( __METHOD__."-media" );
2018                                 # Give extensions a chance to select the file revision for us
2019                                 $time = $sha1 = $descQuery = false;
2020                                 wfRunHooks( 'BeforeParserFetchFileAndTitle',
2021                                         array( $this, $nt, &$time, &$sha1, &$descQuery ) );
2022                                 # Fetch and register the file (file title may be different via hooks)
2023                                 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $time, $sha1 );
2024                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2025                                 $s .= $prefix . $this->armorLinks(
2026                                         Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2027                                 wfProfileOut( __METHOD__."-media" );
2028                                 continue;
2029                         }
2030
2031                         wfProfileIn( __METHOD__."-always_known" );
2032                         # Some titles, such as valid special pages or files in foreign repos, should
2033                         # be shown as bluelinks even though they're not included in the page table
2034                         #
2035                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2036                         # batch file existence checks for NS_FILE and NS_MEDIA
2037                         if ( $iw == '' && $nt->isAlwaysKnown() ) {
2038                                 $this->mOutput->addLink( $nt );
2039                                 $s .= $this->makeKnownLinkHolder( $nt, $text, array(), $trail, $prefix );
2040                         } else {
2041                                 # Links will be added to the output link list after checking
2042                                 $s .= $holders->makeHolder( $nt, $text, array(), $trail, $prefix );
2043                         }
2044                         wfProfileOut( __METHOD__."-always_known" );
2045                 }
2046                 wfProfileOut( __METHOD__ );
2047                 return $holders;
2048         }
2049
2050         /**
2051          * Make a link placeholder. The text returned can be later resolved to a real link with
2052          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
2053          * parsing of interwiki links, and secondly to allow all existence checks and
2054          * article length checks (for stub links) to be bundled into a single query.
2055          *
2056          * @deprecated
2057          */
2058         function makeLinkHolder( &$nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
2059                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
2060         }
2061
2062         /**
2063          * Render a forced-blue link inline; protect against double expansion of
2064          * URLs if we're in a mode that prepends full URL prefixes to internal links.
2065          * Since this little disaster has to split off the trail text to avoid
2066          * breaking URLs in the following text without breaking trails on the
2067          * wiki links, it's been made into a horrible function.
2068          *
2069          * @param $nt Title
2070          * @param $text String
2071          * @param $query Array or String
2072          * @param $trail String
2073          * @param $prefix String
2074          * @return String: HTML-wikitext mix oh yuck
2075          */
2076         function makeKnownLinkHolder( $nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
2077                 list( $inside, $trail ) = Linker::splitTrail( $trail );
2078
2079                 if ( is_string( $query ) ) {
2080                         $query = wfCgiToArray( $query );
2081                 }
2082                 if ( $text == '' ) {
2083                         $text = htmlspecialchars( $nt->getPrefixedText() );
2084                 }
2085
2086                 $link = Linker::linkKnown( $nt, "$prefix$text$inside", array(), $query );
2087
2088                 return $this->armorLinks( $link ) . $trail;
2089         }
2090
2091         /**
2092          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2093          * going to go through further parsing steps before inline URL expansion.
2094          *
2095          * Not needed quite as much as it used to be since free links are a bit
2096          * more sensible these days. But bracketed links are still an issue.
2097          *
2098          * @param $text String: more-or-less HTML
2099          * @return String: less-or-more HTML with NOPARSE bits
2100          */
2101         function armorLinks( $text ) {
2102                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
2103                         "{$this->mUniqPrefix}NOPARSE$1", $text );
2104         }
2105
2106         /**
2107          * Return true if subpage links should be expanded on this page.
2108          * @return Boolean
2109          */
2110         function areSubpagesAllowed() {
2111                 # Some namespaces don't allow subpages
2112                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2113         }
2114
2115         /**
2116          * Handle link to subpage if necessary
2117          *
2118          * @param $target String: the source of the link
2119          * @param &$text String: the link text, modified as necessary
2120          * @return string the full name of the link
2121          * @private
2122          */
2123         function maybeDoSubpageLink( $target, &$text ) {
2124                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2125         }
2126
2127         /**#@+
2128          * Used by doBlockLevels()
2129          * @private
2130          */
2131         function closeParagraph() {
2132                 $result = '';
2133                 if ( $this->mLastSection != '' ) {
2134                         $result = '</' . $this->mLastSection  . ">\n";
2135                 }
2136                 $this->mInPre = false;
2137                 $this->mLastSection = '';
2138                 return $result;
2139         }
2140
2141         /**
2142          * getCommon() returns the length of the longest common substring
2143          * of both arguments, starting at the beginning of both.
2144          * @private
2145          */
2146         function getCommon( $st1, $st2 ) {
2147                 $fl = strlen( $st1 );
2148                 $shorter = strlen( $st2 );
2149                 if ( $fl < $shorter ) {
2150                         $shorter = $fl;
2151                 }
2152
2153                 for ( $i = 0; $i < $shorter; ++$i ) {
2154                         if ( $st1[$i] != $st2[$i] ) {
2155                                 break;
2156                         }
2157                 }
2158                 return $i;
2159         }
2160
2161         /**
2162          * These next three functions open, continue, and close the list
2163          * element appropriate to the prefix character passed into them.
2164          * @private
2165          */
2166         function openList( $char ) {
2167                 $result = $this->closeParagraph();
2168
2169                 if ( '*' === $char ) {
2170                         $result .= '<ul><li>';
2171                 } elseif ( '#' === $char ) {
2172                         $result .= '<ol><li>';
2173                 } elseif ( ':' === $char ) {
2174                         $result .= '<dl><dd>';
2175                 } elseif ( ';' === $char ) {
2176                         $result .= '<dl><dt>';
2177                         $this->mDTopen = true;
2178                 } else {
2179                         $result = '<!-- ERR 1 -->';
2180                 }
2181
2182                 return $result;
2183         }
2184
2185         /**
2186          * TODO: document
2187          * @param $char String
2188          * @private
2189          */
2190         function nextItem( $char ) {
2191                 if ( '*' === $char || '#' === $char ) {
2192                         return '</li><li>';
2193                 } elseif ( ':' === $char || ';' === $char ) {
2194                         $close = '</dd>';
2195                         if ( $this->mDTopen ) {
2196                                 $close = '</dt>';
2197                         }
2198                         if ( ';' === $char ) {
2199                                 $this->mDTopen = true;
2200                                 return $close . '<dt>';
2201                         } else {
2202                                 $this->mDTopen = false;
2203                                 return $close . '<dd>';
2204                         }
2205                 }
2206                 return '<!-- ERR 2 -->';
2207         }
2208
2209         /**
2210          * TODO: document
2211          * @param $char String
2212          * @private
2213          */
2214         function closeList( $char ) {
2215                 if ( '*' === $char ) {
2216                         $text = '</li></ul>';
2217                 } elseif ( '#' === $char ) {
2218                         $text = '</li></ol>';
2219                 } elseif ( ':' === $char ) {
2220                         if ( $this->mDTopen ) {
2221                                 $this->mDTopen = false;
2222                                 $text = '</dt></dl>';
2223                         } else {
2224                                 $text = '</dd></dl>';
2225                         }
2226                 } else {
2227                         return '<!-- ERR 3 -->';
2228                 }
2229                 return $text."\n";
2230         }
2231         /**#@-*/
2232
2233         /**
2234          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2235          *
2236          * @param $text String
2237          * @param $linestart Boolean: whether or not this is at the start of a line.
2238          * @private
2239          * @return string the lists rendered as HTML
2240          */
2241         function doBlockLevels( $text, $linestart ) {
2242                 wfProfileIn( __METHOD__ );
2243
2244                 # Parsing through the text line by line.  The main thing
2245                 # happening here is handling of block-level elements p, pre,
2246                 # and making lists from lines starting with * # : etc.
2247                 #
2248                 $textLines = StringUtils::explode( "\n", $text );
2249
2250                 $lastPrefix = $output = '';
2251                 $this->mDTopen = $inBlockElem = false;
2252                 $prefixLength = 0;
2253                 $paragraphStack = false;
2254
2255                 foreach ( $textLines as $oLine ) {
2256                         # Fix up $linestart
2257                         if ( !$linestart ) {
2258                                 $output .= $oLine;
2259                                 $linestart = true;
2260                                 continue;
2261                         }
2262                         # * = ul
2263                         # # = ol
2264                         # ; = dt
2265                         # : = dd
2266
2267                         $lastPrefixLength = strlen( $lastPrefix );
2268                         $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2269                         $preOpenMatch = preg_match( '/<pre/i', $oLine );
2270                         # If not in a <pre> element, scan for and figure out what prefixes are there.
2271                         if ( !$this->mInPre ) {
2272                                 # Multiple prefixes may abut each other for nested lists.
2273                                 $prefixLength = strspn( $oLine, '*#:;' );
2274                                 $prefix = substr( $oLine, 0, $prefixLength );
2275
2276                                 # eh?
2277                                 # ; and : are both from definition-lists, so they're equivalent
2278                                 #  for the purposes of determining whether or not we need to open/close
2279                                 #  elements.
2280                                 $prefix2 = str_replace( ';', ':', $prefix );
2281                                 $t = substr( $oLine, $prefixLength );
2282                                 $this->mInPre = (bool)$preOpenMatch;
2283                         } else {
2284                                 # Don't interpret any other prefixes in preformatted text
2285                                 $prefixLength = 0;
2286                                 $prefix = $prefix2 = '';
2287                                 $t = $oLine;
2288                         }
2289
2290                         # List generation
2291                         if ( $prefixLength && $lastPrefix === $prefix2 ) {
2292                                 # Same as the last item, so no need to deal with nesting or opening stuff
2293                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2294                                 $paragraphStack = false;
2295
2296                                 if ( substr( $prefix, -1 ) === ';') {
2297                                         # The one nasty exception: definition lists work like this:
2298                                         # ; title : definition text
2299                                         # So we check for : in the remainder text to split up the
2300                                         # title and definition, without b0rking links.
2301                                         $term = $t2 = '';
2302                                         if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2303                                                 $t = $t2;
2304                                                 $output .= $term . $this->nextItem( ':' );
2305                                         }
2306                                 }
2307                         } elseif ( $prefixLength || $lastPrefixLength ) {
2308                                 # We need to open or close prefixes, or both.
2309
2310                                 # Either open or close a level...
2311                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2312                                 $paragraphStack = false;
2313
2314                                 # Close all the prefixes which aren't shared.
2315                                 while ( $commonPrefixLength < $lastPrefixLength ) {
2316                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2317                                         --$lastPrefixLength;
2318                                 }
2319
2320                                 # Continue the current prefix if appropriate.
2321                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2322                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2323                                 }
2324
2325                                 # Open prefixes where appropriate.
2326                                 while ( $prefixLength > $commonPrefixLength ) {
2327                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2328                                         $output .= $this->openList( $char );
2329
2330                                         if ( ';' === $char ) {
2331                                                 # FIXME: This is dupe of code above
2332                                                 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2333                                                         $t = $t2;
2334                                                         $output .= $term . $this->nextItem( ':' );
2335                                                 }
2336                                         }
2337                                         ++$commonPrefixLength;
2338                                 }
2339                                 $lastPrefix = $prefix2;
2340                         }
2341
2342                         # If we have no prefixes, go to paragraph mode.
2343                         if ( 0 == $prefixLength ) {
2344                                 wfProfileIn( __METHOD__."-paragraph" );
2345                                 # No prefix (not in list)--go to paragraph mode
2346                                 # XXX: use a stack for nestable elements like span, table and div
2347                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2348                                 $closematch = preg_match(
2349                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2350                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2351                                 if ( $openmatch or $closematch ) {
2352                                         $paragraphStack = false;
2353                                         # TODO bug 5718: paragraph closed
2354                                         $output .= $this->closeParagraph();
2355                                         if ( $preOpenMatch and !$preCloseMatch ) {
2356                                                 $this->mInPre = true;
2357                                         }
2358                                         $inBlockElem = !$closematch;
2359                                 } elseif ( !$inBlockElem && !$this->mInPre ) {
2360                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) ) {
2361                                                 # pre
2362                                                 if ( $this->mLastSection !== 'pre' ) {
2363                                                         $paragraphStack = false;
2364                                                         $output .= $this->closeParagraph().'<pre>';
2365                                                         $this->mLastSection = 'pre';
2366                                                 }
2367                                                 $t = substr( $t, 1 );
2368                                         } else {
2369                                                 # paragraph
2370                                                 if ( trim( $t ) === '' ) {
2371                                                         if ( $paragraphStack ) {
2372                                                                 $output .= $paragraphStack.'<br />';
2373                                                                 $paragraphStack = false;
2374                                                                 $this->mLastSection = 'p';
2375                                                         } else {
2376                                                                 if ( $this->mLastSection !== 'p' ) {
2377                                                                         $output .= $this->closeParagraph();
2378                                                                         $this->mLastSection = '';
2379                                                                         $paragraphStack = '<p>';
2380                                                                 } else {
2381                                                                         $paragraphStack = '</p><p>';
2382                                                                 }
2383                                                         }
2384                                                 } else {
2385                                                         if ( $paragraphStack ) {
2386                                                                 $output .= $paragraphStack;
2387                                                                 $paragraphStack = false;
2388                                                                 $this->mLastSection = 'p';
2389                                                         } elseif ( $this->mLastSection !== 'p' ) {
2390                                                                 $output .= $this->closeParagraph().'<p>';
2391                                                                 $this->mLastSection = 'p';
2392                                                         }
2393                                                 }
2394                                         }
2395                                 }
2396                                 wfProfileOut( __METHOD__."-paragraph" );
2397                         }
2398                         # somewhere above we forget to get out of pre block (bug 785)
2399                         if ( $preCloseMatch && $this->mInPre ) {
2400                                 $this->mInPre = false;
2401                         }
2402                         if ( $paragraphStack === false ) {
2403                                 $output .= $t."\n";
2404                         }
2405                 }
2406                 while ( $prefixLength ) {
2407                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2408                         --$prefixLength;
2409                 }
2410                 if ( $this->mLastSection != '' ) {
2411                         $output .= '</' . $this->mLastSection . '>';
2412                         $this->mLastSection = '';
2413                 }
2414
2415                 wfProfileOut( __METHOD__ );
2416                 return $output;
2417         }
2418
2419         /**
2420          * Split up a string on ':', ignoring any occurences inside tags
2421          * to prevent illegal overlapping.
2422          *
2423          * @param $str String: the string to split
2424          * @param &$before String: set to everything before the ':'
2425          * @param &$after String: set to everything after the ':'
2426          * return String: the position of the ':', or false if none found
2427          */
2428         function findColonNoLinks( $str, &$before, &$after ) {
2429                 wfProfileIn( __METHOD__ );
2430
2431                 $pos = strpos( $str, ':' );
2432                 if ( $pos === false ) {
2433                         # Nothing to find!
2434                         wfProfileOut( __METHOD__ );
2435                         return false;
2436                 }
2437
2438                 $lt = strpos( $str, '<' );
2439                 if ( $lt === false || $lt > $pos ) {
2440                         # Easy; no tag nesting to worry about
2441                         $before = substr( $str, 0, $pos );
2442                         $after = substr( $str, $pos+1 );
2443                         wfProfileOut( __METHOD__ );
2444                         return $pos;
2445                 }
2446
2447                 # Ugly state machine to walk through avoiding tags.
2448                 $state = self::COLON_STATE_TEXT;
2449                 $stack = 0;
2450                 $len = strlen( $str );
2451                 for( $i = 0; $i < $len; $i++ ) {
2452                         $c = $str[$i];
2453
2454                         switch( $state ) {
2455                         # (Using the number is a performance hack for common cases)
2456                         case 0: # self::COLON_STATE_TEXT:
2457                                 switch( $c ) {
2458                                 case "<":
2459                                         # Could be either a <start> tag or an </end> tag
2460                                         $state = self::COLON_STATE_TAGSTART;
2461                                         break;
2462                                 case ":":
2463                                         if ( $stack == 0 ) {
2464                                                 # We found it!
2465                                                 $before = substr( $str, 0, $i );
2466                                                 $after = substr( $str, $i + 1 );
2467                                                 wfProfileOut( __METHOD__ );
2468                                                 return $i;
2469                                         }
2470                                         # Embedded in a tag; don't break it.
2471                                         break;
2472                                 default:
2473                                         # Skip ahead looking for something interesting
2474                                         $colon = strpos( $str, ':', $i );
2475                                         if ( $colon === false ) {
2476                                                 # Nothing else interesting
2477                                                 wfProfileOut( __METHOD__ );
2478                                                 return false;
2479                                         }
2480                                         $lt = strpos( $str, '<', $i );
2481                                         if ( $stack === 0 ) {
2482                                                 if ( $lt === false || $colon < $lt ) {
2483                                                         # We found it!
2484                                                         $before = substr( $str, 0, $colon );
2485                                                         $after = substr( $str, $colon + 1 );
2486                                                         wfProfileOut( __METHOD__ );
2487                                                         return $i;
2488                                                 }
2489                                         }
2490                                         if ( $lt === false ) {
2491                                                 # Nothing else interesting to find; abort!
2492                                                 # We're nested, but there's no close tags left. Abort!
2493                                                 break 2;
2494                                         }
2495                                         # Skip ahead to next tag start
2496                                         $i = $lt;
2497                                         $state = self::COLON_STATE_TAGSTART;
2498                                 }
2499                                 break;
2500                         case 1: # self::COLON_STATE_TAG:
2501                                 # In a <tag>
2502                                 switch( $c ) {
2503                                 case ">":
2504                                         $stack++;
2505                                         $state = self::COLON_STATE_TEXT;
2506                                         break;
2507                                 case "/":
2508                                         # Slash may be followed by >?
2509                                         $state = self::COLON_STATE_TAGSLASH;
2510                                         break;
2511                                 default:
2512                                         # ignore
2513                                 }
2514                                 break;
2515                         case 2: # self::COLON_STATE_TAGSTART:
2516                                 switch( $c ) {
2517                                 case "/":
2518                                         $state = self::COLON_STATE_CLOSETAG;
2519                                         break;
2520                                 case "!":
2521                                         $state = self::COLON_STATE_COMMENT;
2522                                         break;
2523                                 case ">":
2524                                         # Illegal early close? This shouldn't happen D:
2525                                         $state = self::COLON_STATE_TEXT;
2526                                         break;
2527                                 default:
2528                                         $state = self::COLON_STATE_TAG;
2529                                 }
2530                                 break;
2531                         case 3: # self::COLON_STATE_CLOSETAG:
2532                                 # In a </tag>
2533                                 if ( $c === ">" ) {
2534                                         $stack--;
2535                                         if ( $stack < 0 ) {
2536                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2537                                                 wfProfileOut( __METHOD__ );
2538                                                 return false;
2539                                         }
2540                                         $state = self::COLON_STATE_TEXT;
2541                                 }
2542                                 break;
2543                         case self::COLON_STATE_TAGSLASH:
2544                                 if ( $c === ">" ) {
2545                                         # Yes, a self-closed tag <blah/>
2546                                         $state = self::COLON_STATE_TEXT;
2547                                 } else {
2548                                         # Probably we're jumping the gun, and this is an attribute
2549                                         $state = self::COLON_STATE_TAG;
2550                                 }
2551                                 break;
2552                         case 5: # self::COLON_STATE_COMMENT:
2553                                 if ( $c === "-" ) {
2554                                         $state = self::COLON_STATE_COMMENTDASH;
2555                                 }
2556                                 break;
2557                         case self::COLON_STATE_COMMENTDASH:
2558                                 if ( $c === "-" ) {
2559                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2560                                 } else {
2561                                         $state = self::COLON_STATE_COMMENT;
2562                                 }
2563                                 break;
2564                         case self::COLON_STATE_COMMENTDASHDASH:
2565                                 if ( $c === ">" ) {
2566                                         $state = self::COLON_STATE_TEXT;
2567                                 } else {
2568                                         $state = self::COLON_STATE_COMMENT;
2569                                 }
2570                                 break;
2571                         default:
2572                                 throw new MWException( "State machine error in " . __METHOD__ );
2573                         }
2574                 }
2575                 if ( $stack > 0 ) {
2576                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2577                         wfProfileOut( __METHOD__ );
2578                         return false;
2579                 }
2580                 wfProfileOut( __METHOD__ );
2581                 return false;
2582         }
2583
2584         /**
2585          * Return value of a magic variable (like PAGENAME)
2586          *
2587          * @private
2588          */
2589         function getVariableValue( $index, $frame=false ) {
2590                 global $wgContLang, $wgSitename, $wgServer;
2591                 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2592
2593                 /**
2594                  * Some of these require message or data lookups and can be
2595                  * expensive to check many times.
2596                  */
2597                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2598                         if ( isset( $this->mVarCache[$index] ) ) {
2599                                 return $this->mVarCache[$index];
2600                         }
2601                 }
2602
2603                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2604                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2605
2606                 # Use the time zone
2607                 global $wgLocaltimezone;
2608                 if ( isset( $wgLocaltimezone ) ) {
2609                         $oldtz = date_default_timezone_get();
2610                         date_default_timezone_set( $wgLocaltimezone );
2611                 }
2612
2613                 $localTimestamp = date( 'YmdHis', $ts );
2614                 $localMonth = date( 'm', $ts );
2615                 $localMonth1 = date( 'n', $ts );
2616                 $localMonthName = date( 'n', $ts );
2617                 $localDay = date( 'j', $ts );
2618                 $localDay2 = date( 'd', $ts );
2619                 $localDayOfWeek = date( 'w', $ts );
2620                 $localWeek = date( 'W', $ts );
2621                 $localYear = date( 'Y', $ts );
2622                 $localHour = date( 'H', $ts );
2623                 if ( isset( $wgLocaltimezone ) ) {
2624                         date_default_timezone_set( $oldtz );
2625                 }
2626
2627                 switch ( $index ) {
2628                         case 'currentmonth':
2629                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2630                                 break;
2631                         case 'currentmonth1':
2632                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2633                                 break;
2634                         case 'currentmonthname':
2635                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2636                                 break;
2637                         case 'currentmonthnamegen':
2638                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2639                                 break;
2640                         case 'currentmonthabbrev':
2641                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2642                                 break;
2643                         case 'currentday':
2644                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2645                                 break;
2646                         case 'currentday2':
2647                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2648                                 break;
2649                         case 'localmonth':
2650                                 $value = $wgContLang->formatNum( $localMonth );
2651                                 break;
2652                         case 'localmonth1':
2653                                 $value = $wgContLang->formatNum( $localMonth1 );
2654                                 break;
2655                         case 'localmonthname':
2656                                 $value = $wgContLang->getMonthName( $localMonthName );
2657                                 break;
2658                         case 'localmonthnamegen':
2659                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2660                                 break;
2661                         case 'localmonthabbrev':
2662                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2663                                 break;
2664                         case 'localday':
2665                                 $value = $wgContLang->formatNum( $localDay );
2666                                 break;
2667                         case 'localday2':
2668                                 $value = $wgContLang->formatNum( $localDay2 );
2669                                 break;
2670                         case 'pagename':
2671                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2672                                 break;
2673                         case 'pagenamee':
2674                                 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2675                                 break;
2676                         case 'fullpagename':
2677                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2678                                 break;
2679                         case 'fullpagenamee':
2680                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2681                                 break;
2682                         case 'subpagename':
2683                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2684                                 break;
2685                         case 'subpagenamee':
2686                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2687                                 break;
2688                         case 'basepagename':
2689                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2690                                 break;
2691                         case 'basepagenamee':
2692                                 $value = wfEscapeWikiText( wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) ) );
2693                                 break;
2694                         case 'talkpagename':
2695                                 if ( $this->mTitle->canTalk() ) {
2696                                         $talkPage = $this->mTitle->getTalkPage();
2697                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2698                                 } else {
2699                                         $value = '';
2700                                 }
2701                                 break;
2702                         case 'talkpagenamee':
2703                                 if ( $this->mTitle->canTalk() ) {
2704                                         $talkPage = $this->mTitle->getTalkPage();
2705                                         $value = wfEscapeWikiText( $talkPage->getPrefixedUrl() );
2706                                 } else {
2707                                         $value = '';
2708                                 }
2709                                 break;
2710                         case 'subjectpagename':
2711                                 $subjPage = $this->mTitle->getSubjectPage();
2712                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2713                                 break;
2714                         case 'subjectpagenamee':
2715                                 $subjPage = $this->mTitle->getSubjectPage();
2716                                 $value = wfEscapeWikiText( $subjPage->getPrefixedUrl() );
2717                                 break;
2718                         case 'revisionid':
2719                                 # Let the edit saving system know we should parse the page
2720                                 # *after* a revision ID has been assigned.
2721                                 $this->mOutput->setFlag( 'vary-revision' );
2722                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2723                                 $value = $this->mRevisionId;
2724                                 break;
2725                         case 'revisionday':
2726                                 # Let the edit saving system know we should parse the page
2727                                 # *after* a revision ID has been assigned. This is for null edits.
2728                                 $this->mOutput->setFlag( 'vary-revision' );
2729                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2730                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2731                                 break;
2732                         case 'revisionday2':
2733                                 # Let the edit saving system know we should parse the page
2734                                 # *after* a revision ID has been assigned. This is for null edits.
2735                                 $this->mOutput->setFlag( 'vary-revision' );
2736                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2737                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2738                                 break;
2739                         case 'revisionmonth':
2740                                 # Let the edit saving system know we should parse the page
2741                                 # *after* a revision ID has been assigned. This is for null edits.
2742                                 $this->mOutput->setFlag( 'vary-revision' );
2743                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2744                                 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2745                                 break;
2746                         case 'revisionmonth1':
2747                                 # Let the edit saving system know we should parse the page
2748                                 # *after* a revision ID has been assigned. This is for null edits.
2749                                 $this->mOutput->setFlag( 'vary-revision' );
2750                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2751                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2752                                 break;
2753                         case 'revisionyear':
2754                                 # Let the edit saving system know we should parse the page
2755                                 # *after* a revision ID has been assigned. This is for null edits.
2756                                 $this->mOutput->setFlag( 'vary-revision' );
2757                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2758                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2759                                 break;
2760                         case 'revisiontimestamp':
2761                                 # Let the edit saving system know we should parse the page
2762                                 # *after* a revision ID has been assigned. This is for null edits.
2763                                 $this->mOutput->setFlag( 'vary-revision' );
2764                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2765                                 $value = $this->getRevisionTimestamp();
2766                                 break;
2767                         case 'revisionuser':
2768                                 # Let the edit saving system know we should parse the page
2769                                 # *after* a revision ID has been assigned. This is for null edits.
2770                                 $this->mOutput->setFlag( 'vary-revision' );
2771                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2772                                 $value = $this->getRevisionUser();
2773                                 break;
2774                         case 'namespace':
2775                                 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2776                                 break;
2777                         case 'namespacee':
2778                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2779                                 break;
2780                         case 'talkspace':
2781                                 $value = $this->mTitle->canTalk() ? str_replace( '_',' ',$this->mTitle->getTalkNsText() ) : '';
2782                                 break;
2783                         case 'talkspacee':
2784                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2785                                 break;
2786                         case 'subjectspace':
2787                                 $value = $this->mTitle->getSubjectNsText();
2788                                 break;
2789                         case 'subjectspacee':
2790                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2791                                 break;
2792                         case 'currentdayname':
2793                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2794                                 break;
2795                         case 'currentyear':
2796                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2797                                 break;
2798                         case 'currenttime':
2799                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2800                                 break;
2801                         case 'currenthour':
2802                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2803                                 break;
2804                         case 'currentweek':
2805                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2806                                 # int to remove the padding
2807                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2808                                 break;
2809                         case 'currentdow':
2810                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2811                                 break;
2812                         case 'localdayname':
2813                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2814                                 break;
2815                         case 'localyear':
2816                                 $value = $wgContLang->formatNum( $localYear, true );
2817                                 break;
2818                         case 'localtime':
2819                                 $value = $wgContLang->time( $localTimestamp, false, false );
2820                                 break;
2821                         case 'localhour':
2822                                 $value = $wgContLang->formatNum( $localHour, true );
2823                                 break;
2824                         case 'localweek':
2825                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2826                                 # int to remove the padding
2827                                 $value = $wgContLang->formatNum( (int)$localWeek );
2828                                 break;
2829                         case 'localdow':
2830                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2831                                 break;
2832                         case 'numberofarticles':
2833                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2834                                 break;
2835                         case 'numberoffiles':
2836                                 $value = $wgContLang->formatNum( SiteStats::images() );
2837                                 break;
2838                         case 'numberofusers':
2839                                 $value = $wgContLang->formatNum( SiteStats::users() );
2840                                 break;
2841                         case 'numberofactiveusers':
2842                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2843                                 break;
2844                         case 'numberofpages':
2845                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2846                                 break;
2847                         case 'numberofadmins':
2848                                 $value = $wgContLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2849                                 break;
2850                         case 'numberofedits':
2851                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2852                                 break;
2853                         case 'numberofviews':
2854                                 $value = $wgContLang->formatNum( SiteStats::views() );
2855                                 break;
2856                         case 'currenttimestamp':
2857                                 $value = wfTimestamp( TS_MW, $ts );
2858                                 break;
2859                         case 'localtimestamp':
2860                                 $value = $localTimestamp;
2861                                 break;
2862                         case 'currentversion':
2863                                 $value = SpecialVersion::getVersion();
2864                                 break;
2865                         case 'articlepath':
2866                                 return $wgArticlePath;
2867                         case 'sitename':
2868                                 return $wgSitename;
2869                         case 'server':
2870                                 return $wgServer;
2871                         case 'servername':
2872                                 wfSuppressWarnings(); # May give an E_WARNING in PHP < 5.3.3
2873                                 $serverName = parse_url( $wgServer, PHP_URL_HOST );
2874                                 wfRestoreWarnings();
2875                                 return $serverName ? $serverName : $wgServer;
2876                         case 'scriptpath':
2877                                 return $wgScriptPath;
2878                         case 'stylepath':
2879                                 return $wgStylePath;
2880                         case 'directionmark':
2881                                 return $wgContLang->getDirMark();
2882                         case 'contentlanguage':
2883                                 global $wgLanguageCode;
2884                                 return $wgLanguageCode;
2885                         default:
2886                                 $ret = null;
2887                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
2888                                         return $ret;
2889                                 } else {
2890                                         return null;
2891                                 }
2892                 }
2893
2894                 if ( $index )
2895                         $this->mVarCache[$index] = $value;
2896
2897                 return $value;
2898         }
2899
2900         /**
2901          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2902          *
2903          * @private
2904          */
2905         function initialiseVariables() {
2906                 wfProfileIn( __METHOD__ );
2907                 $variableIDs = MagicWord::getVariableIDs();
2908                 $substIDs = MagicWord::getSubstIDs();
2909
2910                 $this->mVariables = new MagicWordArray( $variableIDs );
2911                 $this->mSubstWords = new MagicWordArray( $substIDs );
2912                 wfProfileOut( __METHOD__ );
2913         }
2914
2915         /**
2916          * Preprocess some wikitext and return the document tree.
2917          * This is the ghost of replace_variables().
2918          *
2919          * @param $text String: The text to parse
2920          * @param $flags Integer: bitwise combination of:
2921          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2922          *                                     included. Default is to assume a direct page view.
2923          *
2924          * The generated DOM tree must depend only on the input text and the flags.
2925          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2926          *
2927          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2928          * change in the DOM tree for a given text, must be passed through the section identifier
2929          * in the section edit link and thus back to extractSections().
2930          *
2931          * The output of this function is currently only cached in process memory, but a persistent
2932          * cache may be implemented at a later date which takes further advantage of these strict
2933          * dependency requirements.
2934          *
2935          * @private
2936          */
2937         function preprocessToDom( $text, $flags = 0 ) {
2938                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2939                 return $dom;
2940         }
2941
2942         /**
2943          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2944          */
2945         public static function splitWhitespace( $s ) {
2946                 $ltrimmed = ltrim( $s );
2947                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2948                 $trimmed = rtrim( $ltrimmed );
2949                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2950                 if ( $diff > 0 ) {
2951                         $w2 = substr( $ltrimmed, -$diff );
2952                 } else {
2953                         $w2 = '';
2954                 }
2955                 return array( $w1, $trimmed, $w2 );
2956         }
2957
2958         /**
2959          * Replace magic variables, templates, and template arguments
2960          * with the appropriate text. Templates are substituted recursively,
2961          * taking care to avoid infinite loops.
2962          *
2963          * Note that the substitution depends on value of $mOutputType:
2964          *  self::OT_WIKI: only {{subst:}} templates
2965          *  self::OT_PREPROCESS: templates but not extension tags
2966          *  self::OT_HTML: all templates and extension tags
2967          *
2968          * @param $text String: the text to transform
2969          * @param $frame PPFrame Object describing the arguments passed to the template.
2970          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2971          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2972          * @param $argsOnly Boolean: only do argument (triple-brace) expansion, not double-brace expansion
2973          * @private
2974          */
2975         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2976                 # Is there any text? Also, Prevent too big inclusions!
2977                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2978                         return $text;
2979                 }
2980                 wfProfileIn( __METHOD__ );
2981
2982                 if ( $frame === false ) {
2983                         $frame = $this->getPreprocessor()->newFrame();
2984                 } elseif ( !( $frame instanceof PPFrame ) ) {
2985                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2986                         $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2987                 }
2988
2989                 $dom = $this->preprocessToDom( $text );
2990                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2991                 $text = $frame->expand( $dom, $flags );
2992
2993                 wfProfileOut( __METHOD__ );
2994                 return $text;
2995         }
2996
2997         # Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2998         static function createAssocArgs( $args ) {
2999                 $assocArgs = array();
3000                 $index = 1;
3001                 foreach ( $args as $arg ) {
3002                         $eqpos = strpos( $arg, '=' );
3003                         if ( $eqpos === false ) {
3004                                 $assocArgs[$index++] = $arg;
3005                         } else {
3006                                 $name = trim( substr( $arg, 0, $eqpos ) );
3007                                 $value = trim( substr( $arg, $eqpos+1 ) );
3008                                 if ( $value === false ) {
3009                                         $value = '';
3010                                 }
3011                                 if ( $name !== false ) {
3012                                         $assocArgs[$name] = $value;
3013                                 }
3014                         }
3015                 }
3016
3017                 return $assocArgs;
3018         }
3019
3020         /**
3021          * Warn the user when a parser limitation is reached
3022          * Will warn at most once the user per limitation type
3023          *
3024          * @param $limitationType String: should be one of:
3025          *   'expensive-parserfunction' (corresponding messages:
3026          *       'expensive-parserfunction-warning',
3027          *       'expensive-parserfunction-category')
3028          *   'post-expand-template-argument' (corresponding messages:
3029          *       'post-expand-template-argument-warning',
3030          *       'post-expand-template-argument-category')
3031          *   'post-expand-template-inclusion' (corresponding messages:
3032          *       'post-expand-template-inclusion-warning',
3033          *       'post-expand-template-inclusion-category')
3034          * @param $current Current value
3035          * @param $max Maximum allowed, when an explicit limit has been
3036          *       exceeded, provide the values (optional)
3037          */
3038         function limitationWarn( $limitationType, $current=null, $max=null) {
3039                 # does no harm if $current and $max are present but are unnecessary for the message
3040                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
3041                 $this->mOutput->addWarning( $warning );
3042                 $this->addTrackingCategory( "$limitationType-category" );
3043         }
3044
3045         /**
3046          * Return the text of a template, after recursively
3047          * replacing any variables or templates within the template.
3048          *
3049          * @param $piece Array: the parts of the template
3050          *  $piece['title']: the title, i.e. the part before the |
3051          *  $piece['parts']: the parameter array
3052          *  $piece['lineStart']: whether the brace was at the start of a line
3053          * @param $frame PPFrame The current frame, contains template arguments
3054          * @return String: the text of the template
3055          * @private
3056          */
3057         function braceSubstitution( $piece, $frame ) {
3058                 global $wgContLang, $wgNonincludableNamespaces;
3059                 wfProfileIn( __METHOD__ );
3060                 wfProfileIn( __METHOD__.'-setup' );
3061
3062                 # Flags
3063                 $found = false;             # $text has been filled
3064                 $nowiki = false;            # wiki markup in $text should be escaped
3065                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
3066                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
3067                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
3068                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
3069
3070                 # Title object, where $text came from
3071                 $title = null;
3072
3073                 # $part1 is the bit before the first |, and must contain only title characters.
3074                 # Various prefixes will be stripped from it later.
3075                 $titleWithSpaces = $frame->expand( $piece['title'] );
3076                 $part1 = trim( $titleWithSpaces );
3077                 $titleText = false;
3078
3079                 # Original title text preserved for various purposes
3080                 $originalTitle = $part1;
3081
3082                 # $args is a list of argument nodes, starting from index 0, not including $part1
3083                 # *** FIXME if piece['parts'] is null then the call to getLength() below won't work b/c this $args isn't an object
3084                 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
3085                 wfProfileOut( __METHOD__.'-setup' );
3086
3087                 # SUBST
3088                 wfProfileIn( __METHOD__.'-modifiers' );
3089                 if ( !$found ) {
3090
3091                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3092
3093                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
3094                         # Decide whether to expand template or keep wikitext as-is.
3095                         if ( $this->ot['wiki'] ) {
3096                                 if ( $substMatch === false ) {
3097                                         $literal = true;  # literal when in PST with no prefix
3098                                 } else {
3099                                         $literal = false; # expand when in PST with subst: or safesubst:
3100                                 }
3101                         } else {
3102                                 if ( $substMatch == 'subst' ) {
3103                                         $literal = true;  # literal when not in PST with plain subst:
3104                                 } else {
3105                                         $literal = false; # expand when not in PST with safesubst: or no prefix
3106                                 }
3107                         }
3108                         if ( $literal ) {
3109                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3110                                 $isLocalObj = true;
3111                                 $found = true;
3112                         }
3113                 }
3114
3115                 # Variables
3116                 if ( !$found && $args->getLength() == 0 ) {
3117                         $id = $this->mVariables->matchStartToEnd( $part1 );
3118                         if ( $id !== false ) {
3119                                 $text = $this->getVariableValue( $id, $frame );
3120                                 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3121                                         $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3122                                 }
3123                                 $found = true;
3124                         }
3125                 }
3126
3127                 # MSG, MSGNW and RAW
3128                 if ( !$found ) {
3129                         # Check for MSGNW:
3130                         $mwMsgnw = MagicWord::get( 'msgnw' );
3131                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3132                                 $nowiki = true;
3133                         } else {
3134                                 # Remove obsolete MSG:
3135                                 $mwMsg = MagicWord::get( 'msg' );
3136                                 $mwMsg->matchStartAndRemove( $part1 );
3137                         }
3138
3139                         # Check for RAW:
3140                         $mwRaw = MagicWord::get( 'raw' );
3141                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3142                                 $forceRawInterwiki = true;
3143                         }
3144                 }
3145                 wfProfileOut( __METHOD__.'-modifiers' );
3146
3147                 # Parser functions
3148                 if ( !$found ) {
3149                         wfProfileIn( __METHOD__ . '-pfunc' );
3150
3151                         $colonPos = strpos( $part1, ':' );
3152                         if ( $colonPos !== false ) {
3153                                 # Case sensitive functions
3154                                 $function = substr( $part1, 0, $colonPos );
3155                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3156                                         $function = $this->mFunctionSynonyms[1][$function];
3157                                 } else {
3158                                         # Case insensitive functions
3159                                         $function = $wgContLang->lc( $function );
3160                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3161                                                 $function = $this->mFunctionSynonyms[0][$function];
3162                                         } else {
3163                                                 $function = false;
3164                                         }
3165                                 }
3166                                 if ( $function ) {
3167                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3168                                         $initialArgs = array( &$this );
3169                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3170                                         if ( $flags & SFH_OBJECT_ARGS ) {
3171                                                 # Add a frame parameter, and pass the arguments as an array
3172                                                 $allArgs = $initialArgs;
3173                                                 $allArgs[] = $frame;
3174                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3175                                                         $funcArgs[] = $args->item( $i );
3176                                                 }
3177                                                 $allArgs[] = $funcArgs;
3178                                         } else {
3179                                                 # Convert arguments to plain text
3180                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3181                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
3182                                                 }
3183                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3184                                         }
3185
3186                                         # Workaround for PHP bug 35229 and similar
3187                                         if ( !is_callable( $callback ) ) {
3188                                                 wfProfileOut( __METHOD__ . '-pfunc' );
3189                                                 wfProfileOut( __METHOD__ );
3190                                                 throw new MWException( "Tag hook for $function is not callable\n" );
3191                                         }
3192                                         $result = call_user_func_array( $callback, $allArgs );
3193                                         $found = true;
3194                                         $noparse = true;
3195                                         $preprocessFlags = 0;
3196
3197                                         if ( is_array( $result ) ) {
3198                                                 if ( isset( $result[0] ) ) {
3199                                                         $text = $result[0];
3200                                                         unset( $result[0] );
3201                                                 }
3202
3203                                                 # Extract flags into the local scope
3204                                                 # This allows callers to set flags such as nowiki, found, etc.
3205                                                 extract( $result );
3206                                         } else {
3207                                                 $text = $result;
3208                                         }
3209                                         if ( !$noparse ) {
3210                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
3211                                                 $isChildObj = true;
3212                                         }
3213                                 }
3214                         }
3215                         wfProfileOut( __METHOD__ . '-pfunc' );
3216                 }
3217
3218                 # Finish mangling title and then check for loops.
3219                 # Set $title to a Title object and $titleText to the PDBK
3220                 if ( !$found ) {
3221                         $ns = NS_TEMPLATE;
3222                         # Split the title into page and subpage
3223                         $subpage = '';
3224                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3225                         if ( $subpage !== '' ) {
3226                                 $ns = $this->mTitle->getNamespace();
3227                         }
3228                         $title = Title::newFromText( $part1, $ns );
3229                         if ( $title ) {
3230                                 $titleText = $title->getPrefixedText();
3231                                 # Check for language variants if the template is not found
3232                                 if ( $wgContLang->hasVariants() && $title->getArticleID() == 0 ) {
3233                                         $wgContLang->findVariantLink( $part1, $title, true );
3234                                 }
3235                                 # Do recursion depth check
3236                                 $limit = $this->mOptions->getMaxTemplateDepth();
3237                                 if ( $frame->depth >= $limit ) {
3238                                         $found = true;
3239                                         $text = '<span class="error">'
3240                                                 . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit )
3241                                                 . '</span>';
3242                                 }
3243                         }
3244                 }
3245
3246                 # Load from database
3247                 if ( !$found && $title ) {
3248                         wfProfileIn( __METHOD__ . '-loadtpl' );
3249                         if ( !$title->isExternal() ) {
3250                                 if ( $title->getNamespace() == NS_SPECIAL
3251                                         && $this->mOptions->getAllowSpecialInclusion()
3252                                         && $this->ot['html'] )
3253                                 {
3254                                         $text = SpecialPageFactory::capturePath( $title );
3255                                         if ( is_string( $text ) ) {
3256                                                 $found = true;
3257                                                 $isHTML = true;
3258                                                 $this->disableCache();
3259                                         }
3260                                 } elseif ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3261                                         $found = false; # access denied
3262                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3263                                 } else {
3264                                         list( $text, $title ) = $this->getTemplateDom( $title );
3265                                         if ( $text !== false ) {
3266                                                 $found = true;
3267                                                 $isChildObj = true;
3268                                         }
3269                                 }
3270
3271                                 # If the title is valid but undisplayable, make a link to it
3272                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3273                                         $text = "[[:$titleText]]";
3274                                         $found = true;
3275                                 }
3276                         } elseif ( $title->isTrans() ) {
3277                                 # Interwiki transclusion
3278                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3279                                         $text = $this->interwikiTransclude( $title, 'render' );
3280                                         $isHTML = true;
3281                                 } else {
3282                                         $text = $this->interwikiTransclude( $title, 'raw' );
3283                                         # Preprocess it like a template
3284                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3285                                         $isChildObj = true;
3286                                 }
3287                                 $found = true;
3288                         }
3289
3290                         # Do infinite loop check
3291                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3292                         if ( !$frame->loopCheck( $title ) ) {
3293                                 $found = true;
3294                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3295                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3296                         }
3297                         wfProfileOut( __METHOD__ . '-loadtpl' );
3298                 }
3299
3300                 # If we haven't found text to substitute by now, we're done
3301                 # Recover the source wikitext and return it
3302                 if ( !$found ) {
3303                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3304                         wfProfileOut( __METHOD__ );
3305                         return array( 'object' => $text );
3306                 }
3307
3308                 # Expand DOM-style return values in a child frame
3309                 if ( $isChildObj ) {
3310                         # Clean up argument array
3311                         $newFrame = $frame->newChild( $args, $title );
3312
3313                         if ( $nowiki ) {
3314                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3315                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3316                                 # Expansion is eligible for the empty-frame cache
3317                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3318                                         $text = $this->mTplExpandCache[$titleText];
3319                                 } else {
3320                                         $text = $newFrame->expand( $text );
3321                                         $this->mTplExpandCache[$titleText] = $text;
3322                                 }
3323                         } else {
3324                                 # Uncached expansion
3325                                 $text = $newFrame->expand( $text );
3326                         }
3327                 }
3328                 if ( $isLocalObj && $nowiki ) {
3329                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3330                         $isLocalObj = false;
3331                 }
3332
3333                 # Replace raw HTML by a placeholder
3334                 # Add a blank line preceding, to prevent it from mucking up
3335                 # immediately preceding headings
3336                 if ( $isHTML ) {
3337                         $text = "\n\n" . $this->insertStripItem( $text );
3338                 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3339                         # Escape nowiki-style return values
3340                         $text = wfEscapeWikiText( $text );
3341                 } elseif ( is_string( $text )
3342                         && !$piece['lineStart']
3343                         && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3344                 {
3345                         # Bug 529: if the template begins with a table or block-level
3346                         # element, it should be treated as beginning a new line.
3347                         # This behaviour is somewhat controversial.
3348                         $text = "\n" . $text;
3349                 }
3350
3351                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3352                         # Error, oversize inclusion
3353                         if ( $titleText !== false ) {
3354                                 # Make a working, properly escaped link if possible (bug 23588)
3355                                 $text = "[[:$titleText]]";
3356                         } else {
3357                                 # This will probably not be a working link, but at least it may
3358                                 # provide some hint of where the problem is
3359                                 preg_replace( '/^:/', '', $originalTitle );
3360                                 $text = "[[:$originalTitle]]";
3361                         }
3362                         $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3363                         $this->limitationWarn( 'post-expand-template-inclusion' );
3364                 }
3365
3366                 if ( $isLocalObj ) {
3367                         $ret = array( 'object' => $text );
3368                 } else {
3369                         $ret = array( 'text' => $text );
3370                 }
3371
3372                 wfProfileOut( __METHOD__ );
3373                 return $ret;
3374         }
3375
3376         /**
3377          * Get the semi-parsed DOM representation of a template with a given title,
3378          * and its redirect destination title. Cached.
3379          */
3380         function getTemplateDom( $title ) {
3381                 $cacheTitle = $title;
3382                 $titleText = $title->getPrefixedDBkey();
3383
3384                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3385                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3386                         $title = Title::makeTitle( $ns, $dbk );
3387                         $titleText = $title->getPrefixedDBkey();
3388                 }
3389                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3390                         return array( $this->mTplDomCache[$titleText], $title );
3391                 }
3392
3393                 # Cache miss, go to the database
3394                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3395
3396                 if ( $text === false ) {
3397                         $this->mTplDomCache[$titleText] = false;
3398                         return array( false, $title );
3399                 }
3400
3401                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3402                 $this->mTplDomCache[ $titleText ] = $dom;
3403
3404                 if ( !$title->equals( $cacheTitle ) ) {
3405                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3406                                 array( $title->getNamespace(), $cdb = $title->getDBkey() );
3407                 }
3408
3409                 return array( $dom, $title );
3410         }
3411
3412         /**
3413          * Fetch the unparsed text of a template and register a reference to it.
3414          * @param Title $title
3415          * @return Array ( string or false, Title )
3416          */
3417         function fetchTemplateAndTitle( $title ) {
3418                 $templateCb = $this->mOptions->getTemplateCallback(); # Defaults to Parser::statelessFetchTemplate()
3419                 $stuff = call_user_func( $templateCb, $title, $this );
3420                 $text = $stuff['text'];
3421                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3422                 if ( isset( $stuff['deps'] ) ) {
3423                         foreach ( $stuff['deps'] as $dep ) {
3424                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3425                         }
3426                 }
3427                 return array( $text, $finalTitle );
3428         }
3429
3430         /**
3431          * Fetch the unparsed text of a template and register a reference to it.
3432          * @param Title $title
3433          * @return mixed string or false
3434          */
3435         function fetchTemplate( $title ) {
3436                 $rv = $this->fetchTemplateAndTitle( $title );
3437                 return $rv[0];
3438         }
3439
3440         /**
3441          * Static function to get a template
3442          * Can be overridden via ParserOptions::setTemplateCallback().
3443          */
3444         static function statelessFetchTemplate( $title, $parser=false ) {
3445                 $text = $skip = false;
3446                 $finalTitle = $title;
3447                 $deps = array();
3448
3449                 # Loop to fetch the article, with up to 1 redirect
3450                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3451                         # Give extensions a chance to select the revision instead
3452                         $id = false; # Assume current
3453                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle',
3454                                 array( $parser, $title, &$skip, &$id ) );
3455
3456                         if ( $skip ) {
3457                                 $text = false;
3458                                 $deps[] = array(
3459                                         'title'         => $title,
3460                                         'page_id'       => $title->getArticleID(),
3461                                         'rev_id'        => null
3462                                 );
3463                                 break;
3464                         }
3465                         # Get the revision
3466                         $rev = $id
3467                                 ? Revision::newFromId( $id )
3468                                 : Revision::newFromTitle( $title );
3469                         $rev_id = $rev ? $rev->getId() : 0;
3470                         # If there is no current revision, there is no page
3471                         if ( $id === false && !$rev ) {
3472                                 $linkCache = LinkCache::singleton();
3473                                 $linkCache->addBadLinkObj( $title );
3474                         }
3475
3476                         $deps[] = array(
3477                                 'title'         => $title,
3478                                 'page_id'       => $title->getArticleID(),
3479                                 'rev_id'        => $rev_id );
3480                         if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3481                                 # We fetched a rev from a different title; register it too...
3482                                 $deps[] = array(
3483                                         'title'         => $rev->getTitle(),
3484                                         'page_id'       => $rev->getPage(),
3485                                         'rev_id'        => $rev_id );
3486                         }
3487
3488                         if ( $rev ) {
3489                                 $text = $rev->getText();
3490                         } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3491                                 global $wgContLang;
3492                                 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3493                                 if ( !$message->exists() ) {
3494                                         $text = false;
3495                                         break;
3496                                 }
3497                                 $text = $message->plain();
3498                         } else {
3499                                 break;
3500                         }
3501                         if ( $text === false ) {
3502                                 break;
3503                         }
3504                         # Redirect?
3505                         $finalTitle = $title;
3506                         $title = Title::newFromRedirect( $text );
3507                 }
3508                 return array(
3509                         'text' => $text,
3510                         'finalTitle' => $finalTitle,
3511                         'deps' => $deps );
3512         }
3513
3514         /**
3515          * Fetch a file and its title and register a reference to it.
3516          * @param Title $title
3517          * @param string $time MW timestamp
3518          * @param string $sha1 base 36 SHA-1
3519          * @return mixed File or false
3520          */
3521         function fetchFile( $title, $time = false, $sha1 = false ) {
3522                 $res = $this->fetchFileAndTitle( $title, $time, $sha1 );
3523                 return $res[0];
3524         }
3525
3526         /**
3527          * Fetch a file and its title and register a reference to it.
3528          * @param Title $title
3529          * @param string $time MW timestamp
3530          * @param string $sha1 base 36 SHA-1
3531          * @return Array ( File or false, Title of file )
3532          */
3533         function fetchFileAndTitle( $title, $time = false, $sha1 = false ) {
3534                 if ( $time === '0' ) {
3535                         $file = false; // broken thumbnail forced by hook
3536                 } elseif ( $sha1 ) { // get by (sha1,timestamp)
3537                         $file = RepoGroup::singleton()->findFileFromKey( $sha1, array( 'time' => $time ) );
3538                 } else { // get by (name,timestamp)
3539                         $file = wfFindFile( $title, array( 'time' => $time ) );
3540                 }
3541                 $time = $file ? $file->getTimestamp() : false;
3542                 $sha1 = $file ? $file->getSha1() : false;
3543                 # Register the file as a dependency...
3544                 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3545                 if ( $file && !$title->equals( $file->getTitle() ) ) {
3546                         # We fetched a rev from a different title; register it too...
3547                         $this->mOutput->addImage( $file->getTitle()->getDBkey(), $time, $sha1 );
3548                         # Update fetched file title
3549                         $title = $file->getTitle();
3550                 }
3551                 return array( $file, $title );
3552         }
3553
3554         /**
3555          * Transclude an interwiki link.
3556          */
3557         function interwikiTransclude( $title, $action ) {
3558                 global $wgEnableScaryTranscluding;
3559
3560                 if ( !$wgEnableScaryTranscluding ) {
3561                         return wfMsgForContent('scarytranscludedisabled');
3562                 }
3563
3564                 $url = $title->getFullUrl( "action=$action" );
3565
3566                 if ( strlen( $url ) > 255 ) {
3567                         return wfMsgForContent( 'scarytranscludetoolong' );
3568                 }
3569                 return $this->fetchScaryTemplateMaybeFromCache( $url );
3570         }
3571
3572         function fetchScaryTemplateMaybeFromCache( $url ) {
3573                 global $wgTranscludeCacheExpiry;
3574                 $dbr = wfGetDB( DB_SLAVE );
3575                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3576                 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3577                                 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3578                 if ( $obj ) {
3579                         return $obj->tc_contents;
3580                 }
3581
3582                 $text = Http::get( $url );
3583                 if ( !$text ) {
3584                         return wfMsgForContent( 'scarytranscludefailed', $url );
3585                 }
3586
3587                 $dbw = wfGetDB( DB_MASTER );
3588                 $dbw->replace( 'transcache', array('tc_url'), array(
3589                         'tc_url' => $url,
3590                         'tc_time' => $dbw->timestamp( time() ),
3591                         'tc_contents' => $text)
3592                 );
3593                 return $text;
3594         }
3595
3596
3597         /**
3598          * Triple brace replacement -- used for template arguments
3599          * @private
3600          */
3601         function argSubstitution( $piece, $frame ) {
3602                 wfProfileIn( __METHOD__ );
3603
3604                 $error = false;
3605                 $parts = $piece['parts'];
3606                 $nameWithSpaces = $frame->expand( $piece['title'] );
3607                 $argName = trim( $nameWithSpaces );
3608                 $object = false;
3609                 $text = $frame->getArgument( $argName );
3610                 if (  $text === false && $parts->getLength() > 0
3611                   && (
3612                         $this->ot['html']
3613                         || $this->ot['pre']
3614                         || ( $this->ot['wiki'] && $frame->isTemplate() )
3615                   )
3616                 ) {
3617                         # No match in frame, use the supplied default
3618                         $object = $parts->item( 0 )->getChildren();
3619                 }
3620                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3621                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3622                         $this->limitationWarn( 'post-expand-template-argument' );
3623                 }
3624
3625                 if ( $text === false && $object === false ) {
3626                         # No match anywhere
3627                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3628                 }
3629                 if ( $error !== false ) {
3630                         $text .= $error;
3631                 }
3632                 if ( $object !== false ) {
3633                         $ret = array( 'object' => $object );
3634                 } else {
3635                         $ret = array( 'text' => $text );
3636                 }
3637
3638                 wfProfileOut( __METHOD__ );
3639                 return $ret;
3640         }
3641
3642         /**
3643          * Return the text to be used for a given extension tag.
3644          * This is the ghost of strip().
3645          *
3646          * @param $params Associative array of parameters:
3647          *     name       PPNode for the tag name
3648          *     attr       PPNode for unparsed text where tag attributes are thought to be
3649          *     attributes Optional associative array of parsed attributes
3650          *     inner      Contents of extension element
3651          *     noClose    Original text did not have a close tag
3652          * @param $frame PPFrame
3653          */
3654         function extensionSubstitution( $params, $frame ) {
3655                 $name = $frame->expand( $params['name'] );
3656                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3657                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3658                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3659
3660                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3661                         ( $this->ot['html'] || $this->ot['pre'] );
3662                 if ( $isFunctionTag ) {
3663                         $markerType = 'none';
3664                 } else {
3665                         $markerType = 'general';
3666                 }
3667                 if ( $this->ot['html'] || $isFunctionTag ) {
3668                         $name = strtolower( $name );
3669                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3670                         if ( isset( $params['attributes'] ) ) {
3671                                 $attributes = $attributes + $params['attributes'];
3672                         }
3673
3674                         if ( isset( $this->mTagHooks[$name] ) ) {
3675                                 # Workaround for PHP bug 35229 and similar
3676                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3677                                         throw new MWException( "Tag hook for $name is not callable\n" );
3678                                 }
3679                                 $output = call_user_func_array( $this->mTagHooks[$name],
3680                                         array( $content, $attributes, $this, $frame ) );
3681                         } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3682                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3683                                 if ( !is_callable( $callback ) ) {
3684                                         throw new MWException( "Tag hook for $name is not callable\n" );
3685                                 }
3686
3687                                 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3688                         } else {
3689                                 $output = '<span class="error">Invalid tag extension name: ' .
3690                                         htmlspecialchars( $name ) . '</span>';
3691                         }
3692
3693                         if ( is_array( $output ) ) {
3694                                 # Extract flags to local scope (to override $markerType)
3695                                 $flags = $output;
3696                                 $output = $flags[0];
3697                                 unset( $flags[0] );
3698                                 extract( $flags );
3699                         }
3700                 } else {
3701                         if ( is_null( $attrText ) ) {
3702                                 $attrText = '';
3703                         }
3704                         if ( isset( $params['attributes'] ) ) {
3705                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3706                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3707                                                 htmlspecialchars( $attrValue ) . '"';
3708                                 }
3709                         }
3710                         if ( $content === null ) {
3711                                 $output = "<$name$attrText/>";
3712                         } else {
3713                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3714                                 $output = "<$name$attrText>$content$close";
3715                         }
3716                 }
3717
3718                 if ( $markerType === 'none' ) {
3719                         return $output;
3720                 } elseif ( $markerType === 'nowiki' ) {
3721                         $this->mStripState->addNoWiki( $marker, $output );
3722                 } elseif ( $markerType === 'general' ) {
3723                         $this->mStripState->addGeneral( $marker, $output );
3724                 } else {
3725                         throw new MWException( __METHOD__.': invalid marker type' );
3726                 }
3727                 return $marker;
3728         }
3729
3730         /**
3731          * Increment an include size counter
3732          *
3733          * @param $type String: the type of expansion
3734          * @param $size Integer: the size of the text
3735          * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
3736          */
3737         function incrementIncludeSize( $type, $size ) {
3738                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3739                         return false;
3740                 } else {
3741                         $this->mIncludeSizes[$type] += $size;
3742                         return true;
3743                 }
3744         }
3745
3746         /**
3747          * Increment the expensive function count
3748          *
3749          * @return Boolean: false if the limit has been exceeded
3750          */
3751         function incrementExpensiveFunctionCount() {
3752                 global $wgExpensiveParserFunctionLimit;
3753                 $this->mExpensiveFunctionCount++;
3754                 if ( $this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit ) {
3755                         return true;
3756                 }
3757                 return false;
3758         }
3759
3760         /**
3761          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3762          * Fills $this->mDoubleUnderscores, returns the modified text
3763          */
3764         function doDoubleUnderscore( $text ) {
3765                 wfProfileIn( __METHOD__ );
3766
3767                 # The position of __TOC__ needs to be recorded
3768                 $mw = MagicWord::get( 'toc' );
3769                 if ( $mw->match( $text ) ) {
3770                         $this->mShowToc = true;
3771                         $this->mForceTocPosition = true;
3772
3773                         # Set a placeholder. At the end we'll fill it in with the TOC.
3774                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3775
3776                         # Only keep the first one.
3777                         $text = $mw->replace( '', $text );
3778                 }
3779
3780                 # Now match and remove the rest of them
3781                 $mwa = MagicWord::getDoubleUnderscoreArray();
3782                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3783
3784                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3785                         $this->mOutput->mNoGallery = true;
3786                 }
3787                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3788                         $this->mShowToc = false;
3789                 }
3790                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3791                         $this->addTrackingCategory( 'hidden-category-category' );
3792                 }
3793                 # (bug 8068) Allow control over whether robots index a page.
3794                 #
3795                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3796                 # is not desirable, the last one on the page should win.
3797                 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3798                         $this->mOutput->setIndexPolicy( 'noindex' );
3799                         $this->addTrackingCategory( 'noindex-category' );
3800                 }
3801                 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3802                         $this->mOutput->setIndexPolicy( 'index' );
3803                         $this->addTrackingCategory( 'index-category' );
3804                 }
3805
3806                 # Cache all double underscores in the database
3807                 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3808                         $this->mOutput->setProperty( $key, '' );
3809                 }
3810
3811                 wfProfileOut( __METHOD__ );
3812                 return $text;
3813         }
3814
3815         /**
3816          * Add a tracking category, getting the title from a system message,
3817          * or print a debug message if the title is invalid.
3818          *
3819          * @param $msg String: message key
3820          * @return Boolean: whether the addition was successful
3821          */
3822         protected function addTrackingCategory( $msg ) {
3823                 $cat = wfMsgForContent( $msg );
3824
3825                 # Allow tracking categories to be disabled by setting them to "-"
3826                 if ( $cat === '-' ) {
3827                         return false;
3828                 }
3829
3830                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3831                 if ( $containerCategory ) {
3832                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3833                         return true;
3834                 } else {
3835                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3836                         return false;
3837                 }
3838         }
3839
3840         /**
3841          * This function accomplishes several tasks:
3842          * 1) Auto-number headings if that option is enabled
3843          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3844          * 3) Add a Table of contents on the top for users who have enabled the option
3845          * 4) Auto-anchor headings
3846          *
3847          * It loops through all headlines, collects the necessary data, then splits up the
3848          * string and re-inserts the newly formatted headlines.
3849          *
3850          * @param $text String
3851          * @param $origText String: original, untouched wikitext
3852          * @param $isMain Boolean
3853          * @private
3854          */
3855         function formatHeadings( $text, $origText, $isMain=true ) {
3856                 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3857
3858                 # Inhibit editsection links if requested in the page
3859                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3860                         $showEditLink = 0;
3861                 } else {
3862                         $showEditLink = $this->mOptions->getEditSection();
3863                 }
3864                 if ( $showEditLink ) {
3865                         $this->mOutput->setEditSectionTokens( true );
3866                 }
3867
3868                 # Get all headlines for numbering them and adding funky stuff like [edit]
3869                 # links - this is for later, but we need the number of headlines right now
3870                 $matches = array();
3871                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3872
3873                 # if there are fewer than 4 headlines in the article, do not show TOC
3874                 # unless it's been explicitly enabled.
3875                 $enoughToc = $this->mShowToc &&
3876                         ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3877
3878                 # Allow user to stipulate that a page should have a "new section"
3879                 # link added via __NEWSECTIONLINK__
3880                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3881                         $this->mOutput->setNewSection( true );
3882                 }
3883
3884                 # Allow user to remove the "new section"
3885                 # link via __NONEWSECTIONLINK__
3886                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3887                         $this->mOutput->hideNewSection( true );
3888                 }
3889
3890                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3891                 # override above conditions and always show TOC above first header
3892                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3893                         $this->mShowToc = true;
3894                         $enoughToc = true;
3895                 }
3896
3897                 # headline counter
3898                 $headlineCount = 0;
3899                 $numVisible = 0;
3900
3901                 # Ugh .. the TOC should have neat indentation levels which can be
3902                 # passed to the skin functions. These are determined here
3903                 $toc = '';
3904                 $full = '';
3905                 $head = array();
3906                 $sublevelCount = array();
3907                 $levelCount = array();
3908                 $level = 0;
3909                 $prevlevel = 0;
3910                 $toclevel = 0;
3911                 $prevtoclevel = 0;
3912                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3913                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3914                 $oldType = $this->mOutputType;
3915                 $this->setOutputType( self::OT_WIKI );
3916                 $frame = $this->getPreprocessor()->newFrame();
3917                 $root = $this->preprocessToDom( $origText );
3918                 $node = $root->getFirstChild();
3919                 $byteOffset = 0;
3920                 $tocraw = array();
3921                 $refers = array();
3922
3923                 foreach ( $matches[3] as $headline ) {
3924                         $isTemplate = false;
3925                         $titleText = false;
3926                         $sectionIndex = false;
3927                         $numbering = '';
3928                         $markerMatches = array();
3929                         if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
3930                                 $serial = $markerMatches[1];
3931                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3932                                 $isTemplate = ( $titleText != $baseTitleText );
3933                                 $headline = preg_replace( "/^$markerRegex/", "", $headline );
3934                         }
3935
3936                         if ( $toclevel ) {
3937                                 $prevlevel = $level;
3938                         }
3939                         $level = $matches[1][$headlineCount];
3940
3941                         if ( $level > $prevlevel ) {
3942                                 # Increase TOC level
3943                                 $toclevel++;
3944                                 $sublevelCount[$toclevel] = 0;
3945                                 if ( $toclevel<$wgMaxTocLevel ) {
3946                                         $prevtoclevel = $toclevel;
3947                                         $toc .= Linker::tocIndent();
3948                                         $numVisible++;
3949                                 }
3950                         } elseif ( $level < $prevlevel && $toclevel > 1 ) {
3951                                 # Decrease TOC level, find level to jump to
3952
3953                                 for ( $i = $toclevel; $i > 0; $i-- ) {
3954                                         if ( $levelCount[$i] == $level ) {
3955                                                 # Found last matching level
3956                                                 $toclevel = $i;
3957                                                 break;
3958                                         } elseif ( $levelCount[$i] < $level ) {
3959                                                 # Found first matching level below current level
3960                                                 $toclevel = $i + 1;
3961                                                 break;
3962                                         }
3963                                 }
3964                                 if ( $i == 0 ) {
3965                                         $toclevel = 1;
3966                                 }
3967                                 if ( $toclevel<$wgMaxTocLevel ) {
3968                                         if ( $prevtoclevel < $wgMaxTocLevel ) {
3969                                                 # Unindent only if the previous toc level was shown :p
3970                                                 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
3971                                                 $prevtoclevel = $toclevel;
3972                                         } else {
3973                                                 $toc .= Linker::tocLineEnd();
3974                                         }
3975                                 }
3976                         } else {
3977                                 # No change in level, end TOC line
3978                                 if ( $toclevel<$wgMaxTocLevel ) {
3979                                         $toc .= Linker::tocLineEnd();
3980                                 }
3981                         }
3982
3983                         $levelCount[$toclevel] = $level;
3984
3985                         # count number of headlines for each level
3986                         @$sublevelCount[$toclevel]++;
3987                         $dot = 0;
3988                         for( $i = 1; $i <= $toclevel; $i++ ) {
3989                                 if ( !empty( $sublevelCount[$i] ) ) {
3990                                         if ( $dot ) {
3991                                                 $numbering .= '.';
3992                                         }
3993                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3994                                         $dot = 1;
3995                                 }
3996                         }
3997
3998                         # The safe header is a version of the header text safe to use for links
3999                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4000                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
4001
4002                         # Remove link placeholders by the link text.
4003                         #     <!--LINK number-->
4004                         # turns into
4005                         #     link text with suffix
4006                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
4007
4008                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
4009                         $tocline = preg_replace(
4010                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
4011                                 array( '',                          '<$1>' ),
4012                                 $safeHeadline
4013                         );
4014                         $tocline = trim( $tocline );
4015
4016                         # For the anchor, strip out HTML-y stuff period
4017                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
4018                         $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4019
4020                         # Save headline for section edit hint before it's escaped
4021                         $headlineHint = $safeHeadline;
4022
4023                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
4024                                 # For reverse compatibility, provide an id that's
4025                                 # HTML4-compatible, like we used to.
4026                                 #
4027                                 # It may be worth noting, academically, that it's possible for
4028                                 # the legacy anchor to conflict with a non-legacy headline
4029                                 # anchor on the page.  In this case likely the "correct" thing
4030                                 # would be to either drop the legacy anchors or make sure
4031                                 # they're numbered first.  However, this would require people
4032                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4033                                 # manually, so let's not bother worrying about it.
4034                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
4035                                         array( 'noninitial', 'legacy' ) );
4036                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4037
4038                                 if ( $legacyHeadline == $safeHeadline ) {
4039                                         # No reason to have both (in fact, we can't)
4040                                         $legacyHeadline = false;
4041                                 }
4042                         } else {
4043                                 $legacyHeadline = false;
4044                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
4045                                         'noninitial' );
4046                         }
4047
4048                         # HTML names must be case-insensitively unique (bug 10721).
4049                         # This does not apply to Unicode characters per
4050                         # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
4051                         # FIXME: We may be changing them depending on the current locale.
4052                         $arrayKey = strtolower( $safeHeadline );
4053                         if ( $legacyHeadline === false ) {
4054                                 $legacyArrayKey = false;
4055                         } else {
4056                                 $legacyArrayKey = strtolower( $legacyHeadline );
4057                         }
4058
4059                         # count how many in assoc. array so we can track dupes in anchors
4060                         if ( isset( $refers[$arrayKey] ) ) {
4061                                 $refers[$arrayKey]++;
4062                         } else {
4063                                 $refers[$arrayKey] = 1;
4064                         }
4065                         if ( isset( $refers[$legacyArrayKey] ) ) {
4066                                 $refers[$legacyArrayKey]++;
4067                         } else {
4068                                 $refers[$legacyArrayKey] = 1;
4069                         }
4070
4071                         # Don't number the heading if it is the only one (looks silly)
4072                         if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4073                                 # the two are different if the line contains a link
4074                                 $headline = $numbering . ' ' . $headline;
4075                         }
4076
4077                         # Create the anchor for linking from the TOC to the section
4078                         $anchor = $safeHeadline;
4079                         $legacyAnchor = $legacyHeadline;
4080                         if ( $refers[$arrayKey] > 1 ) {
4081                                 $anchor .= '_' . $refers[$arrayKey];
4082                         }
4083                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
4084                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
4085                         }
4086                         if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4087                                 $toc .= Linker::tocLine( $anchor, $tocline,
4088                                         $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4089                         }
4090
4091                         # Add the section to the section tree
4092                         # Find the DOM node for this header
4093                         while ( $node && !$isTemplate ) {
4094                                 if ( $node->getName() === 'h' ) {
4095                                         $bits = $node->splitHeading();
4096                                         if ( $bits['i'] == $sectionIndex ) {
4097                                                 break;
4098                                         }
4099                                 }
4100                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4101                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4102                                 $node = $node->getNextSibling();
4103                         }
4104                         $tocraw[] = array(
4105                                 'toclevel' => $toclevel,
4106                                 'level' => $level,
4107                                 'line' => $tocline,
4108                                 'number' => $numbering,
4109                                 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4110                                 'fromtitle' => $titleText,
4111                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
4112                                 'anchor' => $anchor,
4113                         );
4114
4115                         # give headline the correct <h#> tag
4116                         if ( $showEditLink && $sectionIndex !== false ) {
4117                                 // Output edit section links as markers with styles that can be customized by skins
4118                                 if ( $isTemplate ) {
4119                                         # Put a T flag in the section identifier, to indicate to extractSections()
4120                                         # that sections inside <includeonly> should be counted.
4121                                         $editlinkArgs = array( $titleText, "T-$sectionIndex"/*, null */ );
4122                                 } else {
4123                                         $editlinkArgs = array( $this->mTitle->getPrefixedText(), $sectionIndex, $headlineHint );
4124                                 }
4125                                 // We use a bit of pesudo-xml for editsection markers. The language converter is run later on
4126                                 // Using a UNIQ style marker leads to the converter screwing up the tokens when it converts stuff
4127                                 // And trying to insert strip tags fails too. At this point all real inputted tags have already been escaped
4128                                 // so we don't have to worry about a user trying to input one of these markers directly.
4129                                 // We use a page and section attribute to stop the language converter from converting these important bits
4130                                 // of data, but put the headline hint inside a content block because the language converter is supposed to
4131                                 // be able to convert that piece of data.
4132                                 $editlink = '<mw:editsection page="' . htmlspecialchars($editlinkArgs[0]);
4133                                 $editlink .= '" section="' . htmlspecialchars($editlinkArgs[1]) .'"';
4134                                 if ( isset($editlinkArgs[2]) ) {
4135                                         $editlink .= '>' . $editlinkArgs[2] . '</mw:editsection>';
4136                                 } else {
4137                                         $editlink .= '/>';
4138                                 }
4139                         } else {
4140                                 $editlink = '';
4141                         }
4142                         $head[$headlineCount] = Linker::makeHeadline( $level,
4143                                 $matches['attrib'][$headlineCount], $anchor, $headline,
4144                                 $editlink, $legacyAnchor );
4145
4146                         $headlineCount++;
4147                 }
4148
4149                 $this->setOutputType( $oldType );
4150
4151                 # Never ever show TOC if no headers
4152                 if ( $numVisible < 1 ) {
4153                         $enoughToc = false;
4154                 }
4155
4156                 if ( $enoughToc ) {
4157                         if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4158                                 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4159                         }
4160                         $toc = Linker::tocList( $toc, $this->mOptions->getUserLang() );
4161                         $this->mOutput->setTOCHTML( $toc );
4162                 }
4163
4164                 if ( $isMain ) {
4165                         $this->mOutput->setSections( $tocraw );
4166                 }
4167
4168                 # split up and insert constructed headlines
4169
4170                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
4171                 $i = 0;
4172
4173                 foreach ( $blocks as $block ) {
4174                         if ( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
4175                                 # This is the [edit] link that appears for the top block of text when
4176                                 # section editing is enabled
4177
4178                                 # Disabled because it broke block formatting
4179                                 # For example, a bullet point in the top line
4180                                 # $full .= $sk->editSectionLink(0);
4181                         }
4182                         $full .= $block;
4183                         if ( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
4184                                 # Top anchor now in skin
4185                                 $full = $full.$toc;
4186                         }
4187
4188                         if ( !empty( $head[$i] ) ) {
4189                                 $full .= $head[$i];
4190                         }
4191                         $i++;
4192                 }
4193                 if ( $this->mForceTocPosition ) {
4194                         return str_replace( '<!--MWTOC-->', $toc, $full );
4195                 } else {
4196                         return $full;
4197                 }
4198         }
4199
4200         /**
4201          * Transform wiki markup when saving a page by doing \r\n -> \n
4202          * conversion, substitting signatures, {{subst:}} templates, etc.
4203          *
4204          * @param $text String: the text to transform
4205          * @param $title Title: the Title object for the current article
4206          * @param $user User: the User object describing the current user
4207          * @param $options ParserOptions: parsing options
4208          * @param $clearState Boolean: whether to clear the parser state first
4209          * @return String: the altered wiki markup
4210          */
4211         public function preSaveTransform( $text, Title $title, User $user, ParserOptions $options, $clearState = true ) {
4212                 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4213                 $this->setUser( $user );
4214
4215                 $pairs = array(
4216                         "\r\n" => "\n",
4217                 );
4218                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4219                 if( $options->getPreSaveTransform() ) {
4220                         $text = $this->pstPass2( $text, $user );
4221                 }
4222                 $text = $this->mStripState->unstripBoth( $text );
4223
4224                 $this->setUser( null ); #Reset
4225
4226                 return $text;
4227         }
4228
4229         /**
4230          * Pre-save transform helper function
4231          * @private
4232          */
4233         function pstPass2( $text, $user ) {
4234                 global $wgContLang, $wgLocaltimezone;
4235
4236                 # Note: This is the timestamp saved as hardcoded wikitext to
4237                 # the database, we use $wgContLang here in order to give
4238                 # everyone the same signature and use the default one rather
4239                 # than the one selected in each user's preferences.
4240                 # (see also bug 12815)
4241                 $ts = $this->mOptions->getTimestamp();
4242                 if ( isset( $wgLocaltimezone ) ) {
4243                         $tz = $wgLocaltimezone;
4244                 } else {
4245                         $tz = date_default_timezone_get();
4246                 }
4247
4248                 $unixts = wfTimestamp( TS_UNIX, $ts );
4249                 $oldtz = date_default_timezone_get();
4250                 date_default_timezone_set( $tz );
4251                 $ts = date( 'YmdHis', $unixts );
4252                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4253
4254                 # Allow translation of timezones through wiki. date() can return
4255                 # whatever crap the system uses, localised or not, so we cannot
4256                 # ship premade translations.
4257                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4258                 $msg = wfMessage( $key )->inContentLanguage();
4259                 if ( $msg->exists() ) {
4260                         $tzMsg = $msg->text();
4261                 }
4262
4263                 date_default_timezone_set( $oldtz );
4264
4265                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4266
4267                 # Variable replacement
4268                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4269                 $text = $this->replaceVariables( $text );
4270
4271                 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4272                 # which may corrupt this parser instance via its wfMsgExt( parsemag ) call-
4273
4274                 # Signatures
4275                 $sigText = $this->getUserSig( $user );
4276                 $text = strtr( $text, array(
4277                         '~~~~~' => $d,
4278                         '~~~~' => "$sigText $d",
4279                         '~~~' => $sigText
4280                 ) );
4281
4282                 # Context links: [[|name]] and [[name (context)|]]
4283                 global $wgLegalTitleChars;
4284                 $tc = "[$wgLegalTitleChars]";
4285                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4286
4287                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4288                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
4289                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4290                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4291
4292                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4293                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4294                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4295                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4296
4297                 $t = $this->mTitle->getText();
4298                 $m = array();
4299                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4300                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4301                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4302                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4303                 } else {
4304                         # if there's no context, don't bother duplicating the title
4305                         $text = preg_replace( $p2, '[[\\1]]', $text );
4306                 }
4307
4308                 # Trim trailing whitespace
4309                 $text = rtrim( $text );
4310
4311                 return $text;
4312         }
4313
4314         /**
4315          * Fetch the user's signature text, if any, and normalize to
4316          * validated, ready-to-insert wikitext.
4317          * If you have pre-fetched the nickname or the fancySig option, you can
4318          * specify them here to save a database query.
4319          * Do not reuse this parser instance after calling getUserSig(),
4320          * as it may have changed if it's the $wgParser.
4321          *
4322          * @param $user User
4323          * @param $nickname String: nickname to use or false to use user's default nickname
4324          * @param $fancySig Boolean: whether the nicknname is the complete signature
4325          *                  or null to use default value
4326          * @return string
4327          */
4328         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4329                 global $wgMaxSigChars;
4330
4331                 $username = $user->getName();
4332
4333                 # If not given, retrieve from the user object.
4334                 if ( $nickname === false )
4335                         $nickname = $user->getOption( 'nickname' );
4336
4337                 if ( is_null( $fancySig ) ) {
4338                         $fancySig = $user->getBoolOption( 'fancysig' );
4339                 }
4340
4341                 $nickname = $nickname == null ? $username : $nickname;
4342
4343                 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4344                         $nickname = $username;
4345                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4346                 } elseif ( $fancySig !== false ) {
4347                         # Sig. might contain markup; validate this
4348                         if ( $this->validateSig( $nickname ) !== false ) {
4349                                 # Validated; clean up (if needed) and return it
4350                                 return $this->cleanSig( $nickname, true );
4351                         } else {
4352                                 # Failed to validate; fall back to the default
4353                                 $nickname = $username;
4354                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4355                         }
4356                 }
4357
4358                 # Make sure nickname doesnt get a sig in a sig
4359                 $nickname = $this->cleanSigInSig( $nickname );
4360
4361                 # If we're still here, make it a link to the user page
4362                 $userText = wfEscapeWikiText( $username );
4363                 $nickText = wfEscapeWikiText( $nickname );
4364                 if ( $user->isAnon() )  {
4365                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4366                 } else {
4367                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4368                 }
4369         }
4370
4371         /**
4372          * Check that the user's signature contains no bad XML
4373          *
4374          * @param $text String
4375          * @return mixed An expanded string, or false if invalid.
4376          */
4377         function validateSig( $text ) {
4378                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4379         }
4380
4381         /**
4382          * Clean up signature text
4383          *
4384          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4385          * 2) Substitute all transclusions
4386          *
4387          * @param $text String
4388          * @param $parsing Whether we're cleaning (preferences save) or parsing
4389          * @return String: signature text
4390          */
4391         function cleanSig( $text, $parsing = false ) {
4392                 if ( !$parsing ) {
4393                         global $wgTitle;
4394                         $this->mOptions = new ParserOptions;
4395                         $this->clearState();
4396                         $this->setTitle( $wgTitle );
4397                         $this->setOutputType = self::OT_PREPROCESS;
4398                 }
4399
4400                 # Option to disable this feature
4401                 if ( !$this->mOptions->getCleanSignatures() ) {
4402                         return $text;
4403                 }
4404
4405                 # FIXME: regex doesn't respect extension tags or nowiki
4406                 #  => Move this logic to braceSubstitution()
4407                 $substWord = MagicWord::get( 'subst' );
4408                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4409                 $substText = '{{' . $substWord->getSynonym( 0 );
4410
4411                 $text = preg_replace( $substRegex, $substText, $text );
4412                 $text = $this->cleanSigInSig( $text );
4413                 $dom = $this->preprocessToDom( $text );
4414                 $frame = $this->getPreprocessor()->newFrame();
4415                 $text = $frame->expand( $dom );
4416
4417                 if ( !$parsing ) {
4418                         $text = $this->mStripState->unstripBoth( $text );
4419                 }
4420
4421                 return $text;
4422         }
4423
4424         /**
4425          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4426          *
4427          * @param $text String
4428          * @return String: signature text with /~{3,5}/ removed
4429          */
4430         function cleanSigInSig( $text ) {
4431                 $text = preg_replace( '/~{3,5}/', '', $text );
4432                 return $text;
4433         }
4434
4435         /**
4436          * Set up some variables which are usually set up in parse()
4437          * so that an external function can call some class members with confidence
4438          */
4439         public function startExternalParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4440                 $this->startParse( $title, $options, $outputType, $clearState );
4441         }
4442
4443         private function startParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4444                 $this->setTitle( $title );
4445                 $this->mOptions = $options;
4446                 $this->setOutputType( $outputType );
4447                 if ( $clearState ) {
4448                         $this->clearState();
4449                 }
4450         }
4451
4452         /**
4453          * Wrapper for preprocess()
4454          *
4455          * @param $text String: the text to preprocess
4456          * @param $options ParserOptions: options
4457          * @param $title Title object or null to use $wgTitle
4458          * @return String
4459          */
4460         public function transformMsg( $text, $options, $title = null ) {
4461                 static $executing = false;
4462
4463                 # Guard against infinite recursion
4464                 if ( $executing ) {
4465                         return $text;
4466                 }
4467                 $executing = true;
4468
4469                 wfProfileIn( __METHOD__ );
4470                 if ( !$title ) {
4471                         global $wgTitle;
4472                         $title = $wgTitle;
4473                 }
4474                 if ( !$title ) {
4475                         # It's not uncommon having a null $wgTitle in scripts. See r80898
4476                         # Create a ghost title in such case
4477                         $title = Title::newFromText( 'Dwimmerlaik' );
4478                 }
4479                 $text = $this->preprocess( $text, $title, $options );
4480
4481                 $executing = false;
4482                 wfProfileOut( __METHOD__ );
4483                 return $text;
4484         }
4485
4486         /**
4487          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4488          * The callback should have the following form:
4489          *    function myParserHook( $text, $params, $parser, $frame ) { ... }
4490          *
4491          * Transform and return $text. Use $parser for any required context, e.g. use
4492          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4493          *
4494          * Hooks may return extended information by returning an array, of which the
4495          * first numbered element (index 0) must be the return string, and all other
4496          * entries are extracted into local variables within an internal function
4497          * in the Parser class.
4498          *
4499          * This interface (introduced r61913) appears to be undocumented, but
4500          * 'markerName' is used by some core tag hooks to override which strip
4501          * array their results are placed in. **Use great caution if attempting
4502          * this interface, as it is not documented and injudicious use could smash
4503          * private variables.**
4504          *
4505          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4506          * @param $callback Mixed: the callback function (and object) to use for the tag
4507          * @return The old value of the mTagHooks array associated with the hook
4508          */
4509         public function setHook( $tag, $callback ) {
4510                 $tag = strtolower( $tag );
4511                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4512                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4513                 $this->mTagHooks[$tag] = $callback;
4514                 if ( !in_array( $tag, $this->mStripList ) ) {
4515                         $this->mStripList[] = $tag;
4516                 }
4517
4518                 return $oldVal;
4519         }
4520
4521         /**
4522          * As setHook(), but letting the contents be parsed.
4523          *
4524          * Transparent tag hooks are like regular XML-style tag hooks, except they
4525          * operate late in the transformation sequence, on HTML instead of wikitext.
4526          *
4527          * This is probably obsoleted by things dealing with parser frames?
4528          * The only extension currently using it is geoserver.
4529          *
4530          * @since 1.10
4531          * @todo better document or deprecate this
4532          *
4533          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4534          * @param $callback Mixed: the callback function (and object) to use for the tag
4535          * @return The old value of the mTagHooks array associated with the hook
4536          */
4537         function setTransparentTagHook( $tag, $callback ) {
4538                 $tag = strtolower( $tag );
4539                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4540                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4541                 $this->mTransparentTagHooks[$tag] = $callback;
4542
4543                 return $oldVal;
4544         }
4545
4546         /**
4547          * Remove all tag hooks
4548          */
4549         function clearTagHooks() {
4550                 $this->mTagHooks = array();
4551                 $this->mStripList = $this->mDefaultStripList;
4552         }
4553
4554         /**
4555          * Create a function, e.g. {{sum:1|2|3}}
4556          * The callback function should have the form:
4557          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4558          *
4559          * Or with SFH_OBJECT_ARGS:
4560          *    function myParserFunction( $parser, $frame, $args ) { ... }
4561          *
4562          * The callback may either return the text result of the function, or an array with the text
4563          * in element 0, and a number of flags in the other elements. The names of the flags are
4564          * specified in the keys. Valid flags are:
4565          *   found                     The text returned is valid, stop processing the template. This
4566          *                             is on by default.
4567          *   nowiki                    Wiki markup in the return value should be escaped
4568          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4569          *
4570          * @param $id String: The magic word ID
4571          * @param $callback Mixed: the callback function (and object) to use
4572          * @param $flags Integer: a combination of the following flags:
4573          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4574          *
4575          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4576          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4577          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4578          *     the arguments, and to control the way they are expanded.
4579          *
4580          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4581          *     arguments, for instance:
4582          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4583          *
4584          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4585          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4586          *     working if/when this is changed.
4587          *
4588          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4589          *     expansion.
4590          *
4591          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4592          *     about the methods available in PPFrame and PPNode.
4593          *
4594          * @return The old callback function for this name, if any
4595          */
4596         public function setFunctionHook( $id, $callback, $flags = 0 ) {
4597                 global $wgContLang;
4598
4599                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4600                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4601
4602                 # Add to function cache
4603                 $mw = MagicWord::get( $id );
4604                 if ( !$mw )
4605                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4606
4607                 $synonyms = $mw->getSynonyms();
4608                 $sensitive = intval( $mw->isCaseSensitive() );
4609
4610                 foreach ( $synonyms as $syn ) {
4611                         # Case
4612                         if ( !$sensitive ) {
4613                                 $syn = $wgContLang->lc( $syn );
4614                         }
4615                         # Add leading hash
4616                         if ( !( $flags & SFH_NO_HASH ) ) {
4617                                 $syn = '#' . $syn;
4618                         }
4619                         # Remove trailing colon
4620                         if ( substr( $syn, -1, 1 ) === ':' ) {
4621                                 $syn = substr( $syn, 0, -1 );
4622                         }
4623                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4624                 }
4625                 return $oldVal;
4626         }
4627
4628         /**
4629          * Get all registered function hook identifiers
4630          *
4631          * @return Array
4632          */
4633         function getFunctionHooks() {
4634                 return array_keys( $this->mFunctionHooks );
4635         }
4636
4637         /**
4638          * Create a tag function, e.g. <test>some stuff</test>.
4639          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4640          * Unlike parser functions, their content is not preprocessed.
4641          */
4642         function setFunctionTagHook( $tag, $callback, $flags ) {
4643                 $tag = strtolower( $tag );
4644                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4645                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4646                         $this->mFunctionTagHooks[$tag] : null;
4647                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4648
4649                 if ( !in_array( $tag, $this->mStripList ) ) {
4650                         $this->mStripList[] = $tag;
4651                 }
4652
4653                 return $old;
4654         }
4655
4656         /**
4657          * FIXME: update documentation. makeLinkObj() is deprecated.
4658          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4659          * Placeholders created in Skin::makeLinkObj()
4660          * Returns an array of link CSS classes, indexed by PDBK.
4661          */
4662         function replaceLinkHolders( &$text, $options = 0 ) {
4663                 return $this->mLinkHolders->replace( $text );
4664         }
4665
4666         /**
4667          * Replace <!--LINK--> link placeholders with plain text of links
4668          * (not HTML-formatted).
4669          *
4670          * @param $text String
4671          * @return String
4672          */
4673         function replaceLinkHoldersText( $text ) {
4674                 return $this->mLinkHolders->replaceText( $text );
4675         }
4676
4677         /**
4678          * Renders an image gallery from a text with one line per image.
4679          * text labels may be given by using |-style alternative text. E.g.
4680          *   Image:one.jpg|The number "1"
4681          *   Image:tree.jpg|A tree
4682          * given as text will return the HTML of a gallery with two images,
4683          * labeled 'The number "1"' and
4684          * 'A tree'.
4685          *
4686          * @param string $text
4687          * @param array $param
4688          * @return string HTML
4689          */
4690         function renderImageGallery( $text, $params ) {
4691                 $ig = new ImageGallery();
4692                 $ig->setContextTitle( $this->mTitle );
4693                 $ig->setShowBytes( false );
4694                 $ig->setShowFilename( false );
4695                 $ig->setParser( $this );
4696                 $ig->setHideBadImages();
4697                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4698
4699                 if ( isset( $params['showfilename'] ) ) {
4700                         $ig->setShowFilename( true );
4701                 } else {
4702                         $ig->setShowFilename( false );
4703                 }
4704                 if ( isset( $params['caption'] ) ) {
4705                         $caption = $params['caption'];
4706                         $caption = htmlspecialchars( $caption );
4707                         $caption = $this->replaceInternalLinks( $caption );
4708                         $ig->setCaptionHtml( $caption );
4709                 }
4710                 if ( isset( $params['perrow'] ) ) {
4711                         $ig->setPerRow( $params['perrow'] );
4712                 }
4713                 if ( isset( $params['widths'] ) ) {
4714                         $ig->setWidths( $params['widths'] );
4715                 }
4716                 if ( isset( $params['heights'] ) ) {
4717                         $ig->setHeights( $params['heights'] );
4718                 }
4719
4720                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4721
4722                 $lines = StringUtils::explode( "\n", $text );
4723                 foreach ( $lines as $line ) {
4724                         # match lines like these:
4725                         # Image:someimage.jpg|This is some image
4726                         $matches = array();
4727                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4728                         # Skip empty lines
4729                         if ( count( $matches ) == 0 ) {
4730                                 continue;
4731                         }
4732
4733                         if ( strpos( $matches[0], '%' ) !== false ) {
4734                                 $matches[1] = rawurldecode( $matches[1] );
4735                         }
4736                         $title = Title::newFromText( $matches[1], NS_FILE );
4737                         if ( is_null( $title ) ) {
4738                                 # Bogus title. Ignore these so we don't bomb out later.
4739                                 continue;
4740                         }
4741
4742                         $label = '';
4743                         $alt = '';
4744                         if ( isset( $matches[3] ) ) {
4745                                 // look for an |alt= definition while trying not to break existing
4746                                 // captions with multiple pipes (|) in it, until a more sensible grammar
4747                                 // is defined for images in galleries
4748
4749                                 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
4750                                 $altmatches = StringUtils::explode('|', $matches[3]);
4751                                 $magicWordAlt = MagicWord::get( 'img_alt' );
4752
4753                                 foreach ( $altmatches as $altmatch ) {
4754                                         $match = $magicWordAlt->matchVariableStartToEnd( $altmatch );
4755                                         if ( $match ) {
4756                                                 $alt = $this->stripAltText( $match, false );
4757                                         }
4758                                         else {
4759                                                 // concatenate all other pipes
4760                                                 $label .= '|' . $altmatch;
4761                                         }
4762                                 }
4763                                 // remove the first pipe
4764                                 $label = substr( $label, 1 );
4765                         }
4766
4767                         $ig->add( $title, $label, $alt );
4768                 }
4769                 return $ig->toHTML();
4770         }
4771
4772         function getImageParams( $handler ) {
4773                 if ( $handler ) {
4774                         $handlerClass = get_class( $handler );
4775                 } else {
4776                         $handlerClass = '';
4777                 }
4778                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4779                         # Initialise static lists
4780                         static $internalParamNames = array(
4781                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4782                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4783                                         'bottom', 'text-bottom' ),
4784                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4785                                         'upright', 'border', 'link', 'alt' ),
4786                         );
4787                         static $internalParamMap;
4788                         if ( !$internalParamMap ) {
4789                                 $internalParamMap = array();
4790                                 foreach ( $internalParamNames as $type => $names ) {
4791                                         foreach ( $names as $name ) {
4792                                                 $magicName = str_replace( '-', '_', "img_$name" );
4793                                                 $internalParamMap[$magicName] = array( $type, $name );
4794                                         }
4795                                 }
4796                         }
4797
4798                         # Add handler params
4799                         $paramMap = $internalParamMap;
4800                         if ( $handler ) {
4801                                 $handlerParamMap = $handler->getParamMap();
4802                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4803                                         $paramMap[$magic] = array( 'handler', $paramName );
4804                                 }
4805                         }
4806                         $this->mImageParams[$handlerClass] = $paramMap;
4807                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4808                 }
4809                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4810         }
4811
4812         /**
4813          * Parse image options text and use it to make an image
4814          *
4815          * @param $title Title
4816          * @param $options String
4817          * @param $holders LinkHolderArray
4818          * @return string HTML
4819          */
4820         function makeImage( $title, $options, $holders = false ) {
4821                 # Check if the options text is of the form "options|alt text"
4822                 # Options are:
4823                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4824                 #  * left       no resizing, just left align. label is used for alt= only
4825                 #  * right      same, but right aligned
4826                 #  * none       same, but not aligned
4827                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4828                 #  * center     center the image
4829                 #  * frame      Keep original image size, no magnify-button.
4830                 #  * framed     Same as "frame"
4831                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4832                 #  * upright    reduce width for upright images, rounded to full __0 px
4833                 #  * border     draw a 1px border around the image
4834                 #  * alt        Text for HTML alt attribute (defaults to empty)
4835                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4836                 # vertical-align values (no % or length right now):
4837                 #  * baseline
4838                 #  * sub
4839                 #  * super
4840                 #  * top
4841                 #  * text-top
4842                 #  * middle
4843                 #  * bottom
4844                 #  * text-bottom
4845
4846                 $parts = StringUtils::explode( "|", $options );
4847
4848                 # Give extensions a chance to select the file revision for us
4849                 $time = $sha1 = $descQuery = false;
4850                 wfRunHooks( 'BeforeParserFetchFileAndTitle',
4851                         array( $this, $title, &$time, &$sha1, &$descQuery ) );
4852                 # Fetch and register the file (file title may be different via hooks)
4853                 list( $file, $title ) = $this->fetchFileAndTitle( $title, $time, $sha1 );
4854
4855                 # Get parameter map
4856                 $handler = $file ? $file->getHandler() : false;
4857
4858                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4859
4860                 if ( !$file ) {
4861                         $this->addTrackingCategory( 'broken-file-category' );
4862                 }
4863
4864                 # Process the input parameters
4865                 $caption = '';
4866                 $params = array( 'frame' => array(), 'handler' => array(),
4867                         'horizAlign' => array(), 'vertAlign' => array() );
4868                 foreach ( $parts as $part ) {
4869                         $part = trim( $part );
4870                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4871                         $validated = false;
4872                         if ( isset( $paramMap[$magicName] ) ) {
4873                                 list( $type, $paramName ) = $paramMap[$magicName];
4874
4875                                 # Special case; width and height come in one variable together
4876                                 if ( $type === 'handler' && $paramName === 'width' ) {
4877                                         $m = array();
4878                                         # (bug 13500) In both cases (width/height and width only),
4879                                         # permit trailing "px" for backward compatibility.
4880                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4881                                                 $width = intval( $m[1] );
4882                                                 $height = intval( $m[2] );
4883                                                 if ( $handler->validateParam( 'width', $width ) ) {
4884                                                         $params[$type]['width'] = $width;
4885                                                         $validated = true;
4886                                                 }
4887                                                 if ( $handler->validateParam( 'height', $height ) ) {
4888                                                         $params[$type]['height'] = $height;
4889                                                         $validated = true;
4890                                                 }
4891                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4892                                                 $width = intval( $value );
4893                                                 if ( $handler->validateParam( 'width', $width ) ) {
4894                                                         $params[$type]['width'] = $width;
4895                                                         $validated = true;
4896                                                 }
4897                                         } # else no validation -- bug 13436
4898                                 } else {
4899                                         if ( $type === 'handler' ) {
4900                                                 # Validate handler parameter
4901                                                 $validated = $handler->validateParam( $paramName, $value );
4902                                         } else {
4903                                                 # Validate internal parameters
4904                                                 switch( $paramName ) {
4905                                                 case 'manualthumb':
4906                                                 case 'alt':
4907                                                         # @todo Fixme: possibly check validity here for
4908                                                         # manualthumb? downstream behavior seems odd with
4909                                                         # missing manual thumbs.
4910                                                         $validated = true;
4911                                                         $value = $this->stripAltText( $value, $holders );
4912                                                         break;
4913                                                 case 'link':
4914                                                         $chars = self::EXT_LINK_URL_CLASS;
4915                                                         $prots = $this->mUrlProtocols;
4916                                                         if ( $value === '' ) {
4917                                                                 $paramName = 'no-link';
4918                                                                 $value = true;
4919                                                                 $validated = true;
4920                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4921                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4922                                                                         $paramName = 'link-url';
4923                                                                         $this->mOutput->addExternalLink( $value );
4924                                                                         if ( $this->mOptions->getExternalLinkTarget() ) {
4925                                                                                 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
4926                                                                         }
4927                                                                         $validated = true;
4928                                                                 }
4929                                                         } else {
4930                                                                 $linkTitle = Title::newFromText( $value );
4931                                                                 if ( $linkTitle ) {
4932                                                                         $paramName = 'link-title';
4933                                                                         $value = $linkTitle;
4934                                                                         $this->mOutput->addLink( $linkTitle );
4935                                                                         $validated = true;
4936                                                                 }
4937                                                         }
4938                                                         break;
4939                                                 default:
4940                                                         # Most other things appear to be empty or numeric...
4941                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4942                                                 }
4943                                         }
4944
4945                                         if ( $validated ) {
4946                                                 $params[$type][$paramName] = $value;
4947                                         }
4948                                 }
4949                         }
4950                         if ( !$validated ) {
4951                                 $caption = $part;
4952                         }
4953                 }
4954
4955                 # Process alignment parameters
4956                 if ( $params['horizAlign'] ) {
4957                         $params['frame']['align'] = key( $params['horizAlign'] );
4958                 }
4959                 if ( $params['vertAlign'] ) {
4960                         $params['frame']['valign'] = key( $params['vertAlign'] );
4961                 }
4962
4963                 $params['frame']['caption'] = $caption;
4964
4965                 # Will the image be presented in a frame, with the caption below?
4966                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4967                                                  isset( $params['frame']['framed'] ) ||
4968                                                  isset( $params['frame']['thumbnail'] ) ||
4969                                                  isset( $params['frame']['manualthumb'] );
4970
4971                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4972                 # came to also set the caption, ordinary text after the image -- which
4973                 # makes no sense, because that just repeats the text multiple times in
4974                 # screen readers.  It *also* came to set the title attribute.
4975                 #
4976                 # Now that we have an alt attribute, we should not set the alt text to
4977                 # equal the caption: that's worse than useless, it just repeats the
4978                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4979                 # use the unnamed parameter for alt text as well, just for the time be-
4980                 # ing, if the unnamed param is set and the alt param is not.
4981                 #
4982                 # For the future, we need to figure out if we want to tweak this more,
4983                 # e.g., introducing a title= parameter for the title; ignoring the un-
4984                 # named parameter entirely for images without a caption; adding an ex-
4985                 # plicit caption= parameter and preserving the old magic unnamed para-
4986                 # meter for BC; ...
4987                 if ( $imageIsFramed ) { # Framed image
4988                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4989                                 # No caption or alt text, add the filename as the alt text so
4990                                 # that screen readers at least get some description of the image
4991                                 $params['frame']['alt'] = $title->getText();
4992                         }
4993                         # Do not set $params['frame']['title'] because tooltips don't make sense
4994                         # for framed images
4995                 } else { # Inline image
4996                         if ( !isset( $params['frame']['alt'] ) ) {
4997                                 # No alt text, use the "caption" for the alt text
4998                                 if ( $caption !== '') {
4999                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5000                                 } else {
5001                                         # No caption, fall back to using the filename for the
5002                                         # alt text
5003                                         $params['frame']['alt'] = $title->getText();
5004                                 }
5005                         }
5006                         # Use the "caption" for the tooltip text
5007                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5008                 }
5009
5010                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
5011
5012                 # Linker does the rest
5013                 $ret = Linker::makeImageLink2( $title, $file, $params['frame'], $params['handler'],
5014                         $time, $descQuery, $this->mOptions->getThumbSize() );
5015
5016                 # Give the handler a chance to modify the parser object
5017                 if ( $handler ) {
5018                         $handler->parserTransformHook( $this, $file );
5019                 }
5020
5021                 return $ret;
5022         }
5023
5024         protected function stripAltText( $caption, $holders ) {
5025                 # Strip bad stuff out of the title (tooltip).  We can't just use
5026                 # replaceLinkHoldersText() here, because if this function is called
5027                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5028                 if ( $holders ) {
5029                         $tooltip = $holders->replaceText( $caption );
5030                 } else {
5031                         $tooltip = $this->replaceLinkHoldersText( $caption );
5032                 }
5033
5034                 # make sure there are no placeholders in thumbnail attributes
5035                 # that are later expanded to html- so expand them now and
5036                 # remove the tags
5037                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5038                 $tooltip = Sanitizer::stripAllTags( $tooltip );
5039
5040                 return $tooltip;
5041         }
5042
5043         /**
5044          * Set a flag in the output object indicating that the content is dynamic and
5045          * shouldn't be cached.
5046          */
5047         function disableCache() {
5048                 wfDebug( "Parser output marked as uncacheable.\n" );
5049                 $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
5050                 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5051         }
5052
5053         /**
5054          * Callback from the Sanitizer for expanding items found in HTML attribute
5055          * values, so they can be safely tested and escaped.
5056          *
5057          * @param $text String
5058          * @param $frame PPFrame
5059          * @return String
5060          * @private
5061          */
5062         function attributeStripCallback( &$text, $frame = false ) {
5063                 $text = $this->replaceVariables( $text, $frame );
5064                 $text = $this->mStripState->unstripBoth( $text );
5065                 return $text;
5066         }
5067
5068         /**
5069          * Accessor
5070          */
5071         function getTags() {
5072                 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ) );
5073         }
5074
5075         /**
5076          * Replace transparent tags in $text with the values given by the callbacks.
5077          *
5078          * Transparent tag hooks are like regular XML-style tag hooks, except they
5079          * operate late in the transformation sequence, on HTML instead of wikitext.
5080          */
5081         function replaceTransparentTags( $text ) {
5082                 $matches = array();
5083                 $elements = array_keys( $this->mTransparentTagHooks );
5084                 $text = $this->extractTagsAndParams( $elements, $text, $matches, $this->mUniqPrefix );
5085
5086                 foreach ( $matches as $marker => $data ) {
5087                         list( $element, $content, $params, $tag ) = $data;
5088                         $tagName = strtolower( $element );
5089                         if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5090                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
5091                         } else {
5092                                 $output = $tag;
5093                         }
5094                         $this->mStripState->addGeneral( $marker, $output );
5095                 }
5096                 return $text;
5097         }
5098
5099         /**
5100          * Break wikitext input into sections, and either pull or replace
5101          * some particular section's text.
5102          *
5103          * External callers should use the getSection and replaceSection methods.
5104          *
5105          * @param $text String: Page wikitext
5106          * @param $section String: a section identifier string of the form:
5107          *   <flag1> - <flag2> - ... - <section number>
5108          *
5109          * Currently the only recognised flag is "T", which means the target section number
5110          * was derived during a template inclusion parse, in other words this is a template
5111          * section edit link. If no flags are given, it was an ordinary section edit link.
5112          * This flag is required to avoid a section numbering mismatch when a section is
5113          * enclosed by <includeonly> (bug 6563).
5114          *
5115          * The section number 0 pulls the text before the first heading; other numbers will
5116          * pull the given section along with its lower-level subsections. If the section is
5117          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5118          *
5119          * @param $mode String: one of "get" or "replace"
5120          * @param $newText String: replacement text for section data.
5121          * @return String: for "get", the extracted section text.
5122          *                 for "replace", the whole page with the section replaced.
5123          */
5124         private function extractSections( $text, $section, $mode, $newText='' ) {
5125                 global $wgTitle; # not generally used but removes an ugly failure mode
5126                 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5127                 $outText = '';
5128                 $frame = $this->getPreprocessor()->newFrame();
5129
5130                 # Process section extraction flags
5131                 $flags = 0;
5132                 $sectionParts = explode( '-', $section );
5133                 $sectionIndex = array_pop( $sectionParts );
5134                 foreach ( $sectionParts as $part ) {
5135                         if ( $part === 'T' ) {
5136                                 $flags |= self::PTD_FOR_INCLUSION;
5137                         }
5138                 }
5139                 # Preprocess the text
5140                 $root = $this->preprocessToDom( $text, $flags );
5141
5142                 # <h> nodes indicate section breaks
5143                 # They can only occur at the top level, so we can find them by iterating the root's children
5144                 $node = $root->getFirstChild();
5145
5146                 # Find the target section
5147                 if ( $sectionIndex == 0 ) {
5148                         # Section zero doesn't nest, level=big
5149                         $targetLevel = 1000;
5150                         if ( !$node ) {
5151                                 # The page definitely exists - we checked that earlier - so it must be blank: see bug #14005
5152                                 return $text;
5153                         }
5154                 } else {
5155                         while ( $node ) {
5156                                 if ( $node->getName() === 'h' ) {
5157                                         $bits = $node->splitHeading();
5158                                         if ( $bits['i'] == $sectionIndex ) {
5159                                                 $targetLevel = $bits['level'];
5160                                                 break;
5161                                         }
5162                                 }
5163                                 if ( $mode === 'replace' ) {
5164                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5165                                 }
5166                                 $node = $node->getNextSibling();
5167                         }
5168                 }
5169
5170                 if ( !$node ) {
5171                         # Not found
5172                         if ( $mode === 'get' ) {
5173                                 return $newText;
5174                         } else {
5175                                 return $text;
5176                         }
5177                 }
5178
5179                 # Find the end of the section, including nested sections
5180                 do {
5181                         if ( $node->getName() === 'h' ) {
5182                                 $bits = $node->splitHeading();
5183                                 $curLevel = $bits['level'];
5184                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5185                                         break;
5186                                 }
5187                         }
5188                         if ( $mode === 'get' ) {
5189                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5190                         }
5191                         $node = $node->getNextSibling();
5192                 } while ( $node );
5193
5194                 # Write out the remainder (in replace mode only)
5195                 if ( $mode === 'replace' ) {
5196                         # Output the replacement text
5197                         # Add two newlines on -- trailing whitespace in $newText is conventionally
5198                         # stripped by the editor, so we need both newlines to restore the paragraph gap
5199                         # Only add trailing whitespace if there is newText
5200                         if ( $newText != "" ) {
5201                                 $outText .= $newText . "\n\n";
5202                         }
5203
5204                         while ( $node ) {
5205                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5206                                 $node = $node->getNextSibling();
5207                         }
5208                 }
5209
5210                 if ( is_string( $outText ) ) {
5211                         # Re-insert stripped tags
5212                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5213                 }
5214
5215                 return $outText;
5216         }
5217
5218         /**
5219          * This function returns the text of a section, specified by a number ($section).
5220          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5221          * the first section before any such heading (section 0).
5222          *
5223          * If a section contains subsections, these are also returned.
5224          *
5225          * @param $text String: text to look in
5226          * @param $section String: section identifier
5227          * @param $deftext String: default to return if section is not found
5228          * @return string text of the requested section
5229          */
5230         public function getSection( $text, $section, $deftext='' ) {
5231                 return $this->extractSections( $text, $section, "get", $deftext );
5232         }
5233
5234         /**
5235          * This function returns $oldtext after the content of the section
5236          * specified by $section has been replaced with $text.
5237          *
5238          * @param $text String: former text of the article
5239          * @param $section Numeric: section identifier
5240          * @param $text String: replacing text
5241          * #return String: modified text
5242          */
5243         public function replaceSection( $oldtext, $section, $text ) {
5244                 return $this->extractSections( $oldtext, $section, "replace", $text );
5245         }
5246
5247         /**
5248          * Get the ID of the revision we are parsing
5249          *
5250          * @return Mixed: integer or null
5251          */
5252         function getRevisionId() {
5253                 return $this->mRevisionId;
5254         }
5255
5256         /**
5257          * Get the revision object for $this->mRevisionId
5258          *
5259          * @return either a Revision object or null
5260          */
5261         protected function getRevisionObject() {
5262                 if ( !is_null( $this->mRevisionObject ) ) {
5263                         return $this->mRevisionObject;
5264                 }
5265                 if ( is_null( $this->mRevisionId ) ) {
5266                         return null;
5267                 }
5268
5269                 $this->mRevisionObject = Revision::newFromId( $this->mRevisionId );
5270                 return $this->mRevisionObject;
5271         }
5272
5273         /**
5274          * Get the timestamp associated with the current revision, adjusted for
5275          * the default server-local timestamp
5276          */
5277         function getRevisionTimestamp() {
5278                 if ( is_null( $this->mRevisionTimestamp ) ) {
5279                         wfProfileIn( __METHOD__ );
5280
5281                         $revObject = $this->getRevisionObject();
5282                         $timestamp = $revObject ? $revObject->getTimestamp() : false;
5283
5284                         if( $timestamp !== false ) {
5285                                 global $wgContLang;
5286
5287                                 # The cryptic '' timezone parameter tells to use the site-default
5288                                 # timezone offset instead of the user settings.
5289                                 #
5290                                 # Since this value will be saved into the parser cache, served
5291                                 # to other users, and potentially even used inside links and such,
5292                                 # it needs to be consistent for all visitors.
5293                                 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5294                         }
5295
5296                         wfProfileOut( __METHOD__ );
5297                 }
5298                 return $this->mRevisionTimestamp;
5299         }
5300
5301         /**
5302          * Get the name of the user that edited the last revision
5303          *
5304          * @return String: user name
5305          */
5306         function getRevisionUser() {
5307                 if( is_null( $this->mRevisionUser ) ) {
5308                         $revObject = $this->getRevisionObject();
5309
5310                         # if this template is subst: the revision id will be blank,
5311                         # so just use the current user's name
5312                         if( $revObject ) {
5313                                 $this->mRevisionUser = $revObject->getUserText();
5314                         } elseif( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5315                                 $this->mRevisionUser = $this->getUser()->getName();
5316                         }
5317                 }
5318                 return $this->mRevisionUser;
5319         }
5320
5321         /**
5322          * Mutator for $mDefaultSort
5323          *
5324          * @param $sort New value
5325          */
5326         public function setDefaultSort( $sort ) {
5327                 $this->mDefaultSort = $sort;
5328                 $this->mOutput->setProperty( 'defaultsort', $sort );
5329         }
5330
5331         /**
5332          * Accessor for $mDefaultSort
5333          * Will use the empty string if none is set.
5334          *
5335          * This value is treated as a prefix, so the
5336          * empty string is equivalent to sorting by
5337          * page name.
5338          *
5339          * @return string
5340          */
5341         public function getDefaultSort() {
5342                 if ( $this->mDefaultSort !== false ) {
5343                         return $this->mDefaultSort;
5344                 } else {
5345                         return '';
5346                 }
5347         }
5348
5349         /**
5350          * Accessor for $mDefaultSort
5351          * Unlike getDefaultSort(), will return false if none is set
5352          *
5353          * @return string or false
5354          */
5355         public function getCustomDefaultSort() {
5356                 return $this->mDefaultSort;
5357         }
5358
5359         /**
5360          * Try to guess the section anchor name based on a wikitext fragment
5361          * presumably extracted from a heading, for example "Header" from
5362          * "== Header ==".
5363          */
5364         public function guessSectionNameFromWikiText( $text ) {
5365                 # Strip out wikitext links(they break the anchor)
5366                 $text = $this->stripSectionName( $text );
5367                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5368                 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5369         }
5370
5371         /**
5372          * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5373          * instead.  For use in redirects, since IE6 interprets Redirect: headers
5374          * as something other than UTF-8 (apparently?), resulting in breakage.
5375          *
5376          * @param $text String: The section name
5377          * @return string An anchor
5378          */
5379         public function guessLegacySectionNameFromWikiText( $text ) {
5380                 # Strip out wikitext links(they break the anchor)
5381                 $text = $this->stripSectionName( $text );
5382                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5383                 return '#' . Sanitizer::escapeId( $text, array( 'noninitial', 'legacy' ) );
5384         }
5385
5386         /**
5387          * Strips a text string of wikitext for use in a section anchor
5388          *
5389          * Accepts a text string and then removes all wikitext from the
5390          * string and leaves only the resultant text (i.e. the result of
5391          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5392          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5393          * to create valid section anchors by mimicing the output of the
5394          * parser when headings are parsed.
5395          *
5396          * @param $text String: text string to be stripped of wikitext
5397          * for use in a Section anchor
5398          * @return Filtered text string
5399          */
5400         public function stripSectionName( $text ) {
5401                 # Strip internal link markup
5402                 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5403                 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5404
5405                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5406                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5407                 # on how many empty links there are on the page - need to figure that out.
5408                 $text = preg_replace( '/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5409
5410                 # Parse wikitext quotes (italics & bold)
5411                 $text = $this->doQuotes( $text );
5412
5413                 # Strip HTML tags
5414                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5415                 return $text;
5416         }
5417
5418         /**
5419          * strip/replaceVariables/unstrip for preprocessor regression testing
5420          */
5421         function testSrvus( $text, $title, ParserOptions $options, $outputType = self::OT_HTML ) {
5422                 if ( !$title instanceof Title ) {
5423                         $title = Title::newFromText( $title );
5424                 }
5425                 $this->startParse( $title, $options, $outputType, true );
5426
5427                 $text = $this->replaceVariables( $text );
5428                 $text = $this->mStripState->unstripBoth( $text );
5429                 $text = Sanitizer::removeHTMLtags( $text );
5430                 return $text;
5431         }
5432
5433         function testPst( $text, $title, $options ) {
5434                 global $wgUser;
5435                 if ( !$title instanceof Title ) {
5436                         $title = Title::newFromText( $title );
5437                 }
5438                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5439         }
5440
5441         function testPreprocess( $text, $title, $options ) {
5442                 if ( !$title instanceof Title ) {
5443                         $title = Title::newFromText( $title );
5444                 }
5445                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5446         }
5447
5448         /**
5449          * Call a callback function on all regions of the given text that are not
5450          * inside strip markers, and replace those regions with the return value
5451          * of the callback. For example, with input:
5452          *
5453          *  aaa<MARKER>bbb
5454          *
5455          * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5456          * two strings will be replaced with the value returned by the callback in
5457          * each case.
5458          */
5459         function markerSkipCallback( $s, $callback ) {
5460                 $i = 0;
5461                 $out = '';
5462                 while ( $i < strlen( $s ) ) {
5463                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5464                         if ( $markerStart === false ) {
5465                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5466                                 break;
5467                         } else {
5468                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5469                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5470                                 if ( $markerEnd === false ) {
5471                                         $out .= substr( $s, $markerStart );
5472                                         break;
5473                                 } else {
5474                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5475                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5476                                         $i = $markerEnd;
5477                                 }
5478                         }
5479                 }
5480                 return $out;
5481         }
5482
5483         /**
5484          * Save the parser state required to convert the given half-parsed text to
5485          * HTML. "Half-parsed" in this context means the output of
5486          * recursiveTagParse() or internalParse(). This output has strip markers
5487          * from replaceVariables (extensionSubstitution() etc.), and link
5488          * placeholders from replaceLinkHolders().
5489          *
5490          * Returns an array which can be serialized and stored persistently. This
5491          * array can later be loaded into another parser instance with
5492          * unserializeHalfParsedText(). The text can then be safely incorporated into
5493          * the return value of a parser hook.
5494          */
5495         function serializeHalfParsedText( $text ) {
5496                 wfProfileIn( __METHOD__ );
5497                 $data = array(
5498                         'text' => $text,
5499                         'version' => self::HALF_PARSED_VERSION,
5500                         'stripState' => $this->mStripState->getSubState( $text ),
5501                         'linkHolders' => $this->mLinkHolders->getSubArray( $text )
5502                 );
5503                 wfProfileOut( __METHOD__ );
5504                 return $data;
5505         }
5506
5507         /**
5508          * Load the parser state given in the $data array, which is assumed to
5509          * have been generated by serializeHalfParsedText(). The text contents is
5510          * extracted from the array, and its markers are transformed into markers
5511          * appropriate for the current Parser instance. This transformed text is
5512          * returned, and can be safely included in the return value of a parser
5513          * hook.
5514          *
5515          * If the $data array has been stored persistently, the caller should first
5516          * check whether it is still valid, by calling isValidHalfParsedText().
5517          *
5518          * @param $data Serialized data
5519          * @return String
5520          */
5521         function unserializeHalfParsedText( $data ) {
5522                 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
5523                         throw new MWException( __METHOD__.': invalid version' );
5524                 }
5525
5526                 # First, extract the strip state.
5527                 $texts = array( $data['text'] );
5528                 $texts = $this->mStripState->merge( $data['stripState'], $texts );
5529
5530                 # Now renumber links
5531                 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
5532
5533                 # Should be good to go.
5534                 return $texts[0];
5535         }
5536
5537         /**
5538          * Returns true if the given array, presumed to be generated by
5539          * serializeHalfParsedText(), is compatible with the current version of the
5540          * parser.
5541          *
5542          * @param $data Array.
5543          */
5544         function isValidHalfParsedText( $data ) {
5545                 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
5546         }
5547 }