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