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  *  $wgMaxArticleSize*
  44  *
  45  *  * only within ParserOptions
  46  * </pre>
  47  *
  48  * @ingroup Parser
  49  */
  50 class Parser {
  51         /**
  52          * Update this version number when the ParserOutput format
  53          * changes in an incompatible way, so the parser cache
  54          * can automatically discard old data.
  55          */
  56         const VERSION = '1.6.4';
  57
  58         # Flags for Parser::setFunctionHook
  59         # Also available as global constants from Defines.php
  60         const SFH_NO_HASH = 1;
  61         const SFH_OBJECT_ARGS = 2;
  62
  63         # Constants needed for external link processing
  64         # Everything except bracket, space, or control characters
  65         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  66         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  67                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  68
  69         # State constants for the definition list colon extraction
  70         const COLON_STATE_TEXT = 0;
  71         const COLON_STATE_TAG = 1;
  72         const COLON_STATE_TAGSTART = 2;
  73         const COLON_STATE_CLOSETAG = 3;
  74         const COLON_STATE_TAGSLASH = 4;
  75         const COLON_STATE_COMMENT = 5;
  76         const COLON_STATE_COMMENTDASH = 6;
  77         const COLON_STATE_COMMENTDASHDASH = 7;
  78
  79         # Flags for preprocessToDom
  80         const PTD_FOR_INCLUSION = 1;
  81
  82         # Allowed values for $this->mOutputType
  83         # Parameter to startExternalParse().
  84         const OT_HTML = 1; # like parse()
  85         const OT_WIKI = 2; # like preSaveTransform()
  86         const OT_PREPROCESS = 3; # like preprocess()
  87         const OT_MSG = 3;
  88         const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
  89
  90         # Marker Suffix needs to be accessible staticly.
  91         const MARKER_SUFFIX = "-QINU\x7f";
  92
  93         # Persistent:
  94         var $mTagHooks = array();
  95         var $mTransparentTagHooks = array();
  96         var $mFunctionHooks = array();
  97         var $mFunctionSynonyms = array( 0 => array(), 1 => array() );
  98         var $mFunctionTagHooks = array();
  99         var $mStripList  = array();
 100         var $mDefaultStripList  = array();
 101         var $mVarCache = array();
 102         var $mImageParams = array();
 103         var $mImageParamsMagicArray = array();
 104         var $mMarkerIndex = 0;
 105         var $mFirstCall = true;
 106         var $mVariables, $mSubstWords; # Initialised by initialiseVariables()
 107         var $mConf, $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols; # Initialised in constructor
 108
 109
 110         # Cleared with clearState():
 111         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 112         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 113         var $mLinkHolders, $mLinkID;
 114         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 115         var $mTplExpandCache; # empty-frame expansion cache
 116         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 117         var $mExpensiveFunctionCount; # number of expensive parser function calls
 118         var $mUser; # User object; only used when doing pre-save transform
 119
 120         # Temporary
 121         # These are variables reset at least once per parse regardless of $clearState
 122
 123         /**
 124          * @var ParserOptions
 125          */
 126         var $mOptions;
 127         var $mTitle;        # Title context, used for self-link rendering and similar things
 128         var $mOutputType;   # Output type, one of the OT_xxx constants
 129         var $ot;            # Shortcut alias, see setOutputType()
 130         var $mRevisionObject; # The revision object of the specified revision ID
 131         var $mRevisionId;   # ID to display in {{REVISIONID}} tags
 132         var $mRevisionTimestamp; # The timestamp of the specified revision ID
 133         var $mRevisionUser; # Userto display in {{REVISIONUSER}} tag
 134         var $mRevIdForTs;   # The revision ID which was used to fetch the timestamp
 135
 136         /**
 137          * Constructor
 138          *
 139          * @public
 140          */
 141         function __construct( $conf = array() ) {
 142                 $this->mConf = $conf;
 143                 $this->mUrlProtocols = wfUrlProtocols();
 144                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 145                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/S';
 146                 if ( isset( $conf['preprocessorClass'] ) ) {
 147                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 148                 } elseif ( extension_loaded( 'domxml' ) ) {
 149                         # PECL extension that conflicts with the core DOM extension (bug 13770)
 150                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 151                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 152                 } elseif ( extension_loaded( 'dom' ) ) {
 153                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 154                 } else {
 155                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 156                 }
 157         }
 158
 159         /**
 160          * Reduce memory usage to reduce the impact of circular references
 161          */
 162         function __destruct() {
 163                 if ( isset( $this->mLinkHolders ) ) {
 164                         $this->mLinkHolders->__destruct();
 165                 }
 166                 foreach ( $this as $name => $value ) {
 167                         unset( $this->$name );
 168                 }
 169         }
 170
 171         /**
 172          * Do various kinds of initialisation on the first call of the parser
 173          */
 174         function firstCallInit() {
 175                 if ( !$this->mFirstCall ) {
 176                         return;
 177                 }
 178                 $this->mFirstCall = false;
 179
 180                 wfProfileIn( __METHOD__ );
 181
 182                 CoreParserFunctions::register( $this );
 183                 CoreTagHooks::register( $this );
 184                 $this->initialiseVariables();
 185
 186                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 187                 wfProfileOut( __METHOD__ );
 188         }
 189
 190         /**
 191          * Clear Parser state
 192          *
 193          * @private
 194          */
 195         function clearState() {
 196                 wfProfileIn( __METHOD__ );
 197                 if ( $this->mFirstCall ) {
 198                         $this->firstCallInit();
 199                 }
 200                 $this->mOutput = new ParserOutput;
 201                 $this->mOptions->registerWatcher( array( $this->mOutput, 'recordOption' ) );
 202                 $this->mAutonumber = 0;
 203                 $this->mLastSection = '';
 204                 $this->mDTopen = false;
 205                 $this->mIncludeCount = array();
 206                 $this->mStripState = new StripState;
 207                 $this->mArgStack = false;
 208                 $this->mInPre = false;
 209                 $this->mLinkHolders = new LinkHolderArray( $this );
 210                 $this->mLinkID = 0;
 211                 $this->mRevisionObject = $this->mRevisionTimestamp =
 212                         $this->mRevisionId = $this->mRevisionUser = null;
 213                 $this->mVarCache = array();
 214                 $this->mUser = null;
 215
 216                 /**
 217                  * Prefix for temporary replacement strings for the multipass parser.
 218                  * \x07 should never appear in input as it's disallowed in XML.
 219                  * Using it at the front also gives us a little extra robustness
 220                  * since it shouldn't match when butted up against identifier-like
 221                  * string constructs.
 222                  *
 223                  * Must not consist of all title characters, or else it will change
 224                  * the behaviour of <nowiki> in a link.
 225                  */
 226                 # $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 227                 # Changed to \x7f to allow XML double-parsing -- TS
 228                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 229
 230
 231                 # Clear these on every parse, bug 4549
 232                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 233
 234                 $this->mShowToc = true;
 235                 $this->mForceTocPosition = false;
 236                 $this->mIncludeSizes = array(
 237                         'post-expand' => 0,
 238                         'arg' => 0,
 239                 );
 240                 $this->mPPNodeCount = 0;
 241                 $this->mDefaultSort = false;
 242                 $this->mHeadings = array();
 243                 $this->mDoubleUnderscores = array();
 244                 $this->mExpensiveFunctionCount = 0;
 245
 246                 # Fix cloning
 247                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 248                         $this->mPreprocessor = null;
 249                 }
 250
 251                 wfRunHooks( 'ParserClearState', array( &$this ) );
 252                 wfProfileOut( __METHOD__ );
 253         }
 254
 255         /**
 256          * Convert wikitext to HTML
 257          * Do not call this function recursively.
 258          *
 259          * @param $text String: text we want to parse
 260          * @param $title A title object
 261          * @param $options ParserOptions
 262          * @param $linestart boolean
 263          * @param $clearState boolean
 264          * @param $revid Int: number to pass in {{REVISIONID}}
 265          * @return ParserOutput a ParserOutput
 266          */
 267         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 268                 /**
 269                  * First pass--just handle <nowiki> sections, pass the rest off
 270                  * to internalParse() which does all the real work.
 271                  */
 272
 273                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang, $wgDisableLangConversion, $wgDisableTitleConversion;
 274                 $fname = __METHOD__.'-' . wfGetCaller();
 275                 wfProfileIn( __METHOD__ );
 276                 wfProfileIn( $fname );
 277
 278                 $this->startExternalParse( $title, $options, self::OT_HTML, $clearState );
 279
 280                 $oldRevisionId = $this->mRevisionId;
 281                 $oldRevisionObject = $this->mRevisionObject;
 282                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 283                 $oldRevisionUser = $this->mRevisionUser;
 284                 if ( $revid !== null ) {
 285                         $this->mRevisionId = $revid;
 286                         $this->mRevisionObject = null;
 287                         $this->mRevisionTimestamp = null;
 288                         $this->mRevisionUser = null;
 289                 }
 290
 291                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 292                 # No more strip!
 293                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 294                 $text = $this->internalParse( $text );
 295
 296                 $text = $this->mStripState->unstripGeneral( $text );
 297
 298                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 299                 $fixtags = array(
 300                         # french spaces, last one Guillemet-left
 301                         # only if there is something before the space
 302                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;\\2',
 303                         # french spaces, Guillemet-right
 304                         '/(\\302\\253) /' => '\\1&#160;',
 305                         '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
 306                 );
 307                 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
 308
 309                 $text = $this->doBlockLevels( $text, $linestart );
 310
 311                 $this->replaceLinkHolders( $text );
 312
 313                 /**
 314                  * The page doesn't get language converted if
 315                  * a) It's disabled
 316                  * b) Content isn't converted
 317                  * c) It's a conversion table
 318                  */
 319                 if ( !( $wgDisableLangConversion
 320                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 321                                 || $this->mTitle->isConversionTable() ) ) {
 322
 323                         # The position of the convert() call should not be changed. it
 324                         # assumes that the links are all replaced and the only thing left
 325                         # is the <nowiki> mark.
 326
 327                         $text = $wgContLang->convert( $text );
 328                 }
 329
 330                 /**
 331                  * A converted title will be provided in the output object if title and
 332                  * content conversion are enabled, the article text does not contain
 333                  * a conversion-suppressing double-underscore tag, and no
 334                  * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
 335                  * automatic link conversion.
 336                  */
 337                 if ( !( $wgDisableLangConversion
 338                                 || $wgDisableTitleConversion
 339                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 340                                 || isset( $this->mDoubleUnderscores['notitleconvert'] )
 341                                 || $this->mOutput->getDisplayTitle() !== false ) )
 342                 {
 343                         $convruletitle = $wgContLang->getConvRuleTitle();
 344                         if ( $convruletitle ) {
 345                                 $this->mOutput->setTitleText( $convruletitle );
 346                         } else {
 347                                 $titleText = $wgContLang->convertTitle( $title );
 348                                 $this->mOutput->setTitleText( $titleText );
 349                         }
 350                 }
 351
 352                 $text = $this->mStripState->unstripNoWiki( $text );
 353
 354                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 355
 356 //!JF Move to its own function
 357
 358                 $uniq_prefix = $this->mUniqPrefix;
 359                 $matches = array();
 360                 $elements = array_keys( $this->mTransparentTagHooks );
 361                 $text = $this->extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 362
 363                 foreach ( $matches as $marker => $data ) {
 364                         list( $element, $content, $params, $tag ) = $data;
 365                         $tagName = strtolower( $element );
 366                         if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 367                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
 368                         } else {
 369                                 $output = $tag;
 370                         }
 371                         $this->mStripState->general->setPair( $marker, $output );
 372                 }
 373                 $text = $this->mStripState->unstripGeneral( $text );
 374
 375                 $text = Sanitizer::normalizeCharReferences( $text );
 376
 377                 if ( ( $wgUseTidy && $this->mOptions->getTidy() ) || $wgAlwaysUseTidy ) {
 378                         $text = MWTidy::tidy( $text );
 379                 } else {
 380                         # attempt to sanitize at least some nesting problems
 381                         # (bug #2702 and quite a few others)
 382                         $tidyregs = array(
 383                                 # ''Something [http://www.cool.com cool''] -->
 384                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 385                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 386                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 387                                 # fix up an anchor inside another anchor, only
 388                                 # at least for a single single nested link (bug 3695)
 389                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 390                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 391                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 392                                 # contains a div, so fix it up here; replace
 393                                 # div with escaped text
 394                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 395                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 396                                 # remove empty italic or bold tag pairs, some
 397                                 # introduced by rules above
 398                                 '/<([bi])><\/\\1>/' => '',
 399                         );
 400
 401                         $text = preg_replace(
 402                                 array_keys( $tidyregs ),
 403                                 array_values( $tidyregs ),
 404                                 $text );
 405                 }
 406                 global $wgExpensiveParserFunctionLimit;
 407                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 408                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 409                 }
 410
 411                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 412
 413                 # Information on include size limits, for the benefit of users who try to skirt them
 414                 if ( $this->mOptions->getEnableLimitReport() ) {
 415                         $max = $this->mOptions->getMaxIncludeSize();
 416                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 417                         $limitReport =
 418                                 "NewPP limit report\n" .
 419                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->getMaxPPNodeCount()}\n" .
 420                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 421                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 422                                 $PFreport;
 423                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 424                         $text .= "\n<!-- \n$limitReport-->\n";
 425                 }
 426                 $this->mOutput->setText( $text );
 427
 428                 $this->mRevisionId = $oldRevisionId;
 429                 $this->mRevisionObject = $oldRevisionObject;
 430                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 431                 $this->mRevisionUser = $oldRevisionUser;
 432                 wfProfileOut( $fname );
 433                 wfProfileOut( __METHOD__ );
 434
 435                 return $this->mOutput;
 436         }
 437
 438         /**
 439          * Recursive parser entry point that can be called from an extension tag
 440          * hook.
 441          *
 442          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 443          *
 444          * @param $text String: text extension wants to have parsed
 445          * @param $frame PPFrame: The frame to use for expanding any template variables
 446          */
 447         function recursiveTagParse( $text, $frame=false ) {
 448                 wfProfileIn( __METHOD__ );
 449                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 450                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 451                 $text = $this->internalParse( $text, false, $frame );
 452                 wfProfileOut( __METHOD__ );
 453                 return $text;
 454         }
 455
 456         /**
 457          * Expand templates and variables in the text, producing valid, static wikitext.
 458          * Also removes comments.
 459          */
 460         function preprocess( $text, Title $title, ParserOptions $options, $revid = null ) {
 461                 wfProfileIn( __METHOD__ );
 462                 $this->startExternalParse( $title, $options, self::OT_PREPROCESS, true );
 463                 if ( $revid !== null ) {
 464                         $this->mRevisionId = $revid;
 465                 }
 466                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 467                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 468                 $text = $this->replaceVariables( $text );
 469                 $text = $this->mStripState->unstripBoth( $text );
 470                 wfProfileOut( __METHOD__ );
 471                 return $text;
 472         }
 473
 474         /**
 475          * Process the wikitext for the ?preload= feature. (bug 5210)
 476          *
 477          * <noinclude>, <includeonly> etc. are parsed as for template transclusion,
 478          * comments, templates, arguments, tags hooks and parser functions are untouched.
 479          */
 480         public function getPreloadText( $text, Title $title, ParserOptions $options ) {
 481                 # Parser (re)initialisation
 482                 $this->startExternalParse( $title, $options, self::OT_PLAIN, true );
 483
 484                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 485                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
 486                 return $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
 487         }
 488
 489         /**
 490          * Get a random string
 491          *
 492          * @private
 493          * @static
 494          */
 495         static private function getRandomString() {
 496                 return dechex( mt_rand( 0, 0x7fffffff ) ) . dechex( mt_rand( 0, 0x7fffffff ) );
 497         }
 498
 499         /**
 500          * Set the current user.
 501          * Should only be used when doing pre-save transform.
 502          *
 503          * @param $user Mixed: User object or null (to reset)
 504          */
 505         function setUser( $user ) {
 506                 $this->mUser = $user;
 507         }
 508
 509         /**
 510          * Accessor for mUniqPrefix.
 511          *
 512          * @return String
 513          */
 514         public function uniqPrefix() {
 515                 if ( !isset( $this->mUniqPrefix ) ) {
 516                         # @todo Fixme: this is probably *horribly wrong*
 517                         # LanguageConverter seems to want $wgParser's uniqPrefix, however
 518                         # if this is called for a parser cache hit, the parser may not
 519                         # have ever been initialized in the first place.
 520                         # Not really sure what the heck is supposed to be going on here.
 521                         return '';
 522                         # throw new MWException( "Accessing uninitialized mUniqPrefix" );
 523                 }
 524                 return $this->mUniqPrefix;
 525         }
 526
 527         /**
 528          * Set the context title
 529          */
 530         function setTitle( $t ) {
 531                 if ( !$t || $t instanceof FakeTitle ) {
 532                         $t = Title::newFromText( 'NO TITLE' );
 533                 }
 534
 535                 if ( strval( $t->getFragment() ) !== '' ) {
 536                         # Strip the fragment to avoid various odd effects
 537                         $this->mTitle = clone $t;
 538                         $this->mTitle->setFragment( '' );
 539                 } else {
 540                         $this->mTitle = $t;
 541                 }
 542         }
 543
 544         /**
 545          * Accessor for the Title object
 546          *
 547          * @return Title object
 548          */
 549         function getTitle() {
 550                 return $this->mTitle;
 551         }
 552
 553         /**
 554          * Accessor/mutator for the Title object
 555          *
 556          * @param $x New Title object or null to just get the current one
 557          * @return Title object
 558          */
 559         function Title( $x = null ) {
 560                 return wfSetVar( $this->mTitle, $x );
 561         }
 562
 563         /**
 564          * Set the output type
 565          *
 566          * @param $ot Integer: new value
 567          */
 568         function setOutputType( $ot ) {
 569                 $this->mOutputType = $ot;
 570                 # Shortcut alias
 571                 $this->ot = array(
 572                         'html' => $ot == self::OT_HTML,
 573                         'wiki' => $ot == self::OT_WIKI,
 574                         'pre' => $ot == self::OT_PREPROCESS,
 575                         'plain' => $ot == self::OT_PLAIN,
 576                 );
 577         }
 578
 579         /**
 580          * Accessor/mutator for the output type
 581          *
 582          * @param $x New value or null to just get the current one
 583          * @return Integer
 584          */
 585         function OutputType( $x = null ) {
 586                 return wfSetVar( $this->mOutputType, $x );
 587         }
 588
 589         /**
 590          * Get the ParserOutput object
 591          *
 592          * @return ParserOutput object
 593          */
 594         function getOutput() {
 595                 return $this->mOutput;
 596         }
 597
 598         /**
 599          * Get the ParserOptions object
 600          *
 601          * @return ParserOptions object
 602          */
 603         function getOptions() {
 604                 return $this->mOptions;
 605         }
 606
 607         /**
 608          * Accessor/mutator for the ParserOptions object
 609          *
 610          * @param $x New value or null to just get the current one
 611          * @return Current ParserOptions object
 612          */
 613         function Options( $x = null ) {
 614                 return wfSetVar( $this->mOptions, $x );
 615         }
 616
 617         function nextLinkID() {
 618                 return $this->mLinkID++;
 619         }
 620
 621         /**
 622          * @return Language
 623          */
 624         function getFunctionLang() {
 625                 global $wgLang, $wgContLang;
 626
 627                 $target = $this->mOptions->getTargetLanguage();
 628                 if ( $target !== null ) {
 629                         return $target;
 630                 } else {
 631                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 632                 }
 633         }
 634
 635         /**
 636          * Get a User object either from $this->mUser, if set, or from the
 637          * ParserOptions object otherwise
 638          *
 639          * @return User object
 640          */
 641         function getUser() {
 642                 if ( !is_null( $this->mUser ) ) {
 643                         return $this->mUser;
 644                 }
 645                 return $this->mOptions->getUser();
 646         }
 647
 648         /**
 649          * Get a preprocessor object
 650          *
 651          * @return Preprocessor instance
 652          */
 653         function getPreprocessor() {
 654                 if ( !isset( $this->mPreprocessor ) ) {
 655                         $class = $this->mPreprocessorClass;
 656                         $this->mPreprocessor = new $class( $this );
 657                 }
 658                 return $this->mPreprocessor;
 659         }
 660
 661         /**
 662          * Replaces all occurrences of HTML-style comments and the given tags
 663          * in the text with a random marker and returns the next text. The output
 664          * parameter $matches will be an associative array filled with data in
 665          * the form:
 666          *   'UNIQ-xxxxx' => array(
 667          *     'element',
 668          *     'tag content',
 669          *     array( 'param' => 'x' ),
 670          *     '<element param="x">tag content</element>' ) )
 671          *
 672          * @param $elements list of element names. Comments are always extracted.
 673          * @param $text Source text string.
 674          * @param $matches Out parameter, Array: extracted tags
 675          * @param $uniq_prefix
 676          * @return String: stripped text
 677          *
 678          * @static
 679          */
 680         public function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
 681                 static $n = 1;
 682                 $stripped = '';
 683                 $matches = array();
 684
 685                 $taglist = implode( '|', $elements );
 686                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 687
 688                 while ( $text != '' ) {
 689                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 690                         $stripped .= $p[0];
 691                         if ( count( $p ) < 5 ) {
 692                                 break;
 693                         }
 694                         if ( count( $p ) > 5 ) {
 695                                 # comment
 696                                 $element    = $p[4];
 697                                 $attributes = '';
 698                                 $close      = '';
 699                                 $inside     = $p[5];
 700                         } else {
 701                                 # tag
 702                                 $element    = $p[1];
 703                                 $attributes = $p[2];
 704                                 $close      = $p[3];
 705                                 $inside     = $p[4];
 706                         }
 707
 708                         $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
 709                         $stripped .= $marker;
 710
 711                         if ( $close === '/>' ) {
 712                                 # Empty element tag, <tag />
 713                                 $content = null;
 714                                 $text = $inside;
 715                                 $tail = null;
 716                         } else {
 717                                 if ( $element === '!--' ) {
 718                                         $end = '/(-->)/';
 719                                 } else {
 720                                         $end = "/(<\\/$element\\s*>)/i";
 721                                 }
 722                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 723                                 $content = $q[0];
 724                                 if ( count( $q ) < 3 ) {
 725                                         # No end tag -- let it run out to the end of the text.
 726                                         $tail = '';
 727                                         $text = '';
 728                                 } else {
 729                                         $tail = $q[1];
 730                                         $text = $q[2];
 731                                 }
 732                         }
 733
 734                         $matches[$marker] = array( $element,
 735                                 $content,
 736                                 Sanitizer::decodeTagAttributes( $attributes ),
 737                                 "<$element$attributes$close$content$tail" );
 738                 }
 739                 return $stripped;
 740         }
 741
 742         /**
 743          * Get a list of strippable XML-like elements
 744          */
 745         function getStripList() {
 746                 return $this->mStripList;
 747         }
 748
 749         /**
 750          * @deprecated use replaceVariables
 751          */
 752         function strip( $text, $state, $stripcomments = false , $dontstrip = array() ) {
 753                 return $text;
 754         }
 755
 756         /**
 757          * Restores pre, math, and other extensions removed by strip()
 758          *
 759          * always call unstripNoWiki() after this one
 760          * @private
 761          * @deprecated use $this->mStripState->unstrip()
 762          */
 763         function unstrip( $text, $state ) {
 764                 return $state->unstripGeneral( $text );
 765         }
 766
 767         /**
 768          * Always call this after unstrip() to preserve the order
 769          *
 770          * @private
 771          * @deprecated use $this->mStripState->unstrip()
 772          */
 773         function unstripNoWiki( $text, $state ) {
 774                 return $state->unstripNoWiki( $text );
 775         }
 776
 777         /**
 778          * @deprecated use $this->mStripState->unstripBoth()
 779          */
 780         function unstripForHTML( $text ) {
 781                 return $this->mStripState->unstripBoth( $text );
 782         }
 783
 784         /**
 785          * Add an item to the strip state
 786          * Returns the unique tag which must be inserted into the stripped text
 787          * The tag will be replaced with the original text in unstrip()
 788          *
 789          * @private
 790          */
 791         function insertStripItem( $text ) {
 792                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 793                 $this->mMarkerIndex++;
 794                 $this->mStripState->general->setPair( $rnd, $text );
 795                 return $rnd;
 796         }
 797
 798         /**
 799          * Interface with html tidy
 800          * @deprecated Use MWTidy::tidy()
 801          */
 802         public static function tidy( $text ) {
 803                 wfDeprecated( __METHOD__ );
 804                 return MWTidy::tidy( $text );
 805         }
 806
 807         /**
 808          * parse the wiki syntax used to render tables
 809          *
 810          * @private
 811          */
 812         function doTableStuff( $text ) {
 813                 wfProfileIn( __METHOD__ );
 814
 815                 $lines = StringUtils::explode( "\n", $text );
 816                 $out = '';
 817                 $td_history = array(); # Is currently a td tag open?
 818                 $last_tag_history = array(); # Save history of last lag activated (td, th or caption)
 819                 $tr_history = array(); # Is currently a tr tag open?
 820                 $tr_attributes = array(); # history of tr attributes
 821                 $has_opened_tr = array(); # Did this table open a <tr> element?
 822                 $indent_level = 0; # indent level of the table
 823
 824                 foreach ( $lines as $outLine ) {
 825                         $line = trim( $outLine );
 826
 827                         if ( $line === '' ) { # empty line, go to next line
 828                                 $out .= $outLine."\n";
 829                                 continue;
 830                         }
 831
 832                         $first_character = $line[0];
 833                         $matches = array();
 834
 835                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 836                                 # First check if we are starting a new table
 837                                 $indent_level = strlen( $matches[1] );
 838
 839                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 840                                 $attributes = Sanitizer::fixTagAttributes( $attributes , 'table' );
 841
 842                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 843                                 array_push( $td_history , false );
 844                                 array_push( $last_tag_history , '' );
 845                                 array_push( $tr_history , false );
 846                                 array_push( $tr_attributes , '' );
 847                                 array_push( $has_opened_tr , false );
 848                         } elseif ( count( $td_history ) == 0 ) {
 849                                 # Don't do any of the following
 850                                 $out .= $outLine."\n";
 851                                 continue;
 852                         } elseif ( substr( $line , 0 , 2 ) === '|}' ) {
 853                                 # We are ending a table
 854                                 $line = '</table>' . substr( $line , 2 );
 855                                 $last_tag = array_pop( $last_tag_history );
 856
 857                                 if ( !array_pop( $has_opened_tr ) ) {
 858                                         $line = "<tr><td></td></tr>{$line}";
 859                                 }
 860
 861                                 if ( array_pop( $tr_history ) ) {
 862                                         $line = "</tr>{$line}";
 863                                 }
 864
 865                                 if ( array_pop( $td_history ) ) {
 866                                         $line = "</{$last_tag}>{$line}";
 867                                 }
 868                                 array_pop( $tr_attributes );
 869                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 870                         } elseif ( substr( $line , 0 , 2 ) === '|-' ) {
 871                                 # Now we have a table row
 872                                 $line = preg_replace( '#^\|-+#', '', $line );
 873
 874                                 # Whats after the tag is now only attributes
 875                                 $attributes = $this->mStripState->unstripBoth( $line );
 876                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 877                                 array_pop( $tr_attributes );
 878                                 array_push( $tr_attributes, $attributes );
 879
 880                                 $line = '';
 881                                 $last_tag = array_pop( $last_tag_history );
 882                                 array_pop( $has_opened_tr );
 883                                 array_push( $has_opened_tr , true );
 884
 885                                 if ( array_pop( $tr_history ) ) {
 886                                         $line = '</tr>';
 887                                 }
 888
 889                                 if ( array_pop( $td_history ) ) {
 890                                         $line = "</{$last_tag}>{$line}";
 891                                 }
 892
 893                                 $outLine = $line;
 894                                 array_push( $tr_history , false );
 895                                 array_push( $td_history , false );
 896                                 array_push( $last_tag_history , '' );
 897                         } elseif ( $first_character === '|' || $first_character === '!' || substr( $line , 0 , 2 )  === '|+' ) {
 898                                 # This might be cell elements, td, th or captions
 899                                 if ( substr( $line , 0 , 2 ) === '|+' ) {
 900                                         $first_character = '+';
 901                                         $line = substr( $line , 1 );
 902                                 }
 903
 904                                 $line = substr( $line , 1 );
 905
 906                                 if ( $first_character === '!' ) {
 907                                         $line = str_replace( '!!' , '||' , $line );
 908                                 }
 909
 910                                 # Split up multiple cells on the same line.
 911                                 # FIXME : This can result in improper nesting of tags processed
 912                                 # by earlier parser steps, but should avoid splitting up eg
 913                                 # attribute values containing literal "||".
 914                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 915
 916                                 $outLine = '';
 917
 918                                 # Loop through each table cell
 919                                 foreach ( $cells as $cell ) {
 920                                         $previous = '';
 921                                         if ( $first_character !== '+' ) {
 922                                                 $tr_after = array_pop( $tr_attributes );
 923                                                 if ( !array_pop( $tr_history ) ) {
 924                                                         $previous = "<tr{$tr_after}>\n";
 925                                                 }
 926                                                 array_push( $tr_history , true );
 927                                                 array_push( $tr_attributes , '' );
 928                                                 array_pop( $has_opened_tr );
 929                                                 array_push( $has_opened_tr , true );
 930                                         }
 931
 932                                         $last_tag = array_pop( $last_tag_history );
 933
 934                                         if ( array_pop( $td_history ) ) {
 935                                                 $previous = "</{$last_tag}>\n{$previous}";
 936                                         }
 937
 938                                         if ( $first_character === '|' ) {
 939                                                 $last_tag = 'td';
 940                                         } elseif ( $first_character === '!' ) {
 941                                                 $last_tag = 'th';
 942                                         } elseif ( $first_character === '+' ) {
 943                                                 $last_tag = 'caption';
 944                                         } else {
 945                                                 $last_tag = '';
 946                                         }
 947
 948                                         array_push( $last_tag_history , $last_tag );
 949
 950                                         # A cell could contain both parameters and data
 951                                         $cell_data = explode( '|' , $cell , 2 );
 952
 953                                         # Bug 553: Note that a '|' inside an invalid link should not
 954                                         # be mistaken as delimiting cell parameters
 955                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 956                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 957                                         } elseif ( count( $cell_data ) == 1 ) {
 958                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 959                                         } else {
 960                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 961                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 962                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 963                                         }
 964
 965                                         $outLine .= $cell;
 966                                         array_push( $td_history , true );
 967                                 }
 968                         }
 969                         $out .= $outLine . "\n";
 970                 }
 971
 972                 # Closing open td, tr && table
 973                 while ( count( $td_history ) > 0 ) {
 974                         if ( array_pop( $td_history ) ) {
 975                                 $out .= "</td>\n";
 976                         }
 977                         if ( array_pop( $tr_history ) ) {
 978                                 $out .= "</tr>\n";
 979                         }
 980                         if ( !array_pop( $has_opened_tr ) ) {
 981                                 $out .= "<tr><td></td></tr>\n" ;
 982                         }
 983
 984                         $out .= "</table>\n";
 985                 }
 986
 987                 # Remove trailing line-ending (b/c)
 988                 if ( substr( $out, -1 ) === "\n" ) {
 989                         $out = substr( $out, 0, -1 );
 990                 }
 991
 992                 # special case: don't return empty table
 993                 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 994                         $out = '';
 995                 }
 996
 997                 wfProfileOut( __METHOD__ );
 998
 999                 return $out;
1000         }
1001
1002         /**
1003          * Helper function for parse() that transforms wiki markup into
1004          * HTML. Only called for $mOutputType == self::OT_HTML.
1005          *
1006          * @private
1007          */
1008         function internalParse( $text, $isMain = true, $frame=false ) {
1009                 wfProfileIn( __METHOD__ );
1010
1011                 $origText = $text;
1012
1013                 # Hook to suspend the parser in this state
1014                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1015                         wfProfileOut( __METHOD__ );
1016                         return $text ;
1017                 }
1018
1019                 # if $frame is provided, then use $frame for replacing any variables
1020                 if ( $frame ) {
1021                         # use frame depth to infer how include/noinclude tags should be handled
1022                         # depth=0 means this is the top-level document; otherwise it's an included document
1023                         if ( !$frame->depth ) {
1024                                 $flag = 0;
1025                         } else {
1026                                 $flag = Parser::PTD_FOR_INCLUSION;
1027                         }
1028                         $dom = $this->preprocessToDom( $text, $flag );
1029                         $text = $frame->expand( $dom );
1030                 } else {
1031                         # if $frame is not provided, then use old-style replaceVariables
1032                         $text = $this->replaceVariables( $text );
1033                 }
1034
1035                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
1036                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1037
1038                 # Tables need to come after variable replacement for things to work
1039                 # properly; putting them before other transformations should keep
1040                 # exciting things like link expansions from showing up in surprising
1041                 # places.
1042                 $text = $this->doTableStuff( $text );
1043
1044                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1045
1046                 $text = $this->doDoubleUnderscore( $text );
1047
1048                 $text = $this->doHeadings( $text );
1049                 if ( $this->mOptions->getUseDynamicDates() ) {
1050                         $df = DateFormatter::getInstance();
1051                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1052                 }
1053                 $text = $this->replaceInternalLinks( $text );
1054                 $text = $this->doAllQuotes( $text );
1055                 $text = $this->replaceExternalLinks( $text );
1056
1057                 # replaceInternalLinks may sometimes leave behind
1058                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1059                 $text = str_replace( $this->mUniqPrefix.'NOPARSE', '', $text );
1060
1061                 $text = $this->doMagicLinks( $text );
1062                 $text = $this->formatHeadings( $text, $origText, $isMain );
1063
1064                 wfProfileOut( __METHOD__ );
1065                 return $text;
1066         }
1067
1068         /**
1069          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1070          * magic external links.
1071          *
1072          * DML
1073          * @private
1074          */
1075         function doMagicLinks( $text ) {
1076                 wfProfileIn( __METHOD__ );
1077                 $prots = $this->mUrlProtocols;
1078                 $urlChar = self::EXT_LINK_URL_CLASS;
1079                 $text = preg_replace_callback(
1080                         '!(?:                           # Start cases
1081                                 (<a[ \t\r\n>].*?</a>) |     # m[1]: Skip link text
1082                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1083                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
1084                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1085                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1086                                         (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1087                                         (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1088                                         [0-9Xx]                 # check digit
1089                                         \b)
1090                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1091                 wfProfileOut( __METHOD__ );
1092                 return $text;
1093         }
1094
1095         function magicLinkCallback( $m ) {
1096                 if ( isset( $m[1] ) && $m[1] !== '' ) {
1097                         # Skip anchor
1098                         return $m[0];
1099                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1100                         # Skip HTML element
1101                         return $m[0];
1102                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1103                         # Free external link
1104                         return $this->makeFreeExternalLink( $m[0] );
1105                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1106                         # RFC or PMID
1107                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1108                                 $keyword = 'RFC';
1109                                 $urlmsg = 'rfcurl';
1110                                 $CssClass = 'mw-magiclink-rfc';
1111                                 $id = $m[4];
1112                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1113                                 $keyword = 'PMID';
1114                                 $urlmsg = 'pubmedurl';
1115                                 $CssClass = 'mw-magiclink-pmid';
1116                                 $id = $m[4];
1117                         } else {
1118                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1119                                         substr( $m[0], 0, 20 ) . '"' );
1120                         }
1121                         $url = wfMsgForContent( $urlmsg, $id );
1122                         $sk = $this->mOptions->getSkin( $this->mTitle );
1123                         return $sk->makeExternalLink( $url, "{$keyword} {$id}", true, $CssClass );
1124                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1125                         # ISBN
1126                         $isbn = $m[5];
1127                         $num = strtr( $isbn, array(
1128                                 '-' => '',
1129                                 ' ' => '',
1130                                 'x' => 'X',
1131                         ));
1132                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1133                         return'<a href="' .
1134                                 $titleObj->escapeLocalUrl() .
1135                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1136                 } else {
1137                         return $m[0];
1138                 }
1139         }
1140
1141         /**
1142          * Make a free external link, given a user-supplied URL
1143          * @return HTML
1144          * @private
1145          */
1146         function makeFreeExternalLink( $url ) {
1147                 global $wgContLang;
1148                 wfProfileIn( __METHOD__ );
1149
1150                 $sk = $this->mOptions->getSkin( $this->mTitle );
1151                 $trail = '';
1152
1153                 # The characters '<' and '>' (which were escaped by
1154                 # removeHTMLtags()) should not be included in
1155                 # URLs, per RFC 2396.
1156                 $m2 = array();
1157                 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1158                         $trail = substr( $url, $m2[0][1] ) . $trail;
1159                         $url = substr( $url, 0, $m2[0][1] );
1160                 }
1161
1162                 # Move trailing punctuation to $trail
1163                 $sep = ',;\.:!?';
1164                 # If there is no left bracket, then consider right brackets fair game too
1165                 if ( strpos( $url, '(' ) === false ) {
1166                         $sep .= ')';
1167                 }
1168
1169                 $numSepChars = strspn( strrev( $url ), $sep );
1170                 if ( $numSepChars ) {
1171                         $trail = substr( $url, -$numSepChars ) . $trail;
1172                         $url = substr( $url, 0, -$numSepChars );
1173                 }
1174
1175                 $url = Sanitizer::cleanUrl( $url );
1176
1177                 # Is this an external image?
1178                 $text = $this->maybeMakeExternalImage( $url );
1179                 if ( $text === false ) {
1180                         # Not an image, make a link
1181                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1182                                 $this->getExternalLinkAttribs( $url ) );
1183                         # Register it in the output object...
1184                         # Replace unnecessary URL escape codes with their equivalent characters
1185                         $pasteurized = self::replaceUnusualEscapes( $url );
1186                         $this->mOutput->addExternalLink( $pasteurized );
1187                 }
1188                 wfProfileOut( __METHOD__ );
1189                 return $text . $trail;
1190         }
1191
1192
1193         /**
1194          * Parse headers and return html
1195          *
1196          * @private
1197          */
1198         function doHeadings( $text ) {
1199                 wfProfileIn( __METHOD__ );
1200                 for ( $i = 6; $i >= 1; --$i ) {
1201                         $h = str_repeat( '=', $i );
1202                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1203                           "<h$i>\\1</h$i>", $text );
1204                 }
1205                 wfProfileOut( __METHOD__ );
1206                 return $text;
1207         }
1208
1209         /**
1210          * Replace single quotes with HTML markup
1211          * @private
1212          * @return string the altered text
1213          */
1214         function doAllQuotes( $text ) {
1215                 wfProfileIn( __METHOD__ );
1216                 $outtext = '';
1217                 $lines = StringUtils::explode( "\n", $text );
1218                 foreach ( $lines as $line ) {
1219                         $outtext .= $this->doQuotes( $line ) . "\n";
1220                 }
1221                 $outtext = substr( $outtext, 0,-1 );
1222                 wfProfileOut( __METHOD__ );
1223                 return $outtext;
1224         }
1225
1226         /**
1227          * Helper function for doAllQuotes()
1228          */
1229         public function doQuotes( $text ) {
1230                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1231                 if ( count( $arr ) == 1 ) {
1232                         return $text;
1233                 } else {
1234                         # First, do some preliminary work. This may shift some apostrophes from
1235                         # being mark-up to being text. It also counts the number of occurrences
1236                         # of bold and italics mark-ups.
1237                         $numbold = 0;
1238                         $numitalics = 0;
1239                         for ( $i = 0; $i < count( $arr ); $i++ ) {
1240                                 if ( ( $i % 2 ) == 1 ) {
1241                                         # If there are ever four apostrophes, assume the first is supposed to
1242                                         # be text, and the remaining three constitute mark-up for bold text.
1243                                         if ( strlen( $arr[$i] ) == 4 ) {
1244                                                 $arr[$i-1] .= "'";
1245                                                 $arr[$i] = "'''";
1246                                         } elseif ( strlen( $arr[$i] ) > 5 ) {
1247                                                 # If there are more than 5 apostrophes in a row, assume they're all
1248                                                 # text except for the last 5.
1249                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1250                                                 $arr[$i] = "'''''";
1251                                         }
1252                                         # Count the number of occurrences of bold and italics mark-ups.
1253                                         # We are not counting sequences of five apostrophes.
1254                                         if ( strlen( $arr[$i] ) == 2 ) {
1255                                                 $numitalics++;
1256                                         } elseif ( strlen( $arr[$i] ) == 3 ) {
1257                                                 $numbold++;
1258                                         } elseif ( strlen( $arr[$i] ) == 5 ) {
1259                                                 $numitalics++;
1260                                                 $numbold++;
1261                                         }
1262                                 }
1263                         }
1264
1265                         # If there is an odd number of both bold and italics, it is likely
1266                         # that one of the bold ones was meant to be an apostrophe followed
1267                         # by italics. Which one we cannot know for certain, but it is more
1268                         # likely to be one that has a single-letter word before it.
1269                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1270                                 $i = 0;
1271                                 $firstsingleletterword = -1;
1272                                 $firstmultiletterword = -1;
1273                                 $firstspace = -1;
1274                                 foreach ( $arr as $r ) {
1275                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1276                                                 $x1 = substr( $arr[$i-1], -1 );
1277                                                 $x2 = substr( $arr[$i-1], -2, 1 );
1278                                                 if ( $x1 === ' ' ) {
1279                                                         if ( $firstspace == -1 ) {
1280                                                                 $firstspace = $i;
1281                                                         }
1282                                                 } elseif ( $x2 === ' ') {
1283                                                         if ( $firstsingleletterword == -1 ) {
1284                                                                 $firstsingleletterword = $i;
1285                                                         }
1286                                                 } else {
1287                                                         if ( $firstmultiletterword == -1 ) {
1288                                                                 $firstmultiletterword = $i;
1289                                                         }
1290                                                 }
1291                                         }
1292                                         $i++;
1293                                 }
1294
1295                                 # If there is a single-letter word, use it!
1296                                 if ( $firstsingleletterword > -1 ) {
1297                                         $arr[$firstsingleletterword] = "''";
1298                                         $arr[$firstsingleletterword-1] .= "'";
1299                                 } elseif ( $firstmultiletterword > -1 ) {
1300                                         # If not, but there's a multi-letter word, use that one.
1301                                         $arr[$firstmultiletterword] = "''";
1302                                         $arr[$firstmultiletterword-1] .= "'";
1303                                 } elseif ( $firstspace > -1 ) {
1304                                         # ... otherwise use the first one that has neither.
1305                                         # (notice that it is possible for all three to be -1 if, for example,
1306                                         # there is only one pentuple-apostrophe in the line)
1307                                         $arr[$firstspace] = "''";
1308                                         $arr[$firstspace-1] .= "'";
1309                                 }
1310                         }
1311
1312                         # Now let's actually convert our apostrophic mush to HTML!
1313                         $output = '';
1314                         $buffer = '';
1315                         $state = '';
1316                         $i = 0;
1317                         foreach ( $arr as $r ) {
1318                                 if ( ( $i % 2 ) == 0 ) {
1319                                         if ( $state === 'both' ) {
1320                                                 $buffer .= $r;
1321                                         } else {
1322                                                 $output .= $r;
1323                                         }
1324                                 } else {
1325                                         if ( strlen( $r ) == 2 ) {
1326                                                 if ( $state === 'i' ) {
1327                                                         $output .= '</i>'; $state = '';
1328                                                 } elseif ( $state === 'bi' ) {
1329                                                         $output .= '</i>'; $state = 'b';
1330                                                 } elseif ( $state === 'ib' ) {
1331                                                         $output .= '</b></i><b>'; $state = 'b';
1332                                                 } elseif ( $state === 'both' ) {
1333                                                         $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1334                                                 } else { # $state can be 'b' or ''
1335                                                         $output .= '<i>'; $state .= 'i';
1336                                                 }
1337                                         } elseif ( strlen( $r ) == 3 ) {
1338                                                 if ( $state === 'b' ) {
1339                                                         $output .= '</b>'; $state = '';
1340                                                 } elseif ( $state === 'bi' ) {
1341                                                         $output .= '</i></b><i>'; $state = 'i';
1342                                                 } elseif ( $state === 'ib' ) {
1343                                                         $output .= '</b>'; $state = 'i';
1344                                                 } elseif ( $state === 'both' ) {
1345                                                         $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1346                                                 } else { # $state can be 'i' or ''
1347                                                         $output .= '<b>'; $state .= 'b';
1348                                                 }
1349                                         } elseif ( strlen( $r ) == 5 ) {
1350                                                 if ( $state === 'b' ) {
1351                                                         $output .= '</b><i>'; $state = 'i';
1352                                                 } elseif ( $state === 'i' ) {
1353                                                         $output .= '</i><b>'; $state = 'b';
1354                                                 } elseif ( $state === 'bi' ) {
1355                                                         $output .= '</i></b>'; $state = '';
1356                                                 } elseif ( $state === 'ib' ) {
1357                                                         $output .= '</b></i>'; $state = '';
1358                                                 } elseif ( $state === 'both' ) {
1359                                                         $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1360                                                 } else { # ($state == '')
1361                                                         $buffer = ''; $state = 'both';
1362                                                 }
1363                                         }
1364                                 }
1365                                 $i++;
1366                         }
1367                         # Now close all remaining tags.  Notice that the order is important.
1368                         if ( $state === 'b' || $state === 'ib' ) {
1369                                 $output .= '</b>';
1370                         }
1371                         if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1372                                 $output .= '</i>';
1373                         }
1374                         if ( $state === 'bi' ) {
1375                                 $output .= '</b>';
1376                         }
1377                         # There might be lonely ''''', so make sure we have a buffer
1378                         if ( $state === 'both' && $buffer ) {
1379                                 $output .= '<b><i>'.$buffer.'</i></b>';
1380                         }
1381                         return $output;
1382                 }
1383         }
1384
1385         /**
1386          * Replace external links (REL)
1387          *
1388          * Note: this is all very hackish and the order of execution matters a lot.
1389          * Make sure to run maintenance/parserTests.php if you change this code.
1390          *
1391          * @private
1392          */
1393         function replaceExternalLinks( $text ) {
1394                 global $wgContLang;
1395                 wfProfileIn( __METHOD__ );
1396
1397                 $sk = $this->mOptions->getSkin( $this->mTitle );
1398
1399                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1400                 $s = array_shift( $bits );
1401
1402                 $i = 0;
1403                 while ( $i<count( $bits ) ) {
1404                         $url = $bits[$i++];
1405                         $protocol = $bits[$i++];
1406                         $text = $bits[$i++];
1407                         $trail = $bits[$i++];
1408
1409                         # The characters '<' and '>' (which were escaped by
1410                         # removeHTMLtags()) should not be included in
1411                         # URLs, per RFC 2396.
1412                         $m2 = array();
1413                         if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1414                                 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1415                                 $url = substr( $url, 0, $m2[0][1] );
1416                         }
1417
1418                         # If the link text is an image URL, replace it with an <img> tag
1419                         # This happened by accident in the original parser, but some people used it extensively
1420                         $img = $this->maybeMakeExternalImage( $text );
1421                         if ( $img !== false ) {
1422                                 $text = $img;
1423                         }
1424
1425                         $dtrail = '';
1426
1427                         # Set linktype for CSS - if URL==text, link is essentially free
1428                         $linktype = ( $text === $url ) ? 'free' : 'text';
1429
1430                         # No link text, e.g. [http://domain.tld/some.link]
1431                         if ( $text == '' ) {
1432                                 # Autonumber if allowed. See bug #5918
1433                                 if ( strpos( wfUrlProtocols(), substr( $protocol, 0, strpos( $protocol, ':' ) ) ) !== false ) {
1434                                         $langObj = $this->getFunctionLang();
1435                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1436                                         $linktype = 'autonumber';
1437                                 } else {
1438                                         # Otherwise just use the URL
1439                                         $text = htmlspecialchars( $url );
1440                                         $linktype = 'free';
1441                                 }
1442                         } else {
1443                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1444                                 # Check for trail
1445                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1446                         }
1447
1448                         $text = $wgContLang->markNoConversion( $text );
1449
1450                         $url = Sanitizer::cleanUrl( $url );
1451
1452                         # Use the encoded URL
1453                         # This means that users can paste URLs directly into the text
1454                         # Funny characters like ö aren't valid in URLs anyway
1455                         # This was changed in August 2004
1456                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1457                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1458
1459                         # Register link in the output object.
1460                         # Replace unnecessary URL escape codes with the referenced character
1461                         # This prevents spammers from hiding links from the filters
1462                         $pasteurized = self::replaceUnusualEscapes( $url );
1463                         $this->mOutput->addExternalLink( $pasteurized );
1464                 }
1465
1466                 wfProfileOut( __METHOD__ );
1467                 return $s;
1468         }
1469
1470         /**
1471          * Get an associative array of additional HTML attributes appropriate for a
1472          * particular external link.  This currently may include rel => nofollow
1473          * (depending on configuration, namespace, and the URL's domain) and/or a
1474          * target attribute (depending on configuration).
1475          *
1476          * @param $url String: optional URL, to extract the domain from for rel =>
1477          *   nofollow if appropriate
1478          * @return Array: associative array of HTML attributes
1479          */
1480         function getExternalLinkAttribs( $url = false ) {
1481                 $attribs = array();
1482                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1483                 $ns = $this->mTitle->getNamespace();
1484                 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) ) {
1485                         $attribs['rel'] = 'nofollow';
1486
1487                         global $wgNoFollowDomainExceptions;
1488                         if ( $wgNoFollowDomainExceptions ) {
1489                                 $bits = wfParseUrl( $url );
1490                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1491                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1492                                                 if ( substr( $bits['host'], -strlen( $domain ) ) == $domain ) {
1493                                                         unset( $attribs['rel'] );
1494                                                         break;
1495                                                 }
1496                                         }
1497                                 }
1498                         }
1499                 }
1500                 if ( $this->mOptions->getExternalLinkTarget() ) {
1501                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1502                 }
1503                 return $attribs;
1504         }
1505
1506
1507         /**
1508          * Replace unusual URL escape codes with their equivalent characters
1509          *
1510          * @param $url String
1511          * @return String
1512          *
1513          * @todo  This can merge genuinely required bits in the path or query string,
1514          *        breaking legit URLs. A proper fix would treat the various parts of
1515          *        the URL differently; as a workaround, just use the output for
1516          *        statistical records, not for actual linking/output.
1517          */
1518         static function replaceUnusualEscapes( $url ) {
1519                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1520                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1521         }
1522
1523         /**
1524          * Callback function used in replaceUnusualEscapes().
1525          * Replaces unusual URL escape codes with their equivalent character
1526          */
1527         private static function replaceUnusualEscapesCallback( $matches ) {
1528                 $char = urldecode( $matches[0] );
1529                 $ord = ord( $char );
1530                 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1531                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1532                         # No, shouldn't be escaped
1533                         return $char;
1534                 } else {
1535                         # Yes, leave it escaped
1536                         return $matches[0];
1537                 }
1538         }
1539
1540         /**
1541          * make an image if it's allowed, either through the global
1542          * option, through the exception, or through the on-wiki whitelist
1543          * @private
1544          */
1545         function maybeMakeExternalImage( $url ) {
1546                 $sk = $this->mOptions->getSkin( $this->mTitle );
1547                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1548                 $imagesexception = !empty( $imagesfrom );
1549                 $text = false;
1550                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1551                 if ( $imagesexception && is_array( $imagesfrom ) ) {
1552                         $imagematch = false;
1553                         foreach ( $imagesfrom as $match ) {
1554                                 if ( strpos( $url, $match ) === 0 ) {
1555                                         $imagematch = true;
1556                                         break;
1557                                 }
1558                         }
1559                 } elseif ( $imagesexception ) {
1560                         $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1561                 } else {
1562                         $imagematch = false;
1563                 }
1564                 if ( $this->mOptions->getAllowExternalImages()
1565                          || ( $imagesexception && $imagematch ) ) {
1566                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1567                                 # Image found
1568                                 $text = $sk->makeExternalImage( $url );
1569                         }
1570                 }
1571                 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1572                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1573                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1574                         foreach ( $whitelist as $entry ) {
1575                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1576                                 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1577                                         continue;
1578                                 }
1579                                 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1580                                         # Image matches a whitelist entry
1581                                         $text = $sk->makeExternalImage( $url );
1582                                         break;
1583                                 }
1584                         }
1585                 }
1586                 return $text;
1587         }
1588
1589         /**
1590          * Process [[ ]] wikilinks
1591          * @return String: processed text
1592          *
1593          * @private
1594          */
1595         function replaceInternalLinks( $s ) {
1596                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1597                 return $s;
1598         }
1599
1600         /**
1601          * Process [[ ]] wikilinks (RIL)
1602          * @return LinkHolderArray
1603          *
1604          * @private
1605          */
1606         function replaceInternalLinks2( &$s ) {
1607                 global $wgContLang;
1608
1609                 wfProfileIn( __METHOD__ );
1610
1611                 wfProfileIn( __METHOD__.'-setup' );
1612                 static $tc = FALSE, $e1, $e1_img;
1613                 # the % is needed to support urlencoded titles as well
1614                 if ( !$tc ) {
1615                         $tc = Title::legalChars() . '#%';
1616                         # Match a link having the form [[namespace:link|alternate]]trail
1617                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1618                         # Match cases where there is no "]]", which might still be images
1619                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1620                 }
1621
1622                 $sk = $this->mOptions->getSkin( $this->mTitle );
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( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1858                                         } else {
1859                                                 $s .= $prefix . $trail;
1860                                         }
1861                                         $this->mOutput->addImage( $nt->getDBkey() );
1862                                         wfProfileOut( __METHOD__."-image" );
1863                                         continue;
1864
1865                                 }
1866
1867                                 if ( $ns == NS_CATEGORY ) {
1868                                         wfProfileIn( __METHOD__."-category" );
1869                                         $s = rtrim( $s . "\n" ); # bug 87
1870
1871                                         if ( $wasblank ) {
1872                                                 $sortkey = $this->getDefaultSort();
1873                                         } else {
1874                                                 $sortkey = $text;
1875                                         }
1876                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1877                                         $sortkey = str_replace( "\n", '', $sortkey );
1878                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1879                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1880
1881                                         /**
1882                                          * Strip the whitespace Category links produce, see bug 87
1883                                          * @todo We might want to use trim($tmp, "\n") here.
1884                                          */
1885                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
1886
1887                                         wfProfileOut( __METHOD__."-category" );
1888                                         continue;
1889                                 }
1890                         }
1891
1892                         # Self-link checking
1893                         if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1894                                 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1895                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1896                                         continue;
1897                                 }
1898                         }
1899
1900                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1901                         # FIXME: Should do batch file existence checks, see comment below
1902                         if ( $ns == NS_MEDIA ) {
1903                                 wfProfileIn( __METHOD__."-media" );
1904                                 # Give extensions a chance to select the file revision for us
1905                                 $skip = $time = false;
1906                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1907                                 if ( $skip ) {
1908                                         $link = $sk->link( $nt );
1909                                 } else {
1910                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1911                                 }
1912                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1913                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1914                                 $this->mOutput->addImage( $nt->getDBkey() );
1915                                 wfProfileOut( __METHOD__."-media" );
1916                                 continue;
1917                         }
1918
1919                         wfProfileIn( __METHOD__."-always_known" );
1920                         # Some titles, such as valid special pages or files in foreign repos, should
1921                         # be shown as bluelinks even though they're not included in the page table
1922                         #
1923                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1924                         # batch file existence checks for NS_FILE and NS_MEDIA
1925                         if ( $iw == '' && $nt->isAlwaysKnown() ) {
1926                                 $this->mOutput->addLink( $nt );
1927                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1928                         } else {
1929                                 # Links will be added to the output link list after checking
1930                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1931                         }
1932                         wfProfileOut( __METHOD__."-always_known" );
1933                 }
1934                 wfProfileOut( __METHOD__ );
1935                 return $holders;
1936         }
1937
1938         /**
1939          * Make a link placeholder. The text returned can be later resolved to a real link with
1940          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1941          * parsing of interwiki links, and secondly to allow all existence checks and
1942          * article length checks (for stub links) to be bundled into a single query.
1943          *
1944          * @deprecated
1945          */
1946         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1947                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1948         }
1949
1950         /**
1951          * Render a forced-blue link inline; protect against double expansion of
1952          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1953          * Since this little disaster has to split off the trail text to avoid
1954          * breaking URLs in the following text without breaking trails on the
1955          * wiki links, it's been made into a horrible function.
1956          *
1957          * @param $nt Title
1958          * @param $text String
1959          * @param $query String
1960          * @param $trail String
1961          * @param $prefix String
1962          * @return String: HTML-wikitext mix oh yuck
1963          */
1964         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1965                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1966                 $sk = $this->mOptions->getSkin( $this->mTitle );
1967                 # FIXME: use link() instead of deprecated makeKnownLinkObj()
1968                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1969                 return $this->armorLinks( $link ) . $trail;
1970         }
1971
1972         /**
1973          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1974          * going to go through further parsing steps before inline URL expansion.
1975          *
1976          * Not needed quite as much as it used to be since free links are a bit
1977          * more sensible these days. But bracketed links are still an issue.
1978          *
1979          * @param $text String: more-or-less HTML
1980          * @return String: less-or-more HTML with NOPARSE bits
1981          */
1982         function armorLinks( $text ) {
1983                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1984                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1985         }
1986
1987         /**
1988          * Return true if subpage links should be expanded on this page.
1989          * @return Boolean
1990          */
1991         function areSubpagesAllowed() {
1992                 # Some namespaces don't allow subpages
1993                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1994         }
1995
1996         /**
1997          * Handle link to subpage if necessary
1998          *
1999          * @param $target String: the source of the link
2000          * @param &$text String: the link text, modified as necessary
2001          * @return string the full name of the link
2002          * @private
2003          */
2004         function maybeDoSubpageLink( $target, &$text ) {
2005                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2006         }
2007
2008         /**#@+
2009          * Used by doBlockLevels()
2010          * @private
2011          */
2012         function closeParagraph() {
2013                 $result = '';
2014                 if ( $this->mLastSection != '' ) {
2015                         $result = '</' . $this->mLastSection  . ">\n";
2016                 }
2017                 $this->mInPre = false;
2018                 $this->mLastSection = '';
2019                 return $result;
2020         }
2021
2022         /**
2023          * getCommon() returns the length of the longest common substring
2024          * of both arguments, starting at the beginning of both.
2025          * @private
2026          */
2027         function getCommon( $st1, $st2 ) {
2028                 $fl = strlen( $st1 );
2029                 $shorter = strlen( $st2 );
2030                 if ( $fl < $shorter ) {
2031                         $shorter = $fl;
2032                 }
2033
2034                 for ( $i = 0; $i < $shorter; ++$i ) {
2035                         if ( $st1{$i} != $st2{$i} ) {
2036                                 break;
2037                         }
2038                 }
2039                 return $i;
2040         }
2041
2042         /**
2043          * These next three functions open, continue, and close the list
2044          * element appropriate to the prefix character passed into them.
2045          * @private
2046          */
2047         function openList( $char ) {
2048                 $result = $this->closeParagraph();
2049
2050                 if ( '*' === $char ) {
2051                         $result .= '<ul><li>';
2052                 } elseif ( '#' === $char ) {
2053                         $result .= '<ol><li>';
2054                 } elseif ( ':' === $char ) {
2055                         $result .= '<dl><dd>';
2056                 } elseif ( ';' === $char ) {
2057                         $result .= '<dl><dt>';
2058                         $this->mDTopen = true;
2059                 } else {
2060                         $result = '<!-- ERR 1 -->';
2061                 }
2062
2063                 return $result;
2064         }
2065
2066         /**
2067          * TODO: document
2068          * @param $char String
2069          * @private
2070          */
2071         function nextItem( $char ) {
2072                 if ( '*' === $char || '#' === $char ) {
2073                         return '</li><li>';
2074                 } elseif ( ':' === $char || ';' === $char ) {
2075                         $close = '</dd>';
2076                         if ( $this->mDTopen ) {
2077                                 $close = '</dt>';
2078                         }
2079                         if ( ';' === $char ) {
2080                                 $this->mDTopen = true;
2081                                 return $close . '<dt>';
2082                         } else {
2083                                 $this->mDTopen = false;
2084                                 return $close . '<dd>';
2085                         }
2086                 }
2087                 return '<!-- ERR 2 -->';
2088         }
2089
2090         /**
2091          * TODO: document
2092          * @param $char String
2093          * @private
2094          */
2095         function closeList( $char ) {
2096                 if ( '*' === $char ) {
2097                         $text = '</li></ul>';
2098                 } elseif ( '#' === $char ) {
2099                         $text = '</li></ol>';
2100                 } elseif ( ':' === $char ) {
2101                         if ( $this->mDTopen ) {
2102                                 $this->mDTopen = false;
2103                                 $text = '</dt></dl>';
2104                         } else {
2105                                 $text = '</dd></dl>';
2106                         }
2107                 } else {
2108                         return '<!-- ERR 3 -->';
2109                 }
2110                 return $text."\n";
2111         }
2112         /**#@-*/
2113
2114         /**
2115          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2116          *
2117          * @param $text String
2118          * @param $linestart Boolean: whether or not this is at the start of a line.
2119          * @private
2120          * @return string the lists rendered as HTML
2121          */
2122         function doBlockLevels( $text, $linestart ) {
2123                 wfProfileIn( __METHOD__ );
2124
2125                 # Parsing through the text line by line.  The main thing
2126                 # happening here is handling of block-level elements p, pre,
2127                 # and making lists from lines starting with * # : etc.
2128                 #
2129                 $textLines = StringUtils::explode( "\n", $text );
2130
2131                 $lastPrefix = $output = '';
2132                 $this->mDTopen = $inBlockElem = false;
2133                 $prefixLength = 0;
2134                 $paragraphStack = false;
2135
2136                 foreach ( $textLines as $oLine ) {
2137                         # Fix up $linestart
2138                         if ( !$linestart ) {
2139                                 $output .= $oLine;
2140                                 $linestart = true;
2141                                 continue;
2142                         }
2143                         # * = ul
2144                         # # = ol
2145                         # ; = dt
2146                         # : = dd
2147
2148                         $lastPrefixLength = strlen( $lastPrefix );
2149                         $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2150                         $preOpenMatch = preg_match( '/<pre/i', $oLine );
2151                         # If not in a <pre> element, scan for and figure out what prefixes are there.
2152                         if ( !$this->mInPre ) {
2153                                 # Multiple prefixes may abut each other for nested lists.
2154                                 $prefixLength = strspn( $oLine, '*#:;' );
2155                                 $prefix = substr( $oLine, 0, $prefixLength );
2156
2157                                 # eh?
2158                                 # ; and : are both from definition-lists, so they're equivalent
2159                                 #  for the purposes of determining whether or not we need to open/close
2160                                 #  elements.
2161                                 $prefix2 = str_replace( ';', ':', $prefix );
2162                                 $t = substr( $oLine, $prefixLength );
2163                                 $this->mInPre = (bool)$preOpenMatch;
2164                         } else {
2165                                 # Don't interpret any other prefixes in preformatted text
2166                                 $prefixLength = 0;
2167                                 $prefix = $prefix2 = '';
2168                                 $t = $oLine;
2169                         }
2170
2171                         # List generation
2172                         if ( $prefixLength && $lastPrefix === $prefix2 ) {
2173                                 # Same as the last item, so no need to deal with nesting or opening stuff
2174                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2175                                 $paragraphStack = false;
2176
2177                                 if ( substr( $prefix, -1 ) === ';') {
2178                                         # The one nasty exception: definition lists work like this:
2179                                         # ; title : definition text
2180                                         # So we check for : in the remainder text to split up the
2181                                         # title and definition, without b0rking links.
2182                                         $term = $t2 = '';
2183                                         if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2184                                                 $t = $t2;
2185                                                 $output .= $term . $this->nextItem( ':' );
2186                                         }
2187                                 }
2188                         } elseif ( $prefixLength || $lastPrefixLength ) {
2189                                 # We need to open or close prefixes, or both.
2190
2191                                 # Either open or close a level...
2192                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2193                                 $paragraphStack = false;
2194
2195                                 # Close all the prefixes which aren't shared.
2196                                 while ( $commonPrefixLength < $lastPrefixLength ) {
2197                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2198                                         --$lastPrefixLength;
2199                                 }
2200
2201                                 # Continue the current prefix if appropriate.
2202                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2203                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2204                                 }
2205
2206                                 # Open prefixes where appropriate.
2207                                 while ( $prefixLength > $commonPrefixLength ) {
2208                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2209                                         $output .= $this->openList( $char );
2210
2211                                         if ( ';' === $char ) {
2212                                                 # FIXME: This is dupe of code above
2213                                                 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2214                                                         $t = $t2;
2215                                                         $output .= $term . $this->nextItem( ':' );
2216                                                 }
2217                                         }
2218                                         ++$commonPrefixLength;
2219                                 }
2220                                 $lastPrefix = $prefix2;
2221                         }
2222
2223                         # If we have no prefixes, go to paragraph mode.
2224                         if ( 0 == $prefixLength ) {
2225                                 wfProfileIn( __METHOD__."-paragraph" );
2226                                 # No prefix (not in list)--go to paragraph mode
2227                                 # XXX: use a stack for nestable elements like span, table and div
2228                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2229                                 $closematch = preg_match(
2230                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2231                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2232                                 if ( $openmatch or $closematch ) {
2233                                         $paragraphStack = false;
2234                                         # TODO bug 5718: paragraph closed
2235                                         $output .= $this->closeParagraph();
2236                                         if ( $preOpenMatch and !$preCloseMatch ) {
2237                                                 $this->mInPre = true;
2238                                         }
2239                                         $inBlockElem = !$closematch;
2240                                 } elseif ( !$inBlockElem && !$this->mInPre ) {
2241                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) ) {
2242                                                 # pre
2243                                                 if ( $this->mLastSection !== 'pre' ) {
2244                                                         $paragraphStack = false;
2245                                                         $output .= $this->closeParagraph().'<pre>';
2246                                                         $this->mLastSection = 'pre';
2247                                                 }
2248                                                 $t = substr( $t, 1 );
2249                                         } else {
2250                                                 # paragraph
2251                                                 if ( trim( $t ) === '' ) {
2252                                                         if ( $paragraphStack ) {
2253                                                                 $output .= $paragraphStack.'<br />';
2254                                                                 $paragraphStack = false;
2255                                                                 $this->mLastSection = 'p';
2256                                                         } else {
2257                                                                 if ( $this->mLastSection !== 'p' ) {
2258                                                                         $output .= $this->closeParagraph();
2259                                                                         $this->mLastSection = '';
2260                                                                         $paragraphStack = '<p>';
2261                                                                 } else {
2262                                                                         $paragraphStack = '</p><p>';
2263                                                                 }
2264                                                         }
2265                                                 } else {
2266                                                         if ( $paragraphStack ) {
2267                                                                 $output .= $paragraphStack;
2268                                                                 $paragraphStack = false;
2269                                                                 $this->mLastSection = 'p';
2270                                                         } elseif ( $this->mLastSection !== 'p' ) {
2271                                                                 $output .= $this->closeParagraph().'<p>';
2272                                                                 $this->mLastSection = 'p';
2273                                                         }
2274                                                 }
2275                                         }
2276                                 }
2277                                 wfProfileOut( __METHOD__."-paragraph" );
2278                         }
2279                         # somewhere above we forget to get out of pre block (bug 785)
2280                         if ( $preCloseMatch && $this->mInPre ) {
2281                                 $this->mInPre = false;
2282                         }
2283                         if ( $paragraphStack === false ) {
2284                                 $output .= $t."\n";
2285                         }
2286                 }
2287                 while ( $prefixLength ) {
2288                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2289                         --$prefixLength;
2290                 }
2291                 if ( $this->mLastSection != '' ) {
2292                         $output .= '</' . $this->mLastSection . '>';
2293                         $this->mLastSection = '';
2294                 }
2295
2296                 wfProfileOut( __METHOD__ );
2297                 return $output;
2298         }
2299
2300         /**
2301          * Split up a string on ':', ignoring any occurences inside tags
2302          * to prevent illegal overlapping.
2303          *
2304          * @param $str String: the string to split
2305          * @param &$before String: set to everything before the ':'
2306          * @param &$after String: set to everything after the ':'
2307          * return String: the position of the ':', or false if none found
2308          */
2309         function findColonNoLinks( $str, &$before, &$after ) {
2310                 wfProfileIn( __METHOD__ );
2311
2312                 $pos = strpos( $str, ':' );
2313                 if ( $pos === false ) {
2314                         # Nothing to find!
2315                         wfProfileOut( __METHOD__ );
2316                         return false;
2317                 }
2318
2319                 $lt = strpos( $str, '<' );
2320                 if ( $lt === false || $lt > $pos ) {
2321                         # Easy; no tag nesting to worry about
2322                         $before = substr( $str, 0, $pos );
2323                         $after = substr( $str, $pos+1 );
2324                         wfProfileOut( __METHOD__ );
2325                         return $pos;
2326                 }
2327
2328                 # Ugly state machine to walk through avoiding tags.
2329                 $state = self::COLON_STATE_TEXT;
2330                 $stack = 0;
2331                 $len = strlen( $str );
2332                 for( $i = 0; $i < $len; $i++ ) {
2333                         $c = $str{$i};
2334
2335                         switch( $state ) {
2336                         # (Using the number is a performance hack for common cases)
2337                         case 0: # self::COLON_STATE_TEXT:
2338                                 switch( $c ) {
2339                                 case "<":
2340                                         # Could be either a <start> tag or an </end> tag
2341                                         $state = self::COLON_STATE_TAGSTART;
2342                                         break;
2343                                 case ":":
2344                                         if ( $stack == 0 ) {
2345                                                 # We found it!
2346                                                 $before = substr( $str, 0, $i );
2347                                                 $after = substr( $str, $i + 1 );
2348                                                 wfProfileOut( __METHOD__ );
2349                                                 return $i;
2350                                         }
2351                                         # Embedded in a tag; don't break it.
2352                                         break;
2353                                 default:
2354                                         # Skip ahead looking for something interesting
2355                                         $colon = strpos( $str, ':', $i );
2356                                         if ( $colon === false ) {
2357                                                 # Nothing else interesting
2358                                                 wfProfileOut( __METHOD__ );
2359                                                 return false;
2360                                         }
2361                                         $lt = strpos( $str, '<', $i );
2362                                         if ( $stack === 0 ) {
2363                                                 if ( $lt === false || $colon < $lt ) {
2364                                                         # We found it!
2365                                                         $before = substr( $str, 0, $colon );
2366                                                         $after = substr( $str, $colon + 1 );
2367                                                         wfProfileOut( __METHOD__ );
2368                                                         return $i;
2369                                                 }
2370                                         }
2371                                         if ( $lt === false ) {
2372                                                 # Nothing else interesting to find; abort!
2373                                                 # We're nested, but there's no close tags left. Abort!
2374                                                 break 2;
2375                                         }
2376                                         # Skip ahead to next tag start
2377                                         $i = $lt;
2378                                         $state = self::COLON_STATE_TAGSTART;
2379                                 }
2380                                 break;
2381                         case 1: # self::COLON_STATE_TAG:
2382                                 # In a <tag>
2383                                 switch( $c ) {
2384                                 case ">":
2385                                         $stack++;
2386                                         $state = self::COLON_STATE_TEXT;
2387                                         break;
2388                                 case "/":
2389                                         # Slash may be followed by >?
2390                                         $state = self::COLON_STATE_TAGSLASH;
2391                                         break;
2392                                 default:
2393                                         # ignore
2394                                 }
2395                                 break;
2396                         case 2: # self::COLON_STATE_TAGSTART:
2397                                 switch( $c ) {
2398                                 case "/":
2399                                         $state = self::COLON_STATE_CLOSETAG;
2400                                         break;
2401                                 case "!":
2402                                         $state = self::COLON_STATE_COMMENT;
2403                                         break;
2404                                 case ">":
2405                                         # Illegal early close? This shouldn't happen D:
2406                                         $state = self::COLON_STATE_TEXT;
2407                                         break;
2408                                 default:
2409                                         $state = self::COLON_STATE_TAG;
2410                                 }
2411                                 break;
2412                         case 3: # self::COLON_STATE_CLOSETAG:
2413                                 # In a </tag>
2414                                 if ( $c === ">" ) {
2415                                         $stack--;
2416                                         if ( $stack < 0 ) {
2417                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2418                                                 wfProfileOut( __METHOD__ );
2419                                                 return false;
2420                                         }
2421                                         $state = self::COLON_STATE_TEXT;
2422                                 }
2423                                 break;
2424                         case self::COLON_STATE_TAGSLASH:
2425                                 if ( $c === ">" ) {
2426                                         # Yes, a self-closed tag <blah/>
2427                                         $state = self::COLON_STATE_TEXT;
2428                                 } else {
2429                                         # Probably we're jumping the gun, and this is an attribute
2430                                         $state = self::COLON_STATE_TAG;
2431                                 }
2432                                 break;
2433                         case 5: # self::COLON_STATE_COMMENT:
2434                                 if ( $c === "-" ) {
2435                                         $state = self::COLON_STATE_COMMENTDASH;
2436                                 }
2437                                 break;
2438                         case self::COLON_STATE_COMMENTDASH:
2439                                 if ( $c === "-" ) {
2440                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2441                                 } else {
2442                                         $state = self::COLON_STATE_COMMENT;
2443                                 }
2444                                 break;
2445                         case self::COLON_STATE_COMMENTDASHDASH:
2446                                 if ( $c === ">" ) {
2447                                         $state = self::COLON_STATE_TEXT;
2448                                 } else {
2449                                         $state = self::COLON_STATE_COMMENT;
2450                                 }
2451                                 break;
2452                         default:
2453                                 throw new MWException( "State machine error in " . __METHOD__ );
2454                         }
2455                 }
2456                 if ( $stack > 0 ) {
2457                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2458                         wfProfileOut( __METHOD__ );
2459                         return false;
2460                 }
2461                 wfProfileOut( __METHOD__ );
2462                 return false;
2463         }
2464
2465         /**
2466          * Return value of a magic variable (like PAGENAME)
2467          *
2468          * @private
2469          */
2470         function getVariableValue( $index, $frame=false ) {
2471                 global $wgContLang, $wgSitename, $wgServer;
2472                 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2473
2474                 /**
2475                  * Some of these require message or data lookups and can be
2476                  * expensive to check many times.
2477                  */
2478                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2479                         if ( isset( $this->mVarCache[$index] ) ) {
2480                                 return $this->mVarCache[$index];
2481                         }
2482                 }
2483
2484                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2485                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2486
2487                 # Use the time zone
2488                 global $wgLocaltimezone;
2489                 if ( isset( $wgLocaltimezone ) ) {
2490                         $oldtz = date_default_timezone_get();
2491                         date_default_timezone_set( $wgLocaltimezone );
2492                 }
2493
2494                 $localTimestamp = date( 'YmdHis', $ts );
2495                 $localMonth = date( 'm', $ts );
2496                 $localMonth1 = date( 'n', $ts );
2497                 $localMonthName = date( 'n', $ts );
2498                 $localDay = date( 'j', $ts );
2499                 $localDay2 = date( 'd', $ts );
2500                 $localDayOfWeek = date( 'w', $ts );
2501                 $localWeek = date( 'W', $ts );
2502                 $localYear = date( 'Y', $ts );
2503                 $localHour = date( 'H', $ts );
2504                 if ( isset( $wgLocaltimezone ) ) {
2505                         date_default_timezone_set( $oldtz );
2506                 }
2507
2508                 switch ( $index ) {
2509                         case 'currentmonth':
2510                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2511                                 break;
2512                         case 'currentmonth1':
2513                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2514                                 break;
2515                         case 'currentmonthname':
2516                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2517                                 break;
2518                         case 'currentmonthnamegen':
2519                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2520                                 break;
2521                         case 'currentmonthabbrev':
2522                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2523                                 break;
2524                         case 'currentday':
2525                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2526                                 break;
2527                         case 'currentday2':
2528                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2529                                 break;
2530                         case 'localmonth':
2531                                 $value = $wgContLang->formatNum( $localMonth );
2532                                 break;
2533                         case 'localmonth1':
2534                                 $value = $wgContLang->formatNum( $localMonth1 );
2535                                 break;
2536                         case 'localmonthname':
2537                                 $value = $wgContLang->getMonthName( $localMonthName );
2538                                 break;
2539                         case 'localmonthnamegen':
2540                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2541                                 break;
2542                         case 'localmonthabbrev':
2543                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2544                                 break;
2545                         case 'localday':
2546                                 $value = $wgContLang->formatNum( $localDay );
2547                                 break;
2548                         case 'localday2':
2549                                 $value = $wgContLang->formatNum( $localDay2 );
2550                                 break;
2551                         case 'pagename':
2552                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2553                                 break;
2554                         case 'pagenamee':
2555                                 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2556                                 break;
2557                         case 'fullpagename':
2558                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2559                                 break;
2560                         case 'fullpagenamee':
2561                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2562                                 break;
2563                         case 'subpagename':
2564                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2565                                 break;
2566                         case 'subpagenamee':
2567                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2568                                 break;
2569                         case 'basepagename':
2570                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2571                                 break;
2572                         case 'basepagenamee':
2573                                 $value = wfEscapeWikiText( wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) ) );
2574                                 break;
2575                         case 'talkpagename':
2576                                 if ( $this->mTitle->canTalk() ) {
2577                                         $talkPage = $this->mTitle->getTalkPage();
2578                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2579                                 } else {
2580                                         $value = '';
2581                                 }
2582                                 break;
2583                         case 'talkpagenamee':
2584                                 if ( $this->mTitle->canTalk() ) {
2585                                         $talkPage = $this->mTitle->getTalkPage();
2586                                         $value = wfEscapeWikiText( $talkPage->getPrefixedUrl() );
2587                                 } else {
2588                                         $value = '';
2589                                 }
2590                                 break;
2591                         case 'subjectpagename':
2592                                 $subjPage = $this->mTitle->getSubjectPage();
2593                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2594                                 break;
2595                         case 'subjectpagenamee':
2596                                 $subjPage = $this->mTitle->getSubjectPage();
2597                                 $value = wfEscapeWikiText( $subjPage->getPrefixedUrl() );
2598                                 break;
2599                         case 'revisionid':
2600                                 # Let the edit saving system know we should parse the page
2601                                 # *after* a revision ID has been assigned.
2602                                 $this->mOutput->setFlag( 'vary-revision' );
2603                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2604                                 $value = $this->mRevisionId;
2605                                 break;
2606                         case 'revisionday':
2607                                 # Let the edit saving system know we should parse the page
2608                                 # *after* a revision ID has been assigned. This is for null edits.
2609                                 $this->mOutput->setFlag( 'vary-revision' );
2610                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2611                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2612                                 break;
2613                         case 'revisionday2':
2614                                 # Let the edit saving system know we should parse the page
2615                                 # *after* a revision ID has been assigned. This is for null edits.
2616                                 $this->mOutput->setFlag( 'vary-revision' );
2617                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2618                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2619                                 break;
2620                         case 'revisionmonth':
2621                                 # Let the edit saving system know we should parse the page
2622                                 # *after* a revision ID has been assigned. This is for null edits.
2623                                 $this->mOutput->setFlag( 'vary-revision' );
2624                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2625                                 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2626                                 break;
2627                         case 'revisionmonth1':
2628                                 # Let the edit saving system know we should parse the page
2629                                 # *after* a revision ID has been assigned. This is for null edits.
2630                                 $this->mOutput->setFlag( 'vary-revision' );
2631                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2632                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2633                                 break;
2634                         case 'revisionyear':
2635                                 # Let the edit saving system know we should parse the page
2636                                 # *after* a revision ID has been assigned. This is for null edits.
2637                                 $this->mOutput->setFlag( 'vary-revision' );
2638                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2639                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2640                                 break;
2641                         case 'revisiontimestamp':
2642                                 # Let the edit saving system know we should parse the page
2643                                 # *after* a revision ID has been assigned. This is for null edits.
2644                                 $this->mOutput->setFlag( 'vary-revision' );
2645                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2646                                 $value = $this->getRevisionTimestamp();
2647                                 break;
2648                         case 'revisionuser':
2649                                 # Let the edit saving system know we should parse the page
2650                                 # *after* a revision ID has been assigned. This is for null edits.
2651                                 $this->mOutput->setFlag( 'vary-revision' );
2652                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2653                                 $value = $this->getRevisionUser();
2654                                 break;
2655                         case 'namespace':
2656                                 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2657                                 break;
2658                         case 'namespacee':
2659                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2660                                 break;
2661                         case 'talkspace':
2662                                 $value = $this->mTitle->canTalk() ? str_replace( '_',' ',$this->mTitle->getTalkNsText() ) : '';
2663                                 break;
2664                         case 'talkspacee':
2665                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2666                                 break;
2667                         case 'subjectspace':
2668                                 $value = $this->mTitle->getSubjectNsText();
2669                                 break;
2670                         case 'subjectspacee':
2671                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2672                                 break;
2673                         case 'currentdayname':
2674                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2675                                 break;
2676                         case 'currentyear':
2677                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2678                                 break;
2679                         case 'currenttime':
2680                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2681                                 break;
2682                         case 'currenthour':
2683                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2684                                 break;
2685                         case 'currentweek':
2686                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2687                                 # int to remove the padding
2688                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2689                                 break;
2690                         case 'currentdow':
2691                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2692                                 break;
2693                         case 'localdayname':
2694                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2695                                 break;
2696                         case 'localyear':
2697                                 $value = $wgContLang->formatNum( $localYear, true );
2698                                 break;
2699                         case 'localtime':
2700                                 $value = $wgContLang->time( $localTimestamp, false, false );
2701                                 break;
2702                         case 'localhour':
2703                                 $value = $wgContLang->formatNum( $localHour, true );
2704                                 break;
2705                         case 'localweek':
2706                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2707                                 # int to remove the padding
2708                                 $value = $wgContLang->formatNum( (int)$localWeek );
2709                                 break;
2710                         case 'localdow':
2711                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2712                                 break;
2713                         case 'numberofarticles':
2714                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2715                                 break;
2716                         case 'numberoffiles':
2717                                 $value = $wgContLang->formatNum( SiteStats::images() );
2718                                 break;
2719                         case 'numberofusers':
2720                                 $value = $wgContLang->formatNum( SiteStats::users() );
2721                                 break;
2722                         case 'numberofactiveusers':
2723                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2724                                 break;
2725                         case 'numberofpages':
2726                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2727                                 break;
2728                         case 'numberofadmins':
2729                                 $value = $wgContLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2730                                 break;
2731                         case 'numberofedits':
2732                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2733                                 break;
2734                         case 'numberofviews':
2735                                 $value = $wgContLang->formatNum( SiteStats::views() );
2736                                 break;
2737                         case 'currenttimestamp':
2738                                 $value = wfTimestamp( TS_MW, $ts );
2739                                 break;
2740                         case 'localtimestamp':
2741                                 $value = $localTimestamp;
2742                                 break;
2743                         case 'currentversion':
2744                                 $value = SpecialVersion::getVersion();
2745                                 break;
2746                         case 'articlepath':
2747                                 return $wgArticlePath;
2748                         case 'sitename':
2749                                 return $wgSitename;
2750                         case 'server':
2751                                 return $wgServer;
2752                         case 'servername':
2753                                 wfSuppressWarnings(); # May give an E_WARNING in PHP < 5.3.3
2754                                 $serverName = parse_url( $wgServer, PHP_URL_HOST );
2755                                 wfRestoreWarnings();
2756                                 return $serverName ? $serverName : $wgServer;
2757                         case 'scriptpath':
2758                                 return $wgScriptPath;
2759                         case 'stylepath':
2760                                 return $wgStylePath;
2761                         case 'directionmark':
2762                                 return $wgContLang->getDirMark();
2763                         case 'contentlanguage':
2764                                 global $wgLanguageCode;
2765                                 return $wgLanguageCode;
2766                         default:
2767                                 $ret = null;
2768                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
2769                                         return $ret;
2770                                 } else {
2771                                         return null;
2772                                 }
2773                 }
2774
2775                 if ( $index )
2776                         $this->mVarCache[$index] = $value;
2777
2778                 return $value;
2779         }
2780
2781         /**
2782          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2783          *
2784          * @private
2785          */
2786         function initialiseVariables() {
2787                 wfProfileIn( __METHOD__ );
2788                 $variableIDs = MagicWord::getVariableIDs();
2789                 $substIDs = MagicWord::getSubstIDs();
2790
2791                 $this->mVariables = new MagicWordArray( $variableIDs );
2792                 $this->mSubstWords = new MagicWordArray( $substIDs );
2793                 wfProfileOut( __METHOD__ );
2794         }
2795
2796         /**
2797          * Preprocess some wikitext and return the document tree.
2798          * This is the ghost of replace_variables().
2799          *
2800          * @param $text String: The text to parse
2801          * @param $flags Integer: bitwise combination of:
2802          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2803          *                                     included. Default is to assume a direct page view.
2804          *
2805          * The generated DOM tree must depend only on the input text and the flags.
2806          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2807          *
2808          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2809          * change in the DOM tree for a given text, must be passed through the section identifier
2810          * in the section edit link and thus back to extractSections().
2811          *
2812          * The output of this function is currently only cached in process memory, but a persistent
2813          * cache may be implemented at a later date which takes further advantage of these strict
2814          * dependency requirements.
2815          *
2816          * @private
2817          */
2818         function preprocessToDom( $text, $flags = 0 ) {
2819                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2820                 return $dom;
2821         }
2822
2823         /**
2824          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2825          */
2826         public static function splitWhitespace( $s ) {
2827                 $ltrimmed = ltrim( $s );
2828                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2829                 $trimmed = rtrim( $ltrimmed );
2830                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2831                 if ( $diff > 0 ) {
2832                         $w2 = substr( $ltrimmed, -$diff );
2833                 } else {
2834                         $w2 = '';
2835                 }
2836                 return array( $w1, $trimmed, $w2 );
2837         }
2838
2839         /**
2840          * Replace magic variables, templates, and template arguments
2841          * with the appropriate text. Templates are substituted recursively,
2842          * taking care to avoid infinite loops.
2843          *
2844          * Note that the substitution depends on value of $mOutputType:
2845          *  self::OT_WIKI: only {{subst:}} templates
2846          *  self::OT_PREPROCESS: templates but not extension tags
2847          *  self::OT_HTML: all templates and extension tags
2848          *
2849          * @param $text String: the text to transform
2850          * @param $frame PPFrame Object describing the arguments passed to the template.
2851          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2852          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2853          * @param $argsOnly Boolean: only do argument (triple-brace) expansion, not double-brace expansion
2854          * @private
2855          */
2856         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2857                 # Is there any text? Also, Prevent too big inclusions!
2858                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2859                         return $text;
2860                 }
2861                 wfProfileIn( __METHOD__ );
2862
2863                 if ( $frame === false ) {
2864                         $frame = $this->getPreprocessor()->newFrame();
2865                 } elseif ( !( $frame instanceof PPFrame ) ) {
2866                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2867                         $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2868                 }
2869
2870                 $dom = $this->preprocessToDom( $text );
2871                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2872                 $text = $frame->expand( $dom, $flags );
2873
2874                 wfProfileOut( __METHOD__ );
2875                 return $text;
2876         }
2877
2878         # Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2879         static function createAssocArgs( $args ) {
2880                 $assocArgs = array();
2881                 $index = 1;
2882                 foreach ( $args as $arg ) {
2883                         $eqpos = strpos( $arg, '=' );
2884                         if ( $eqpos === false ) {
2885                                 $assocArgs[$index++] = $arg;
2886                         } else {
2887                                 $name = trim( substr( $arg, 0, $eqpos ) );
2888                                 $value = trim( substr( $arg, $eqpos+1 ) );
2889                                 if ( $value === false ) {
2890                                         $value = '';
2891                                 }
2892                                 if ( $name !== false ) {
2893                                         $assocArgs[$name] = $value;
2894                                 }
2895                         }
2896                 }
2897
2898                 return $assocArgs;
2899         }
2900
2901         /**
2902          * Warn the user when a parser limitation is reached
2903          * Will warn at most once the user per limitation type
2904          *
2905          * @param $limitationType String: should be one of:
2906          *   'expensive-parserfunction' (corresponding messages:
2907          *       'expensive-parserfunction-warning',
2908          *       'expensive-parserfunction-category')
2909          *   'post-expand-template-argument' (corresponding messages:
2910          *       'post-expand-template-argument-warning',
2911          *       'post-expand-template-argument-category')
2912          *   'post-expand-template-inclusion' (corresponding messages:
2913          *       'post-expand-template-inclusion-warning',
2914          *       'post-expand-template-inclusion-category')
2915          * @param $current Current value
2916          * @param $max Maximum allowed, when an explicit limit has been
2917          *       exceeded, provide the values (optional)
2918          */
2919         function limitationWarn( $limitationType, $current=null, $max=null) {
2920                 # does no harm if $current and $max are present but are unnecessary for the message
2921                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
2922                 $this->mOutput->addWarning( $warning );
2923                 $this->addTrackingCategory( "$limitationType-category" );
2924         }
2925
2926         /**
2927          * Return the text of a template, after recursively
2928          * replacing any variables or templates within the template.
2929          *
2930          * @param $piece Array: the parts of the template
2931          *  $piece['title']: the title, i.e. the part before the |
2932          *  $piece['parts']: the parameter array
2933          *  $piece['lineStart']: whether the brace was at the start of a line
2934          * @param $frame PPFrame The current frame, contains template arguments
2935          * @return String: the text of the template
2936          * @private
2937          */
2938         function braceSubstitution( $piece, $frame ) {
2939                 global $wgContLang, $wgNonincludableNamespaces;
2940                 wfProfileIn( __METHOD__ );
2941                 wfProfileIn( __METHOD__.'-setup' );
2942
2943                 # Flags
2944                 $found = false;             # $text has been filled
2945                 $nowiki = false;            # wiki markup in $text should be escaped
2946                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2947                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2948                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2949                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2950
2951                 # Title object, where $text came from
2952                 $title = null;
2953
2954                 # $part1 is the bit before the first |, and must contain only title characters.
2955                 # Various prefixes will be stripped from it later.
2956                 $titleWithSpaces = $frame->expand( $piece['title'] );
2957                 $part1 = trim( $titleWithSpaces );
2958                 $titleText = false;
2959
2960                 # Original title text preserved for various purposes
2961                 $originalTitle = $part1;
2962
2963                 # $args is a list of argument nodes, starting from index 0, not including $part1
2964                 # *** FIXME if piece['parts'] is null then the call to getLength() below won't work b/c this $args isn't an object
2965                 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
2966                 wfProfileOut( __METHOD__.'-setup' );
2967
2968                 # SUBST
2969                 wfProfileIn( __METHOD__.'-modifiers' );
2970                 if ( !$found ) {
2971
2972                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
2973
2974                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
2975                         # Decide whether to expand template or keep wikitext as-is.
2976                         if ( $this->ot['wiki'] ) {
2977                                 if ( $substMatch === false ) {
2978                                         $literal = true;  # literal when in PST with no prefix
2979                                 } else {
2980                                         $literal = false; # expand when in PST with subst: or safesubst:
2981                                 }
2982                         } else {
2983                                 if ( $substMatch == 'subst' ) {
2984                                         $literal = true;  # literal when not in PST with plain subst:
2985                                 } else {
2986                                         $literal = false; # expand when not in PST with safesubst: or no prefix
2987                                 }
2988                         }
2989                         if ( $literal ) {
2990                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2991                                 $isLocalObj = true;
2992                                 $found = true;
2993                         }
2994                 }
2995
2996                 # Variables
2997                 if ( !$found && $args->getLength() == 0 ) {
2998                         $id = $this->mVariables->matchStartToEnd( $part1 );
2999                         if ( $id !== false ) {
3000                                 $text = $this->getVariableValue( $id, $frame );
3001                                 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3002                                         $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3003                                 }
3004                                 $found = true;
3005                         }
3006                 }
3007
3008                 # MSG, MSGNW and RAW
3009                 if ( !$found ) {
3010                         # Check for MSGNW:
3011                         $mwMsgnw = MagicWord::get( 'msgnw' );
3012                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3013                                 $nowiki = true;
3014                         } else {
3015                                 # Remove obsolete MSG:
3016                                 $mwMsg = MagicWord::get( 'msg' );
3017                                 $mwMsg->matchStartAndRemove( $part1 );
3018                         }
3019
3020                         # Check for RAW:
3021                         $mwRaw = MagicWord::get( 'raw' );
3022                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3023                                 $forceRawInterwiki = true;
3024                         }
3025                 }
3026                 wfProfileOut( __METHOD__.'-modifiers' );
3027
3028                 # Parser functions
3029                 if ( !$found ) {
3030                         wfProfileIn( __METHOD__ . '-pfunc' );
3031
3032                         $colonPos = strpos( $part1, ':' );
3033                         if ( $colonPos !== false ) {
3034                                 # Case sensitive functions
3035                                 $function = substr( $part1, 0, $colonPos );
3036                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3037                                         $function = $this->mFunctionSynonyms[1][$function];
3038                                 } else {
3039                                         # Case insensitive functions
3040                                         $function = $wgContLang->lc( $function );
3041                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3042                                                 $function = $this->mFunctionSynonyms[0][$function];
3043                                         } else {
3044                                                 $function = false;
3045                                         }
3046                                 }
3047                                 if ( $function ) {
3048                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3049                                         $initialArgs = array( &$this );
3050                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3051                                         if ( $flags & SFH_OBJECT_ARGS ) {
3052                                                 # Add a frame parameter, and pass the arguments as an array
3053                                                 $allArgs = $initialArgs;
3054                                                 $allArgs[] = $frame;
3055                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3056                                                         $funcArgs[] = $args->item( $i );
3057                                                 }
3058                                                 $allArgs[] = $funcArgs;
3059                                         } else {
3060                                                 # Convert arguments to plain text
3061                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3062                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
3063                                                 }
3064                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3065                                         }
3066
3067                                         # Workaround for PHP bug 35229 and similar
3068                                         if ( !is_callable( $callback ) ) {
3069                                                 wfProfileOut( __METHOD__ . '-pfunc' );
3070                                                 wfProfileOut( __METHOD__ );
3071                                                 throw new MWException( "Tag hook for $function is not callable\n" );
3072                                         }
3073                                         $result = call_user_func_array( $callback, $allArgs );
3074                                         $found = true;
3075                                         $noparse = true;
3076                                         $preprocessFlags = 0;
3077
3078                                         if ( is_array( $result ) ) {
3079                                                 if ( isset( $result[0] ) ) {
3080                                                         $text = $result[0];
3081                                                         unset( $result[0] );
3082                                                 }
3083
3084                                                 # Extract flags into the local scope
3085                                                 # This allows callers to set flags such as nowiki, found, etc.
3086                                                 extract( $result );
3087                                         } else {
3088                                                 $text = $result;
3089                                         }
3090                                         if ( !$noparse ) {
3091                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
3092                                                 $isChildObj = true;
3093                                         }
3094                                 }
3095                         }
3096                         wfProfileOut( __METHOD__ . '-pfunc' );
3097                 }
3098
3099                 # Finish mangling title and then check for loops.
3100                 # Set $title to a Title object and $titleText to the PDBK
3101                 if ( !$found ) {
3102                         $ns = NS_TEMPLATE;
3103                         # Split the title into page and subpage
3104                         $subpage = '';
3105                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3106                         if ( $subpage !== '' ) {
3107                                 $ns = $this->mTitle->getNamespace();
3108                         }
3109                         $title = Title::newFromText( $part1, $ns );
3110                         if ( $title ) {
3111                                 $titleText = $title->getPrefixedText();
3112                                 # Check for language variants if the template is not found
3113                                 if ( $wgContLang->hasVariants() && $title->getArticleID() == 0 ) {
3114                                         $wgContLang->findVariantLink( $part1, $title, true );
3115                                 }
3116                                 # Do recursion depth check
3117                                 $limit = $this->mOptions->getMaxTemplateDepth();
3118                                 if ( $frame->depth >= $limit ) {
3119                                         $found = true;
3120                                         $text = '<span class="error">'
3121                                                 . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit )
3122                                                 . '</span>';
3123                                 }
3124                         }
3125                 }
3126
3127                 # Load from database
3128                 if ( !$found && $title ) {
3129                         wfProfileIn( __METHOD__ . '-loadtpl' );
3130                         if ( !$title->isExternal() ) {
3131                                 if ( $title->getNamespace() == NS_SPECIAL
3132                                         && $this->mOptions->getAllowSpecialInclusion()
3133                                         && $this->ot['html'] )
3134                                 {
3135                                         $text = SpecialPage::capturePath( $title );
3136                                         if ( is_string( $text ) ) {
3137                                                 $found = true;
3138                                                 $isHTML = true;
3139                                                 $this->disableCache();
3140                                         }
3141                                 } elseif ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3142                                         $found = false; # access denied
3143                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3144                                 } else {
3145                                         list( $text, $title ) = $this->getTemplateDom( $title );
3146                                         if ( $text !== false ) {
3147                                                 $found = true;
3148                                                 $isChildObj = true;
3149                                         }
3150                                 }
3151
3152                                 # If the title is valid but undisplayable, make a link to it
3153                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3154                                         $text = "[[:$titleText]]";
3155                                         $found = true;
3156                                 }
3157                         } elseif ( $title->isTrans() ) {
3158                                 # Interwiki transclusion
3159                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3160                                         $text = $this->interwikiTransclude( $title, 'render' );
3161                                         $isHTML = true;
3162                                 } else {
3163                                         $text = $this->interwikiTransclude( $title, 'raw' );
3164                                         # Preprocess it like a template
3165                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3166                                         $isChildObj = true;
3167                                 }
3168                                 $found = true;
3169                         }
3170
3171                         # Do infinite loop check
3172                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3173                         if ( !$frame->loopCheck( $title ) ) {
3174                                 $found = true;
3175                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3176                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3177                         }
3178                         wfProfileOut( __METHOD__ . '-loadtpl' );
3179                 }
3180
3181                 # If we haven't found text to substitute by now, we're done
3182                 # Recover the source wikitext and return it
3183                 if ( !$found ) {
3184                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3185                         wfProfileOut( __METHOD__ );
3186                         return array( 'object' => $text );
3187                 }
3188
3189                 # Expand DOM-style return values in a child frame
3190                 if ( $isChildObj ) {
3191                         # Clean up argument array
3192                         $newFrame = $frame->newChild( $args, $title );
3193
3194                         if ( $nowiki ) {
3195                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3196                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3197                                 # Expansion is eligible for the empty-frame cache
3198                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3199                                         $text = $this->mTplExpandCache[$titleText];
3200                                 } else {
3201                                         $text = $newFrame->expand( $text );
3202                                         $this->mTplExpandCache[$titleText] = $text;
3203                                 }
3204                         } else {
3205                                 # Uncached expansion
3206                                 $text = $newFrame->expand( $text );
3207                         }
3208                 }
3209                 if ( $isLocalObj && $nowiki ) {
3210                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3211                         $isLocalObj = false;
3212                 }
3213
3214                 # Replace raw HTML by a placeholder
3215                 # Add a blank line preceding, to prevent it from mucking up
3216                 # immediately preceding headings
3217                 if ( $isHTML ) {
3218                         $text = "\n\n" . $this->insertStripItem( $text );
3219                 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3220                         # Escape nowiki-style return values
3221                         $text = wfEscapeWikiText( $text );
3222                 } elseif ( is_string( $text )
3223                         && !$piece['lineStart']
3224                         && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3225                 {
3226                         # Bug 529: if the template begins with a table or block-level
3227                         # element, it should be treated as beginning a new line.
3228                         # This behaviour is somewhat controversial.
3229                         $text = "\n" . $text;
3230                 }
3231
3232                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3233                         # Error, oversize inclusion
3234                         if ( $titleText !== false ) {
3235                                 # Make a working, properly escaped link if possible (bug 23588)
3236                                 $text = "[[:$titleText]]";
3237                         } else {
3238                                 # This will probably not be a working link, but at least it may
3239                                 # provide some hint of where the problem is
3240                                 preg_replace( '/^:/', '', $originalTitle );
3241                                 $text = "[[:$originalTitle]]";
3242                         }
3243                         $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3244                         $this->limitationWarn( 'post-expand-template-inclusion' );
3245                 }
3246
3247                 if ( $isLocalObj ) {
3248                         $ret = array( 'object' => $text );
3249                 } else {
3250                         $ret = array( 'text' => $text );
3251                 }
3252
3253                 wfProfileOut( __METHOD__ );
3254                 return $ret;
3255         }
3256
3257         /**
3258          * Get the semi-parsed DOM representation of a template with a given title,
3259          * and its redirect destination title. Cached.
3260          */
3261         function getTemplateDom( $title ) {
3262                 $cacheTitle = $title;
3263                 $titleText = $title->getPrefixedDBkey();
3264
3265                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3266                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3267                         $title = Title::makeTitle( $ns, $dbk );
3268                         $titleText = $title->getPrefixedDBkey();
3269                 }
3270                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3271                         return array( $this->mTplDomCache[$titleText], $title );
3272                 }
3273
3274                 # Cache miss, go to the database
3275                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3276
3277                 if ( $text === false ) {
3278                         $this->mTplDomCache[$titleText] = false;
3279                         return array( false, $title );
3280                 }
3281
3282                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3283                 $this->mTplDomCache[ $titleText ] = $dom;
3284
3285                 if ( !$title->equals( $cacheTitle ) ) {
3286                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3287                                 array( $title->getNamespace(), $cdb = $title->getDBkey() );
3288                 }
3289
3290                 return array( $dom, $title );
3291         }
3292
3293         /**
3294          * Fetch the unparsed text of a template and register a reference to it.
3295          */
3296         function fetchTemplateAndTitle( $title ) {
3297                 $templateCb = $this->mOptions->getTemplateCallback(); # Defaults to Parser::statelessFetchTemplate()
3298                 $stuff = call_user_func( $templateCb, $title, $this );
3299                 $text = $stuff['text'];
3300                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3301                 if ( isset( $stuff['deps'] ) ) {
3302                         foreach ( $stuff['deps'] as $dep ) {
3303                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3304                         }
3305                 }
3306                 return array( $text, $finalTitle );
3307         }
3308
3309         function fetchTemplate( $title ) {
3310                 $rv = $this->fetchTemplateAndTitle( $title );
3311                 return $rv[0];
3312         }
3313
3314         /**
3315          * Static function to get a template
3316          * Can be overridden via ParserOptions::setTemplateCallback().
3317          */
3318         static function statelessFetchTemplate( $title, $parser=false ) {
3319                 $text = $skip = false;
3320                 $finalTitle = $title;
3321                 $deps = array();
3322
3323                 # Loop to fetch the article, with up to 1 redirect
3324                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3325                         # Give extensions a chance to select the revision instead
3326                         $id = false; # Assume current
3327                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3328
3329                         if ( $skip ) {
3330                                 $text = false;
3331                                 $deps[] = array(
3332                                         'title' => $title,
3333                                         'page_id' => $title->getArticleID(),
3334                                         'rev_id' => null );
3335                                 break;
3336                         }
3337                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3338                         $rev_id = $rev ? $rev->getId() : 0;
3339                         # If there is no current revision, there is no page
3340                         if ( $id === false && !$rev ) {
3341                                 $linkCache = LinkCache::singleton();
3342                                 $linkCache->addBadLinkObj( $title );
3343                         }
3344
3345                         $deps[] = array(
3346                                 'title' => $title,
3347                                 'page_id' => $title->getArticleID(),
3348                                 'rev_id' => $rev_id );
3349
3350                         if ( $rev ) {
3351                                 $text = $rev->getText();
3352                         } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3353                                 global $wgContLang;
3354                                 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3355                                 if ( !$message->exists() ) {
3356                                         $text = false;
3357                                         break;
3358                                 }
3359                                 $text = $message->plain();
3360                         } else {
3361                                 break;
3362                         }
3363                         if ( $text === false ) {
3364                                 break;
3365                         }
3366                         # Redirect?
3367                         $finalTitle = $title;
3368                         $title = Title::newFromRedirect( $text );
3369                 }
3370                 return array(
3371                         'text' => $text,
3372                         'finalTitle' => $finalTitle,
3373                         'deps' => $deps );
3374         }
3375
3376         /**
3377          * Transclude an interwiki link.
3378          */
3379         function interwikiTransclude( $title, $action ) {
3380                 global $wgEnableScaryTranscluding;
3381
3382                 if ( !$wgEnableScaryTranscluding ) {
3383                         return wfMsgForContent('scarytranscludedisabled');
3384                 }
3385
3386                 $url = $title->getFullUrl( "action=$action" );
3387
3388                 if ( strlen( $url ) > 255 ) {
3389                         return wfMsgForContent( 'scarytranscludetoolong' );
3390                 }
3391                 return $this->fetchScaryTemplateMaybeFromCache( $url );
3392         }
3393
3394         function fetchScaryTemplateMaybeFromCache( $url ) {
3395                 global $wgTranscludeCacheExpiry;
3396                 $dbr = wfGetDB( DB_SLAVE );
3397                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3398                 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3399                                 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3400                 if ( $obj ) {
3401                         return $obj->tc_contents;
3402                 }
3403
3404                 $text = Http::get( $url );
3405                 if ( !$text ) {
3406                         return wfMsgForContent( 'scarytranscludefailed', $url );
3407                 }
3408
3409                 $dbw = wfGetDB( DB_MASTER );
3410                 $dbw->replace( 'transcache', array('tc_url'), array(
3411                         'tc_url' => $url,
3412                         'tc_time' => $dbw->timestamp( time() ),
3413                         'tc_contents' => $text)
3414                 );
3415                 return $text;
3416         }
3417
3418
3419         /**
3420          * Triple brace replacement -- used for template arguments
3421          * @private
3422          */
3423         function argSubstitution( $piece, $frame ) {
3424                 wfProfileIn( __METHOD__ );
3425
3426                 $error = false;
3427                 $parts = $piece['parts'];
3428                 $nameWithSpaces = $frame->expand( $piece['title'] );
3429                 $argName = trim( $nameWithSpaces );
3430                 $object = false;
3431                 $text = $frame->getArgument( $argName );
3432                 if (  $text === false && $parts->getLength() > 0
3433                   && (
3434                         $this->ot['html']
3435                         || $this->ot['pre']
3436                         || ( $this->ot['wiki'] && $frame->isTemplate() )
3437                   )
3438                 ) {
3439                         # No match in frame, use the supplied default
3440                         $object = $parts->item( 0 )->getChildren();
3441                 }
3442                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3443                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3444                         $this->limitationWarn( 'post-expand-template-argument' );
3445                 }
3446
3447                 if ( $text === false && $object === false ) {
3448                         # No match anywhere
3449                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3450                 }
3451                 if ( $error !== false ) {
3452                         $text .= $error;
3453                 }
3454                 if ( $object !== false ) {
3455                         $ret = array( 'object' => $object );
3456                 } else {
3457                         $ret = array( 'text' => $text );
3458                 }
3459
3460                 wfProfileOut( __METHOD__ );
3461                 return $ret;
3462         }
3463
3464         /**
3465          * Return the text to be used for a given extension tag.
3466          * This is the ghost of strip().
3467          *
3468          * @param $params Associative array of parameters:
3469          *     name       PPNode for the tag name
3470          *     attr       PPNode for unparsed text where tag attributes are thought to be
3471          *     attributes Optional associative array of parsed attributes
3472          *     inner      Contents of extension element
3473          *     noClose    Original text did not have a close tag
3474          * @param $frame PPFrame
3475          */
3476         function extensionSubstitution( $params, $frame ) {
3477                 $name = $frame->expand( $params['name'] );
3478                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3479                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3480                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3481
3482                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3483                         ( $this->ot['html'] || $this->ot['pre'] );
3484                 if ( $isFunctionTag ) {
3485                         $markerType = 'none';
3486                 } else {
3487                         $markerType = 'general';
3488                 }
3489                 if ( $this->ot['html'] || $isFunctionTag ) {
3490                         $name = strtolower( $name );
3491                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3492                         if ( isset( $params['attributes'] ) ) {
3493                                 $attributes = $attributes + $params['attributes'];
3494                         }
3495
3496                         if ( isset( $this->mTagHooks[$name] ) ) {
3497                                 # Workaround for PHP bug 35229 and similar
3498                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3499                                         throw new MWException( "Tag hook for $name is not callable\n" );
3500                                 }
3501                                 $output = call_user_func_array( $this->mTagHooks[$name],
3502                                         array( $content, $attributes, $this, $frame ) );
3503                         } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3504                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3505                                 if ( !is_callable( $callback ) ) {
3506                                         throw new MWException( "Tag hook for $name is not callable\n" );
3507                                 }
3508
3509                                 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3510                         } else {
3511                                 $output = '<span class="error">Invalid tag extension name: ' .
3512                                         htmlspecialchars( $name ) . '</span>';
3513                         }
3514
3515                         if ( is_array( $output ) ) {
3516                                 # Extract flags to local scope (to override $markerType)
3517                                 $flags = $output;
3518                                 $output = $flags[0];
3519                                 unset( $flags[0] );
3520                                 extract( $flags );
3521                         }
3522                 } else {
3523                         if ( is_null( $attrText ) ) {
3524                                 $attrText = '';
3525                         }
3526                         if ( isset( $params['attributes'] ) ) {
3527                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3528                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3529                                                 htmlspecialchars( $attrValue ) . '"';
3530                                 }
3531                         }
3532                         if ( $content === null ) {
3533                                 $output = "<$name$attrText/>";
3534                         } else {
3535                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3536                                 $output = "<$name$attrText>$content$close";
3537                         }
3538                 }
3539
3540                 if ( $markerType === 'none' ) {
3541                         return $output;
3542                 } elseif ( $markerType === 'nowiki' ) {
3543                         $this->mStripState->nowiki->setPair( $marker, $output );
3544                 } elseif ( $markerType === 'general' ) {
3545                         $this->mStripState->general->setPair( $marker, $output );
3546                 } else {
3547                         throw new MWException( __METHOD__.': invalid marker type' );
3548                 }
3549                 return $marker;
3550         }
3551
3552         /**
3553          * Increment an include size counter
3554          *
3555          * @param $type String: the type of expansion
3556          * @param $size Integer: the size of the text
3557          * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
3558          */
3559         function incrementIncludeSize( $type, $size ) {
3560                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3561                         return false;
3562                 } else {
3563                         $this->mIncludeSizes[$type] += $size;
3564                         return true;
3565                 }
3566         }
3567
3568         /**
3569          * Increment the expensive function count
3570          *
3571          * @return Boolean: false if the limit has been exceeded
3572          */
3573         function incrementExpensiveFunctionCount() {
3574                 global $wgExpensiveParserFunctionLimit;
3575                 $this->mExpensiveFunctionCount++;
3576                 if ( $this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit ) {
3577                         return true;
3578                 }
3579                 return false;
3580         }
3581
3582         /**
3583          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3584          * Fills $this->mDoubleUnderscores, returns the modified text
3585          */
3586         function doDoubleUnderscore( $text ) {
3587                 wfProfileIn( __METHOD__ );
3588
3589                 # The position of __TOC__ needs to be recorded
3590                 $mw = MagicWord::get( 'toc' );
3591                 if ( $mw->match( $text ) ) {
3592                         $this->mShowToc = true;
3593                         $this->mForceTocPosition = true;
3594
3595                         # Set a placeholder. At the end we'll fill it in with the TOC.
3596                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3597
3598                         # Only keep the first one.
3599                         $text = $mw->replace( '', $text );
3600                 }
3601
3602                 # Now match and remove the rest of them
3603                 $mwa = MagicWord::getDoubleUnderscoreArray();
3604                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3605
3606                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3607                         $this->mOutput->mNoGallery = true;
3608                 }
3609                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3610                         $this->mShowToc = false;
3611                 }
3612                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3613                         $this->addTrackingCategory( 'hidden-category-category' );
3614                 }
3615                 # (bug 8068) Allow control over whether robots index a page.
3616                 #
3617                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3618                 # is not desirable, the last one on the page should win.
3619                 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3620                         $this->mOutput->setIndexPolicy( 'noindex' );
3621                         $this->addTrackingCategory( 'noindex-category' );
3622                 }
3623                 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3624                         $this->mOutput->setIndexPolicy( 'index' );
3625                         $this->addTrackingCategory( 'index-category' );
3626                 }
3627
3628                 # Cache all double underscores in the database
3629                 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3630                         $this->mOutput->setProperty( $key, '' );
3631                 }
3632
3633                 wfProfileOut( __METHOD__ );
3634                 return $text;
3635         }
3636
3637         /**
3638          * Add a tracking category, getting the title from a system message,
3639          * or print a debug message if the title is invalid.
3640          *
3641          * @param $msg String: message key
3642          * @return Boolean: whether the addition was successful
3643          */
3644         protected function addTrackingCategory( $msg ) {
3645                 $cat = wfMsgForContent( $msg );
3646
3647                 # Allow tracking categories to be disabled by setting them to "-"
3648                 if ( $cat === '-' ) {
3649                         return false;
3650                 }
3651
3652                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3653                 if ( $containerCategory ) {
3654                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3655                         return true;
3656                 } else {
3657                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3658                         return false;
3659                 }
3660         }
3661
3662         /**
3663          * This function accomplishes several tasks:
3664          * 1) Auto-number headings if that option is enabled
3665          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3666          * 3) Add a Table of contents on the top for users who have enabled the option
3667          * 4) Auto-anchor headings
3668          *
3669          * It loops through all headlines, collects the necessary data, then splits up the
3670          * string and re-inserts the newly formatted headlines.
3671          *
3672          * @param $text String
3673          * @param $origText String: original, untouched wikitext
3674          * @param $isMain Boolean
3675          * @private
3676          */
3677         function formatHeadings( $text, $origText, $isMain=true ) {
3678                 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3679
3680                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3681
3682                 # Inhibit editsection links if requested in the page
3683                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3684                         $showEditLink = 0;
3685                 } else {
3686                         $showEditLink = $this->mOptions->getEditSection();
3687                 }
3688                 if ( $showEditLink ) {
3689                         $this->mOutput->setEditSectionTokens( true );
3690                 }
3691
3692                 # Get all headlines for numbering them and adding funky stuff like [edit]
3693                 # links - this is for later, but we need the number of headlines right now
3694                 $matches = array();
3695                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3696
3697                 # if there are fewer than 4 headlines in the article, do not show TOC
3698                 # unless it's been explicitly enabled.
3699                 $enoughToc = $this->mShowToc &&
3700                         ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3701
3702                 # Allow user to stipulate that a page should have a "new section"
3703                 # link added via __NEWSECTIONLINK__
3704                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3705                         $this->mOutput->setNewSection( true );
3706                 }
3707
3708                 # Allow user to remove the "new section"
3709                 # link via __NONEWSECTIONLINK__
3710                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3711                         $this->mOutput->hideNewSection( true );
3712                 }
3713
3714                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3715                 # override above conditions and always show TOC above first header
3716                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3717                         $this->mShowToc = true;
3718                         $enoughToc = true;
3719                 }
3720
3721                 # We need this to perform operations on the HTML
3722                 $sk = $this->mOptions->getSkin( $this->mTitle );
3723
3724                 # headline counter
3725                 $headlineCount = 0;
3726                 $numVisible = 0;
3727
3728                 # Ugh .. the TOC should have neat indentation levels which can be
3729                 # passed to the skin functions. These are determined here
3730                 $toc = '';
3731                 $full = '';
3732                 $head = array();
3733                 $sublevelCount = array();
3734                 $levelCount = array();
3735                 $level = 0;
3736                 $prevlevel = 0;
3737                 $toclevel = 0;
3738                 $prevtoclevel = 0;
3739                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3740                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3741                 $oldType = $this->mOutputType;
3742                 $this->setOutputType( self::OT_WIKI );
3743                 $frame = $this->getPreprocessor()->newFrame();
3744                 $root = $this->preprocessToDom( $origText );
3745                 $node = $root->getFirstChild();
3746                 $byteOffset = 0;
3747                 $tocraw = array();
3748                 $refers = array();
3749
3750                 foreach ( $matches[3] as $headline ) {
3751                         $isTemplate = false;
3752                         $titleText = false;
3753                         $sectionIndex = false;
3754                         $numbering = '';
3755                         $markerMatches = array();
3756                         if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
3757                                 $serial = $markerMatches[1];
3758                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3759                                 $isTemplate = ( $titleText != $baseTitleText );
3760                                 $headline = preg_replace( "/^$markerRegex/", "", $headline );
3761                         }
3762
3763                         if ( $toclevel ) {
3764                                 $prevlevel = $level;
3765                         }
3766                         $level = $matches[1][$headlineCount];
3767
3768                         if ( $level > $prevlevel ) {
3769                                 # Increase TOC level
3770                                 $toclevel++;
3771                                 $sublevelCount[$toclevel] = 0;
3772                                 if ( $toclevel<$wgMaxTocLevel ) {
3773                                         $prevtoclevel = $toclevel;
3774                                         $toc .= $sk->tocIndent();
3775                                         $numVisible++;
3776                                 }
3777                         } elseif ( $level < $prevlevel && $toclevel > 1 ) {
3778                                 # Decrease TOC level, find level to jump to
3779
3780                                 for ( $i = $toclevel; $i > 0; $i-- ) {
3781                                         if ( $levelCount[$i] == $level ) {
3782                                                 # Found last matching level
3783                                                 $toclevel = $i;
3784                                                 break;
3785                                         } elseif ( $levelCount[$i] < $level ) {
3786                                                 # Found first matching level below current level
3787                                                 $toclevel = $i + 1;
3788                                                 break;
3789                                         }
3790                                 }
3791                                 if ( $i == 0 ) {
3792                                         $toclevel = 1;
3793                                 }
3794                                 if ( $toclevel<$wgMaxTocLevel ) {
3795                                         if ( $prevtoclevel < $wgMaxTocLevel ) {
3796                                                 # Unindent only if the previous toc level was shown :p
3797                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3798                                                 $prevtoclevel = $toclevel;
3799                                         } else {
3800                                                 $toc .= $sk->tocLineEnd();
3801                                         }
3802                                 }
3803                         } else {
3804                                 # No change in level, end TOC line
3805                                 if ( $toclevel<$wgMaxTocLevel ) {
3806                                         $toc .= $sk->tocLineEnd();
3807                                 }
3808                         }
3809
3810                         $levelCount[$toclevel] = $level;
3811
3812                         # count number of headlines for each level
3813                         @$sublevelCount[$toclevel]++;
3814                         $dot = 0;
3815                         for( $i = 1; $i <= $toclevel; $i++ ) {
3816                                 if ( !empty( $sublevelCount[$i] ) ) {
3817                                         if ( $dot ) {
3818                                                 $numbering .= '.';
3819                                         }
3820                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3821                                         $dot = 1;
3822                                 }
3823                         }
3824
3825                         # The safe header is a version of the header text safe to use for links
3826                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3827                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3828
3829                         # Remove link placeholders by the link text.
3830                         #     <!--LINK number-->
3831                         # turns into
3832                         #     link text with suffix
3833                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3834
3835                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3836                         $tocline = preg_replace(
3837                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3838                                 array( '',                          '<$1>' ),
3839                                 $safeHeadline
3840                         );
3841                         $tocline = trim( $tocline );
3842
3843                         # For the anchor, strip out HTML-y stuff period
3844                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3845                         $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
3846
3847                         # Save headline for section edit hint before it's escaped
3848                         $headlineHint = $safeHeadline;
3849
3850                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
3851                                 # For reverse compatibility, provide an id that's
3852                                 # HTML4-compatible, like we used to.
3853                                 #
3854                                 # It may be worth noting, academically, that it's possible for
3855                                 # the legacy anchor to conflict with a non-legacy headline
3856                                 # anchor on the page.  In this case likely the "correct" thing
3857                                 # would be to either drop the legacy anchors or make sure
3858                                 # they're numbered first.  However, this would require people
3859                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3860                                 # manually, so let's not bother worrying about it.
3861                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3862                                         array( 'noninitial', 'legacy' ) );
3863                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3864
3865                                 if ( $legacyHeadline == $safeHeadline ) {
3866                                         # No reason to have both (in fact, we can't)
3867                                         $legacyHeadline = false;
3868                                 }
3869                         } else {
3870                                 $legacyHeadline = false;
3871                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3872                                         'noninitial' );
3873                         }
3874
3875                         # HTML names must be case-insensitively unique (bug 10721).
3876                         # This does not apply to Unicode characters per
3877                         # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
3878                         # FIXME: We may be changing them depending on the current locale.
3879                         $arrayKey = strtolower( $safeHeadline );
3880                         if ( $legacyHeadline === false ) {
3881                                 $legacyArrayKey = false;
3882                         } else {
3883                                 $legacyArrayKey = strtolower( $legacyHeadline );
3884                         }
3885
3886                         # count how many in assoc. array so we can track dupes in anchors
3887                         if ( isset( $refers[$arrayKey] ) ) {
3888                                 $refers[$arrayKey]++;
3889                         } else {
3890                                 $refers[$arrayKey] = 1;
3891                         }
3892                         if ( isset( $refers[$legacyArrayKey] ) ) {
3893                                 $refers[$legacyArrayKey]++;
3894                         } else {
3895                                 $refers[$legacyArrayKey] = 1;
3896                         }
3897
3898                         # Don't number the heading if it is the only one (looks silly)
3899                         if ( $doNumberHeadings && count( $matches[3] ) > 1) {
3900                                 # the two are different if the line contains a link
3901                                 $headline = $numbering . ' ' . $headline;
3902                         }
3903
3904                         # Create the anchor for linking from the TOC to the section
3905                         $anchor = $safeHeadline;
3906                         $legacyAnchor = $legacyHeadline;
3907                         if ( $refers[$arrayKey] > 1 ) {
3908                                 $anchor .= '_' . $refers[$arrayKey];
3909                         }
3910                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3911                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3912                         }
3913                         if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
3914                                 $toc .= $sk->tocLine( $anchor, $tocline,
3915                                         $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
3916                         }
3917
3918                         # Add the section to the section tree
3919                         # Find the DOM node for this header
3920                         while ( $node && !$isTemplate ) {
3921                                 if ( $node->getName() === 'h' ) {
3922                                         $bits = $node->splitHeading();
3923                                         if ( $bits['i'] == $sectionIndex ) {
3924                                                 break;
3925                                         }
3926                                 }
3927                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
3928                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
3929                                 $node = $node->getNextSibling();
3930                         }
3931                         $tocraw[] = array(
3932                                 'toclevel' => $toclevel,
3933                                 'level' => $level,
3934                                 'line' => $tocline,
3935                                 'number' => $numbering,
3936                                 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
3937                                 'fromtitle' => $titleText,
3938                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
3939                                 'anchor' => $anchor,
3940                         );
3941
3942                         # give headline the correct <h#> tag
3943                         if ( $showEditLink && $sectionIndex !== false ) {
3944                                 // Output edit section links as markers with styles that can be customized by skins
3945                                 if ( $isTemplate ) {
3946                                         # Put a T flag in the section identifier, to indicate to extractSections()
3947                                         # that sections inside <includeonly> should be counted.
3948                                         $editlinkArgs = array( $titleText, "T-$sectionIndex"/*, null */ );
3949                                 } else {
3950                                         $editlinkArgs = array( $this->mTitle->getPrefixedText(), $sectionIndex, $headlineHint );
3951                                 }
3952                                 // We use a bit of pesudo-xml for editsection markers. The language converter is run later on
3953                                 // Using a UNIQ style marker leads to the converter screwing up the tokens when it converts stuff
3954                                 // And trying to insert strip tags fails too. At this point all real inputted tags have already been escaped
3955                                 // so we don't have to worry about a user trying to input one of these markers directly.
3956                                 // We use a page and section attribute to stop the language converter from converting these important bits
3957                                 // of data, but put the headline hint inside a content block because the language converter is supposed to
3958                                 // be able to convert that piece of data.
3959                                 $editlink = '<mw:editsection page="' . htmlspecialchars($editlinkArgs[0]);
3960                                 $editlink .= '" section="' . htmlspecialchars($editlinkArgs[1]) .'"';
3961                                 if ( isset($editlinkArgs[2]) ) {
3962                                         $editlink .= '>' . $editlinkArgs[2] . '</mw:editsection>';
3963                                 } else {
3964                                         $editlink .= '/>';
3965                                 }
3966                         } else {
3967                                 $editlink = '';
3968                         }
3969                         $head[$headlineCount] = $sk->makeHeadline( $level,
3970                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3971                                 $editlink, $legacyAnchor );
3972
3973                         $headlineCount++;
3974                 }
3975
3976                 $this->setOutputType( $oldType );
3977
3978                 # Never ever show TOC if no headers
3979                 if ( $numVisible < 1 ) {
3980                         $enoughToc = false;
3981                 }
3982
3983                 if ( $enoughToc ) {
3984                         if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3985                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3986                         }
3987                         $toc = $sk->tocList( $toc );
3988                         $this->mOutput->setTOCHTML( $toc );
3989                 }
3990
3991                 if ( $isMain ) {
3992                         $this->mOutput->setSections( $tocraw );
3993                 }
3994
3995                 # split up and insert constructed headlines
3996
3997                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3998                 $i = 0;
3999
4000                 foreach ( $blocks as $block ) {
4001                         if ( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
4002                                 # This is the [edit] link that appears for the top block of text when
4003                                 # section editing is enabled
4004
4005                                 # Disabled because it broke block formatting
4006                                 # For example, a bullet point in the top line
4007                                 # $full .= $sk->editSectionLink(0);
4008                         }
4009                         $full .= $block;
4010                         if ( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
4011                                 # Top anchor now in skin
4012                                 $full = $full.$toc;
4013                         }
4014
4015                         if ( !empty( $head[$i] ) ) {
4016                                 $full .= $head[$i];
4017                         }
4018                         $i++;
4019                 }
4020                 if ( $this->mForceTocPosition ) {
4021                         return str_replace( '<!--MWTOC-->', $toc, $full );
4022                 } else {
4023                         return $full;
4024                 }
4025         }
4026
4027         /**
4028          * Transform wiki markup when saving a page by doing \r\n -> \n
4029          * conversion, substitting signatures, {{subst:}} templates, etc.
4030          *
4031          * @param $text String: the text to transform
4032          * @param $title Title: the Title object for the current article
4033          * @param $user User: the User object describing the current user
4034          * @param $options ParserOptions: parsing options
4035          * @param $clearState Boolean: whether to clear the parser state first
4036          * @return String: the altered wiki markup
4037          */
4038         public function preSaveTransform( $text, Title $title, User $user, ParserOptions $options, $clearState = true ) {
4039                 $this->startExternalParse( $title, $options, self::OT_WIKI, $clearState );
4040                 $this->setUser( $user );
4041
4042                 $pairs = array(
4043                         "\r\n" => "\n",
4044                 );
4045                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4046                 if( $options->getPreSaveTransform() ) {
4047                         $text = $this->pstPass2( $text, $user );
4048                 }
4049                 $text = $this->mStripState->unstripBoth( $text );
4050
4051                 $this->setUser( null ); #Reset
4052
4053                 return $text;
4054         }
4055
4056         /**
4057          * Pre-save transform helper function
4058          * @private
4059          */
4060         function pstPass2( $text, $user ) {
4061                 global $wgContLang, $wgLocaltimezone;
4062
4063                 # Note: This is the timestamp saved as hardcoded wikitext to
4064                 # the database, we use $wgContLang here in order to give
4065                 # everyone the same signature and use the default one rather
4066                 # than the one selected in each user's preferences.
4067                 # (see also bug 12815)
4068                 $ts = $this->mOptions->getTimestamp();
4069                 if ( isset( $wgLocaltimezone ) ) {
4070                         $tz = $wgLocaltimezone;
4071                 } else {
4072                         $tz = date_default_timezone_get();
4073                 }
4074
4075                 $unixts = wfTimestamp( TS_UNIX, $ts );
4076                 $oldtz = date_default_timezone_get();
4077                 date_default_timezone_set( $tz );
4078                 $ts = date( 'YmdHis', $unixts );
4079                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4080
4081                 # Allow translation of timezones through wiki. date() can return
4082                 # whatever crap the system uses, localised or not, so we cannot
4083                 # ship premade translations.
4084                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4085                 $msg = wfMessage( $key )->inContentLanguage();
4086                 if ( $msg->exists() ) {
4087                         $tzMsg = $msg->text();
4088                 }
4089
4090                 date_default_timezone_set( $oldtz );
4091
4092                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4093
4094                 # Variable replacement
4095                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4096                 $text = $this->replaceVariables( $text );
4097
4098                 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4099                 # which may corrupt this parser instance via its wfMsgExt( parsemag ) call-
4100
4101                 # Signatures
4102                 $sigText = $this->getUserSig( $user );
4103                 $text = strtr( $text, array(
4104                         '~~~~~' => $d,
4105                         '~~~~' => "$sigText $d",
4106                         '~~~' => $sigText
4107                 ) );
4108
4109                 # Context links: [[|name]] and [[name (context)|]]
4110                 global $wgLegalTitleChars;
4111                 $tc = "[$wgLegalTitleChars]";
4112                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4113
4114                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4115                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
4116                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4117                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4118
4119                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4120                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4121                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4122                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4123
4124                 $t = $this->mTitle->getText();
4125                 $m = array();
4126                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4127                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4128                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4129                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4130                 } else {
4131                         # if there's no context, don't bother duplicating the title
4132                         $text = preg_replace( $p2, '[[\\1]]', $text );
4133                 }
4134
4135                 # Trim trailing whitespace
4136                 $text = rtrim( $text );
4137
4138                 return $text;
4139         }
4140
4141         /**
4142          * Fetch the user's signature text, if any, and normalize to
4143          * validated, ready-to-insert wikitext.
4144          * If you have pre-fetched the nickname or the fancySig option, you can
4145          * specify them here to save a database query.
4146          * Do not reuse this parser instance after calling getUserSig(),
4147          * as it may have changed if it's the $wgParser.
4148          *
4149          * @param $user User
4150          * @param $nickname String: nickname to use or false to use user's default nickname
4151          * @param $fancySig Boolean: whether the nicknname is the complete signature
4152          *                  or null to use default value
4153          * @return string
4154          */
4155         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4156                 global $wgMaxSigChars;
4157
4158                 $username = $user->getName();
4159
4160                 # If not given, retrieve from the user object.
4161                 if ( $nickname === false )
4162                         $nickname = $user->getOption( 'nickname' );
4163
4164                 if ( is_null( $fancySig ) ) {
4165                         $fancySig = $user->getBoolOption( 'fancysig' );
4166                 }
4167
4168                 $nickname = $nickname == null ? $username : $nickname;
4169
4170                 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4171                         $nickname = $username;
4172                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4173                 } elseif ( $fancySig !== false ) {
4174                         # Sig. might contain markup; validate this
4175                         if ( $this->validateSig( $nickname ) !== false ) {
4176                                 # Validated; clean up (if needed) and return it
4177                                 return $this->cleanSig( $nickname, true );
4178                         } else {
4179                                 # Failed to validate; fall back to the default
4180                                 $nickname = $username;
4181                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4182                         }
4183                 }
4184
4185                 # Make sure nickname doesnt get a sig in a sig
4186                 $nickname = $this->cleanSigInSig( $nickname );
4187
4188                 # If we're still here, make it a link to the user page
4189                 $userText = wfEscapeWikiText( $username );
4190                 $nickText = wfEscapeWikiText( $nickname );
4191                 if ( $user->isAnon() )  {
4192                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4193                 } else {
4194                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4195                 }
4196         }
4197
4198         /**
4199          * Check that the user's signature contains no bad XML
4200          *
4201          * @param $text String
4202          * @return mixed An expanded string, or false if invalid.
4203          */
4204         function validateSig( $text ) {
4205                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4206         }
4207
4208         /**
4209          * Clean up signature text
4210          *
4211          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4212          * 2) Substitute all transclusions
4213          *
4214          * @param $text String
4215          * @param $parsing Whether we're cleaning (preferences save) or parsing
4216          * @return String: signature text
4217          */
4218         function cleanSig( $text, $parsing = false ) {
4219                 if ( !$parsing ) {
4220                         global $wgTitle;
4221                         $this->mOptions = new ParserOptions;
4222                         $this->clearState();
4223                         $this->setTitle( $wgTitle );
4224                         $this->setOutputType = self::OT_PREPROCESS;
4225                 }
4226
4227                 # Option to disable this feature
4228                 if ( !$this->mOptions->getCleanSignatures() ) {
4229                         return $text;
4230                 }
4231
4232                 # FIXME: regex doesn't respect extension tags or nowiki
4233                 #  => Move this logic to braceSubstitution()
4234                 $substWord = MagicWord::get( 'subst' );
4235                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4236                 $substText = '{{' . $substWord->getSynonym( 0 );
4237
4238                 $text = preg_replace( $substRegex, $substText, $text );
4239                 $text = $this->cleanSigInSig( $text );
4240                 $dom = $this->preprocessToDom( $text );
4241                 $frame = $this->getPreprocessor()->newFrame();
4242                 $text = $frame->expand( $dom );
4243
4244                 if ( !$parsing ) {
4245                         $text = $this->mStripState->unstripBoth( $text );
4246                 }
4247
4248                 return $text;
4249         }
4250
4251         /**
4252          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4253          *
4254          * @param $text String
4255          * @return String: signature text with /~{3,5}/ removed
4256          */
4257         function cleanSigInSig( $text ) {
4258                 $text = preg_replace( '/~{3,5}/', '', $text );
4259                 return $text;
4260         }
4261
4262         /**
4263          * Set up some variables which are usually set up in parse()
4264          * so that an external function can call some class members with confidence
4265          */
4266         public function startExternalParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4267                 $this->setTitle( $title );
4268                 $this->mOptions = $options;
4269                 $this->setOutputType( $outputType );
4270                 if ( $clearState ) {
4271                         $this->clearState();
4272                 }
4273         }
4274
4275         /**
4276          * Wrapper for preprocess()
4277          *
4278          * @param $text String: the text to preprocess
4279          * @param $options ParserOptions: options
4280          * @param $title Title object or null to use $wgTitle
4281          * @return String
4282          */
4283         public function transformMsg( $text, $options, $title = null ) {
4284                 static $executing = false;
4285
4286                 # Guard against infinite recursion
4287                 if ( $executing ) {
4288                         return $text;
4289                 }
4290                 $executing = true;
4291
4292                 wfProfileIn( __METHOD__ );
4293                 if ( !$title ) {
4294                         global $wgTitle;
4295                         $title = $wgTitle;
4296                 }
4297                 if ( !$title ) {
4298                         # It's not uncommon having a null $wgTitle in scripts. See r80898
4299                         # Create a ghost title in such case
4300                         $title = Title::newFromText( 'Dwimmerlaik' );
4301                 }
4302                 $text = $this->preprocess( $text, $title, $options );
4303
4304                 $executing = false;
4305                 wfProfileOut( __METHOD__ );
4306                 return $text;
4307         }
4308
4309         /**
4310          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4311          * The callback should have the following form:
4312          *    function myParserHook( $text, $params, $parser ) { ... }
4313          *
4314          * Transform and return $text. Use $parser for any required context, e.g. use
4315          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4316          *
4317          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4318          * @param $callback Mixed: the callback function (and object) to use for the tag
4319          * @return The old value of the mTagHooks array associated with the hook
4320          */
4321         public function setHook( $tag, $callback ) {
4322                 $tag = strtolower( $tag );
4323                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4324                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4325                 $this->mTagHooks[$tag] = $callback;
4326                 if ( !in_array( $tag, $this->mStripList ) ) {
4327                         $this->mStripList[] = $tag;
4328                 }
4329
4330                 return $oldVal;
4331         }
4332
4333         function setTransparentTagHook( $tag, $callback ) {
4334                 $tag = strtolower( $tag );
4335                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4336                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4337                 $this->mTransparentTagHooks[$tag] = $callback;
4338
4339                 return $oldVal;
4340         }
4341
4342         /**
4343          * Remove all tag hooks
4344          */
4345         function clearTagHooks() {
4346                 $this->mTagHooks = array();
4347                 $this->mStripList = $this->mDefaultStripList;
4348         }
4349
4350         /**
4351          * Create a function, e.g. {{sum:1|2|3}}
4352          * The callback function should have the form:
4353          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4354          *
4355          * Or with SFH_OBJECT_ARGS:
4356          *    function myParserFunction( $parser, $frame, $args ) { ... }
4357          *
4358          * The callback may either return the text result of the function, or an array with the text
4359          * in element 0, and a number of flags in the other elements. The names of the flags are
4360          * specified in the keys. Valid flags are:
4361          *   found                     The text returned is valid, stop processing the template. This
4362          *                             is on by default.
4363          *   nowiki                    Wiki markup in the return value should be escaped
4364          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4365          *
4366          * @param $id String: The magic word ID
4367          * @param $callback Mixed: the callback function (and object) to use
4368          * @param $flags Integer: a combination of the following flags:
4369          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4370          *
4371          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4372          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4373          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4374          *     the arguments, and to control the way they are expanded.
4375          *
4376          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4377          *     arguments, for instance:
4378          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4379          *
4380          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4381          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4382          *     working if/when this is changed.
4383          *
4384          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4385          *     expansion.
4386          *
4387          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4388          *     about the methods available in PPFrame and PPNode.
4389          *
4390          * @return The old callback function for this name, if any
4391          */
4392         public function setFunctionHook( $id, $callback, $flags = 0 ) {
4393                 global $wgContLang;
4394
4395                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4396                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4397
4398                 # Add to function cache
4399                 $mw = MagicWord::get( $id );
4400                 if ( !$mw )
4401                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4402
4403                 $synonyms = $mw->getSynonyms();
4404                 $sensitive = intval( $mw->isCaseSensitive() );
4405
4406                 foreach ( $synonyms as $syn ) {
4407                         # Case
4408                         if ( !$sensitive ) {
4409                                 $syn = $wgContLang->lc( $syn );
4410                         }
4411                         # Add leading hash
4412                         if ( !( $flags & SFH_NO_HASH ) ) {
4413                                 $syn = '#' . $syn;
4414                         }
4415                         # Remove trailing colon
4416                         if ( substr( $syn, -1, 1 ) === ':' ) {
4417                                 $syn = substr( $syn, 0, -1 );
4418                         }
4419                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4420                 }
4421                 return $oldVal;
4422         }
4423
4424         /**
4425          * Get all registered function hook identifiers
4426          *
4427          * @return Array
4428          */
4429         function getFunctionHooks() {
4430                 return array_keys( $this->mFunctionHooks );
4431         }
4432
4433         /**
4434          * Create a tag function, e.g. <test>some stuff</test>.
4435          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4436          * Unlike parser functions, their content is not preprocessed.
4437          */
4438         function setFunctionTagHook( $tag, $callback, $flags ) {
4439                 $tag = strtolower( $tag );
4440                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4441                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4442                         $this->mFunctionTagHooks[$tag] : null;
4443                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4444
4445                 if ( !in_array( $tag, $this->mStripList ) ) {
4446                         $this->mStripList[] = $tag;
4447                 }
4448
4449                 return $old;
4450         }
4451
4452         /**
4453          * FIXME: update documentation. makeLinkObj() is deprecated.
4454          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4455          * Placeholders created in Skin::makeLinkObj()
4456          * Returns an array of link CSS classes, indexed by PDBK.
4457          */
4458         function replaceLinkHolders( &$text, $options = 0 ) {
4459                 return $this->mLinkHolders->replace( $text );
4460         }
4461
4462         /**
4463          * Replace <!--LINK--> link placeholders with plain text of links
4464          * (not HTML-formatted).
4465          *
4466          * @param $text String
4467          * @return String
4468          */
4469         function replaceLinkHoldersText( $text ) {
4470                 return $this->mLinkHolders->replaceText( $text );
4471         }
4472
4473         /**
4474          * Renders an image gallery from a text with one line per image.
4475          * text labels may be given by using |-style alternative text. E.g.
4476          *   Image:one.jpg|The number "1"
4477          *   Image:tree.jpg|A tree
4478          * given as text will return the HTML of a gallery with two images,
4479          * labeled 'The number "1"' and
4480          * 'A tree'.
4481          */
4482         function renderImageGallery( $text, $params ) {
4483                 $ig = new ImageGallery();
4484                 $ig->setContextTitle( $this->mTitle );
4485                 $ig->setShowBytes( false );
4486                 $ig->setShowFilename( false );
4487                 $ig->setParser( $this );
4488                 $ig->setHideBadImages();
4489                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4490                 $ig->useSkin( $this->mOptions->getSkin( $this->mTitle ) );
4491                 $ig->mRevisionId = $this->mRevisionId;
4492
4493                 if ( isset( $params['showfilename'] ) ) {
4494                         $ig->setShowFilename( true );
4495                 } else {
4496                         $ig->setShowFilename( false );
4497                 }
4498                 if ( isset( $params['caption'] ) ) {
4499                         $caption = $params['caption'];
4500                         $caption = htmlspecialchars( $caption );
4501                         $caption = $this->replaceInternalLinks( $caption );
4502                         $ig->setCaptionHtml( $caption );
4503                 }
4504                 if ( isset( $params['perrow'] ) ) {
4505                         $ig->setPerRow( $params['perrow'] );
4506                 }
4507                 if ( isset( $params['widths'] ) ) {
4508                         $ig->setWidths( $params['widths'] );
4509                 }
4510                 if ( isset( $params['heights'] ) ) {
4511                         $ig->setHeights( $params['heights'] );
4512                 }
4513
4514                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4515
4516                 $lines = StringUtils::explode( "\n", $text );
4517                 foreach ( $lines as $line ) {
4518                         # match lines like these:
4519                         # Image:someimage.jpg|This is some image
4520                         $matches = array();
4521                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4522                         # Skip empty lines
4523                         if ( count( $matches ) == 0 ) {
4524                                 continue;
4525                         }
4526
4527                         if ( strpos( $matches[0], '%' ) !== false ) {
4528                                 $matches[1] = rawurldecode( $matches[1] );
4529                         }
4530                         $tp = Title::newFromText( $matches[1], NS_FILE );
4531                         $nt =& $tp;
4532                         if ( is_null( $nt ) ) {
4533                                 # Bogus title. Ignore these so we don't bomb out later.
4534                                 continue;
4535                         }
4536                         if ( isset( $matches[3] ) ) {
4537                                 $label = $matches[3];
4538                         } else {
4539                                 $label = '';
4540                         }
4541
4542                         $html = $this->recursiveTagParse( trim( $label ) );
4543
4544                         $ig->add( $nt, $html );
4545
4546                         # Only add real images (bug #5586)
4547                         if ( $nt->getNamespace() == NS_FILE ) {
4548                                 $this->mOutput->addImage( $nt->getDBkey() );
4549                         }
4550                 }
4551                 return $ig->toHTML();
4552         }
4553
4554         function getImageParams( $handler ) {
4555                 if ( $handler ) {
4556                         $handlerClass = get_class( $handler );
4557                 } else {
4558                         $handlerClass = '';
4559                 }
4560                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4561                         # Initialise static lists
4562                         static $internalParamNames = array(
4563                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4564                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4565                                         'bottom', 'text-bottom' ),
4566                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4567                                         'upright', 'border', 'link', 'alt' ),
4568                         );
4569                         static $internalParamMap;
4570                         if ( !$internalParamMap ) {
4571                                 $internalParamMap = array();
4572                                 foreach ( $internalParamNames as $type => $names ) {
4573                                         foreach ( $names as $name ) {
4574                                                 $magicName = str_replace( '-', '_', "img_$name" );
4575                                                 $internalParamMap[$magicName] = array( $type, $name );
4576                                         }
4577                                 }
4578                         }
4579
4580                         # Add handler params
4581                         $paramMap = $internalParamMap;
4582                         if ( $handler ) {
4583                                 $handlerParamMap = $handler->getParamMap();
4584                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4585                                         $paramMap[$magic] = array( 'handler', $paramName );
4586                                 }
4587                         }
4588                         $this->mImageParams[$handlerClass] = $paramMap;
4589                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4590                 }
4591                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4592         }
4593
4594         /**
4595          * Parse image options text and use it to make an image
4596          *
4597          * @param $title Title
4598          * @param $options String
4599          * @param $holders LinkHolderArray
4600          */
4601         function makeImage( $title, $options, $holders = false ) {
4602                 # Check if the options text is of the form "options|alt text"
4603                 # Options are:
4604                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4605                 #  * left       no resizing, just left align. label is used for alt= only
4606                 #  * right      same, but right aligned
4607                 #  * none       same, but not aligned
4608                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4609                 #  * center     center the image
4610                 #  * frame      Keep original image size, no magnify-button.
4611                 #  * framed     Same as "frame"
4612                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4613                 #  * upright    reduce width for upright images, rounded to full __0 px
4614                 #  * border     draw a 1px border around the image
4615                 #  * alt        Text for HTML alt attribute (defaults to empty)
4616                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4617                 # vertical-align values (no % or length right now):
4618                 #  * baseline
4619                 #  * sub
4620                 #  * super
4621                 #  * top
4622                 #  * text-top
4623                 #  * middle
4624                 #  * bottom
4625                 #  * text-bottom
4626
4627                 $parts = StringUtils::explode( "|", $options );
4628                 $sk = $this->mOptions->getSkin( $this->mTitle );
4629
4630                 # Give extensions a chance to select the file revision for us
4631                 $skip = $time = $descQuery = false;
4632                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4633
4634                 if ( $skip ) {
4635                         return $sk->link( $title );
4636                 }
4637
4638                 # Get the file
4639                 $file = wfFindFile( $title, array( 'time' => $time ) );
4640                 # Get parameter map
4641                 $handler = $file ? $file->getHandler() : false;
4642
4643                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4644
4645                 # Process the input parameters
4646                 $caption = '';
4647                 $params = array( 'frame' => array(), 'handler' => array(),
4648                         'horizAlign' => array(), 'vertAlign' => array() );
4649                 foreach ( $parts as $part ) {
4650                         $part = trim( $part );
4651                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4652                         $validated = false;
4653                         if ( isset( $paramMap[$magicName] ) ) {
4654                                 list( $type, $paramName ) = $paramMap[$magicName];
4655
4656                                 # Special case; width and height come in one variable together
4657                                 if ( $type === 'handler' && $paramName === 'width' ) {
4658                                         $m = array();
4659                                         # (bug 13500) In both cases (width/height and width only),
4660                                         # permit trailing "px" for backward compatibility.
4661                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4662                                                 $width = intval( $m[1] );
4663                                                 $height = intval( $m[2] );
4664                                                 if ( $handler->validateParam( 'width', $width ) ) {
4665                                                         $params[$type]['width'] = $width;
4666                                                         $validated = true;
4667                                                 }
4668                                                 if ( $handler->validateParam( 'height', $height ) ) {
4669                                                         $params[$type]['height'] = $height;
4670                                                         $validated = true;
4671                                                 }
4672                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4673                                                 $width = intval( $value );
4674                                                 if ( $handler->validateParam( 'width', $width ) ) {
4675                                                         $params[$type]['width'] = $width;
4676                                                         $validated = true;
4677                                                 }
4678                                         } # else no validation -- bug 13436
4679                                 } else {
4680                                         if ( $type === 'handler' ) {
4681                                                 # Validate handler parameter
4682                                                 $validated = $handler->validateParam( $paramName, $value );
4683                                         } else {
4684                                                 # Validate internal parameters
4685                                                 switch( $paramName ) {
4686                                                 case 'manualthumb':
4687                                                 case 'alt':
4688                                                         # @todo Fixme: possibly check validity here for
4689                                                         # manualthumb? downstream behavior seems odd with
4690                                                         # missing manual thumbs.
4691                                                         $validated = true;
4692                                                         $value = $this->stripAltText( $value, $holders );
4693                                                         break;
4694                                                 case 'link':
4695                                                         $chars = self::EXT_LINK_URL_CLASS;
4696                                                         $prots = $this->mUrlProtocols;
4697                                                         if ( $value === '' ) {
4698                                                                 $paramName = 'no-link';
4699                                                                 $value = true;
4700                                                                 $validated = true;
4701                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4702                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4703                                                                         $paramName = 'link-url';
4704                                                                         $this->mOutput->addExternalLink( $value );
4705                                                                         if ( $this->mOptions->getExternalLinkTarget() ) {
4706                                                                                 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
4707                                                                         }
4708                                                                         $validated = true;
4709                                                                 }
4710                                                         } else {
4711                                                                 $linkTitle = Title::newFromText( $value );
4712                                                                 if ( $linkTitle ) {
4713                                                                         $paramName = 'link-title';
4714                                                                         $value = $linkTitle;
4715                                                                         $this->mOutput->addLink( $linkTitle );
4716                                                                         $validated = true;
4717                                                                 }
4718                                                         }
4719                                                         break;
4720                                                 default:
4721                                                         # Most other things appear to be empty or numeric...
4722                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4723                                                 }
4724                                         }
4725
4726                                         if ( $validated ) {
4727                                                 $params[$type][$paramName] = $value;
4728                                         }
4729                                 }
4730                         }
4731                         if ( !$validated ) {
4732                                 $caption = $part;
4733                         }
4734                 }
4735
4736                 # Process alignment parameters
4737                 if ( $params['horizAlign'] ) {
4738                         $params['frame']['align'] = key( $params['horizAlign'] );
4739                 }
4740                 if ( $params['vertAlign'] ) {
4741                         $params['frame']['valign'] = key( $params['vertAlign'] );
4742                 }
4743
4744                 $params['frame']['caption'] = $caption;
4745
4746                 # Will the image be presented in a frame, with the caption below?
4747                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4748                                                  isset( $params['frame']['framed'] ) ||
4749                                                  isset( $params['frame']['thumbnail'] ) ||
4750                                                  isset( $params['frame']['manualthumb'] );
4751
4752                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4753                 # came to also set the caption, ordinary text after the image -- which
4754                 # makes no sense, because that just repeats the text multiple times in
4755                 # screen readers.  It *also* came to set the title attribute.
4756                 #
4757                 # Now that we have an alt attribute, we should not set the alt text to
4758                 # equal the caption: that's worse than useless, it just repeats the
4759                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4760                 # use the unnamed parameter for alt text as well, just for the time be-
4761                 # ing, if the unnamed param is set and the alt param is not.
4762                 #
4763                 # For the future, we need to figure out if we want to tweak this more,
4764                 # e.g., introducing a title= parameter for the title; ignoring the un-
4765                 # named parameter entirely for images without a caption; adding an ex-
4766                 # plicit caption= parameter and preserving the old magic unnamed para-
4767                 # meter for BC; ...
4768                 if ( $imageIsFramed ) { # Framed image
4769                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4770                                 # No caption or alt text, add the filename as the alt text so
4771                                 # that screen readers at least get some description of the image
4772                                 $params['frame']['alt'] = $title->getText();
4773                         }
4774                         # Do not set $params['frame']['title'] because tooltips don't make sense
4775                         # for framed images
4776                 } else { # Inline image
4777                         if ( !isset( $params['frame']['alt'] ) ) {
4778                                 # No alt text, use the "caption" for the alt text
4779                                 if ( $caption !== '') {
4780                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4781                                 } else {
4782                                         # No caption, fall back to using the filename for the
4783                                         # alt text
4784                                         $params['frame']['alt'] = $title->getText();
4785                                 }
4786                         }
4787                         # Use the "caption" for the tooltip text
4788                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4789                 }
4790
4791                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4792
4793                 # Linker does the rest
4794                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery, $this->mOptions->getThumbSize() );
4795
4796                 # Give the handler a chance to modify the parser object
4797                 if ( $handler ) {
4798                         $handler->parserTransformHook( $this, $file );
4799                 }
4800
4801                 return $ret;
4802         }
4803
4804         protected function stripAltText( $caption, $holders ) {
4805                 # Strip bad stuff out of the title (tooltip).  We can't just use
4806                 # replaceLinkHoldersText() here, because if this function is called
4807                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4808                 if ( $holders ) {
4809                         $tooltip = $holders->replaceText( $caption );
4810                 } else {
4811                         $tooltip = $this->replaceLinkHoldersText( $caption );
4812                 }
4813
4814                 # make sure there are no placeholders in thumbnail attributes
4815                 # that are later expanded to html- so expand them now and
4816                 # remove the tags
4817                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4818                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4819
4820                 return $tooltip;
4821         }
4822
4823         /**
4824          * Set a flag in the output object indicating that the content is dynamic and
4825          * shouldn't be cached.
4826          */
4827         function disableCache() {
4828                 wfDebug( "Parser output marked as uncacheable.\n" );
4829                 $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
4830                 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
4831         }
4832
4833         /**
4834          * Callback from the Sanitizer for expanding items found in HTML attribute
4835          * values, so they can be safely tested and escaped.
4836          *
4837          * @param $text String
4838          * @param $frame PPFrame
4839          * @return String
4840          * @private
4841          */
4842         function attributeStripCallback( &$text, $frame = false ) {
4843                 $text = $this->replaceVariables( $text, $frame );
4844                 $text = $this->mStripState->unstripBoth( $text );
4845                 return $text;
4846         }
4847
4848         /**
4849          * Accessor
4850          */
4851         function getTags() {
4852                 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ) );
4853         }
4854
4855         /**
4856          * Break wikitext input into sections, and either pull or replace
4857          * some particular section's text.
4858          *
4859          * External callers should use the getSection and replaceSection methods.
4860          *
4861          * @param $text String: Page wikitext
4862          * @param $section String: a section identifier string of the form:
4863          *   <flag1> - <flag2> - ... - <section number>
4864          *
4865          * Currently the only recognised flag is "T", which means the target section number
4866          * was derived during a template inclusion parse, in other words this is a template
4867          * section edit link. If no flags are given, it was an ordinary section edit link.
4868          * This flag is required to avoid a section numbering mismatch when a section is
4869          * enclosed by <includeonly> (bug 6563).
4870          *
4871          * The section number 0 pulls the text before the first heading; other numbers will
4872          * pull the given section along with its lower-level subsections. If the section is
4873          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4874          *
4875          * @param $mode String: one of "get" or "replace"
4876          * @param $newText String: replacement text for section data.
4877          * @return String: for "get", the extracted section text.
4878          *                 for "replace", the whole page with the section replaced.
4879          */
4880         private function extractSections( $text, $section, $mode, $newText='' ) {
4881                 global $wgTitle; # not generally used but removes an ugly failure mode
4882                 $this->startExternalParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
4883                 $outText = '';
4884                 $frame = $this->getPreprocessor()->newFrame();
4885
4886                 # Process section extraction flags
4887                 $flags = 0;
4888                 $sectionParts = explode( '-', $section );
4889                 $sectionIndex = array_pop( $sectionParts );
4890                 foreach ( $sectionParts as $part ) {
4891                         if ( $part === 'T' ) {
4892                                 $flags |= self::PTD_FOR_INCLUSION;
4893                         }
4894                 }
4895                 # Preprocess the text
4896                 $root = $this->preprocessToDom( $text, $flags );
4897
4898                 # <h> nodes indicate section breaks
4899                 # They can only occur at the top level, so we can find them by iterating the root's children
4900                 $node = $root->getFirstChild();
4901
4902                 # Find the target section
4903                 if ( $sectionIndex == 0 ) {
4904                         # Section zero doesn't nest, level=big
4905                         $targetLevel = 1000;
4906                 } else {
4907                         while ( $node ) {
4908                                 if ( $node->getName() === 'h' ) {
4909                                         $bits = $node->splitHeading();
4910                                         if ( $bits['i'] == $sectionIndex ) {
4911                                                 $targetLevel = $bits['level'];
4912                                                 break;
4913                                         }
4914                                 }
4915                                 if ( $mode === 'replace' ) {
4916                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4917                                 }
4918                                 $node = $node->getNextSibling();
4919                         }
4920                 }
4921
4922                 if ( !$node ) {
4923                         # Not found
4924                         if ( $mode === 'get' ) {
4925                                 return $newText;
4926                         } else {
4927                                 return $text;
4928                         }
4929                 }
4930
4931                 # Find the end of the section, including nested sections
4932                 do {
4933                         if ( $node->getName() === 'h' ) {
4934                                 $bits = $node->splitHeading();
4935                                 $curLevel = $bits['level'];
4936                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4937                                         break;
4938                                 }
4939                         }
4940                         if ( $mode === 'get' ) {
4941                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4942                         }
4943                         $node = $node->getNextSibling();
4944                 } while ( $node );
4945
4946                 # Write out the remainder (in replace mode only)
4947                 if ( $mode === 'replace' ) {
4948                         # Output the replacement text
4949                         # Add two newlines on -- trailing whitespace in $newText is conventionally
4950                         # stripped by the editor, so we need both newlines to restore the paragraph gap
4951                         # Only add trailing whitespace if there is newText
4952                         if ( $newText != "" ) {
4953                                 $outText .= $newText . "\n\n";
4954                         }
4955
4956                         while ( $node ) {
4957                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4958                                 $node = $node->getNextSibling();
4959                         }
4960                 }
4961
4962                 if ( is_string( $outText ) ) {
4963                         # Re-insert stripped tags
4964                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4965                 }
4966
4967                 return $outText;
4968         }
4969
4970         /**
4971          * This function returns the text of a section, specified by a number ($section).
4972          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4973          * the first section before any such heading (section 0).
4974          *
4975          * If a section contains subsections, these are also returned.
4976          *
4977          * @param $text String: text to look in
4978          * @param $section String: section identifier
4979          * @param $deftext String: default to return if section is not found
4980          * @return string text of the requested section
4981          */
4982         public function getSection( $text, $section, $deftext='' ) {
4983                 return $this->extractSections( $text, $section, "get", $deftext );
4984         }
4985
4986         /**
4987          * This function returns $oldtext after the content of the section
4988          * specified by $section has been replaced with $text.
4989          *
4990          * @param $text String: former text of the article
4991          * @param $section Numeric: section identifier
4992          * @param $text String: replacing text
4993          * #return String: modified text
4994          */
4995         public function replaceSection( $oldtext, $section, $text ) {
4996                 return $this->extractSections( $oldtext, $section, "replace", $text );
4997         }
4998
4999         /**
5000          * Get the ID of the revision we are parsing
5001          *
5002          * @return Mixed: integer or null
5003          */
5004         function getRevisionId() {
5005                 return $this->mRevisionId;
5006         }
5007
5008         /**
5009          * Get the revision object for $this->mRevisionId
5010          *
5011          * @return either a Revision object or null
5012          */
5013         protected function getRevisionObject() {
5014                 if ( !is_null( $this->mRevisionObject ) ) {
5015                         return $this->mRevisionObject;
5016                 }
5017                 if ( is_null( $this->mRevisionId ) ) {
5018                         return null;
5019                 }
5020
5021                 $this->mRevisionObject = Revision::newFromId( $this->mRevisionId );
5022                 return $this->mRevisionObject;
5023         }
5024
5025         /**
5026          * Get the timestamp associated with the current revision, adjusted for
5027          * the default server-local timestamp
5028          */
5029         function getRevisionTimestamp() {
5030                 if ( is_null( $this->mRevisionTimestamp ) ) {
5031                         wfProfileIn( __METHOD__ );
5032
5033                         $revObject = $this->getRevisionObject();
5034                         $timestamp = $revObject ? $revObject->getTimestamp() : false;
5035
5036                         if( $timestamp !== false ) {
5037                                 global $wgContLang;
5038
5039                                 # The cryptic '' timezone parameter tells to use the site-default
5040                                 # timezone offset instead of the user settings.
5041                                 #
5042                                 # Since this value will be saved into the parser cache, served
5043                                 # to other users, and potentially even used inside links and such,
5044                                 # it needs to be consistent for all visitors.
5045                                 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5046                         }
5047
5048                         wfProfileOut( __METHOD__ );
5049                 }
5050                 return $this->mRevisionTimestamp;
5051         }
5052
5053         /**
5054          * Get the name of the user that edited the last revision
5055          *
5056          * @return String: user name
5057          */
5058         function getRevisionUser() {
5059                 if( is_null( $this->mRevisionUser ) ) {
5060                         $revObject = $this->getRevisionObject();
5061
5062                         # if this template is subst: the revision id will be blank,
5063                         # so just use the current user's name
5064                         if( $revObject ) {
5065                                 $this->mRevisionUser = $revObject->getUserText();
5066                         } elseif( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5067                                 $this->mRevisionUser = $this->getUser()->getName();
5068                         }
5069                 }
5070                 return $this->mRevisionUser;
5071         }
5072
5073         /**
5074          * Mutator for $mDefaultSort
5075          *
5076          * @param $sort New value
5077          */
5078         public function setDefaultSort( $sort ) {
5079                 $this->mDefaultSort = $sort;
5080                 $this->mOutput->setProperty( 'defaultsort', $sort );
5081         }
5082
5083         /**
5084          * Accessor for $mDefaultSort
5085          * Will use the empty string if none is set.
5086          *
5087          * This value is treated as a prefix, so the
5088          * empty string is equivalent to sorting by
5089          * page name.
5090          *
5091          * @return string
5092          */
5093         public function getDefaultSort() {
5094                 if ( $this->mDefaultSort !== false ) {
5095                         return $this->mDefaultSort;
5096                 } else {
5097                         return '';
5098                 }
5099         }
5100
5101         /**
5102          * Accessor for $mDefaultSort
5103          * Unlike getDefaultSort(), will return false if none is set
5104          *
5105          * @return string or false
5106          */
5107         public function getCustomDefaultSort() {
5108                 return $this->mDefaultSort;
5109         }
5110
5111         /**
5112          * Try to guess the section anchor name based on a wikitext fragment
5113          * presumably extracted from a heading, for example "Header" from
5114          * "== Header ==".
5115          */
5116         public function guessSectionNameFromWikiText( $text ) {
5117                 # Strip out wikitext links(they break the anchor)
5118                 $text = $this->stripSectionName( $text );
5119                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5120                 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5121         }
5122
5123         /**
5124          * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5125          * instead.  For use in redirects, since IE6 interprets Redirect: headers
5126          * as something other than UTF-8 (apparently?), resulting in breakage.
5127          *
5128          * @param $text String: The section name
5129          * @return string An anchor
5130          */
5131         public function guessLegacySectionNameFromWikiText( $text ) {
5132                 # Strip out wikitext links(they break the anchor)
5133                 $text = $this->stripSectionName( $text );
5134                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5135                 return '#' . Sanitizer::escapeId( $text, array( 'noninitial', 'legacy' ) );
5136         }
5137
5138         /**
5139          * Strips a text string of wikitext for use in a section anchor
5140          *
5141          * Accepts a text string and then removes all wikitext from the
5142          * string and leaves only the resultant text (i.e. the result of
5143          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5144          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5145          * to create valid section anchors by mimicing the output of the
5146          * parser when headings are parsed.
5147          *
5148          * @param $text String: text string to be stripped of wikitext
5149          * for use in a Section anchor
5150          * @return Filtered text string
5151          */
5152         public function stripSectionName( $text ) {
5153                 # Strip internal link markup
5154                 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5155                 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5156
5157                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5158                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5159                 # on how many empty links there are on the page - need to figure that out.
5160                 $text = preg_replace( '/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5161
5162                 # Parse wikitext quotes (italics & bold)
5163                 $text = $this->doQuotes( $text );
5164
5165                 # Strip HTML tags
5166                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5167                 return $text;
5168         }
5169
5170         /**
5171          * strip/replaceVariables/unstrip for preprocessor regression testing
5172          */
5173         function testSrvus( $text, $title, ParserOptions $options, $outputType = self::OT_HTML ) {
5174                 if ( !$title instanceof Title ) {
5175                         $title = Title::newFromText( $title );
5176                 }
5177                 $this->startExternalParse( $title, $options, $outputType, true );
5178
5179                 $text = $this->replaceVariables( $text );
5180                 $text = $this->mStripState->unstripBoth( $text );
5181                 $text = Sanitizer::removeHTMLtags( $text );
5182                 return $text;
5183         }
5184
5185         function testPst( $text, $title, $options ) {
5186                 global $wgUser;
5187                 if ( !$title instanceof Title ) {
5188                         $title = Title::newFromText( $title );
5189                 }
5190                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5191         }
5192
5193         function testPreprocess( $text, $title, $options ) {
5194                 if ( !$title instanceof Title ) {
5195                         $title = Title::newFromText( $title );
5196                 }
5197                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5198         }
5199
5200         function markerSkipCallback( $s, $callback ) {
5201                 $i = 0;
5202                 $out = '';
5203                 while ( $i < strlen( $s ) ) {
5204                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5205                         if ( $markerStart === false ) {
5206                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5207                                 break;
5208                         } else {
5209                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5210                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5211                                 if ( $markerEnd === false ) {
5212                                         $out .= substr( $s, $markerStart );
5213                                         break;
5214                                 } else {
5215                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5216                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5217                                         $i = $markerEnd;
5218                                 }
5219                         }
5220                 }
5221                 return $out;
5222         }
5223
5224         function serialiseHalfParsedText( $text ) {
5225                 $data = array();
5226                 $data['text'] = $text;
5227
5228                 # First, find all strip markers, and store their
5229                 #  data in an array.
5230                 $stripState = new StripState;
5231                 $pos = 0;
5232                 while ( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) )
5233                         && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) )
5234                 {
5235                         $end_pos += strlen( self::MARKER_SUFFIX );
5236                         $marker = substr( $text, $start_pos, $end_pos-$start_pos );
5237
5238                         if ( !empty( $this->mStripState->general->data[$marker] ) ) {
5239                                 $replaceArray = $stripState->general;
5240                                 $stripText = $this->mStripState->general->data[$marker];
5241                         } elseif ( !empty( $this->mStripState->nowiki->data[$marker] ) ) {
5242                                 $replaceArray = $stripState->nowiki;
5243                                 $stripText = $this->mStripState->nowiki->data[$marker];
5244                         } else {
5245                                 throw new MWException( "Hanging strip marker: '$marker'." );
5246                         }
5247
5248                         $replaceArray->setPair( $marker, $stripText );
5249                         $pos = $end_pos;
5250                 }
5251                 $data['stripstate'] = $stripState;
5252
5253                 # Now, find all of our links, and store THEIR
5254                 #  data in an array! :)
5255                 $links = array( 'internal' => array(), 'interwiki' => array() );
5256                 $pos = 0;
5257
5258                 # Internal links
5259                 while ( ( $start_pos = strpos( $text, '<!--LINK ', $pos ) ) ) {
5260                         list( $ns, $trail ) = explode( ':', substr( $text, $start_pos + strlen( '<!--LINK ' ) ), 2 );
5261
5262                         $ns = trim( $ns );
5263                         if ( empty( $links['internal'][$ns] ) ) {
5264                                 $links['internal'][$ns] = array();
5265                         }
5266
5267                         $key = trim( substr( $trail, 0, strpos( $trail, '-->' ) ) );
5268                         $links['internal'][$ns][] = $this->mLinkHolders->internals[$ns][$key];
5269                         $pos = $start_pos + strlen( "<!--LINK $ns:$key-->" );
5270                 }
5271
5272                 $pos = 0;
5273
5274                 # Interwiki links
5275                 while ( ( $start_pos = strpos( $text, '<!--IWLINK ', $pos ) ) ) {
5276                         $data = substr( $text, $start_pos );
5277                         $key = trim( substr( $data, 0, strpos( $data, '-->' ) ) );
5278                         $links['interwiki'][] = $this->mLinkHolders->interwiki[$key];
5279                         $pos = $start_pos + strlen( "<!--IWLINK $key-->" );
5280                 }
5281
5282                 $data['linkholder'] = $links;
5283
5284                 return $data;
5285         }
5286
5287         /**
5288          * TODO: document
5289          * @param $data Array
5290          * @param $intPrefix String unique identifying prefix
5291          * @return String
5292          */
5293         function unserialiseHalfParsedText( $data, $intPrefix = null ) {
5294                 if ( !$intPrefix ) {
5295                         $intPrefix = self::getRandomString();
5296                 }
5297
5298                 # First, extract the strip state.
5299                 $stripState = $data['stripstate'];
5300                 $this->mStripState->general->merge( $stripState->general );
5301                 $this->mStripState->nowiki->merge( $stripState->nowiki );
5302
5303                 # Now, extract the text, and renumber links
5304                 $text = $data['text'];
5305                 $links = $data['linkholder'];
5306
5307                 # Internal...
5308                 foreach ( $links['internal'] as $ns => $nsLinks ) {
5309                         foreach ( $nsLinks as $key => $entry ) {
5310                                 $newKey = $intPrefix . '-' . $key;
5311                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
5312
5313                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
5314                         }
5315                 }
5316
5317                 # Interwiki...
5318                 foreach ( $links['interwiki'] as $key => $entry ) {
5319                         $newKey = "$intPrefix-$key";
5320                         $this->mLinkHolders->interwikis[$newKey] = $entry;
5321
5322                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
5323                 }
5324
5325                 # Should be good to go.
5326                 return $text;
5327         }
5328 }
5329
5330 /**
5331  * @todo document, briefly.
5332  * @ingroup Parser
5333  */
5334 class StripState {
5335         var $general, $nowiki;
5336
5337         function __construct() {
5338                 $this->general = new ReplacementArray;
5339                 $this->nowiki = new ReplacementArray;
5340         }
5341
5342         function unstripGeneral( $text ) {
5343                 wfProfileIn( __METHOD__ );
5344                 do {
5345                         $oldText = $text;
5346                         $text = $this->general->replace( $text );
5347                 } while ( $text !== $oldText );
5348                 wfProfileOut( __METHOD__ );
5349                 return $text;
5350         }
5351
5352         function unstripNoWiki( $text ) {
5353                 wfProfileIn( __METHOD__ );
5354                 do {
5355                         $oldText = $text;
5356                         $text = $this->nowiki->replace( $text );
5357                 } while ( $text !== $oldText );
5358                 wfProfileOut( __METHOD__ );
5359                 return $text;
5360         }
5361
5362         function unstripBoth( $text ) {
5363                 wfProfileIn( __METHOD__ );
5364                 do {
5365                         $oldText = $text;
5366                         $text = $this->general->replace( $text );
5367                         $text = $this->nowiki->replace( $text );
5368                 } while ( $text !== $oldText );
5369                 wfProfileOut( __METHOD__ );
5370                 return $text;
5371         }
5372 }
5373
5374 /**
5375  * @todo document, briefly.
5376  * @ingroup Parser
5377  */
5378 class OnlyIncludeReplacer {
5379         var $output = '';
5380
5381         function replace( $matches ) {
5382                 if ( substr( $matches[1], -1 ) === "\n" ) {
5383                         $this->output .= substr( $matches[1], 0, -1 );
5384                 } else {
5385                         $this->output .= $matches[1];
5386                 }
5387         }
5388 }