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 $wgUser or $wgTitle. Keep them away!
  38  *
  39  * settings:
  40  *  $wgUseDynamicDates*, $wgInterwikiMagic*,
  41  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  42  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  43  *  $wgMaxArticleSize*
  44  *
  45  *  * only within ParserOptions
  46  * </pre>
  47  *
  48  * @ingroup Parser
  49  */
  50 class Parser {
  51         /**
  52          * Update this version number when the ParserOutput format
  53          * changes in an incompatible way, so the parser cache
  54          * can automatically discard old data.
  55          */
  56         const VERSION = '1.6.4';
  57
  58         /**
  59          * Update this version number when the output of serialiseHalfParsedText()
  60          * changes in an incompatible way
  61          */
  62         const HALF_PARSED_VERSION = 2;
  63
  64         # Flags for Parser::setFunctionHook
  65         # Also available as global constants from Defines.php
  66         const SFH_NO_HASH = 1;
  67         const SFH_OBJECT_ARGS = 2;
  68
  69         # Constants needed for external link processing
  70         # Everything except bracket, space, or control characters
  71         # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
  72         # as well as U+3000 is IDEOGRAPHIC SPACE for bug 19052
  73         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}]';
  74         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F\p{Zs}]+)
  75                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
  76
  77         # State constants for the definition list colon extraction
  78         const COLON_STATE_TEXT = 0;
  79         const COLON_STATE_TAG = 1;
  80         const COLON_STATE_TAGSTART = 2;
  81         const COLON_STATE_CLOSETAG = 3;
  82         const COLON_STATE_TAGSLASH = 4;
  83         const COLON_STATE_COMMENT = 5;
  84         const COLON_STATE_COMMENTDASH = 6;
  85         const COLON_STATE_COMMENTDASHDASH = 7;
  86
  87         # Flags for preprocessToDom
  88         const PTD_FOR_INCLUSION = 1;
  89
  90         # Allowed values for $this->mOutputType
  91         # Parameter to startExternalParse().
  92         const OT_HTML = 1; # like parse()
  93         const OT_WIKI = 2; # like preSaveTransform()
  94         const OT_PREPROCESS = 3; # like preprocess()
  95         const OT_MSG = 3;
  96         const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
  97
  98         # Marker Suffix needs to be accessible staticly.
  99         const MARKER_SUFFIX = "-QINU\x7f";
 100
 101         # Persistent:
 102         var $mTagHooks = array();
 103         var $mTransparentTagHooks = array();
 104         var $mFunctionHooks = array();
 105         var $mFunctionSynonyms = array( 0 => array(), 1 => array() );
 106         var $mFunctionTagHooks = array();
 107         var $mStripList  = array();
 108         var $mDefaultStripList  = array();
 109         var $mVarCache = array();
 110         var $mImageParams = array();
 111         var $mImageParamsMagicArray = array();
 112         var $mMarkerIndex = 0;
 113         var $mFirstCall = true;
 114
 115         # Initialised by initialiseVariables()
 116
 117         /**
 118          * @var MagicWordArray
 119          */
 120         var $mVariables;
 121
 122         /**
 123          * @var MagicWordArray
 124          */
 125         var $mSubstWords;
 126         var $mConf, $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols; # Initialised in constructor
 127
 128         # Cleared with clearState():
 129         /**
 130          * @var ParserOutput
 131          */
 132         var $mOutput;
 133         var $mAutonumber, $mDTopen;
 134
 135         /**
 136          * @var StripState
 137          */
 138         var $mStripState;
 139
 140         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 141         /**
 142          * @var LinkHolderArray
 143          */
 144         var $mLinkHolders;
 145
 146         var $mLinkID;
 147         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 148         var $mTplExpandCache; # empty-frame expansion cache
 149         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 150         var $mExpensiveFunctionCount; # number of expensive parser function calls
 151         var $mShowToc, $mForceTocPosition;
 152
 153         /**
 154          * @var User
 155          */
 156         var $mUser; # User object; only used when doing pre-save transform
 157
 158         # Temporary
 159         # These are variables reset at least once per parse regardless of $clearState
 160
 161         /**
 162          * @var ParserOptions
 163          */
 164         var $mOptions;
 165
 166         /**
 167          * @var Title
 168          */
 169         var $mTitle;        # Title context, used for self-link rendering and similar things
 170         var $mOutputType;   # Output type, one of the OT_xxx constants
 171         var $ot;            # Shortcut alias, see setOutputType()
 172         var $mRevisionObject; # The revision object of the specified revision ID
 173         var $mRevisionId;   # ID to display in {{REVISIONID}} tags
 174         var $mRevisionTimestamp; # The timestamp of the specified revision ID
 175         var $mRevisionUser; # User to display in {{REVISIONUSER}} tag
 176         var $mRevIdForTs;   # The revision ID which was used to fetch the timestamp
 177
 178         /**
 179          * @var string
 180          */
 181         var $mUniqPrefix;
 182
 183         /**
 184          * Constructor
 185          *
 186          * @param $conf array
 187          */
 188         public function __construct( $conf = array() ) {
 189                 $this->mConf = $conf;
 190                 $this->mUrlProtocols = wfUrlProtocols();
 191                 $this->mExtLinkBracketedRegex = '/\[((' . wfUrlProtocols() . ')'.
 192                         self::EXT_LINK_URL_CLASS.'+)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/Su';
 193                 if ( isset( $conf['preprocessorClass'] ) ) {
 194                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 195                 } elseif ( defined( 'MW_COMPILED' ) ) {
 196                         # Preprocessor_Hash is much faster than Preprocessor_DOM in compiled mode
 197                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 198                 } elseif ( extension_loaded( 'domxml' ) ) {
 199                         # PECL extension that conflicts with the core DOM extension (bug 13770)
 200                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 201                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 202                 } elseif ( extension_loaded( 'dom' ) ) {
 203                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 204                 } else {
 205                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 206                 }
 207                 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
 208         }
 209
 210         /**
 211          * Reduce memory usage to reduce the impact of circular references
 212          */
 213         function __destruct() {
 214                 if ( isset( $this->mLinkHolders ) ) {
 215                         unset( $this->mLinkHolders );
 216                 }
 217                 foreach ( $this as $name => $value ) {
 218                         unset( $this->$name );
 219                 }
 220         }
 221
 222         /**
 223          * Do various kinds of initialisation on the first call of the parser
 224          */
 225         function firstCallInit() {
 226                 if ( !$this->mFirstCall ) {
 227                         return;
 228                 }
 229                 $this->mFirstCall = false;
 230
 231                 wfProfileIn( __METHOD__ );
 232
 233                 CoreParserFunctions::register( $this );
 234                 CoreTagHooks::register( $this );
 235                 $this->initialiseVariables();
 236
 237                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 238                 wfProfileOut( __METHOD__ );
 239         }
 240
 241         /**
 242          * Clear Parser state
 243          *
 244          * @private
 245          */
 246         function clearState() {
 247                 wfProfileIn( __METHOD__ );
 248                 if ( $this->mFirstCall ) {
 249                         $this->firstCallInit();
 250                 }
 251                 $this->mOutput = new ParserOutput;
 252                 $this->mOptions->registerWatcher( array( $this->mOutput, 'recordOption' ) );
 253                 $this->mAutonumber = 0;
 254                 $this->mLastSection = '';
 255                 $this->mDTopen = false;
 256                 $this->mIncludeCount = array();
 257                 $this->mArgStack = false;
 258                 $this->mInPre = false;
 259                 $this->mLinkHolders = new LinkHolderArray( $this );
 260                 $this->mLinkID = 0;
 261                 $this->mRevisionObject = $this->mRevisionTimestamp =
 262                         $this->mRevisionId = $this->mRevisionUser = null;
 263                 $this->mVarCache = array();
 264                 $this->mUser = null;
 265
 266                 /**
 267                  * Prefix for temporary replacement strings for the multipass parser.
 268                  * \x07 should never appear in input as it's disallowed in XML.
 269                  * Using it at the front also gives us a little extra robustness
 270                  * since it shouldn't match when butted up against identifier-like
 271                  * string constructs.
 272                  *
 273                  * Must not consist of all title characters, or else it will change
 274                  * the behaviour of <nowiki> in a link.
 275                  */
 276                 # $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 277                 # Changed to \x7f to allow XML double-parsing -- TS
 278                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 279                 $this->mStripState = new StripState( $this->mUniqPrefix );
 280
 281
 282                 # Clear these on every parse, bug 4549
 283                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 284
 285                 $this->mShowToc = true;
 286                 $this->mForceTocPosition = false;
 287                 $this->mIncludeSizes = array(
 288                         'post-expand' => 0,
 289                         'arg' => 0,
 290                 );
 291                 $this->mPPNodeCount = 0;
 292                 $this->mDefaultSort = false;
 293                 $this->mHeadings = array();
 294                 $this->mDoubleUnderscores = array();
 295                 $this->mExpensiveFunctionCount = 0;
 296
 297                 # Fix cloning
 298                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 299                         $this->mPreprocessor = null;
 300                 }
 301
 302                 wfRunHooks( 'ParserClearState', array( &$this ) );
 303                 wfProfileOut( __METHOD__ );
 304         }
 305
 306         /**
 307          * Convert wikitext to HTML
 308          * Do not call this function recursively.
 309          *
 310          * @param $text String: text we want to parse
 311          * @param $title Title object
 312          * @param $options ParserOptions
 313          * @param $linestart boolean
 314          * @param $clearState boolean
 315          * @param $revid Int: number to pass in {{REVISIONID}}
 316          * @return ParserOutput a ParserOutput
 317          */
 318         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 319                 /**
 320                  * First pass--just handle <nowiki> sections, pass the rest off
 321                  * to internalParse() which does all the real work.
 322                  */
 323
 324                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang, $wgDisableLangConversion, $wgDisableTitleConversion;
 325                 $fname = __METHOD__.'-' . wfGetCaller();
 326                 wfProfileIn( __METHOD__ );
 327                 wfProfileIn( $fname );
 328
 329                 $this->startParse( $title, $options, self::OT_HTML, $clearState );
 330
 331                 $oldRevisionId = $this->mRevisionId;
 332                 $oldRevisionObject = $this->mRevisionObject;
 333                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 334                 $oldRevisionUser = $this->mRevisionUser;
 335                 if ( $revid !== null ) {
 336                         $this->mRevisionId = $revid;
 337                         $this->mRevisionObject = null;
 338                         $this->mRevisionTimestamp = null;
 339                         $this->mRevisionUser = null;
 340                 }
 341
 342                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 343                 # No more strip!
 344                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 345                 $text = $this->internalParse( $text );
 346
 347                 $text = $this->mStripState->unstripGeneral( $text );
 348
 349                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 350                 $fixtags = array(
 351                         # french spaces, last one Guillemet-left
 352                         # only if there is something before the space
 353                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;',
 354                         # french spaces, Guillemet-right
 355                         '/(\\302\\253) /' => '\\1&#160;',
 356                         '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
 357                 );
 358                 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
 359
 360                 $text = $this->doBlockLevels( $text, $linestart );
 361
 362                 $this->replaceLinkHolders( $text );
 363
 364                 /**
 365                  * The input doesn't get language converted if
 366                  * a) It's disabled
 367                  * b) Content isn't converted
 368                  * c) It's a conversion table
 369                  * d) it is an interface message (which is in the user language)
 370                  */
 371                 if ( !( $wgDisableLangConversion
 372                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 373                                 || $this->mTitle->isConversionTable()
 374                                 || $this->mOptions->getInterfaceMessage() ) ) {
 375
 376                         # The position of the convert() call should not be changed. it
 377                         # assumes that the links are all replaced and the only thing left
 378                         # is the <nowiki> mark.
 379
 380                         $text = $wgContLang->convert( $text );
 381                 }
 382
 383                 /**
 384                  * A converted title will be provided in the output object if title and
 385                  * content conversion are enabled, the article text does not contain
 386                  * a conversion-suppressing double-underscore tag, and no
 387                  * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
 388                  * automatic link conversion.
 389                  */
 390                 if ( !( $wgDisableLangConversion
 391                                 || $wgDisableTitleConversion
 392                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 393                                 || isset( $this->mDoubleUnderscores['notitleconvert'] )
 394                                 || $this->mOutput->getDisplayTitle() !== false ) )
 395                 {
 396                         $convruletitle = $wgContLang->getConvRuleTitle();
 397                         if ( $convruletitle ) {
 398                                 $this->mOutput->setTitleText( $convruletitle );
 399                         } else {
 400                                 $titleText = $wgContLang->convertTitle( $title );
 401                                 $this->mOutput->setTitleText( $titleText );
 402                         }
 403                 }
 404
 405                 $text = $this->mStripState->unstripNoWiki( $text );
 406
 407                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 408
 409                 $text = $this->replaceTransparentTags( $text );
 410                 $text = $this->mStripState->unstripGeneral( $text );
 411
 412                 $text = Sanitizer::normalizeCharReferences( $text );
 413
 414                 if ( ( $wgUseTidy && $this->mOptions->getTidy() ) || $wgAlwaysUseTidy ) {
 415                         $text = MWTidy::tidy( $text );
 416                 } else {
 417                         # attempt to sanitize at least some nesting problems
 418                         # (bug #2702 and quite a few others)
 419                         $tidyregs = array(
 420                                 # ''Something [http://www.cool.com cool''] -->
 421                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 422                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 423                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 424                                 # fix up an anchor inside another anchor, only
 425                                 # at least for a single single nested link (bug 3695)
 426                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 427                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 428                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 429                                 # contains a div, so fix it up here; replace
 430                                 # div with escaped text
 431                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 432                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 433                                 # remove empty italic or bold tag pairs, some
 434                                 # introduced by rules above
 435                                 '/<([bi])><\/\\1>/' => '',
 436                         );
 437
 438                         $text = preg_replace(
 439                                 array_keys( $tidyregs ),
 440                                 array_values( $tidyregs ),
 441                                 $text );
 442                 }
 443                 global $wgExpensiveParserFunctionLimit;
 444                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 445                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 446                 }
 447
 448                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 449
 450                 # Information on include size limits, for the benefit of users who try to skirt them
 451                 if ( $this->mOptions->getEnableLimitReport() ) {
 452                         $max = $this->mOptions->getMaxIncludeSize();
 453                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 454                         $limitReport =
 455                                 "NewPP limit report\n" .
 456                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->getMaxPPNodeCount()}\n" .
 457                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 458                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 459                                 $PFreport;
 460                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 461                         $text .= "\n<!-- \n$limitReport-->\n";
 462                 }
 463                 $this->mOutput->setText( $text );
 464
 465                 $this->mRevisionId = $oldRevisionId;
 466                 $this->mRevisionObject = $oldRevisionObject;
 467                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 468                 $this->mRevisionUser = $oldRevisionUser;
 469                 wfProfileOut( $fname );
 470                 wfProfileOut( __METHOD__ );
 471
 472                 return $this->mOutput;
 473         }
 474
 475         /**
 476          * Recursive parser entry point that can be called from an extension tag
 477          * hook.
 478          *
 479          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 480          *
 481          * @param $text String: text extension wants to have parsed
 482          * @param $frame PPFrame: The frame to use for expanding any template variables
 483          *
 484          * @return string
 485          */
 486         function recursiveTagParse( $text, $frame=false ) {
 487                 wfProfileIn( __METHOD__ );
 488                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 489                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 490                 $text = $this->internalParse( $text, false, $frame );
 491                 wfProfileOut( __METHOD__ );
 492                 return $text;
 493         }
 494
 495         /**
 496          * Expand templates and variables in the text, producing valid, static wikitext.
 497          * Also removes comments.
 498          */
 499         function preprocess( $text, Title $title, ParserOptions $options, $revid = null ) {
 500                 wfProfileIn( __METHOD__ );
 501                 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
 502                 if ( $revid !== null ) {
 503                         $this->mRevisionId = $revid;
 504                 }
 505                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 506                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 507                 $text = $this->replaceVariables( $text );
 508                 $text = $this->mStripState->unstripBoth( $text );
 509                 wfProfileOut( __METHOD__ );
 510                 return $text;
 511         }
 512
 513         /**
 514          * Recursive parser entry point that can be called from an extension tag
 515          * hook.
 516          *
 517          * @param $text String: text to be expanded
 518          * @param $frame PPFrame: The frame to use for expanding any template variables
 519          * @return String
 520          */
 521         public function recursivePreprocess( $text, $frame = false ) {
 522                 wfProfileIn( __METHOD__ );
 523                 $text = $this->replaceVariables( $text, $frame );
 524                 $text = $this->mStripState->unstripBoth( $text );
 525                 wfProfileOut( __METHOD__ );
 526                 return $text;
 527         }
 528
 529         /**
 530          * Process the wikitext for the ?preload= feature. (bug 5210)
 531          *
 532          * <noinclude>, <includeonly> etc. are parsed as for template transclusion,
 533          * comments, templates, arguments, tags hooks and parser functions are untouched.
 534          *
 535          * @param $text String
 536          * @param $title Title
 537          * @param $options ParserOptions
 538          * @return String
 539          */
 540         public function getPreloadText( $text, Title $title, ParserOptions $options ) {
 541                 # Parser (re)initialisation
 542                 $this->startParse( $title, $options, self::OT_PLAIN, true );
 543
 544                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 545                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
 546                 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
 547                 $text = $this->mStripState->unstripBoth( $text );
 548                 return $text;
 549         }
 550
 551         /**
 552          * Get a random string
 553          *
 554          * @return string
 555          */
 556         static public function getRandomString() {
 557                 return dechex( mt_rand( 0, 0x7fffffff ) ) . dechex( mt_rand( 0, 0x7fffffff ) );
 558         }
 559
 560         /**
 561          * Set the current user.
 562          * Should only be used when doing pre-save transform.
 563          *
 564          * @param $user Mixed: User object or null (to reset)
 565          */
 566         function setUser( $user ) {
 567                 $this->mUser = $user;
 568         }
 569
 570         /**
 571          * Accessor for mUniqPrefix.
 572          *
 573          * @return String
 574          */
 575         public function uniqPrefix() {
 576                 if ( !isset( $this->mUniqPrefix ) ) {
 577                         # @todo FIXME: This is probably *horribly wrong*
 578                         # LanguageConverter seems to want $wgParser's uniqPrefix, however
 579                         # if this is called for a parser cache hit, the parser may not
 580                         # have ever been initialized in the first place.
 581                         # Not really sure what the heck is supposed to be going on here.
 582                         return '';
 583                         # throw new MWException( "Accessing uninitialized mUniqPrefix" );
 584                 }
 585                 return $this->mUniqPrefix;
 586         }
 587
 588         /**
 589          * Set the context title
 590          *
 591          * @param $t Title
 592          */
 593         function setTitle( $t ) {
 594                 if ( !$t || $t instanceof FakeTitle ) {
 595                         $t = Title::newFromText( 'NO TITLE' );
 596                 }
 597
 598                 if ( strval( $t->getFragment() ) !== '' ) {
 599                         # Strip the fragment to avoid various odd effects
 600                         $this->mTitle = clone $t;
 601                         $this->mTitle->setFragment( '' );
 602                 } else {
 603                         $this->mTitle = $t;
 604                 }
 605         }
 606
 607         /**
 608          * Accessor for the Title object
 609          *
 610          * @return Title object
 611          */
 612         function getTitle() {
 613                 return $this->mTitle;
 614         }
 615
 616         /**
 617          * Accessor/mutator for the Title object
 618          *
 619          * @param $x New Title object or null to just get the current one
 620          * @return Title object
 621          */
 622         function Title( $x = null ) {
 623                 return wfSetVar( $this->mTitle, $x );
 624         }
 625
 626         /**
 627          * Set the output type
 628          *
 629          * @param $ot Integer: new value
 630          */
 631         function setOutputType( $ot ) {
 632                 $this->mOutputType = $ot;
 633                 # Shortcut alias
 634                 $this->ot = array(
 635                         'html' => $ot == self::OT_HTML,
 636                         'wiki' => $ot == self::OT_WIKI,
 637                         'pre' => $ot == self::OT_PREPROCESS,
 638                         'plain' => $ot == self::OT_PLAIN,
 639                 );
 640         }
 641
 642         /**
 643          * Accessor/mutator for the output type
 644          *
 645          * @param $x New value or null to just get the current one
 646          * @return Integer
 647          */
 648         function OutputType( $x = null ) {
 649                 return wfSetVar( $this->mOutputType, $x );
 650         }
 651
 652         /**
 653          * Get the ParserOutput object
 654          *
 655          * @return ParserOutput object
 656          */
 657         function getOutput() {
 658                 return $this->mOutput;
 659         }
 660
 661         /**
 662          * Get the ParserOptions object
 663          *
 664          * @return ParserOptions object
 665          */
 666         function getOptions() {
 667                 return $this->mOptions;
 668         }
 669
 670         /**
 671          * Accessor/mutator for the ParserOptions object
 672          *
 673          * @param $x New value or null to just get the current one
 674          * @return Current ParserOptions object
 675          */
 676         function Options( $x = null ) {
 677                 return wfSetVar( $this->mOptions, $x );
 678         }
 679
 680         /**
 681          * @return int
 682          */
 683         function nextLinkID() {
 684                 return $this->mLinkID++;
 685         }
 686
 687         /**
 688          * @param $id int
 689          */
 690         function setLinkID( $id ) {
 691                 $this->mLinkID = $id;
 692         }
 693
 694         /**
 695          * @return Language
 696          */
 697         function getFunctionLang() {
 698                 $target = $this->mOptions->getTargetLanguage();
 699                 if ( $target !== null ) {
 700                         return $target;
 701                 } elseif( $this->mOptions->getInterfaceMessage() ) {
 702                         global $wgLang;
 703                         return $wgLang;
 704                 } elseif( is_null( $this->mTitle ) ) {
 705                         throw new MWException( __METHOD__.': $this->mTitle is null' );
 706                 }
 707                 return $this->mTitle->getPageLanguage();
 708         }
 709
 710         /**
 711          * Get a User object either from $this->mUser, if set, or from the
 712          * ParserOptions object otherwise
 713          *
 714          * @return User object
 715          */
 716         function getUser() {
 717                 if ( !is_null( $this->mUser ) ) {
 718                         return $this->mUser;
 719                 }
 720                 return $this->mOptions->getUser();
 721         }
 722
 723         /**
 724          * Get a preprocessor object
 725          *
 726          * @return Preprocessor instance
 727          */
 728         function getPreprocessor() {
 729                 if ( !isset( $this->mPreprocessor ) ) {
 730                         $class = $this->mPreprocessorClass;
 731                         $this->mPreprocessor = new $class( $this );
 732                 }
 733                 return $this->mPreprocessor;
 734         }
 735
 736         /**
 737          * Replaces all occurrences of HTML-style comments and the given tags
 738          * in the text with a random marker and returns the next text. The output
 739          * parameter $matches will be an associative array filled with data in
 740          * the form:
 741          *   'UNIQ-xxxxx' => array(
 742          *     'element',
 743          *     'tag content',
 744          *     array( 'param' => 'x' ),
 745          *     '<element param="x">tag content</element>' ) )
 746          *
 747          * @param $elements array list of element names. Comments are always extracted.
 748          * @param $text string Source text string.
 749          * @param $matches array Out parameter, Array: extracted tags
 750          * @param $uniq_prefix string
 751          * @return String: stripped text
 752          */
 753         public static function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
 754                 static $n = 1;
 755                 $stripped = '';
 756                 $matches = array();
 757
 758                 $taglist = implode( '|', $elements );
 759                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 760
 761                 while ( $text != '' ) {
 762                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 763                         $stripped .= $p[0];
 764                         if ( count( $p ) < 5 ) {
 765                                 break;
 766                         }
 767                         if ( count( $p ) > 5 ) {
 768                                 # comment
 769                                 $element    = $p[4];
 770                                 $attributes = '';
 771                                 $close      = '';
 772                                 $inside     = $p[5];
 773                         } else {
 774                                 # tag
 775                                 $element    = $p[1];
 776                                 $attributes = $p[2];
 777                                 $close      = $p[3];
 778                                 $inside     = $p[4];
 779                         }
 780
 781                         $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
 782                         $stripped .= $marker;
 783
 784                         if ( $close === '/>' ) {
 785                                 # Empty element tag, <tag />
 786                                 $content = null;
 787                                 $text = $inside;
 788                                 $tail = null;
 789                         } else {
 790                                 if ( $element === '!--' ) {
 791                                         $end = '/(-->)/';
 792                                 } else {
 793                                         $end = "/(<\\/$element\\s*>)/i";
 794                                 }
 795                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 796                                 $content = $q[0];
 797                                 if ( count( $q ) < 3 ) {
 798                                         # No end tag -- let it run out to the end of the text.
 799                                         $tail = '';
 800                                         $text = '';
 801                                 } else {
 802                                         $tail = $q[1];
 803                                         $text = $q[2];
 804                                 }
 805                         }
 806
 807                         $matches[$marker] = array( $element,
 808                                 $content,
 809                                 Sanitizer::decodeTagAttributes( $attributes ),
 810                                 "<$element$attributes$close$content$tail" );
 811                 }
 812                 return $stripped;
 813         }
 814
 815         /**
 816          * Get a list of strippable XML-like elements
 817          *
 818          * @return array
 819          */
 820         function getStripList() {
 821                 return $this->mStripList;
 822         }
 823
 824         /**
 825          * Add an item to the strip state
 826          * Returns the unique tag which must be inserted into the stripped text
 827          * The tag will be replaced with the original text in unstrip()
 828          *
 829          * @param $text string
 830          *
 831          * @return string
 832          */
 833         function insertStripItem( $text ) {
 834                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 835                 $this->mMarkerIndex++;
 836                 $this->mStripState->addGeneral( $rnd, $text );
 837                 return $rnd;
 838         }
 839
 840         /**
 841          * parse the wiki syntax used to render tables
 842          *
 843          * @private
 844          */
 845         function doTableStuff( $text ) {
 846                 wfProfileIn( __METHOD__ );
 847
 848                 $lines = StringUtils::explode( "\n", $text );
 849                 $out = '';
 850                 $td_history = array(); # Is currently a td tag open?
 851                 $last_tag_history = array(); # Save history of last lag activated (td, th or caption)
 852                 $tr_history = array(); # Is currently a tr tag open?
 853                 $tr_attributes = array(); # history of tr attributes
 854                 $has_opened_tr = array(); # Did this table open a <tr> element?
 855                 $indent_level = 0; # indent level of the table
 856
 857                 foreach ( $lines as $outLine ) {
 858                         $line = trim( $outLine );
 859
 860                         if ( $line === '' ) { # empty line, go to next line
 861                                 $out .= $outLine."\n";
 862                                 continue;
 863                         }
 864
 865                         $first_character = $line[0];
 866                         $matches = array();
 867
 868                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 869                                 # First check if we are starting a new table
 870                                 $indent_level = strlen( $matches[1] );
 871
 872                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 873                                 $attributes = Sanitizer::fixTagAttributes( $attributes , 'table' );
 874
 875                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 876                                 array_push( $td_history , false );
 877                                 array_push( $last_tag_history , '' );
 878                                 array_push( $tr_history , false );
 879                                 array_push( $tr_attributes , '' );
 880                                 array_push( $has_opened_tr , false );
 881                         } elseif ( count( $td_history ) == 0 ) {
 882                                 # Don't do any of the following
 883                                 $out .= $outLine."\n";
 884                                 continue;
 885                         } elseif ( substr( $line , 0 , 2 ) === '|}' ) {
 886                                 # We are ending a table
 887                                 $line = '</table>' . substr( $line , 2 );
 888                                 $last_tag = array_pop( $last_tag_history );
 889
 890                                 if ( !array_pop( $has_opened_tr ) ) {
 891                                         $line = "<tr><td></td></tr>{$line}";
 892                                 }
 893
 894                                 if ( array_pop( $tr_history ) ) {
 895                                         $line = "</tr>{$line}";
 896                                 }
 897
 898                                 if ( array_pop( $td_history ) ) {
 899                                         $line = "</{$last_tag}>{$line}";
 900                                 }
 901                                 array_pop( $tr_attributes );
 902                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 903                         } elseif ( substr( $line , 0 , 2 ) === '|-' ) {
 904                                 # Now we have a table row
 905                                 $line = preg_replace( '#^\|-+#', '', $line );
 906
 907                                 # Whats after the tag is now only attributes
 908                                 $attributes = $this->mStripState->unstripBoth( $line );
 909                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 910                                 array_pop( $tr_attributes );
 911                                 array_push( $tr_attributes, $attributes );
 912
 913                                 $line = '';
 914                                 $last_tag = array_pop( $last_tag_history );
 915                                 array_pop( $has_opened_tr );
 916                                 array_push( $has_opened_tr , true );
 917
 918                                 if ( array_pop( $tr_history ) ) {
 919                                         $line = '</tr>';
 920                                 }
 921
 922                                 if ( array_pop( $td_history ) ) {
 923                                         $line = "</{$last_tag}>{$line}";
 924                                 }
 925
 926                                 $outLine = $line;
 927                                 array_push( $tr_history , false );
 928                                 array_push( $td_history , false );
 929                                 array_push( $last_tag_history , '' );
 930                         } elseif ( $first_character === '|' || $first_character === '!' || substr( $line , 0 , 2 )  === '|+' ) {
 931                                 # This might be cell elements, td, th or captions
 932                                 if ( substr( $line , 0 , 2 ) === '|+' ) {
 933                                         $first_character = '+';
 934                                         $line = substr( $line , 1 );
 935                                 }
 936
 937                                 $line = substr( $line , 1 );
 938
 939                                 if ( $first_character === '!' ) {
 940                                         $line = str_replace( '!!' , '||' , $line );
 941                                 }
 942
 943                                 # Split up multiple cells on the same line.
 944                                 # FIXME : This can result in improper nesting of tags processed
 945                                 # by earlier parser steps, but should avoid splitting up eg
 946                                 # attribute values containing literal "||".
 947                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 948
 949                                 $outLine = '';
 950
 951                                 # Loop through each table cell
 952                                 foreach ( $cells as $cell ) {
 953                                         $previous = '';
 954                                         if ( $first_character !== '+' ) {
 955                                                 $tr_after = array_pop( $tr_attributes );
 956                                                 if ( !array_pop( $tr_history ) ) {
 957                                                         $previous = "<tr{$tr_after}>\n";
 958                                                 }
 959                                                 array_push( $tr_history , true );
 960                                                 array_push( $tr_attributes , '' );
 961                                                 array_pop( $has_opened_tr );
 962                                                 array_push( $has_opened_tr , true );
 963                                         }
 964
 965                                         $last_tag = array_pop( $last_tag_history );
 966
 967                                         if ( array_pop( $td_history ) ) {
 968                                                 $previous = "</{$last_tag}>\n{$previous}";
 969                                         }
 970
 971                                         if ( $first_character === '|' ) {
 972                                                 $last_tag = 'td';
 973                                         } elseif ( $first_character === '!' ) {
 974                                                 $last_tag = 'th';
 975                                         } elseif ( $first_character === '+' ) {
 976                                                 $last_tag = 'caption';
 977                                         } else {
 978                                                 $last_tag = '';
 979                                         }
 980
 981                                         array_push( $last_tag_history , $last_tag );
 982
 983                                         # A cell could contain both parameters and data
 984                                         $cell_data = explode( '|' , $cell , 2 );
 985
 986                                         # Bug 553: Note that a '|' inside an invalid link should not
 987                                         # be mistaken as delimiting cell parameters
 988                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 989                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 990                                         } elseif ( count( $cell_data ) == 1 ) {
 991                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 992                                         } else {
 993                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 994                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 995                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 996                                         }
 997
 998                                         $outLine .= $cell;
 999                                         array_push( $td_history , true );
1000                                 }
1001                         }
1002                         $out .= $outLine . "\n";
1003                 }
1004
1005                 # Closing open td, tr && table
1006                 while ( count( $td_history ) > 0 ) {
1007                         if ( array_pop( $td_history ) ) {
1008                                 $out .= "</td>\n";
1009                         }
1010                         if ( array_pop( $tr_history ) ) {
1011                                 $out .= "</tr>\n";
1012                         }
1013                         if ( !array_pop( $has_opened_tr ) ) {
1014                                 $out .= "<tr><td></td></tr>\n" ;
1015                         }
1016
1017                         $out .= "</table>\n";
1018                 }
1019
1020                 # Remove trailing line-ending (b/c)
1021                 if ( substr( $out, -1 ) === "\n" ) {
1022                         $out = substr( $out, 0, -1 );
1023                 }
1024
1025                 # special case: don't return empty table
1026                 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1027                         $out = '';
1028                 }
1029
1030                 wfProfileOut( __METHOD__ );
1031
1032                 return $out;
1033         }
1034
1035         /**
1036          * Helper function for parse() that transforms wiki markup into
1037          * HTML. Only called for $mOutputType == self::OT_HTML.
1038          *
1039          * @private
1040          *
1041          * @param $text string
1042          * @param $isMain bool
1043          * @param $frame bool
1044          *
1045          * @return string
1046          */
1047         function internalParse( $text, $isMain = true, $frame = false ) {
1048                 wfProfileIn( __METHOD__ );
1049
1050                 $origText = $text;
1051
1052                 # Hook to suspend the parser in this state
1053                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1054                         wfProfileOut( __METHOD__ );
1055                         return $text ;
1056                 }
1057
1058                 # if $frame is provided, then use $frame for replacing any variables
1059                 if ( $frame ) {
1060                         # use frame depth to infer how include/noinclude tags should be handled
1061                         # depth=0 means this is the top-level document; otherwise it's an included document
1062                         if ( !$frame->depth ) {
1063                                 $flag = 0;
1064                         } else {
1065                                 $flag = Parser::PTD_FOR_INCLUSION;
1066                         }
1067                         $dom = $this->preprocessToDom( $text, $flag );
1068                         $text = $frame->expand( $dom );
1069                 } else {
1070                         # if $frame is not provided, then use old-style replaceVariables
1071                         $text = $this->replaceVariables( $text );
1072                 }
1073
1074                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
1075                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1076
1077                 # Tables need to come after variable replacement for things to work
1078                 # properly; putting them before other transformations should keep
1079                 # exciting things like link expansions from showing up in surprising
1080                 # places.
1081                 $text = $this->doTableStuff( $text );
1082
1083                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1084
1085                 $text = $this->doDoubleUnderscore( $text );
1086
1087                 $text = $this->doHeadings( $text );
1088                 if ( $this->mOptions->getUseDynamicDates() ) {
1089                         $df = DateFormatter::getInstance();
1090                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1091                 }
1092                 $text = $this->replaceInternalLinks( $text );
1093                 $text = $this->doAllQuotes( $text );
1094                 $text = $this->replaceExternalLinks( $text );
1095
1096                 # replaceInternalLinks may sometimes leave behind
1097                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1098                 $text = str_replace( $this->mUniqPrefix.'NOPARSE', '', $text );
1099
1100                 $text = $this->doMagicLinks( $text );
1101                 $text = $this->formatHeadings( $text, $origText, $isMain );
1102
1103                 wfProfileOut( __METHOD__ );
1104                 return $text;
1105         }
1106
1107         /**
1108          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1109          * magic external links.
1110          *
1111          * DML
1112          * @private
1113          *
1114          * @param $text string
1115          *
1116          * @return string
1117          */
1118         function doMagicLinks( $text ) {
1119                 wfProfileIn( __METHOD__ );
1120                 $prots = wfUrlProtocolsWithoutProtRel();
1121                 $urlChar = self::EXT_LINK_URL_CLASS;
1122                 $text = preg_replace_callback(
1123                         '!(?:                           # Start cases
1124                                 (<a[ \t\r\n>].*?</a>) |     # m[1]: Skip link text
1125                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1126                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
1127                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1128                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1129                                         (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1130                                         (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1131                                         [0-9Xx]                 # check digit
1132                                         \b)
1133                         )!xu', array( &$this, 'magicLinkCallback' ), $text );
1134                 wfProfileOut( __METHOD__ );
1135                 return $text;
1136         }
1137
1138         /**
1139          * @throws MWException
1140          * @param $m array
1141          * @return HTML|string
1142          */
1143         function magicLinkCallback( $m ) {
1144                 if ( isset( $m[1] ) && $m[1] !== '' ) {
1145                         # Skip anchor
1146                         return $m[0];
1147                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1148                         # Skip HTML element
1149                         return $m[0];
1150                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1151                         # Free external link
1152                         return $this->makeFreeExternalLink( $m[0] );
1153                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1154                         # RFC or PMID
1155                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1156                                 $keyword = 'RFC';
1157                                 $urlmsg = 'rfcurl';
1158                                 $CssClass = 'mw-magiclink-rfc';
1159                                 $id = $m[4];
1160                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1161                                 $keyword = 'PMID';
1162                                 $urlmsg = 'pubmedurl';
1163                                 $CssClass = 'mw-magiclink-pmid';
1164                                 $id = $m[4];
1165                         } else {
1166                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1167                                         substr( $m[0], 0, 20 ) . '"' );
1168                         }
1169                         $url = wfMsgForContent( $urlmsg, $id );
1170                         return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $CssClass );
1171                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1172                         # ISBN
1173                         $isbn = $m[5];
1174                         $num = strtr( $isbn, array(
1175                                 '-' => '',
1176                                 ' ' => '',
1177                                 'x' => 'X',
1178                         ));
1179                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1180                         return'<a href="' .
1181                                 $titleObj->escapeLocalUrl() .
1182                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1183                 } else {
1184                         return $m[0];
1185                 }
1186         }
1187
1188         /**
1189          * Make a free external link, given a user-supplied URL
1190          *
1191          * @param $url string
1192          *
1193          * @return string HTML
1194          * @private
1195          */
1196         function makeFreeExternalLink( $url ) {
1197                 global $wgContLang;
1198                 wfProfileIn( __METHOD__ );
1199
1200                 $trail = '';
1201
1202                 # The characters '<' and '>' (which were escaped by
1203                 # removeHTMLtags()) should not be included in
1204                 # URLs, per RFC 2396.
1205                 $m2 = array();
1206                 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1207                         $trail = substr( $url, $m2[0][1] ) . $trail;
1208                         $url = substr( $url, 0, $m2[0][1] );
1209                 }
1210
1211                 # Move trailing punctuation to $trail
1212                 $sep = ',;\.:!?';
1213                 # If there is no left bracket, then consider right brackets fair game too
1214                 if ( strpos( $url, '(' ) === false ) {
1215                         $sep .= ')';
1216                 }
1217
1218                 $numSepChars = strspn( strrev( $url ), $sep );
1219                 if ( $numSepChars ) {
1220                         $trail = substr( $url, -$numSepChars ) . $trail;
1221                         $url = substr( $url, 0, -$numSepChars );
1222                 }
1223
1224                 $url = Sanitizer::cleanUrl( $url );
1225
1226                 # Is this an external image?
1227                 $text = $this->maybeMakeExternalImage( $url );
1228                 if ( $text === false ) {
1229                         # Not an image, make a link
1230                         $text = Linker::makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1231                                 $this->getExternalLinkAttribs( $url ) );
1232                         # Register it in the output object...
1233                         # Replace unnecessary URL escape codes with their equivalent characters
1234                         $pasteurized = self::replaceUnusualEscapes( $url );
1235                         $this->mOutput->addExternalLink( $pasteurized );
1236                 }
1237                 wfProfileOut( __METHOD__ );
1238                 return $text . $trail;
1239         }
1240
1241
1242         /**
1243          * Parse headers and return html
1244          *
1245          * @private
1246          *
1247          * @param $text string
1248          *
1249          * @return string
1250          */
1251         function doHeadings( $text ) {
1252                 wfProfileIn( __METHOD__ );
1253                 for ( $i = 6; $i >= 1; --$i ) {
1254                         $h = str_repeat( '=', $i );
1255                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1256                           "<h$i>\\1</h$i>", $text );
1257                 }
1258                 wfProfileOut( __METHOD__ );
1259                 return $text;
1260         }
1261
1262         /**
1263          * Replace single quotes with HTML markup
1264          * @private
1265          *
1266          * @param $text string
1267          *
1268          * @return string the altered text
1269          */
1270         function doAllQuotes( $text ) {
1271                 wfProfileIn( __METHOD__ );
1272                 $outtext = '';
1273                 $lines = StringUtils::explode( "\n", $text );
1274                 foreach ( $lines as $line ) {
1275                         $outtext .= $this->doQuotes( $line ) . "\n";
1276                 }
1277                 $outtext = substr( $outtext, 0,-1 );
1278                 wfProfileOut( __METHOD__ );
1279                 return $outtext;
1280         }
1281
1282         /**
1283          * Helper function for doAllQuotes()
1284          *
1285          * @param $text string
1286          *
1287          * @return string
1288          */
1289         public function doQuotes( $text ) {
1290                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1291                 if ( count( $arr ) == 1 ) {
1292                         return $text;
1293                 } else {
1294                         # First, do some preliminary work. This may shift some apostrophes from
1295                         # being mark-up to being text. It also counts the number of occurrences
1296                         # of bold and italics mark-ups.
1297                         $numbold = 0;
1298                         $numitalics = 0;
1299                         for ( $i = 0; $i < count( $arr ); $i++ ) {
1300                                 if ( ( $i % 2 ) == 1 ) {
1301                                         # If there are ever four apostrophes, assume the first is supposed to
1302                                         # be text, and the remaining three constitute mark-up for bold text.
1303                                         if ( strlen( $arr[$i] ) == 4 ) {
1304                                                 $arr[$i-1] .= "'";
1305                                                 $arr[$i] = "'''";
1306                                         } elseif ( strlen( $arr[$i] ) > 5 ) {
1307                                                 # If there are more than 5 apostrophes in a row, assume they're all
1308                                                 # text except for the last 5.
1309                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1310                                                 $arr[$i] = "'''''";
1311                                         }
1312                                         # Count the number of occurrences of bold and italics mark-ups.
1313                                         # We are not counting sequences of five apostrophes.
1314                                         if ( strlen( $arr[$i] ) == 2 ) {
1315                                                 $numitalics++;
1316                                         } elseif ( strlen( $arr[$i] ) == 3 ) {
1317                                                 $numbold++;
1318                                         } elseif ( strlen( $arr[$i] ) == 5 ) {
1319                                                 $numitalics++;
1320                                                 $numbold++;
1321                                         }
1322                                 }
1323                         }
1324
1325                         # If there is an odd number of both bold and italics, it is likely
1326                         # that one of the bold ones was meant to be an apostrophe followed
1327                         # by italics. Which one we cannot know for certain, but it is more
1328                         # likely to be one that has a single-letter word before it.
1329                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1330                                 $i = 0;
1331                                 $firstsingleletterword = -1;
1332                                 $firstmultiletterword = -1;
1333                                 $firstspace = -1;
1334                                 foreach ( $arr as $r ) {
1335                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1336                                                 $x1 = substr( $arr[$i-1], -1 );
1337                                                 $x2 = substr( $arr[$i-1], -2, 1 );
1338                                                 if ( $x1 === ' ' ) {
1339                                                         if ( $firstspace == -1 ) {
1340                                                                 $firstspace = $i;
1341                                                         }
1342                                                 } elseif ( $x2 === ' ') {
1343                                                         if ( $firstsingleletterword == -1 ) {
1344                                                                 $firstsingleletterword = $i;
1345                                                         }
1346                                                 } else {
1347                                                         if ( $firstmultiletterword == -1 ) {
1348                                                                 $firstmultiletterword = $i;
1349                                                         }
1350                                                 }
1351                                         }
1352                                         $i++;
1353                                 }
1354
1355                                 # If there is a single-letter word, use it!
1356                                 if ( $firstsingleletterword > -1 ) {
1357                                         $arr[$firstsingleletterword] = "''";
1358                                         $arr[$firstsingleletterword-1] .= "'";
1359                                 } elseif ( $firstmultiletterword > -1 ) {
1360                                         # If not, but there's a multi-letter word, use that one.
1361                                         $arr[$firstmultiletterword] = "''";
1362                                         $arr[$firstmultiletterword-1] .= "'";
1363                                 } elseif ( $firstspace > -1 ) {
1364                                         # ... otherwise use the first one that has neither.
1365                                         # (notice that it is possible for all three to be -1 if, for example,
1366                                         # there is only one pentuple-apostrophe in the line)
1367                                         $arr[$firstspace] = "''";
1368                                         $arr[$firstspace-1] .= "'";
1369                                 }
1370                         }
1371
1372                         # Now let's actually convert our apostrophic mush to HTML!
1373                         $output = '';
1374                         $buffer = '';
1375                         $state = '';
1376                         $i = 0;
1377                         foreach ( $arr as $r ) {
1378                                 if ( ( $i % 2 ) == 0 ) {
1379                                         if ( $state === 'both' ) {
1380                                                 $buffer .= $r;
1381                                         } else {
1382                                                 $output .= $r;
1383                                         }
1384                                 } else {
1385                                         if ( strlen( $r ) == 2 ) {
1386                                                 if ( $state === 'i' ) {
1387                                                         $output .= '</i>'; $state = '';
1388                                                 } elseif ( $state === 'bi' ) {
1389                                                         $output .= '</i>'; $state = 'b';
1390                                                 } elseif ( $state === 'ib' ) {
1391                                                         $output .= '</b></i><b>'; $state = 'b';
1392                                                 } elseif ( $state === 'both' ) {
1393                                                         $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1394                                                 } else { # $state can be 'b' or ''
1395                                                         $output .= '<i>'; $state .= 'i';
1396                                                 }
1397                                         } elseif ( strlen( $r ) == 3 ) {
1398                                                 if ( $state === 'b' ) {
1399                                                         $output .= '</b>'; $state = '';
1400                                                 } elseif ( $state === 'bi' ) {
1401                                                         $output .= '</i></b><i>'; $state = 'i';
1402                                                 } elseif ( $state === 'ib' ) {
1403                                                         $output .= '</b>'; $state = 'i';
1404                                                 } elseif ( $state === 'both' ) {
1405                                                         $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1406                                                 } else { # $state can be 'i' or ''
1407                                                         $output .= '<b>'; $state .= 'b';
1408                                                 }
1409                                         } elseif ( strlen( $r ) == 5 ) {
1410                                                 if ( $state === 'b' ) {
1411                                                         $output .= '</b><i>'; $state = 'i';
1412                                                 } elseif ( $state === 'i' ) {
1413                                                         $output .= '</i><b>'; $state = 'b';
1414                                                 } elseif ( $state === 'bi' ) {
1415                                                         $output .= '</i></b>'; $state = '';
1416                                                 } elseif ( $state === 'ib' ) {
1417                                                         $output .= '</b></i>'; $state = '';
1418                                                 } elseif ( $state === 'both' ) {
1419                                                         $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1420                                                 } else { # ($state == '')
1421                                                         $buffer = ''; $state = 'both';
1422                                                 }
1423                                         }
1424                                 }
1425                                 $i++;
1426                         }
1427                         # Now close all remaining tags.  Notice that the order is important.
1428                         if ( $state === 'b' || $state === 'ib' ) {
1429                                 $output .= '</b>';
1430                         }
1431                         if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1432                                 $output .= '</i>';
1433                         }
1434                         if ( $state === 'bi' ) {
1435                                 $output .= '</b>';
1436                         }
1437                         # There might be lonely ''''', so make sure we have a buffer
1438                         if ( $state === 'both' && $buffer ) {
1439                                 $output .= '<b><i>'.$buffer.'</i></b>';
1440                         }
1441                         return $output;
1442                 }
1443         }
1444
1445         /**
1446          * Replace external links (REL)
1447          *
1448          * Note: this is all very hackish and the order of execution matters a lot.
1449          * Make sure to run maintenance/parserTests.php if you change this code.
1450          *
1451          * @private
1452          *
1453          * @param $text string
1454          *
1455          * @return string
1456          */
1457         function replaceExternalLinks( $text ) {
1458                 global $wgContLang;
1459                 wfProfileIn( __METHOD__ );
1460
1461                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1462                 $s = array_shift( $bits );
1463
1464                 $i = 0;
1465                 while ( $i<count( $bits ) ) {
1466                         $url = $bits[$i++];
1467                         $protocol = $bits[$i++];
1468                         $text = $bits[$i++];
1469                         $trail = $bits[$i++];
1470
1471                         # The characters '<' and '>' (which were escaped by
1472                         # removeHTMLtags()) should not be included in
1473                         # URLs, per RFC 2396.
1474                         $m2 = array();
1475                         if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1476                                 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1477                                 $url = substr( $url, 0, $m2[0][1] );
1478                         }
1479
1480                         # If the link text is an image URL, replace it with an <img> tag
1481                         # This happened by accident in the original parser, but some people used it extensively
1482                         $img = $this->maybeMakeExternalImage( $text );
1483                         if ( $img !== false ) {
1484                                 $text = $img;
1485                         }
1486
1487                         $dtrail = '';
1488
1489                         # Set linktype for CSS - if URL==text, link is essentially free
1490                         $linktype = ( $text === $url ) ? 'free' : 'text';
1491
1492                         # No link text, e.g. [http://domain.tld/some.link]
1493                         if ( $text == '' ) {
1494                                 # Autonumber
1495                                 $langObj = $this->getFunctionLang();
1496                                 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1497                                 $linktype = 'autonumber';
1498                         } else {
1499                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1500                                 # Check for trail
1501                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1502                         }
1503
1504                         $text = $wgContLang->markNoConversion( $text );
1505
1506                         $url = Sanitizer::cleanUrl( $url );
1507
1508                         # Use the encoded URL
1509                         # This means that users can paste URLs directly into the text
1510                         # Funny characters like ö aren't valid in URLs anyway
1511                         # This was changed in August 2004
1512                         $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1513                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1514
1515                         # Register link in the output object.
1516                         # Replace unnecessary URL escape codes with the referenced character
1517                         # This prevents spammers from hiding links from the filters
1518                         $pasteurized = self::replaceUnusualEscapes( $url );
1519                         $this->mOutput->addExternalLink( $pasteurized );
1520                 }
1521
1522                 wfProfileOut( __METHOD__ );
1523                 return $s;
1524         }
1525
1526         /**
1527          * Get an associative array of additional HTML attributes appropriate for a
1528          * particular external link.  This currently may include rel => nofollow
1529          * (depending on configuration, namespace, and the URL's domain) and/or a
1530          * target attribute (depending on configuration).
1531          *
1532          * @param $url String|bool optional URL, to extract the domain from for rel =>
1533          *   nofollow if appropriate
1534          * @return Array associative array of HTML attributes
1535          */
1536         function getExternalLinkAttribs( $url = false ) {
1537                 $attribs = array();
1538                 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1539                 $ns = $this->mTitle->getNamespace();
1540                 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) &&
1541                                 !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions ) )
1542                 {
1543                         $attribs['rel'] = 'nofollow';
1544                 }
1545                 if ( $this->mOptions->getExternalLinkTarget() ) {
1546                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1547                 }
1548                 return $attribs;
1549         }
1550
1551         /**
1552          * Replace unusual URL escape codes with their equivalent characters
1553          *
1554          * @param $url String
1555          * @return String
1556          *
1557          * @todo  This can merge genuinely required bits in the path or query string,
1558          *        breaking legit URLs. A proper fix would treat the various parts of
1559          *        the URL differently; as a workaround, just use the output for
1560          *        statistical records, not for actual linking/output.
1561          */
1562         static function replaceUnusualEscapes( $url ) {
1563                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1564                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1565         }
1566
1567         /**
1568          * Callback function used in replaceUnusualEscapes().
1569          * Replaces unusual URL escape codes with their equivalent character
1570          *
1571          * @param $matches array
1572          *
1573          * @return string
1574          */
1575         private static function replaceUnusualEscapesCallback( $matches ) {
1576                 $char = urldecode( $matches[0] );
1577                 $ord = ord( $char );
1578                 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1579                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1580                         # No, shouldn't be escaped
1581                         return $char;
1582                 } else {
1583                         # Yes, leave it escaped
1584                         return $matches[0];
1585                 }
1586         }
1587
1588         /**
1589          * make an image if it's allowed, either through the global
1590          * option, through the exception, or through the on-wiki whitelist
1591          * @private
1592          *
1593          * $param $url string
1594          *
1595          * @return string
1596          */
1597         function maybeMakeExternalImage( $url ) {
1598                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1599                 $imagesexception = !empty( $imagesfrom );
1600                 $text = false;
1601                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1602                 if ( $imagesexception && is_array( $imagesfrom ) ) {
1603                         $imagematch = false;
1604                         foreach ( $imagesfrom as $match ) {
1605                                 if ( strpos( $url, $match ) === 0 ) {
1606                                         $imagematch = true;
1607                                         break;
1608                                 }
1609                         }
1610                 } elseif ( $imagesexception ) {
1611                         $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1612                 } else {
1613                         $imagematch = false;
1614                 }
1615                 if ( $this->mOptions->getAllowExternalImages()
1616                          || ( $imagesexception && $imagematch ) ) {
1617                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1618                                 # Image found
1619                                 $text = Linker::makeExternalImage( $url );
1620                         }
1621                 }
1622                 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1623                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1624                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1625                         foreach ( $whitelist as $entry ) {
1626                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1627                                 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1628                                         continue;
1629                                 }
1630                                 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1631                                         # Image matches a whitelist entry
1632                                         $text = Linker::makeExternalImage( $url );
1633                                         break;
1634                                 }
1635                         }
1636                 }
1637                 return $text;
1638         }
1639
1640         /**
1641          * Process [[ ]] wikilinks
1642          *
1643          * @param $s string
1644          *
1645          * @return String: processed text
1646          *
1647          * @private
1648          */
1649         function replaceInternalLinks( $s ) {
1650                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1651                 return $s;
1652         }
1653
1654         /**
1655          * Process [[ ]] wikilinks (RIL)
1656          * @return LinkHolderArray
1657          *
1658          * @private
1659          */
1660         function replaceInternalLinks2( &$s ) {
1661                 global $wgContLang;
1662
1663                 wfProfileIn( __METHOD__ );
1664
1665                 wfProfileIn( __METHOD__.'-setup' );
1666                 static $tc = FALSE, $e1, $e1_img;
1667                 # the % is needed to support urlencoded titles as well
1668                 if ( !$tc ) {
1669                         $tc = Title::legalChars() . '#%';
1670                         # Match a link having the form [[namespace:link|alternate]]trail
1671                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1672                         # Match cases where there is no "]]", which might still be images
1673                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1674                 }
1675
1676                 $holders = new LinkHolderArray( $this );
1677
1678                 # split the entire text string on occurences of [[
1679                 $a = StringUtils::explode( '[[', ' ' . $s );
1680                 # get the first element (all text up to first [[), and remove the space we added
1681                 $s = $a->current();
1682                 $a->next();
1683                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1684                 $s = substr( $s, 1 );
1685
1686                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1687                 $e2 = null;
1688                 if ( $useLinkPrefixExtension ) {
1689                         # Match the end of a line for a word that's not followed by whitespace,
1690                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1691                         $e2 = wfMsgForContent( 'linkprefix' );
1692                 }
1693
1694                 if ( is_null( $this->mTitle ) ) {
1695                         wfProfileOut( __METHOD__.'-setup' );
1696                         wfProfileOut( __METHOD__ );
1697                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1698                 }
1699                 $nottalk = !$this->mTitle->isTalkPage();
1700
1701                 if ( $useLinkPrefixExtension ) {
1702                         $m = array();
1703                         if ( preg_match( $e2, $s, $m ) ) {
1704                                 $first_prefix = $m[2];
1705                         } else {
1706                                 $first_prefix = false;
1707                         }
1708                 } else {
1709                         $prefix = '';
1710                 }
1711
1712                 if ( $wgContLang->hasVariants() ) {
1713                         $selflink = $wgContLang->autoConvertToAllVariants( $this->mTitle->getPrefixedText() );
1714                 } else {
1715                         $selflink = array( $this->mTitle->getPrefixedText() );
1716                 }
1717                 $useSubpages = $this->areSubpagesAllowed();
1718                 wfProfileOut( __METHOD__.'-setup' );
1719
1720                 # Loop for each link
1721                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1722                         # Check for excessive memory usage
1723                         if ( $holders->isBig() ) {
1724                                 # Too big
1725                                 # Do the existence check, replace the link holders and clear the array
1726                                 $holders->replace( $s );
1727                                 $holders->clear();
1728                         }
1729
1730                         if ( $useLinkPrefixExtension ) {
1731                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1732                                 if ( preg_match( $e2, $s, $m ) ) {
1733                                         $prefix = $m[2];
1734                                         $s = $m[1];
1735                                 } else {
1736                                         $prefix='';
1737                                 }
1738                                 # first link
1739                                 if ( $first_prefix ) {
1740                                         $prefix = $first_prefix;
1741                                         $first_prefix = false;
1742                                 }
1743                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1744                         }
1745
1746                         $might_be_img = false;
1747
1748                         wfProfileIn( __METHOD__."-e1" );
1749                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1750                                 $text = $m[2];
1751                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1752                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1753                                 # the real problem is with the $e1 regex
1754                                 # See bug 1300.
1755                                 #
1756                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1757                                 # and no image is in sight. See bug 2095.
1758                                 #
1759                                 if ( $text !== '' &&
1760                                         substr( $m[3], 0, 1 ) === ']' &&
1761                                         strpos( $text, '[' ) !== false
1762                                 )
1763                                 {
1764                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1765                                         $m[3] = substr( $m[3], 1 );
1766                                 }
1767                                 # fix up urlencoded title texts
1768                                 if ( strpos( $m[1], '%' ) !== false ) {
1769                                         # Should anchors '#' also be rejected?
1770                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), rawurldecode( $m[1] ) );
1771                                 }
1772                                 $trail = $m[3];
1773                         } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1774                                 $might_be_img = true;
1775                                 $text = $m[2];
1776                                 if ( strpos( $m[1], '%' ) !== false ) {
1777                                         $m[1] = rawurldecode( $m[1] );
1778                                 }
1779                                 $trail = "";
1780                         } else { # Invalid form; output directly
1781                                 $s .= $prefix . '[[' . $line ;
1782                                 wfProfileOut( __METHOD__."-e1" );
1783                                 continue;
1784                         }
1785                         wfProfileOut( __METHOD__."-e1" );
1786                         wfProfileIn( __METHOD__."-misc" );
1787
1788                         # Don't allow internal links to pages containing
1789                         # PROTO: where PROTO is a valid URL protocol; these
1790                         # should be external links.
1791                         if ( preg_match( '/^(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1792                                 $s .= $prefix . '[[' . $line ;
1793                                 wfProfileOut( __METHOD__."-misc" );
1794                                 continue;
1795                         }
1796
1797                         # Make subpage if necessary
1798                         if ( $useSubpages ) {
1799                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1800                         } else {
1801                                 $link = $m[1];
1802                         }
1803
1804                         $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1805                         if ( !$noforce ) {
1806                                 # Strip off leading ':'
1807                                 $link = substr( $link, 1 );
1808                         }
1809
1810                         wfProfileOut( __METHOD__."-misc" );
1811                         wfProfileIn( __METHOD__."-title" );
1812                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1813                         if ( $nt === null ) {
1814                                 $s .= $prefix . '[[' . $line;
1815                                 wfProfileOut( __METHOD__."-title" );
1816                                 continue;
1817                         }
1818
1819                         $ns = $nt->getNamespace();
1820                         $iw = $nt->getInterWiki();
1821                         wfProfileOut( __METHOD__."-title" );
1822
1823                         if ( $might_be_img ) { # if this is actually an invalid link
1824                                 wfProfileIn( __METHOD__."-might_be_img" );
1825                                 if ( $ns == NS_FILE && $noforce ) { # but might be an image
1826                                         $found = false;
1827                                         while ( true ) {
1828                                                 # look at the next 'line' to see if we can close it there
1829                                                 $a->next();
1830                                                 $next_line = $a->current();
1831                                                 if ( $next_line === false || $next_line === null ) {
1832                                                         break;
1833                                                 }
1834                                                 $m = explode( ']]', $next_line, 3 );
1835                                                 if ( count( $m ) == 3 ) {
1836                                                         # the first ]] closes the inner link, the second the image
1837                                                         $found = true;
1838                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1839                                                         $trail = $m[2];
1840                                                         break;
1841                                                 } elseif ( count( $m ) == 2 ) {
1842                                                         # if there's exactly one ]] that's fine, we'll keep looking
1843                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1844                                                 } else {
1845                                                         # if $next_line is invalid too, we need look no further
1846                                                         $text .= '[[' . $next_line;
1847                                                         break;
1848                                                 }
1849                                         }
1850                                         if ( !$found ) {
1851                                                 # we couldn't find the end of this imageLink, so output it raw
1852                                                 # but don't ignore what might be perfectly normal links in the text we've examined
1853                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1854                                                 $s .= "{$prefix}[[$link|$text";
1855                                                 # note: no $trail, because without an end, there *is* no trail
1856                                                 wfProfileOut( __METHOD__."-might_be_img" );
1857                                                 continue;
1858                                         }
1859                                 } else { # it's not an image, so output it raw
1860                                         $s .= "{$prefix}[[$link|$text";
1861                                         # note: no $trail, because without an end, there *is* no trail
1862                                         wfProfileOut( __METHOD__."-might_be_img" );
1863                                         continue;
1864                                 }
1865                                 wfProfileOut( __METHOD__."-might_be_img" );
1866                         }
1867
1868                         $wasblank = ( $text  == '' );
1869                         if ( $wasblank ) {
1870                                 $text = $link;
1871                         } else {
1872                                 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
1873                                 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
1874                                 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
1875                                 #    -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
1876                                 $text = $this->doQuotes( $text );
1877                         }
1878
1879                         # Link not escaped by : , create the various objects
1880                         if ( $noforce ) {
1881
1882                                 # Interwikis
1883                                 wfProfileIn( __METHOD__."-interwiki" );
1884                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1885                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1886                                         $s = rtrim( $s . $prefix );
1887                                         $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
1888                                         wfProfileOut( __METHOD__."-interwiki" );
1889                                         continue;
1890                                 }
1891                                 wfProfileOut( __METHOD__."-interwiki" );
1892
1893                                 if ( $ns == NS_FILE ) {
1894                                         wfProfileIn( __METHOD__."-image" );
1895                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1896                                                 if ( $wasblank ) {
1897                                                         # if no parameters were passed, $text
1898                                                         # becomes something like "File:Foo.png",
1899                                                         # which we don't want to pass on to the
1900                                                         # image generator
1901                                                         $text = '';
1902                                                 } else {
1903                                                         # recursively parse links inside the image caption
1904                                                         # actually, this will parse them in any other parameters, too,
1905                                                         # but it might be hard to fix that, and it doesn't matter ATM
1906                                                         $text = $this->replaceExternalLinks( $text );
1907                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1908                                                 }
1909                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1910                                                 $s .= $prefix . $this->armorLinks(
1911                                                         $this->makeImage( $nt, $text, $holders ) ) . $trail;
1912                                         } else {
1913                                                 $s .= $prefix . $trail;
1914                                         }
1915                                         wfProfileOut( __METHOD__."-image" );
1916                                         continue;
1917                                 }
1918
1919                                 if ( $ns == NS_CATEGORY ) {
1920                                         wfProfileIn( __METHOD__."-category" );
1921                                         $s = rtrim( $s . "\n" ); # bug 87
1922
1923                                         if ( $wasblank ) {
1924                                                 $sortkey = $this->getDefaultSort();
1925                                         } else {
1926                                                 $sortkey = $text;
1927                                         }
1928                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1929                                         $sortkey = str_replace( "\n", '', $sortkey );
1930                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1931                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1932
1933                                         /**
1934                                          * Strip the whitespace Category links produce, see bug 87
1935                                          * @todo We might want to use trim($tmp, "\n") here.
1936                                          */
1937                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
1938
1939                                         wfProfileOut( __METHOD__."-category" );
1940                                         continue;
1941                                 }
1942                         }
1943
1944                         # Self-link checking
1945                         if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1946                                 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1947                                         $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
1948                                         continue;
1949                                 }
1950                         }
1951
1952                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1953                         # @todo FIXME: Should do batch file existence checks, see comment below
1954                         if ( $ns == NS_MEDIA ) {
1955                                 wfProfileIn( __METHOD__."-media" );
1956                                 # Give extensions a chance to select the file revision for us
1957                                 $options = array();
1958                                 $descQuery = false;
1959                                 wfRunHooks( 'BeforeParserFetchFileAndTitle',
1960                                         array( $this, $nt, &$options, &$descQuery ) );
1961                                 # Fetch and register the file (file title may be different via hooks)
1962                                 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
1963                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1964                                 $s .= $prefix . $this->armorLinks(
1965                                         Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
1966                                 wfProfileOut( __METHOD__."-media" );
1967                                 continue;
1968                         }
1969
1970                         wfProfileIn( __METHOD__."-always_known" );
1971                         # Some titles, such as valid special pages or files in foreign repos, should
1972                         # be shown as bluelinks even though they're not included in the page table
1973                         #
1974                         # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1975                         # batch file existence checks for NS_FILE and NS_MEDIA
1976                         if ( $iw == '' && $nt->isAlwaysKnown() ) {
1977                                 $this->mOutput->addLink( $nt );
1978                                 $s .= $this->makeKnownLinkHolder( $nt, $text, array(), $trail, $prefix );
1979                         } else {
1980                                 # Links will be added to the output link list after checking
1981                                 $s .= $holders->makeHolder( $nt, $text, array(), $trail, $prefix );
1982                         }
1983                         wfProfileOut( __METHOD__."-always_known" );
1984                 }
1985                 wfProfileOut( __METHOD__ );
1986                 return $holders;
1987         }
1988
1989         /**
1990          * Render a forced-blue link inline; protect against double expansion of
1991          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1992          * Since this little disaster has to split off the trail text to avoid
1993          * breaking URLs in the following text without breaking trails on the
1994          * wiki links, it's been made into a horrible function.
1995          *
1996          * @param $nt Title
1997          * @param $text String
1998          * @param $query Array or String
1999          * @param $trail String
2000          * @param $prefix String
2001          * @return String: HTML-wikitext mix oh yuck
2002          */
2003         function makeKnownLinkHolder( $nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
2004                 list( $inside, $trail ) = Linker::splitTrail( $trail );
2005
2006                 if ( is_string( $query ) ) {
2007                         $query = wfCgiToArray( $query );
2008                 }
2009                 if ( $text == '' ) {
2010                         $text = htmlspecialchars( $nt->getPrefixedText() );
2011                 }
2012
2013                 $link = Linker::linkKnown( $nt, "$prefix$text$inside", array(), $query );
2014
2015                 return $this->armorLinks( $link ) . $trail;
2016         }
2017
2018         /**
2019          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2020          * going to go through further parsing steps before inline URL expansion.
2021          *
2022          * Not needed quite as much as it used to be since free links are a bit
2023          * more sensible these days. But bracketed links are still an issue.
2024          *
2025          * @param $text String: more-or-less HTML
2026          * @return String: less-or-more HTML with NOPARSE bits
2027          */
2028         function armorLinks( $text ) {
2029                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
2030                         "{$this->mUniqPrefix}NOPARSE$1", $text );
2031         }
2032
2033         /**
2034          * Return true if subpage links should be expanded on this page.
2035          * @return Boolean
2036          */
2037         function areSubpagesAllowed() {
2038                 # Some namespaces don't allow subpages
2039                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2040         }
2041
2042         /**
2043          * Handle link to subpage if necessary
2044          *
2045          * @param $target String: the source of the link
2046          * @param &$text String: the link text, modified as necessary
2047          * @return string the full name of the link
2048          * @private
2049          */
2050         function maybeDoSubpageLink( $target, &$text ) {
2051                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2052         }
2053
2054         /**#@+
2055          * Used by doBlockLevels()
2056          * @private
2057          *
2058          * @return string
2059          */
2060         function closeParagraph() {
2061                 $result = '';
2062                 if ( $this->mLastSection != '' ) {
2063                         $result = '</' . $this->mLastSection  . ">\n";
2064                 }
2065                 $this->mInPre = false;
2066                 $this->mLastSection = '';
2067                 return $result;
2068         }
2069
2070         /**
2071          * getCommon() returns the length of the longest common substring
2072          * of both arguments, starting at the beginning of both.
2073          * @private
2074          *
2075          * @param $st1 string
2076          * @param $st2 string
2077          *
2078          * @return int
2079          */
2080         function getCommon( $st1, $st2 ) {
2081                 $fl = strlen( $st1 );
2082                 $shorter = strlen( $st2 );
2083                 if ( $fl < $shorter ) {
2084                         $shorter = $fl;
2085                 }
2086
2087                 for ( $i = 0; $i < $shorter; ++$i ) {
2088                         if ( $st1[$i] != $st2[$i] ) {
2089                                 break;
2090                         }
2091                 }
2092                 return $i;
2093         }
2094
2095         /**
2096          * These next three functions open, continue, and close the list
2097          * element appropriate to the prefix character passed into them.
2098          * @private
2099          *
2100          * @param $char char
2101          *
2102          * @return string
2103          */
2104         function openList( $char ) {
2105                 $result = $this->closeParagraph();
2106
2107                 if ( '*' === $char ) {
2108                         $result .= '<ul><li>';
2109                 } elseif ( '#' === $char ) {
2110                         $result .= '<ol><li>';
2111                 } elseif ( ':' === $char ) {
2112                         $result .= '<dl><dd>';
2113                 } elseif ( ';' === $char ) {
2114                         $result .= '<dl><dt>';
2115                         $this->mDTopen = true;
2116                 } else {
2117                         $result = '<!-- ERR 1 -->';
2118                 }
2119
2120                 return $result;
2121         }
2122
2123         /**
2124          * TODO: document
2125          * @param $char String
2126          * @private
2127          *
2128          * @return string
2129          */
2130         function nextItem( $char ) {
2131                 if ( '*' === $char || '#' === $char ) {
2132                         return '</li><li>';
2133                 } elseif ( ':' === $char || ';' === $char ) {
2134                         $close = '</dd>';
2135                         if ( $this->mDTopen ) {
2136                                 $close = '</dt>';
2137                         }
2138                         if ( ';' === $char ) {
2139                                 $this->mDTopen = true;
2140                                 return $close . '<dt>';
2141                         } else {
2142                                 $this->mDTopen = false;
2143                                 return $close . '<dd>';
2144                         }
2145                 }
2146                 return '<!-- ERR 2 -->';
2147         }
2148
2149         /**
2150          * TODO: document
2151          * @param $char String
2152          * @private
2153          *
2154          * @return string
2155          */
2156         function closeList( $char ) {
2157                 if ( '*' === $char ) {
2158                         $text = '</li></ul>';
2159                 } elseif ( '#' === $char ) {
2160                         $text = '</li></ol>';
2161                 } elseif ( ':' === $char ) {
2162                         if ( $this->mDTopen ) {
2163                                 $this->mDTopen = false;
2164                                 $text = '</dt></dl>';
2165                         } else {
2166                                 $text = '</dd></dl>';
2167                         }
2168                 } else {
2169                         return '<!-- ERR 3 -->';
2170                 }
2171                 return $text."\n";
2172         }
2173         /**#@-*/
2174
2175         /**
2176          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2177          *
2178          * @param $text String
2179          * @param $linestart Boolean: whether or not this is at the start of a line.
2180          * @private
2181          * @return string the lists rendered as HTML
2182          */
2183         function doBlockLevels( $text, $linestart ) {
2184                 wfProfileIn( __METHOD__ );
2185
2186                 # Parsing through the text line by line.  The main thing
2187                 # happening here is handling of block-level elements p, pre,
2188                 # and making lists from lines starting with * # : etc.
2189                 #
2190                 $textLines = StringUtils::explode( "\n", $text );
2191
2192                 $lastPrefix = $output = '';
2193                 $this->mDTopen = $inBlockElem = false;
2194                 $prefixLength = 0;
2195                 $paragraphStack = false;
2196
2197                 foreach ( $textLines as $oLine ) {
2198                         # Fix up $linestart
2199                         if ( !$linestart ) {
2200                                 $output .= $oLine;
2201                                 $linestart = true;
2202                                 continue;
2203                         }
2204                         # * = ul
2205                         # # = ol
2206                         # ; = dt
2207                         # : = dd
2208
2209                         $lastPrefixLength = strlen( $lastPrefix );
2210                         $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2211                         $preOpenMatch = preg_match( '/<pre/i', $oLine );
2212                         # If not in a <pre> element, scan for and figure out what prefixes are there.
2213                         if ( !$this->mInPre ) {
2214                                 # Multiple prefixes may abut each other for nested lists.
2215                                 $prefixLength = strspn( $oLine, '*#:;' );
2216                                 $prefix = substr( $oLine, 0, $prefixLength );
2217
2218                                 # eh?
2219                                 # ; and : are both from definition-lists, so they're equivalent
2220                                 #  for the purposes of determining whether or not we need to open/close
2221                                 #  elements.
2222                                 $prefix2 = str_replace( ';', ':', $prefix );
2223                                 $t = substr( $oLine, $prefixLength );
2224                                 $this->mInPre = (bool)$preOpenMatch;
2225                         } else {
2226                                 # Don't interpret any other prefixes in preformatted text
2227                                 $prefixLength = 0;
2228                                 $prefix = $prefix2 = '';
2229                                 $t = $oLine;
2230                         }
2231
2232                         # List generation
2233                         if ( $prefixLength && $lastPrefix === $prefix2 ) {
2234                                 # Same as the last item, so no need to deal with nesting or opening stuff
2235                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2236                                 $paragraphStack = false;
2237
2238                                 if ( substr( $prefix, -1 ) === ';') {
2239                                         # The one nasty exception: definition lists work like this:
2240                                         # ; title : definition text
2241                                         # So we check for : in the remainder text to split up the
2242                                         # title and definition, without b0rking links.
2243                                         $term = $t2 = '';
2244                                         if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2245                                                 $t = $t2;
2246                                                 $output .= $term . $this->nextItem( ':' );
2247                                         }
2248                                 }
2249                         } elseif ( $prefixLength || $lastPrefixLength ) {
2250                                 # We need to open or close prefixes, or both.
2251
2252                                 # Either open or close a level...
2253                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2254                                 $paragraphStack = false;
2255
2256                                 # Close all the prefixes which aren't shared.
2257                                 while ( $commonPrefixLength < $lastPrefixLength ) {
2258                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2259                                         --$lastPrefixLength;
2260                                 }
2261
2262                                 # Continue the current prefix if appropriate.
2263                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2264                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2265                                 }
2266
2267                                 # Open prefixes where appropriate.
2268                                 while ( $prefixLength > $commonPrefixLength ) {
2269                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2270                                         $output .= $this->openList( $char );
2271
2272                                         if ( ';' === $char ) {
2273                                                 # @todo FIXME: This is dupe of code above
2274                                                 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2275                                                         $t = $t2;
2276                                                         $output .= $term . $this->nextItem( ':' );
2277                                                 }
2278                                         }
2279                                         ++$commonPrefixLength;
2280                                 }
2281                                 $lastPrefix = $prefix2;
2282                         }
2283
2284                         # If we have no prefixes, go to paragraph mode.
2285                         if ( 0 == $prefixLength ) {
2286                                 wfProfileIn( __METHOD__."-paragraph" );
2287                                 # No prefix (not in list)--go to paragraph mode
2288                                 # XXX: use a stack for nestable elements like span, table and div
2289                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2290                                 $closematch = preg_match(
2291                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2292                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2293                                 if ( $openmatch or $closematch ) {
2294                                         $paragraphStack = false;
2295                                         # TODO bug 5718: paragraph closed
2296                                         $output .= $this->closeParagraph();
2297                                         if ( $preOpenMatch and !$preCloseMatch ) {
2298                                                 $this->mInPre = true;
2299                                         }
2300                                         $inBlockElem = !$closematch;
2301                                 } elseif ( !$inBlockElem && !$this->mInPre ) {
2302                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) ) {
2303                                                 # pre
2304                                                 if ( $this->mLastSection !== 'pre' ) {
2305                                                         $paragraphStack = false;
2306                                                         $output .= $this->closeParagraph().'<pre>';
2307                                                         $this->mLastSection = 'pre';
2308                                                 }
2309                                                 $t = substr( $t, 1 );
2310                                         } else {
2311                                                 # paragraph
2312                                                 if ( trim( $t ) === '' ) {
2313                                                         if ( $paragraphStack ) {
2314                                                                 $output .= $paragraphStack.'<br />';
2315                                                                 $paragraphStack = false;
2316                                                                 $this->mLastSection = 'p';
2317                                                         } else {
2318                                                                 if ( $this->mLastSection !== 'p' ) {
2319                                                                         $output .= $this->closeParagraph();
2320                                                                         $this->mLastSection = '';
2321                                                                         $paragraphStack = '<p>';
2322                                                                 } else {
2323                                                                         $paragraphStack = '</p><p>';
2324                                                                 }
2325                                                         }
2326                                                 } else {
2327                                                         if ( $paragraphStack ) {
2328                                                                 $output .= $paragraphStack;
2329                                                                 $paragraphStack = false;
2330                                                                 $this->mLastSection = 'p';
2331                                                         } elseif ( $this->mLastSection !== 'p' ) {
2332                                                                 $output .= $this->closeParagraph().'<p>';
2333                                                                 $this->mLastSection = 'p';
2334                                                         }
2335                                                 }
2336                                         }
2337                                 }
2338                                 wfProfileOut( __METHOD__."-paragraph" );
2339                         }
2340                         # somewhere above we forget to get out of pre block (bug 785)
2341                         if ( $preCloseMatch && $this->mInPre ) {
2342                                 $this->mInPre = false;
2343                         }
2344                         if ( $paragraphStack === false ) {
2345                                 $output .= $t."\n";
2346                         }
2347                 }
2348                 while ( $prefixLength ) {
2349                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2350                         --$prefixLength;
2351                 }
2352                 if ( $this->mLastSection != '' ) {
2353                         $output .= '</' . $this->mLastSection . '>';
2354                         $this->mLastSection = '';
2355                 }
2356
2357                 wfProfileOut( __METHOD__ );
2358                 return $output;
2359         }
2360
2361         /**
2362          * Split up a string on ':', ignoring any occurences inside tags
2363          * to prevent illegal overlapping.
2364          *
2365          * @param $str String the string to split
2366          * @param &$before String set to everything before the ':'
2367          * @param &$after String set to everything after the ':'
2368          * @return String the position of the ':', or false if none found
2369          */
2370         function findColonNoLinks( $str, &$before, &$after ) {
2371                 wfProfileIn( __METHOD__ );
2372
2373                 $pos = strpos( $str, ':' );
2374                 if ( $pos === false ) {
2375                         # Nothing to find!
2376                         wfProfileOut( __METHOD__ );
2377                         return false;
2378                 }
2379
2380                 $lt = strpos( $str, '<' );
2381                 if ( $lt === false || $lt > $pos ) {
2382                         # Easy; no tag nesting to worry about
2383                         $before = substr( $str, 0, $pos );
2384                         $after = substr( $str, $pos+1 );
2385                         wfProfileOut( __METHOD__ );
2386                         return $pos;
2387                 }
2388
2389                 # Ugly state machine to walk through avoiding tags.
2390                 $state = self::COLON_STATE_TEXT;
2391                 $stack = 0;
2392                 $len = strlen( $str );
2393                 for( $i = 0; $i < $len; $i++ ) {
2394                         $c = $str[$i];
2395
2396                         switch( $state ) {
2397                         # (Using the number is a performance hack for common cases)
2398                         case 0: # self::COLON_STATE_TEXT:
2399                                 switch( $c ) {
2400                                 case "<":
2401                                         # Could be either a <start> tag or an </end> tag
2402                                         $state = self::COLON_STATE_TAGSTART;
2403                                         break;
2404                                 case ":":
2405                                         if ( $stack == 0 ) {
2406                                                 # We found it!
2407                                                 $before = substr( $str, 0, $i );
2408                                                 $after = substr( $str, $i + 1 );
2409                                                 wfProfileOut( __METHOD__ );
2410                                                 return $i;
2411                                         }
2412                                         # Embedded in a tag; don't break it.
2413                                         break;
2414                                 default:
2415                                         # Skip ahead looking for something interesting
2416                                         $colon = strpos( $str, ':', $i );
2417                                         if ( $colon === false ) {
2418                                                 # Nothing else interesting
2419                                                 wfProfileOut( __METHOD__ );
2420                                                 return false;
2421                                         }
2422                                         $lt = strpos( $str, '<', $i );
2423                                         if ( $stack === 0 ) {
2424                                                 if ( $lt === false || $colon < $lt ) {
2425                                                         # We found it!
2426                                                         $before = substr( $str, 0, $colon );
2427                                                         $after = substr( $str, $colon + 1 );
2428                                                         wfProfileOut( __METHOD__ );
2429                                                         return $i;
2430                                                 }
2431                                         }
2432                                         if ( $lt === false ) {
2433                                                 # Nothing else interesting to find; abort!
2434                                                 # We're nested, but there's no close tags left. Abort!
2435                                                 break 2;
2436                                         }
2437                                         # Skip ahead to next tag start
2438                                         $i = $lt;
2439                                         $state = self::COLON_STATE_TAGSTART;
2440                                 }
2441                                 break;
2442                         case 1: # self::COLON_STATE_TAG:
2443                                 # In a <tag>
2444                                 switch( $c ) {
2445                                 case ">":
2446                                         $stack++;
2447                                         $state = self::COLON_STATE_TEXT;
2448                                         break;
2449                                 case "/":
2450                                         # Slash may be followed by >?
2451                                         $state = self::COLON_STATE_TAGSLASH;
2452                                         break;
2453                                 default:
2454                                         # ignore
2455                                 }
2456                                 break;
2457                         case 2: # self::COLON_STATE_TAGSTART:
2458                                 switch( $c ) {
2459                                 case "/":
2460                                         $state = self::COLON_STATE_CLOSETAG;
2461                                         break;
2462                                 case "!":
2463                                         $state = self::COLON_STATE_COMMENT;
2464                                         break;
2465                                 case ">":
2466                                         # Illegal early close? This shouldn't happen D:
2467                                         $state = self::COLON_STATE_TEXT;
2468                                         break;
2469                                 default:
2470                                         $state = self::COLON_STATE_TAG;
2471                                 }
2472                                 break;
2473                         case 3: # self::COLON_STATE_CLOSETAG:
2474                                 # In a </tag>
2475                                 if ( $c === ">" ) {
2476                                         $stack--;
2477                                         if ( $stack < 0 ) {
2478                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2479                                                 wfProfileOut( __METHOD__ );
2480                                                 return false;
2481                                         }
2482                                         $state = self::COLON_STATE_TEXT;
2483                                 }
2484                                 break;
2485                         case self::COLON_STATE_TAGSLASH:
2486                                 if ( $c === ">" ) {
2487                                         # Yes, a self-closed tag <blah/>
2488                                         $state = self::COLON_STATE_TEXT;
2489                                 } else {
2490                                         # Probably we're jumping the gun, and this is an attribute
2491                                         $state = self::COLON_STATE_TAG;
2492                                 }
2493                                 break;
2494                         case 5: # self::COLON_STATE_COMMENT:
2495                                 if ( $c === "-" ) {
2496                                         $state = self::COLON_STATE_COMMENTDASH;
2497                                 }
2498                                 break;
2499                         case self::COLON_STATE_COMMENTDASH:
2500                                 if ( $c === "-" ) {
2501                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2502                                 } else {
2503                                         $state = self::COLON_STATE_COMMENT;
2504                                 }
2505                                 break;
2506                         case self::COLON_STATE_COMMENTDASHDASH:
2507                                 if ( $c === ">" ) {
2508                                         $state = self::COLON_STATE_TEXT;
2509                                 } else {
2510                                         $state = self::COLON_STATE_COMMENT;
2511                                 }
2512                                 break;
2513                         default:
2514                                 throw new MWException( "State machine error in " . __METHOD__ );
2515                         }
2516                 }
2517                 if ( $stack > 0 ) {
2518                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2519                         wfProfileOut( __METHOD__ );
2520                         return false;
2521                 }
2522                 wfProfileOut( __METHOD__ );
2523                 return false;
2524         }
2525
2526         /**
2527          * Return value of a magic variable (like PAGENAME)
2528          *
2529          * @private
2530          *
2531          * @param $index integer
2532          * @param $frame PPFrame
2533          *
2534          * @return string
2535          */
2536         function getVariableValue( $index, $frame = false ) {
2537                 global $wgContLang, $wgSitename, $wgServer;
2538                 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2539
2540                 if ( is_null( $this->mTitle ) ) {
2541                         // If no title set, bad things are going to happen
2542                         // later. Title should always be set since this
2543                         // should only be called in the middle of a parse
2544                         // operation (but the unit-tests do funky stuff)
2545                         throw new MWException( __METHOD__ . ' Should only be '
2546                                 . ' called while parsing (no title set)' );
2547                 }
2548
2549                 /**
2550                  * Some of these require message or data lookups and can be
2551                  * expensive to check many times.
2552                  */
2553                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2554                         if ( isset( $this->mVarCache[$index] ) ) {
2555                                 return $this->mVarCache[$index];
2556                         }
2557                 }
2558
2559                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2560                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2561
2562                 # Use the time zone
2563                 global $wgLocaltimezone;
2564                 if ( isset( $wgLocaltimezone ) ) {
2565                         $oldtz = date_default_timezone_get();
2566                         date_default_timezone_set( $wgLocaltimezone );
2567                 }
2568
2569                 $localTimestamp = date( 'YmdHis', $ts );
2570                 $localMonth = date( 'm', $ts );
2571                 $localMonth1 = date( 'n', $ts );
2572                 $localMonthName = date( 'n', $ts );
2573                 $localDay = date( 'j', $ts );
2574                 $localDay2 = date( 'd', $ts );
2575                 $localDayOfWeek = date( 'w', $ts );
2576                 $localWeek = date( 'W', $ts );
2577                 $localYear = date( 'Y', $ts );
2578                 $localHour = date( 'H', $ts );
2579                 if ( isset( $wgLocaltimezone ) ) {
2580                         date_default_timezone_set( $oldtz );
2581                 }
2582
2583                 $pageLang = $this->getFunctionLang();
2584
2585                 switch ( $index ) {
2586                         case 'currentmonth':
2587                                 $value = $pageLang->formatNum( gmdate( 'm', $ts ) );
2588                                 break;
2589                         case 'currentmonth1':
2590                                 $value = $pageLang->formatNum( gmdate( 'n', $ts ) );
2591                                 break;
2592                         case 'currentmonthname':
2593                                 $value = $pageLang->getMonthName( gmdate( 'n', $ts ) );
2594                                 break;
2595                         case 'currentmonthnamegen':
2596                                 $value = $pageLang->getMonthNameGen( gmdate( 'n', $ts ) );
2597                                 break;
2598                         case 'currentmonthabbrev':
2599                                 $value = $pageLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2600                                 break;
2601                         case 'currentday':
2602                                 $value = $pageLang->formatNum( gmdate( 'j', $ts ) );
2603                                 break;
2604                         case 'currentday2':
2605                                 $value = $pageLang->formatNum( gmdate( 'd', $ts ) );
2606                                 break;
2607                         case 'localmonth':
2608                                 $value = $pageLang->formatNum( $localMonth );
2609                                 break;
2610                         case 'localmonth1':
2611                                 $value = $pageLang->formatNum( $localMonth1 );
2612                                 break;
2613                         case 'localmonthname':
2614                                 $value = $pageLang->getMonthName( $localMonthName );
2615                                 break;
2616                         case 'localmonthnamegen':
2617                                 $value = $pageLang->getMonthNameGen( $localMonthName );
2618                                 break;
2619                         case 'localmonthabbrev':
2620                                 $value = $pageLang->getMonthAbbreviation( $localMonthName );
2621                                 break;
2622                         case 'localday':
2623                                 $value = $pageLang->formatNum( $localDay );
2624                                 break;
2625                         case 'localday2':
2626                                 $value = $pageLang->formatNum( $localDay2 );
2627                                 break;
2628                         case 'pagename':
2629                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2630                                 break;
2631                         case 'pagenamee':
2632                                 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2633                                 break;
2634                         case 'fullpagename':
2635                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2636                                 break;
2637                         case 'fullpagenamee':
2638                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2639                                 break;
2640                         case 'subpagename':
2641                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2642                                 break;
2643                         case 'subpagenamee':
2644                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2645                                 break;
2646                         case 'basepagename':
2647                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2648                                 break;
2649                         case 'basepagenamee':
2650                                 $value = wfEscapeWikiText( wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) ) );
2651                                 break;
2652                         case 'talkpagename':
2653                                 if ( $this->mTitle->canTalk() ) {
2654                                         $talkPage = $this->mTitle->getTalkPage();
2655                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2656                                 } else {
2657                                         $value = '';
2658                                 }
2659                                 break;
2660                         case 'talkpagenamee':
2661                                 if ( $this->mTitle->canTalk() ) {
2662                                         $talkPage = $this->mTitle->getTalkPage();
2663                                         $value = wfEscapeWikiText( $talkPage->getPrefixedUrl() );
2664                                 } else {
2665                                         $value = '';
2666                                 }
2667                                 break;
2668                         case 'subjectpagename':
2669                                 $subjPage = $this->mTitle->getSubjectPage();
2670                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2671                                 break;
2672                         case 'subjectpagenamee':
2673                                 $subjPage = $this->mTitle->getSubjectPage();
2674                                 $value = wfEscapeWikiText( $subjPage->getPrefixedUrl() );
2675                                 break;
2676                         case 'revisionid':
2677                                 # Let the edit saving system know we should parse the page
2678                                 # *after* a revision ID has been assigned.
2679                                 $this->mOutput->setFlag( 'vary-revision' );
2680                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2681                                 $value = $this->mRevisionId;
2682                                 break;
2683                         case 'revisionday':
2684                                 # Let the edit saving system know we should parse the page
2685                                 # *after* a revision ID has been assigned. This is for null edits.
2686                                 $this->mOutput->setFlag( 'vary-revision' );
2687                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2688                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2689                                 break;
2690                         case 'revisionday2':
2691                                 # Let the edit saving system know we should parse the page
2692                                 # *after* a revision ID has been assigned. This is for null edits.
2693                                 $this->mOutput->setFlag( 'vary-revision' );
2694                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2695                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2696                                 break;
2697                         case 'revisionmonth':
2698                                 # Let the edit saving system know we should parse the page
2699                                 # *after* a revision ID has been assigned. This is for null edits.
2700                                 $this->mOutput->setFlag( 'vary-revision' );
2701                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2702                                 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2703                                 break;
2704                         case 'revisionmonth1':
2705                                 # Let the edit saving system know we should parse the page
2706                                 # *after* a revision ID has been assigned. This is for null edits.
2707                                 $this->mOutput->setFlag( 'vary-revision' );
2708                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2709                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2710                                 break;
2711                         case 'revisionyear':
2712                                 # Let the edit saving system know we should parse the page
2713                                 # *after* a revision ID has been assigned. This is for null edits.
2714                                 $this->mOutput->setFlag( 'vary-revision' );
2715                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2716                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2717                                 break;
2718                         case 'revisiontimestamp':
2719                                 # Let the edit saving system know we should parse the page
2720                                 # *after* a revision ID has been assigned. This is for null edits.
2721                                 $this->mOutput->setFlag( 'vary-revision' );
2722                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2723                                 $value = $this->getRevisionTimestamp();
2724                                 break;
2725                         case 'revisionuser':
2726                                 # Let the edit saving system know we should parse the page
2727                                 # *after* a revision ID has been assigned. This is for null edits.
2728                                 $this->mOutput->setFlag( 'vary-revision' );
2729                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2730                                 $value = $this->getRevisionUser();
2731                                 break;
2732                         case 'namespace':
2733                                 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2734                                 break;
2735                         case 'namespacee':
2736                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2737                                 break;
2738                         case 'talkspace':
2739                                 $value = $this->mTitle->canTalk() ? str_replace( '_',' ',$this->mTitle->getTalkNsText() ) : '';
2740                                 break;
2741                         case 'talkspacee':
2742                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2743                                 break;
2744                         case 'subjectspace':
2745                                 $value = $this->mTitle->getSubjectNsText();
2746                                 break;
2747                         case 'subjectspacee':
2748                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2749                                 break;
2750                         case 'currentdayname':
2751                                 $value = $pageLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2752                                 break;
2753                         case 'currentyear':
2754                                 $value = $pageLang->formatNum( gmdate( 'Y', $ts ), true );
2755                                 break;
2756                         case 'currenttime':
2757                                 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2758                                 break;
2759                         case 'currenthour':
2760                                 $value = $pageLang->formatNum( gmdate( 'H', $ts ), true );
2761                                 break;
2762                         case 'currentweek':
2763                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2764                                 # int to remove the padding
2765                                 $value = $pageLang->formatNum( (int)gmdate( 'W', $ts ) );
2766                                 break;
2767                         case 'currentdow':
2768                                 $value = $pageLang->formatNum( gmdate( 'w', $ts ) );
2769                                 break;
2770                         case 'localdayname':
2771                                 $value = $pageLang->getWeekdayName( $localDayOfWeek + 1 );
2772                                 break;
2773                         case 'localyear':
2774                                 $value = $pageLang->formatNum( $localYear, true );
2775                                 break;
2776                         case 'localtime':
2777                                 $value = $pageLang->time( $localTimestamp, false, false );
2778                                 break;
2779                         case 'localhour':
2780                                 $value = $pageLang->formatNum( $localHour, true );
2781                                 break;
2782                         case 'localweek':
2783                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2784                                 # int to remove the padding
2785                                 $value = $pageLang->formatNum( (int)$localWeek );
2786                                 break;
2787                         case 'localdow':
2788                                 $value = $pageLang->formatNum( $localDayOfWeek );
2789                                 break;
2790                         case 'numberofarticles':
2791                                 $value = $pageLang->formatNum( SiteStats::articles() );
2792                                 break;
2793                         case 'numberoffiles':
2794                                 $value = $pageLang->formatNum( SiteStats::images() );
2795                                 break;
2796                         case 'numberofusers':
2797                                 $value = $pageLang->formatNum( SiteStats::users() );
2798                                 break;
2799                         case 'numberofactiveusers':
2800                                 $value = $pageLang->formatNum( SiteStats::activeUsers() );
2801                                 break;
2802                         case 'numberofpages':
2803                                 $value = $pageLang->formatNum( SiteStats::pages() );
2804                                 break;
2805                         case 'numberofadmins':
2806                                 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2807                                 break;
2808                         case 'numberofedits':
2809                                 $value = $pageLang->formatNum( SiteStats::edits() );
2810                                 break;
2811                         case 'numberofviews':
2812                                 $value = $pageLang->formatNum( SiteStats::views() );
2813                                 break;
2814                         case 'currenttimestamp':
2815                                 $value = wfTimestamp( TS_MW, $ts );
2816                                 break;
2817                         case 'localtimestamp':
2818                                 $value = $localTimestamp;
2819                                 break;
2820                         case 'currentversion':
2821                                 $value = SpecialVersion::getVersion();
2822                                 break;
2823                         case 'articlepath':
2824                                 return $wgArticlePath;
2825                         case 'sitename':
2826                                 return $wgSitename;
2827                         case 'server':
2828                                 return $wgServer;
2829                         case 'servername':
2830                                 wfSuppressWarnings(); # May give an E_WARNING in PHP < 5.3.3
2831                                 $serverName = parse_url( $wgServer, PHP_URL_HOST );
2832                                 wfRestoreWarnings();
2833                                 return $serverName ? $serverName : $wgServer;
2834                         case 'scriptpath':
2835                                 return $wgScriptPath;
2836                         case 'stylepath':
2837                                 return $wgStylePath;
2838                         case 'directionmark':
2839                                 return $pageLang->getDirMark();
2840                         case 'contentlanguage':
2841                                 global $wgLanguageCode;
2842                                 return $wgLanguageCode;
2843                         default:
2844                                 $ret = null;
2845                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
2846                                         return $ret;
2847                                 } else {
2848                                         return null;
2849                                 }
2850                 }
2851
2852                 if ( $index ) {
2853                         $this->mVarCache[$index] = $value;
2854                 }
2855
2856                 return $value;
2857         }
2858
2859         /**
2860          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2861          *
2862          * @private
2863          */
2864         function initialiseVariables() {
2865                 wfProfileIn( __METHOD__ );
2866                 $variableIDs = MagicWord::getVariableIDs();
2867                 $substIDs = MagicWord::getSubstIDs();
2868
2869                 $this->mVariables = new MagicWordArray( $variableIDs );
2870                 $this->mSubstWords = new MagicWordArray( $substIDs );
2871                 wfProfileOut( __METHOD__ );
2872         }
2873
2874         /**
2875          * Preprocess some wikitext and return the document tree.
2876          * This is the ghost of replace_variables().
2877          *
2878          * @param $text String: The text to parse
2879          * @param $flags Integer: bitwise combination of:
2880          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2881          *                                     included. Default is to assume a direct page view.
2882          *
2883          * The generated DOM tree must depend only on the input text and the flags.
2884          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2885          *
2886          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2887          * change in the DOM tree for a given text, must be passed through the section identifier
2888          * in the section edit link and thus back to extractSections().
2889          *
2890          * The output of this function is currently only cached in process memory, but a persistent
2891          * cache may be implemented at a later date which takes further advantage of these strict
2892          * dependency requirements.
2893          *
2894          * @private
2895          *
2896          * @return PPNode
2897          */
2898         function preprocessToDom( $text, $flags = 0 ) {
2899                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2900                 return $dom;
2901         }
2902
2903         /**
2904          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2905          *
2906          * @param $s string
2907          *
2908          * @return array
2909          */
2910         public static function splitWhitespace( $s ) {
2911                 $ltrimmed = ltrim( $s );
2912                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2913                 $trimmed = rtrim( $ltrimmed );
2914                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2915                 if ( $diff > 0 ) {
2916                         $w2 = substr( $ltrimmed, -$diff );
2917                 } else {
2918                         $w2 = '';
2919                 }
2920                 return array( $w1, $trimmed, $w2 );
2921         }
2922
2923         /**
2924          * Replace magic variables, templates, and template arguments
2925          * with the appropriate text. Templates are substituted recursively,
2926          * taking care to avoid infinite loops.
2927          *
2928          * Note that the substitution depends on value of $mOutputType:
2929          *  self::OT_WIKI: only {{subst:}} templates
2930          *  self::OT_PREPROCESS: templates but not extension tags
2931          *  self::OT_HTML: all templates and extension tags
2932          *
2933          * @param $text String the text to transform
2934          * @param $frame PPFrame Object describing the arguments passed to the template.
2935          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2936          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2937          * @param $argsOnly Boolean only do argument (triple-brace) expansion, not double-brace expansion
2938          * @private
2939          *
2940          * @return string
2941          */
2942         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2943                 # Is there any text? Also, Prevent too big inclusions!
2944                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2945                         return $text;
2946                 }
2947                 wfProfileIn( __METHOD__ );
2948
2949                 if ( $frame === false ) {
2950                         $frame = $this->getPreprocessor()->newFrame();
2951                 } elseif ( !( $frame instanceof PPFrame ) ) {
2952                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2953                         $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2954                 }
2955
2956                 $dom = $this->preprocessToDom( $text );
2957                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2958                 $text = $frame->expand( $dom, $flags );
2959
2960                 wfProfileOut( __METHOD__ );
2961                 return $text;
2962         }
2963
2964         /**
2965          * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2966          *
2967          * @param $args array
2968          *
2969          * @return array
2970          */
2971         static function createAssocArgs( $args ) {
2972                 $assocArgs = array();
2973                 $index = 1;
2974                 foreach ( $args as $arg ) {
2975                         $eqpos = strpos( $arg, '=' );
2976                         if ( $eqpos === false ) {
2977                                 $assocArgs[$index++] = $arg;
2978                         } else {
2979                                 $name = trim( substr( $arg, 0, $eqpos ) );
2980                                 $value = trim( substr( $arg, $eqpos+1 ) );
2981                                 if ( $value === false ) {
2982                                         $value = '';
2983                                 }
2984                                 if ( $name !== false ) {
2985                                         $assocArgs[$name] = $value;
2986                                 }
2987                         }
2988                 }
2989
2990                 return $assocArgs;
2991         }
2992
2993         /**
2994          * Warn the user when a parser limitation is reached
2995          * Will warn at most once the user per limitation type
2996          *
2997          * @param $limitationType String: should be one of:
2998          *   'expensive-parserfunction' (corresponding messages:
2999          *       'expensive-parserfunction-warning',
3000          *       'expensive-parserfunction-category')
3001          *   'post-expand-template-argument' (corresponding messages:
3002          *       'post-expand-template-argument-warning',
3003          *       'post-expand-template-argument-category')
3004          *   'post-expand-template-inclusion' (corresponding messages:
3005          *       'post-expand-template-inclusion-warning',
3006          *       'post-expand-template-inclusion-category')
3007          * @param $current Current value
3008          * @param $max Maximum allowed, when an explicit limit has been
3009          *       exceeded, provide the values (optional)
3010          */
3011         function limitationWarn( $limitationType, $current=null, $max=null) {
3012                 # does no harm if $current and $max are present but are unnecessary for the message
3013                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
3014                 $this->mOutput->addWarning( $warning );
3015                 $this->addTrackingCategory( "$limitationType-category" );
3016         }
3017
3018         /**
3019          * Return the text of a template, after recursively
3020          * replacing any variables or templates within the template.
3021          *
3022          * @param $piece Array: the parts of the template
3023          *  $piece['title']: the title, i.e. the part before the |
3024          *  $piece['parts']: the parameter array
3025          *  $piece['lineStart']: whether the brace was at the start of a line
3026          * @param $frame PPFrame The current frame, contains template arguments
3027          * @return String: the text of the template
3028          * @private
3029          */
3030         function braceSubstitution( $piece, $frame ) {
3031                 global $wgContLang, $wgNonincludableNamespaces, $wgEnableInterwikiTranscluding, $wgEnableInterwikiTemplatesTracking;
3032                 wfProfileIn( __METHOD__ );
3033                 wfProfileIn( __METHOD__.'-setup' );
3034
3035                 # Flags
3036                 $found = false;             # $text has been filled
3037                 $nowiki = false;            # wiki markup in $text should be escaped
3038                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
3039                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
3040                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
3041
3042                 # Title object, where $text came from
3043                 $title = null;
3044
3045                 # $part1 is the bit before the first |, and must contain only title characters.
3046                 # Various prefixes will be stripped from it later.
3047                 $titleWithSpaces = $frame->expand( $piece['title'] );
3048                 $part1 = trim( $titleWithSpaces );
3049                 $titleText = false;
3050
3051                 # Original title text preserved for various purposes
3052                 $originalTitle = $part1;
3053
3054                 # $args is a list of argument nodes, starting from index 0, not including $part1
3055                 # @todo FIXME: If piece['parts'] is null then the call to getLength() below won't work b/c this $args isn't an object
3056                 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
3057                 wfProfileOut( __METHOD__.'-setup' );
3058                 wfProfileIn( __METHOD__."-title-$originalTitle" );
3059
3060                 # SUBST
3061                 wfProfileIn( __METHOD__.'-modifiers' );
3062                 if ( !$found ) {
3063
3064                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3065
3066                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
3067                         # Decide whether to expand template or keep wikitext as-is.
3068                         if ( $this->ot['wiki'] ) {
3069                                 if ( $substMatch === false ) {
3070                                         $literal = true;  # literal when in PST with no prefix
3071                                 } else {
3072                                         $literal = false; # expand when in PST with subst: or safesubst:
3073                                 }
3074                         } else {
3075                                 if ( $substMatch == 'subst' ) {
3076                                         $literal = true;  # literal when not in PST with plain subst:
3077                                 } else {
3078                                         $literal = false; # expand when not in PST with safesubst: or no prefix
3079                                 }
3080                         }
3081                         if ( $literal ) {
3082                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3083                                 $isLocalObj = true;
3084                                 $found = true;
3085                         }
3086                 }
3087
3088                 # Variables
3089                 if ( !$found && $args->getLength() == 0 ) {
3090                         $id = $this->mVariables->matchStartToEnd( $part1 );
3091                         if ( $id !== false ) {
3092                                 $text = $this->getVariableValue( $id, $frame );
3093                                 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3094                                         $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3095                                 }
3096                                 $found = true;
3097                         }
3098                 }
3099
3100                 # MSG, MSGNW and RAW
3101                 if ( !$found ) {
3102                         # Check for MSGNW:
3103                         $mwMsgnw = MagicWord::get( 'msgnw' );
3104                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3105                                 $nowiki = true;
3106                         } else {
3107                                 # Remove obsolete MSG:
3108                                 $mwMsg = MagicWord::get( 'msg' );
3109                                 $mwMsg->matchStartAndRemove( $part1 );
3110                         }
3111
3112                         # Check for RAW:
3113                         $mwRaw = MagicWord::get( 'raw' );
3114                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3115                                 $forceRawInterwiki = true;
3116                         }
3117                 }
3118                 wfProfileOut( __METHOD__.'-modifiers' );
3119
3120                 # Parser functions
3121                 if ( !$found ) {
3122                         wfProfileIn( __METHOD__ . '-pfunc' );
3123
3124                         $colonPos = strpos( $part1, ':' );
3125                         if ( $colonPos !== false ) {
3126                                 # Case sensitive functions
3127                                 $function = substr( $part1, 0, $colonPos );
3128                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3129                                         $function = $this->mFunctionSynonyms[1][$function];
3130                                 } else {
3131                                         # Case insensitive functions
3132                                         $function = $wgContLang->lc( $function );
3133                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3134                                                 $function = $this->mFunctionSynonyms[0][$function];
3135                                         } else {
3136                                                 $function = false;
3137                                         }
3138                                 }
3139                                 if ( $function ) {
3140                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3141                                         $initialArgs = array( &$this );
3142                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3143                                         if ( $flags & SFH_OBJECT_ARGS ) {
3144                                                 # Add a frame parameter, and pass the arguments as an array
3145                                                 $allArgs = $initialArgs;
3146                                                 $allArgs[] = $frame;
3147                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3148                                                         $funcArgs[] = $args->item( $i );
3149                                                 }
3150                                                 $allArgs[] = $funcArgs;
3151                                         } else {
3152                                                 # Convert arguments to plain text
3153                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3154                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
3155                                                 }
3156                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3157                                         }
3158
3159                                         # Workaround for PHP bug 35229 and similar
3160                                         if ( !is_callable( $callback ) ) {
3161                                                 wfProfileOut( __METHOD__ . '-pfunc' );
3162                                                 wfProfileOut( __METHOD__ );
3163                                                 throw new MWException( "Tag hook for $function is not callable\n" );
3164                                         }
3165                                         $result = call_user_func_array( $callback, $allArgs );
3166                                         $found = true;
3167                                         $noparse = true;
3168                                         $preprocessFlags = 0;
3169
3170                                         if ( is_array( $result ) ) {
3171                                                 if ( isset( $result[0] ) ) {
3172                                                         $text = $result[0];
3173                                                         unset( $result[0] );
3174                                                 }
3175
3176                                                 # Extract flags into the local scope
3177                                                 # This allows callers to set flags such as nowiki, found, etc.
3178                                                 extract( $result );
3179                                         } else {
3180                                                 $text = $result;
3181                                         }
3182                                         if ( !$noparse ) {
3183                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
3184                                                 $isChildObj = true;
3185                                         }
3186                                 }
3187                         }
3188                         wfProfileOut( __METHOD__ . '-pfunc' );
3189                 }
3190
3191                 # Finish mangling title and then check for loops.
3192                 # Set $title to a Title object and $titleText to the PDBK
3193                 if ( !$found ) {
3194                         $ns = NS_TEMPLATE;
3195                         # Split the title into page and subpage
3196                         $subpage = '';
3197                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3198                         if ( $subpage !== '' ) {
3199                                 $ns = $this->mTitle->getNamespace();
3200                         }
3201                         $title = Title::newFromText( $part1, $ns );
3202                         if ( $title ) {
3203                                 if ( !$title->isExternal() && $piece['interwiki'] !== '' ) {
3204                                         $title->setInterwiki( $piece['interwiki'] );
3205                                 }
3206                                 $titleText = $title->getPrefixedText();
3207                                 # Check for language variants if the template is not found
3208                                 if ( $wgContLang->hasVariants() && $title->getArticleID() == 0 ) {
3209                                         $wgContLang->findVariantLink( $part1, $title, true );
3210                                 }
3211                                 # Do recursion depth check
3212                                 $limit = $this->mOptions->getMaxTemplateDepth();
3213                                 if ( $frame->depth >= $limit ) {
3214                                         $found = true;
3215                                         $text = '<span class="error">'
3216                                                 . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit )
3217                                                 . '</span>';
3218                                 }
3219                         }
3220                 }
3221
3222                 # Load from database
3223                 if ( !$found && $title ) {
3224                         wfProfileIn( __METHOD__ . '-loadtpl' );
3225                         if ( !$title->isExternal() ) {
3226                                 if ( $title->getNamespace() == NS_SPECIAL
3227                                         && $this->mOptions->getAllowSpecialInclusion()
3228                                         && $this->ot['html'] )
3229                                 {
3230                                         $pageArgs = array();
3231                                         for ( $i = 0; $i < $args->getLength(); $i++ ) {
3232                                                 $bits = $args->item( $i )->splitArg();
3233                                                 if ( strval( $bits['index'] ) === '' ) {
3234                                                         $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3235                                                         $value = trim( $frame->expand( $bits['value'] ) );
3236                                                         $pageArgs[$name] = $value;
3237                                                 }
3238                                         }
3239                                         $context = new RequestContext;
3240                                         $context->setTitle( $title );
3241                                         $context->setRequest( new FauxRequest( $pageArgs ) );
3242                                         $context->setUser( $this->getUser() );
3243                                         $context->setLang( Language::factory( $this->mOptions->getUserLang() ) );
3244                                         $ret = SpecialPageFactory::capturePath( $title, $context );
3245                                         if ( $ret ) {
3246                                                 $text = $context->getOutput()->getHTML();
3247                                                 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3248                                                 $found = true;
3249                                                 $isHTML = true;
3250                                                 $this->disableCache();
3251                                         }
3252                                 } elseif ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3253                                         $found = false; # access denied
3254                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3255                                 } else {
3256                                         list( $text, $title ) = $this->getTemplateDom( $title );
3257                                         if ( $text !== false ) {
3258                                                 $found = true;
3259                                                 $isChildObj = true;
3260                                         }
3261                                 }
3262
3263                                 # If the title is valid but undisplayable, make a link to it
3264                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3265                                         $text = "[[:$titleText]]";
3266                                         $found = true;
3267                                 }
3268                         } elseif ( $wgEnableInterwikiTranscluding && $title->isTrans() ) {
3269
3270                                 $text = Interwiki::interwikiTransclude( $title );
3271                                 $this->registerDistantTemplate( $title );
3272
3273                                 if ( $wgEnableInterwikiTemplatesTracking ) {
3274                                         $this->registerDistantTemplate( $title );
3275                                 }
3276
3277                                 if ( $text !== false ) {
3278                                         # Preprocess it like a template
3279                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3280                                         $found = true;
3281                                         $isChildObj = true;
3282                                 }
3283
3284                         }
3285
3286                         # Do infinite loop check
3287                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3288                         if ( !$frame->loopCheck( $title ) ) {
3289                                 $found = true;
3290                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3291                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3292                         }
3293                         wfProfileOut( __METHOD__ . '-loadtpl' );
3294                 }
3295
3296                 # If we haven't found text to substitute by now, we're done
3297                 # Recover the source wikitext and return it
3298                 if ( !$found ) {
3299                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3300                         wfProfileOut( __METHOD__."-title-$originalTitle" );
3301                         wfProfileOut( __METHOD__ );
3302                         return array( 'object' => $text );
3303                 }
3304
3305                 # Expand DOM-style return values in a child frame
3306                 if ( $isChildObj ) {
3307                         # Clean up argument array
3308                         $newFrame = $frame->newChild( $args, $title );
3309
3310                         if ( $nowiki ) {
3311                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3312                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3313                                 # Expansion is eligible for the empty-frame cache
3314                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3315                                         $text = $this->mTplExpandCache[$titleText];
3316                                 } else {
3317                                         $text = $newFrame->expand( $text );
3318                                         $this->mTplExpandCache[$titleText] = $text;
3319                                 }
3320                         } else {
3321                                 # Uncached expansion
3322                                 $text = $newFrame->expand( $text );
3323                         }
3324                 }
3325                 if ( $isLocalObj && $nowiki ) {
3326                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3327                         $isLocalObj = false;
3328                 }
3329
3330                 # Replace raw HTML by a placeholder
3331                 # Add a blank line preceding, to prevent it from mucking up
3332                 # immediately preceding headings
3333                 if ( $isHTML ) {
3334                         $text = "\n\n" . $this->insertStripItem( $text );
3335                 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3336                         # Escape nowiki-style return values
3337                         $text = wfEscapeWikiText( $text );
3338                 } elseif ( is_string( $text )
3339                         && !$piece['lineStart']
3340                         && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3341                 {
3342                         # Bug 529: if the template begins with a table or block-level
3343                         # element, it should be treated as beginning a new line.
3344                         # This behaviour is somewhat controversial.
3345                         $text = "\n" . $text;
3346                 }
3347
3348                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3349                         # Error, oversize inclusion
3350                         if ( $titleText !== false ) {
3351                                 # Make a working, properly escaped link if possible (bug 23588)
3352                                 $text = "[[:$titleText]]";
3353                         } else {
3354                                 # This will probably not be a working link, but at least it may
3355                                 # provide some hint of where the problem is
3356                                 preg_replace( '/^:/', '', $originalTitle );
3357                                 $text = "[[:$originalTitle]]";
3358                         }
3359                         $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3360                         $this->limitationWarn( 'post-expand-template-inclusion' );
3361                 }
3362
3363                 if ( $isLocalObj ) {
3364                         $ret = array( 'object' => $text );
3365                 } else {
3366                         $ret = array( 'text' => $text );
3367                 }
3368
3369                 wfProfileOut( __METHOD__."-title-$originalTitle" );
3370                 wfProfileOut( __METHOD__ );
3371                 return $ret;
3372         }
3373
3374         /**
3375          * Get the semi-parsed DOM representation of a template with a given title,
3376          * and its redirect destination title. Cached.
3377          *
3378          * @param $title Title
3379          *
3380          * @return array
3381          */
3382         function getTemplateDom( $title ) {
3383                 $cacheTitle = $title;
3384                 $titleText = $title->getPrefixedDBkey();
3385
3386                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3387                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3388                         $title = Title::makeTitle( $ns, $dbk );
3389                         $titleText = $title->getPrefixedDBkey();
3390                 }
3391                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3392                         return array( $this->mTplDomCache[$titleText], $title );
3393                 }
3394
3395                 # Cache miss, go to the database
3396                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3397
3398                 if ( $text === false ) {
3399                         $this->mTplDomCache[$titleText] = false;
3400                         return array( false, $title );
3401                 }
3402
3403                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3404                 $this->mTplDomCache[ $titleText ] = $dom;
3405
3406                 if ( !$title->equals( $cacheTitle ) ) {
3407                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3408                                 array( $title->getNamespace(), $cdb = $title->getDBkey() );
3409                 }
3410
3411                 return array( $dom, $title );
3412         }
3413
3414         /**
3415          * Fetch the unparsed text of a template and register a reference to it.
3416          * @param Title $title
3417          * @return Array ( string or false, Title )
3418          */
3419         function fetchTemplateAndTitle( $title ) {
3420                 $templateCb = $this->mOptions->getTemplateCallback(); # Defaults to Parser::statelessFetchTemplate()
3421                 $stuff = call_user_func( $templateCb, $title, $this );
3422                 $text = $stuff['text'];
3423                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3424                 if ( isset( $stuff['deps'] ) ) {
3425                         foreach ( $stuff['deps'] as $dep ) {
3426                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3427                         }
3428                 }
3429                 return array( $text, $finalTitle );
3430         }
3431
3432         /**
3433          * Register a distant template as used
3434          */
3435         function registerDistantTemplate( $title ) {
3436                 $stuff = Parser::distantTemplateCallback( $title, $this );
3437                 $text = $stuff['text'];
3438                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3439                 if ( isset( $stuff['deps'] ) ) {
3440                         foreach ( $stuff['deps'] as $dep ) {
3441                                 $this->mOutput->addDistantTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3442                         }
3443                 }
3444         }
3445
3446         function fetchTemplate( $title ) {
3447                 $rv = $this->fetchTemplateAndTitle( $title );
3448                 return $rv[0];
3449         }
3450
3451         /**
3452          * Static function to get a template
3453          * Can be overridden via ParserOptions::setTemplateCallback().
3454          *
3455          * @parma $title Title
3456          * @param $parser Parser
3457          *
3458          * @return array
3459          */
3460         static function statelessFetchTemplate( $title, $parser = false ) {
3461                 $text = $skip = false;
3462                 $finalTitle = $title;
3463                 $deps = array();
3464
3465                 # Loop to fetch the article, with up to 1 redirect
3466                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3467                         # Give extensions a chance to select the revision instead
3468                         $id = false; # Assume current
3469                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle',
3470                                 array( $parser, $title, &$skip, &$id ) );
3471
3472                         if ( $skip ) {
3473                                 $text = false;
3474                                 $deps[] = array(
3475                                         'title'         => $title,
3476                                         'page_id'       => $title->getArticleID(),
3477                                         'rev_id'        => null
3478                                 );
3479                                 break;
3480                         }
3481                         # Get the revision
3482                         $rev = $id
3483                                 ? Revision::newFromId( $id )
3484                                 : Revision::newFromTitle( $title );
3485                         $rev_id = $rev ? $rev->getId() : 0;
3486                         # If there is no current revision, there is no page
3487                         if ( $id === false && !$rev ) {
3488                                 $linkCache = LinkCache::singleton();
3489                                 $linkCache->addBadLinkObj( $title );
3490                         }
3491
3492                         $deps[] = array(
3493                                 'title'         => $title,
3494                                 'page_id'       => $title->getArticleID(),
3495                                 'rev_id'        => $rev_id );
3496                         if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3497                                 # We fetched a rev from a different title; register it too...
3498                                 $deps[] = array(
3499                                         'title'         => $rev->getTitle(),
3500                                         'page_id'       => $rev->getPage(),
3501                                         'rev_id'        => $rev_id );
3502                         }
3503
3504                         if ( $rev ) {
3505                                 $text = $rev->getText();
3506                         } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3507                                 global $wgContLang;
3508                                 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3509                                 if ( !$message->exists() ) {
3510                                         $text = false;
3511                                         break;
3512                                 }
3513                                 $text = $message->plain();
3514                         } else {
3515                                 break;
3516                         }
3517                         if ( $text === false ) {
3518                                 break;
3519                         }
3520                         # Redirect?
3521                         $finalTitle = $title;
3522                         $title = Title::newFromRedirect( $text );
3523                 }
3524                 return array(
3525                         'text' => $text,
3526                         'finalTitle' => $finalTitle,
3527                         'deps' => $deps );
3528         }
3529
3530         /**
3531          * Fetch a file and its title and register a reference to it.
3532          * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3533          * @param Title $title
3534          * @param Array $options Array of options to RepoGroup::findFile
3535          * @return File|false
3536          */
3537         function fetchFile( $title, $options = array() ) {
3538                 $res = $this->fetchFileAndTitle( $title, $options );
3539                 return $res[0];
3540         }
3541
3542         /**
3543          * Fetch a file and its title and register a reference to it.
3544          * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3545          * @param Title $title
3546          * @param Array $options Array of options to RepoGroup::findFile
3547          * @return Array ( File or false, Title of file )
3548          */
3549         function fetchFileAndTitle( $title, $options = array() ) {
3550                 if ( isset( $options['broken'] ) ) {
3551                         $file = false; // broken thumbnail forced by hook
3552                 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3553                         $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3554                 } else { // get by (name,timestamp)
3555                         $file = wfFindFile( $title, $options );
3556                 }
3557                 $time = $file ? $file->getTimestamp() : false;
3558                 $sha1 = $file ? $file->getSha1() : false;
3559                 # Register the file as a dependency...
3560                 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3561                 if ( $file && !$title->equals( $file->getTitle() ) ) {
3562                         # Update fetched file title
3563                         $title = $file->getTitle();
3564                         if ( is_null( $file->getRedirectedTitle() ) ) {
3565                                 # This file was not a redirect, but the title does not match.
3566                                 # Register under the new name because otherwise the link will
3567                                 # get lost.
3568                                 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3569                         }
3570                 }
3571                 return array( $file, $title );
3572         }
3573
3574         static function distantTemplateCallback( $title, $parser=false ) {
3575                 $text = '';
3576                 $rev_id = null;
3577                 $deps[] = array(
3578                         'title' => $title,
3579                         'page_id' => $title->getArticleID(),
3580                         'rev_id' => $rev_id );
3581
3582                 $finalTitle = $title;
3583
3584                 return array(
3585                         'text' => $text,
3586                         'finalTitle' => $finalTitle,
3587                         'deps' => $deps );
3588                 }
3589
3590         /**
3591          * Triple brace replacement -- used for template arguments
3592          * @private
3593          *
3594          * @param $peice array
3595          * @param $frame PPFrame
3596          *
3597          * @return array
3598          */
3599         function argSubstitution( $piece, $frame ) {
3600                 wfProfileIn( __METHOD__ );
3601
3602                 $error = false;
3603                 $parts = $piece['parts'];
3604                 $nameWithSpaces = $frame->expand( $piece['title'] );
3605                 $argName = trim( $nameWithSpaces );
3606                 $object = false;
3607                 $text = $frame->getArgument( $argName );
3608                 if (  $text === false && $parts->getLength() > 0
3609                   && (
3610                         $this->ot['html']
3611                         || $this->ot['pre']
3612                         || ( $this->ot['wiki'] && $frame->isTemplate() )
3613                   )
3614                 ) {
3615                         # No match in frame, use the supplied default
3616                         $object = $parts->item( 0 )->getChildren();
3617                 }
3618                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3619                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3620                         $this->limitationWarn( 'post-expand-template-argument' );
3621                 }
3622
3623                 if ( $text === false && $object === false ) {
3624                         # No match anywhere
3625                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3626                 }
3627                 if ( $error !== false ) {
3628                         $text .= $error;
3629                 }
3630                 if ( $object !== false ) {
3631                         $ret = array( 'object' => $object );
3632                 } else {
3633                         $ret = array( 'text' => $text );
3634                 }
3635
3636                 wfProfileOut( __METHOD__ );
3637                 return $ret;
3638         }
3639
3640         /**
3641          * Return the text to be used for a given extension tag.
3642          * This is the ghost of strip().
3643          *
3644          * @param $params Associative array of parameters:
3645          *     name       PPNode for the tag name
3646          *     attr       PPNode for unparsed text where tag attributes are thought to be
3647          *     attributes Optional associative array of parsed attributes
3648          *     inner      Contents of extension element
3649          *     noClose    Original text did not have a close tag
3650          * @param $frame PPFrame
3651          *
3652          * @return string
3653          */
3654         function extensionSubstitution( $params, $frame ) {
3655                 $name = $frame->expand( $params['name'] );
3656                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3657                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3658                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3659
3660                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3661                         ( $this->ot['html'] || $this->ot['pre'] );
3662                 if ( $isFunctionTag ) {
3663                         $markerType = 'none';
3664                 } else {
3665                         $markerType = 'general';
3666                 }
3667                 if ( $this->ot['html'] || $isFunctionTag ) {
3668                         $name = strtolower( $name );
3669                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3670                         if ( isset( $params['attributes'] ) ) {
3671                                 $attributes = $attributes + $params['attributes'];
3672                         }
3673
3674                         if ( isset( $this->mTagHooks[$name] ) ) {
3675                                 # Workaround for PHP bug 35229 and similar
3676                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3677                                         throw new MWException( "Tag hook for $name is not callable\n" );
3678                                 }
3679                                 $output = call_user_func_array( $this->mTagHooks[$name],
3680                                         array( $content, $attributes, $this, $frame ) );
3681                         } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3682                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3683                                 if ( !is_callable( $callback ) ) {
3684                                         throw new MWException( "Tag hook for $name is not callable\n" );
3685                                 }
3686
3687                                 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3688                         } else {
3689                                 $output = '<span class="error">Invalid tag extension name: ' .
3690                                         htmlspecialchars( $name ) . '</span>';
3691                         }
3692
3693                         if ( is_array( $output ) ) {
3694                                 # Extract flags to local scope (to override $markerType)
3695                                 $flags = $output;
3696                                 $output = $flags[0];
3697                                 unset( $flags[0] );
3698                                 extract( $flags );
3699                         }
3700                 } else {
3701                         if ( is_null( $attrText ) ) {
3702                                 $attrText = '';
3703                         }
3704                         if ( isset( $params['attributes'] ) ) {
3705                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3706                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3707                                                 htmlspecialchars( $attrValue ) . '"';
3708                                 }
3709                         }
3710                         if ( $content === null ) {
3711                                 $output = "<$name$attrText/>";
3712                         } else {
3713                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3714                                 $output = "<$name$attrText>$content$close";
3715                         }
3716                 }
3717
3718                 if ( $markerType === 'none' ) {
3719                         return $output;
3720                 } elseif ( $markerType === 'nowiki' ) {
3721                         $this->mStripState->addNoWiki( $marker, $output );
3722                 } elseif ( $markerType === 'general' ) {
3723                         $this->mStripState->addGeneral( $marker, $output );
3724                 } else {
3725                         throw new MWException( __METHOD__.': invalid marker type' );
3726                 }
3727                 return $marker;
3728         }
3729
3730         /**
3731          * Increment an include size counter
3732          *
3733          * @param $type String: the type of expansion
3734          * @param $size Integer: the size of the text
3735          * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
3736          */
3737         function incrementIncludeSize( $type, $size ) {
3738                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3739                         return false;
3740                 } else {
3741                         $this->mIncludeSizes[$type] += $size;
3742                         return true;
3743                 }
3744         }
3745
3746         /**
3747          * Increment the expensive function count
3748          *
3749          * @return Boolean: false if the limit has been exceeded
3750          */
3751         function incrementExpensiveFunctionCount() {
3752                 global $wgExpensiveParserFunctionLimit;
3753                 $this->mExpensiveFunctionCount++;
3754                 if ( $this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit ) {
3755                         return true;
3756                 }
3757                 return false;
3758         }
3759
3760         /**
3761          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3762          * Fills $this->mDoubleUnderscores, returns the modified text
3763          *
3764          * @param $text string
3765          *
3766          * @return string
3767          */
3768         function doDoubleUnderscore( $text ) {
3769                 wfProfileIn( __METHOD__ );
3770
3771                 # The position of __TOC__ needs to be recorded
3772                 $mw = MagicWord::get( 'toc' );
3773                 if ( $mw->match( $text ) ) {
3774                         $this->mShowToc = true;
3775                         $this->mForceTocPosition = true;
3776
3777                         # Set a placeholder. At the end we'll fill it in with the TOC.
3778                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3779
3780                         # Only keep the first one.
3781                         $text = $mw->replace( '', $text );
3782                 }
3783
3784                 # Now match and remove the rest of them
3785                 $mwa = MagicWord::getDoubleUnderscoreArray();
3786                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3787
3788                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3789                         $this->mOutput->mNoGallery = true;
3790                 }
3791                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3792                         $this->mShowToc = false;
3793                 }
3794                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3795                         $this->addTrackingCategory( 'hidden-category-category' );
3796                 }
3797                 # (bug 8068) Allow control over whether robots index a page.
3798                 #
3799                 # @todo FIXME: Bug 14899: __INDEX__ always overrides __NOINDEX__ here!  This
3800                 # is not desirable, the last one on the page should win.
3801                 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3802                         $this->mOutput->setIndexPolicy( 'noindex' );
3803                         $this->addTrackingCategory( 'noindex-category' );
3804                 }
3805                 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3806                         $this->mOutput->setIndexPolicy( 'index' );
3807                         $this->addTrackingCategory( 'index-category' );
3808                 }
3809
3810                 # Cache all double underscores in the database
3811                 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3812                         $this->mOutput->setProperty( $key, '' );
3813                 }
3814
3815                 wfProfileOut( __METHOD__ );
3816                 return $text;
3817         }
3818
3819         /**
3820          * Add a tracking category, getting the title from a system message,
3821          * or print a debug message if the title is invalid.
3822          *
3823          * @param $msg String: message key
3824          * @return Boolean: whether the addition was successful
3825          */
3826         protected function addTrackingCategory( $msg ) {
3827                 if ( $this->mTitle->getNamespace() === NS_SPECIAL ) {
3828                         wfDebug( __METHOD__.": Not adding tracking category $msg to special page!\n" );
3829                         return false;
3830                 }
3831                 $cat = wfMsgForContent( $msg );
3832
3833                 # Allow tracking categories to be disabled by setting them to "-"
3834                 if ( $cat === '-' ) {
3835                         return false;
3836                 }
3837
3838                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3839                 if ( $containerCategory ) {
3840                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3841                         return true;
3842                 } else {
3843                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3844                         return false;
3845                 }
3846         }
3847
3848         /**
3849          * This function accomplishes several tasks:
3850          * 1) Auto-number headings if that option is enabled
3851          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3852          * 3) Add a Table of contents on the top for users who have enabled the option
3853          * 4) Auto-anchor headings
3854          *
3855          * It loops through all headlines, collects the necessary data, then splits up the
3856          * string and re-inserts the newly formatted headlines.
3857          *
3858          * @param $text String
3859          * @param $origText String: original, untouched wikitext
3860          * @param $isMain Boolean
3861          * @private
3862          */
3863         function formatHeadings( $text, $origText, $isMain=true ) {
3864                 global $wgMaxTocLevel, $wgHtml5, $wgExperimentalHtmlIds;
3865
3866                 # Inhibit editsection links if requested in the page
3867                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3868                         $showEditLink = 0;
3869                 } else {
3870                         $showEditLink = $this->mOptions->getEditSection();
3871                 }
3872                 if ( $showEditLink ) {
3873                         $this->mOutput->setEditSectionTokens( true );
3874                 }
3875
3876                 # Get all headlines for numbering them and adding funky stuff like [edit]
3877                 # links - this is for later, but we need the number of headlines right now
3878                 $matches = array();
3879                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3880
3881                 # if there are fewer than 4 headlines in the article, do not show TOC
3882                 # unless it's been explicitly enabled.
3883                 $enoughToc = $this->mShowToc &&
3884                         ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3885
3886                 # Allow user to stipulate that a page should have a "new section"
3887                 # link added via __NEWSECTIONLINK__
3888                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3889                         $this->mOutput->setNewSection( true );
3890                 }
3891
3892                 # Allow user to remove the "new section"
3893                 # link via __NONEWSECTIONLINK__
3894                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3895                         $this->mOutput->hideNewSection( true );
3896                 }
3897
3898                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3899                 # override above conditions and always show TOC above first header
3900                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3901                         $this->mShowToc = true;
3902                         $enoughToc = true;
3903                 }
3904
3905                 # headline counter
3906                 $headlineCount = 0;
3907                 $numVisible = 0;
3908
3909                 # Ugh .. the TOC should have neat indentation levels which can be
3910                 # passed to the skin functions. These are determined here
3911                 $toc = '';
3912                 $full = '';
3913                 $head = array();
3914                 $sublevelCount = array();
3915                 $levelCount = array();
3916                 $level = 0;
3917                 $prevlevel = 0;
3918                 $toclevel = 0;
3919                 $prevtoclevel = 0;
3920                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3921                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3922                 $oldType = $this->mOutputType;
3923                 $this->setOutputType( self::OT_WIKI );
3924                 $frame = $this->getPreprocessor()->newFrame();
3925                 $root = $this->preprocessToDom( $origText );
3926                 $node = $root->getFirstChild();
3927                 $byteOffset = 0;
3928                 $tocraw = array();
3929                 $refers = array();
3930
3931                 foreach ( $matches[3] as $headline ) {
3932                         $isTemplate = false;
3933                         $titleText = false;
3934                         $sectionIndex = false;
3935                         $numbering = '';
3936                         $markerMatches = array();
3937                         if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
3938                                 $serial = $markerMatches[1];
3939                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3940                                 $isTemplate = ( $titleText != $baseTitleText );
3941                                 $headline = preg_replace( "/^$markerRegex/", "", $headline );
3942                         }
3943
3944                         if ( $toclevel ) {
3945                                 $prevlevel = $level;
3946                         }
3947                         $level = $matches[1][$headlineCount];
3948
3949                         if ( $level > $prevlevel ) {
3950                                 # Increase TOC level
3951                                 $toclevel++;
3952                                 $sublevelCount[$toclevel] = 0;
3953                                 if ( $toclevel<$wgMaxTocLevel ) {
3954                                         $prevtoclevel = $toclevel;
3955                                         $toc .= Linker::tocIndent();
3956                                         $numVisible++;
3957                                 }
3958                         } elseif ( $level < $prevlevel && $toclevel > 1 ) {
3959                                 # Decrease TOC level, find level to jump to
3960
3961                                 for ( $i = $toclevel; $i > 0; $i-- ) {
3962                                         if ( $levelCount[$i] == $level ) {
3963                                                 # Found last matching level
3964                                                 $toclevel = $i;
3965                                                 break;
3966                                         } elseif ( $levelCount[$i] < $level ) {
3967                                                 # Found first matching level below current level
3968                                                 $toclevel = $i + 1;
3969                                                 break;
3970                                         }
3971                                 }
3972                                 if ( $i == 0 ) {
3973                                         $toclevel = 1;
3974                                 }
3975                                 if ( $toclevel<$wgMaxTocLevel ) {
3976                                         if ( $prevtoclevel < $wgMaxTocLevel ) {
3977                                                 # Unindent only if the previous toc level was shown :p
3978                                                 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
3979                                                 $prevtoclevel = $toclevel;
3980                                         } else {
3981                                                 $toc .= Linker::tocLineEnd();
3982                                         }
3983                                 }
3984                         } else {
3985                                 # No change in level, end TOC line
3986                                 if ( $toclevel<$wgMaxTocLevel ) {
3987                                         $toc .= Linker::tocLineEnd();
3988                                 }
3989                         }
3990
3991                         $levelCount[$toclevel] = $level;
3992
3993                         # count number of headlines for each level
3994                         @$sublevelCount[$toclevel]++;
3995                         $dot = 0;
3996                         for( $i = 1; $i <= $toclevel; $i++ ) {
3997                                 if ( !empty( $sublevelCount[$i] ) ) {
3998                                         if ( $dot ) {
3999                                                 $numbering .= '.';
4000                                         }
4001                                         $numbering .= $this->getFunctionLang()->formatNum( $sublevelCount[$i] );
4002                                         $dot = 1;
4003                                 }
4004                         }
4005
4006                         # The safe header is a version of the header text safe to use for links
4007                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4008                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
4009
4010                         # Remove link placeholders by the link text.
4011                         #     <!--LINK number-->
4012                         # turns into
4013                         #     link text with suffix
4014                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
4015
4016                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
4017                         $tocline = preg_replace(
4018                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
4019                                 array( '',                          '<$1>' ),
4020                                 $safeHeadline
4021                         );
4022                         $tocline = trim( $tocline );
4023
4024                         # For the anchor, strip out HTML-y stuff period
4025                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
4026                         $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4027
4028                         # Save headline for section edit hint before it's escaped
4029                         $headlineHint = $safeHeadline;
4030
4031                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
4032                                 # For reverse compatibility, provide an id that's
4033                                 # HTML4-compatible, like we used to.
4034                                 #
4035                                 # It may be worth noting, academically, that it's possible for
4036                                 # the legacy anchor to conflict with a non-legacy headline
4037                                 # anchor on the page.  In this case likely the "correct" thing
4038                                 # would be to either drop the legacy anchors or make sure
4039                                 # they're numbered first.  However, this would require people
4040                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4041                                 # manually, so let's not bother worrying about it.
4042                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
4043                                         array( 'noninitial', 'legacy' ) );
4044                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4045
4046                                 if ( $legacyHeadline == $safeHeadline ) {
4047                                         # No reason to have both (in fact, we can't)
4048                                         $legacyHeadline = false;
4049                                 }
4050                         } else {
4051                                 $legacyHeadline = false;
4052                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
4053                                         'noninitial' );
4054                         }
4055
4056                         # HTML names must be case-insensitively unique (bug 10721).
4057                         # This does not apply to Unicode characters per
4058                         # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
4059                         # @todo FIXME: We may be changing them depending on the current locale.
4060                         $arrayKey = strtolower( $safeHeadline );
4061                         if ( $legacyHeadline === false ) {
4062                                 $legacyArrayKey = false;
4063                         } else {
4064                                 $legacyArrayKey = strtolower( $legacyHeadline );
4065                         }
4066
4067                         # count how many in assoc. array so we can track dupes in anchors
4068                         if ( isset( $refers[$arrayKey] ) ) {
4069                                 $refers[$arrayKey]++;
4070                         } else {
4071                                 $refers[$arrayKey] = 1;
4072                         }
4073                         if ( isset( $refers[$legacyArrayKey] ) ) {
4074                                 $refers[$legacyArrayKey]++;
4075                         } else {
4076                                 $refers[$legacyArrayKey] = 1;
4077                         }
4078
4079                         # Don't number the heading if it is the only one (looks silly)
4080                         if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4081                                 # the two are different if the line contains a link
4082                                 $headline = $numbering . ' ' . $headline;
4083                         }
4084
4085                         # Create the anchor for linking from the TOC to the section
4086                         $anchor = $safeHeadline;
4087                         $legacyAnchor = $legacyHeadline;
4088                         if ( $refers[$arrayKey] > 1 ) {
4089                                 $anchor .= '_' . $refers[$arrayKey];
4090                         }
4091                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
4092                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
4093                         }
4094                         if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4095                                 $toc .= Linker::tocLine( $anchor, $tocline,
4096                                         $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4097                         }
4098
4099                         # Add the section to the section tree
4100                         # Find the DOM node for this header
4101                         while ( $node && !$isTemplate ) {
4102                                 if ( $node->getName() === 'h' ) {
4103                                         $bits = $node->splitHeading();
4104                                         if ( $bits['i'] == $sectionIndex ) {
4105                                                 break;
4106                                         }
4107                                 }
4108                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4109                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4110                                 $node = $node->getNextSibling();
4111                         }
4112                         $tocraw[] = array(
4113                                 'toclevel' => $toclevel,
4114                                 'level' => $level,
4115                                 'line' => $tocline,
4116                                 'number' => $numbering,
4117                                 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4118                                 'fromtitle' => $titleText,
4119                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
4120                                 'anchor' => $anchor,
4121                         );
4122
4123                         # give headline the correct <h#> tag
4124                         if ( $showEditLink && $sectionIndex !== false ) {
4125                                 // Output edit section links as markers with styles that can be customized by skins
4126                                 if ( $isTemplate ) {
4127                                         # Put a T flag in the section identifier, to indicate to extractSections()
4128                                         # that sections inside <includeonly> should be counted.
4129                                         $editlinkArgs = array( $titleText, "T-$sectionIndex"/*, null */ );
4130                                 } else {
4131                                         $editlinkArgs = array( $this->mTitle->getPrefixedText(), $sectionIndex, $headlineHint );
4132                                 }
4133                                 // We use a bit of pesudo-xml for editsection markers. The language converter is run later on
4134                                 // Using a UNIQ style marker leads to the converter screwing up the tokens when it converts stuff
4135                                 // And trying to insert strip tags fails too. At this point all real inputted tags have already been escaped
4136                                 // so we don't have to worry about a user trying to input one of these markers directly.
4137                                 // We use a page and section attribute to stop the language converter from converting these important bits
4138                                 // of data, but put the headline hint inside a content block because the language converter is supposed to
4139                                 // be able to convert that piece of data.
4140                                 $editlink = '<mw:editsection page="' . htmlspecialchars($editlinkArgs[0]);
4141                                 $editlink .= '" section="' . htmlspecialchars($editlinkArgs[1]) .'"';
4142                                 if ( isset($editlinkArgs[2]) ) {
4143                                         $editlink .= '>' . $editlinkArgs[2] . '</mw:editsection>';
4144                                 } else {
4145                                         $editlink .= '/>';
4146                                 }
4147                         } else {
4148                                 $editlink = '';
4149                         }
4150                         $head[$headlineCount] = Linker::makeHeadline( $level,
4151                                 $matches['attrib'][$headlineCount], $anchor, $headline,
4152                                 $editlink, $legacyAnchor );
4153
4154                         $headlineCount++;
4155                 }
4156
4157                 $this->setOutputType( $oldType );
4158
4159                 # Never ever show TOC if no headers
4160                 if ( $numVisible < 1 ) {
4161                         $enoughToc = false;
4162                 }
4163
4164                 if ( $enoughToc ) {
4165                         if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4166                                 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4167                         }
4168                         $toc = Linker::tocList( $toc, $this->mOptions->getUserLang() );
4169                         $this->mOutput->setTOCHTML( $toc );
4170                 }
4171
4172                 if ( $isMain ) {
4173                         $this->mOutput->setSections( $tocraw );
4174                 }
4175
4176                 # split up and insert constructed headlines
4177                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
4178                 $i = 0;
4179
4180                 // build an array of document sections
4181                 $sections = array();
4182                 foreach ( $blocks as $block ) {
4183                         // $head is zero-based, sections aren't.
4184                         if ( empty( $head[$i - 1] ) ) {
4185                                 $sections[$i] = $block;
4186                         } else {
4187                                 $sections[$i] = $head[$i - 1] . $block;
4188                         }
4189
4190                         /**
4191                          * Send a hook, one per section.
4192                          * The idea here is to be able to make section-level DIVs, but to do so in a
4193                          * lower-impact, more correct way than r50769
4194                          *
4195                          * $this : caller
4196                          * $section : the section number
4197                          * &$sectionContent : ref to the content of the section
4198                          * $showEditLinks : boolean describing whether this section has an edit link
4199                          */
4200                         wfRunHooks( 'ParserSectionCreate', array( $this, $i, &$sections[$i], $showEditLink ) );
4201
4202                         $i++;
4203                 }
4204
4205                 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4206                         // append the TOC at the beginning
4207                         // Top anchor now in skin
4208                         $sections[0] = $sections[0] . $toc . "\n";
4209                 }
4210
4211                 $full .= join( '', $sections );
4212
4213                 if ( $this->mForceTocPosition ) {
4214                         return str_replace( '<!--MWTOC-->', $toc, $full );
4215                 } else {
4216                         return $full;
4217                 }
4218         }
4219
4220         /**
4221          * Transform wiki markup when saving a page by doing \r\n -> \n
4222          * conversion, substitting signatures, {{subst:}} templates, etc.
4223          *
4224          * @param $text String: the text to transform
4225          * @param $title Title: the Title object for the current article
4226          * @param $user User: the User object describing the current user
4227          * @param $options ParserOptions: parsing options
4228          * @param $clearState Boolean: whether to clear the parser state first
4229          * @return String: the altered wiki markup
4230          */
4231         public function preSaveTransform( $text, Title $title, User $user, ParserOptions $options, $clearState = true ) {
4232                 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4233                 $this->setUser( $user );
4234
4235                 $pairs = array(
4236                         "\r\n" => "\n",
4237                 );
4238                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4239                 if( $options->getPreSaveTransform() ) {
4240                         $text = $this->pstPass2( $text, $user );
4241                 }
4242                 $text = $this->mStripState->unstripBoth( $text );
4243
4244                 $this->setUser( null ); #Reset
4245
4246                 return $text;
4247         }
4248
4249         /**
4250          * Pre-save transform helper function
4251          * @private
4252          *
4253          * @param $text string
4254          * @param $user User
4255          *
4256          * @return string
4257          */
4258         function pstPass2( $text, $user ) {
4259                 global $wgContLang, $wgLocaltimezone;
4260
4261                 # Note: This is the timestamp saved as hardcoded wikitext to
4262                 # the database, we use $wgContLang here in order to give
4263                 # everyone the same signature and use the default one rather
4264                 # than the one selected in each user's preferences.
4265                 # (see also bug 12815)
4266                 $ts = $this->mOptions->getTimestamp();
4267                 if ( isset( $wgLocaltimezone ) ) {
4268                         $tz = $wgLocaltimezone;
4269                 } else {
4270                         $tz = date_default_timezone_get();
4271                 }
4272
4273                 $unixts = wfTimestamp( TS_UNIX, $ts );
4274                 $oldtz = date_default_timezone_get();
4275                 date_default_timezone_set( $tz );
4276                 $ts = date( 'YmdHis', $unixts );
4277                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4278
4279                 # Allow translation of timezones through wiki. date() can return
4280                 # whatever crap the system uses, localised or not, so we cannot
4281                 # ship premade translations.
4282                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4283                 $msg = wfMessage( $key )->inContentLanguage();
4284                 if ( $msg->exists() ) {
4285                         $tzMsg = $msg->text();
4286                 }
4287
4288                 date_default_timezone_set( $oldtz );
4289
4290                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4291
4292                 # Variable replacement
4293                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4294                 $text = $this->replaceVariables( $text );
4295
4296                 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4297                 # which may corrupt this parser instance via its wfMsgExt( parsemag ) call-
4298
4299                 # Signatures
4300                 $sigText = $this->getUserSig( $user );
4301                 $text = strtr( $text, array(
4302                         '~~~~~' => $d,
4303                         '~~~~' => "$sigText $d",
4304                         '~~~' => $sigText
4305                 ) );
4306
4307                 # Context links: [[|name]] and [[name (context)|]]
4308                 global $wgLegalTitleChars;
4309                 $tc = "[$wgLegalTitleChars]";
4310                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4311
4312                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";           # [[ns:page (context)|]]
4313                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";           # [[ns:page（context）|]]
4314                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)(, $tc+|)\\|]]/"; # [[ns:page (context), context|]]
4315                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4316
4317                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4318                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4319                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4320                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4321
4322                 $t = $this->mTitle->getText();
4323                 $m = array();
4324                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4325                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4326                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4327                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4328                 } else {
4329                         # if there's no context, don't bother duplicating the title
4330                         $text = preg_replace( $p2, '[[\\1]]', $text );
4331                 }
4332
4333                 # Trim trailing whitespace
4334                 $text = rtrim( $text );
4335
4336                 return $text;
4337         }
4338
4339         /**
4340          * Fetch the user's signature text, if any, and normalize to
4341          * validated, ready-to-insert wikitext.
4342          * If you have pre-fetched the nickname or the fancySig option, you can
4343          * specify them here to save a database query.
4344          * Do not reuse this parser instance after calling getUserSig(),
4345          * as it may have changed if it's the $wgParser.
4346          *
4347          * @param $user User
4348          * @param $nickname String|bool nickname to use or false to use user's default nickname
4349          * @param $fancySig Boolean|null whether the nicknname is the complete signature
4350          *                  or null to use default value
4351          * @return string
4352          */
4353         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4354                 global $wgMaxSigChars;
4355
4356                 $username = $user->getName();
4357
4358                 # If not given, retrieve from the user object.
4359                 if ( $nickname === false )
4360                         $nickname = $user->getOption( 'nickname' );
4361
4362                 if ( is_null( $fancySig ) ) {
4363                         $fancySig = $user->getBoolOption( 'fancysig' );
4364                 }
4365
4366                 $nickname = $nickname == null ? $username : $nickname;
4367
4368                 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4369                         $nickname = $username;
4370                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4371                 } elseif ( $fancySig !== false ) {
4372                         # Sig. might contain markup; validate this
4373                         if ( $this->validateSig( $nickname ) !== false ) {
4374                                 # Validated; clean up (if needed) and return it
4375                                 return $this->cleanSig( $nickname, true );
4376                         } else {
4377                                 # Failed to validate; fall back to the default
4378                                 $nickname = $username;
4379                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4380                         }
4381                 }
4382
4383                 # Make sure nickname doesnt get a sig in a sig
4384                 $nickname = $this->cleanSigInSig( $nickname );
4385
4386                 # If we're still here, make it a link to the user page
4387                 $userText = wfEscapeWikiText( $username );
4388                 $nickText = wfEscapeWikiText( $nickname );
4389                 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4390
4391                 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()->title( $this->getTitle() )->text();
4392         }
4393
4394         /**
4395          * Check that the user's signature contains no bad XML
4396          *
4397          * @param $text String
4398          * @return mixed An expanded string, or false if invalid.
4399          */
4400         function validateSig( $text ) {
4401                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4402         }
4403
4404         /**
4405          * Clean up signature text
4406          *
4407          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4408          * 2) Substitute all transclusions
4409          *
4410          * @param $text String
4411          * @param $parsing bool Whether we're cleaning (preferences save) or parsing
4412          * @return String: signature text
4413          */
4414         function cleanSig( $text, $parsing = false ) {
4415                 if ( !$parsing ) {
4416                         global $wgTitle;
4417                         $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4418                 }
4419
4420                 # Option to disable this feature
4421                 if ( !$this->mOptions->getCleanSignatures() ) {
4422                         return $text;
4423                 }
4424
4425                 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4426                 #  => Move this logic to braceSubstitution()
4427                 $substWord = MagicWord::get( 'subst' );
4428                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4429                 $substText = '{{' . $substWord->getSynonym( 0 );
4430
4431                 $text = preg_replace( $substRegex, $substText, $text );
4432                 $text = $this->cleanSigInSig( $text );
4433                 $dom = $this->preprocessToDom( $text );
4434                 $frame = $this->getPreprocessor()->newFrame();
4435                 $text = $frame->expand( $dom );
4436
4437                 if ( !$parsing ) {
4438                         $text = $this->mStripState->unstripBoth( $text );
4439                 }
4440
4441                 return $text;
4442         }
4443
4444         /**
4445          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4446          *
4447          * @param $text String
4448          * @return String: signature text with /~{3,5}/ removed
4449          */
4450         function cleanSigInSig( $text ) {
4451                 $text = preg_replace( '/~{3,5}/', '', $text );
4452                 return $text;
4453         }
4454
4455         /**
4456          * Set up some variables which are usually set up in parse()
4457          * so that an external function can call some class members with confidence
4458          *
4459          * @param $title Title|null
4460          * @param $options ParserOptions
4461          * @param $outputType
4462          * @param $clearState bool
4463          */
4464         public function startExternalParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4465                 $this->startParse( $title, $options, $outputType, $clearState );
4466         }
4467
4468         /**
4469          * @param $title Title|null
4470          * @param $options ParserOptions
4471          * @param $outputType
4472          * @param $clearState bool
4473          */
4474         private function startParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4475                 $this->setTitle( $title );
4476                 $this->mOptions = $options;
4477                 $this->setOutputType( $outputType );
4478                 if ( $clearState ) {
4479                         $this->clearState();
4480                 }
4481         }
4482
4483         /**
4484          * Wrapper for preprocess()
4485          *
4486          * @param $text String: the text to preprocess
4487          * @param $options ParserOptions: options
4488          * @param $title Title object or null to use $wgTitle
4489          * @return String
4490          */
4491         public function transformMsg( $text, $options, $title = null ) {
4492                 static $executing = false;
4493
4494                 # Guard against infinite recursion
4495                 if ( $executing ) {
4496                         return $text;
4497                 }
4498                 $executing = true;
4499
4500                 wfProfileIn( __METHOD__ );
4501                 if ( !$title ) {
4502                         global $wgTitle;
4503                         $title = $wgTitle;
4504                 }
4505                 if ( !$title ) {
4506                         # It's not uncommon having a null $wgTitle in scripts. See r80898
4507                         # Create a ghost title in such case
4508                         $title = Title::newFromText( 'Dwimmerlaik' );
4509                 }
4510                 $text = $this->preprocess( $text, $title, $options );
4511
4512                 $executing = false;
4513                 wfProfileOut( __METHOD__ );
4514                 return $text;
4515         }
4516
4517         /**
4518          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4519          * The callback should have the following form:
4520          *    function myParserHook( $text, $params, $parser, $frame ) { ... }
4521          *
4522          * Transform and return $text. Use $parser for any required context, e.g. use
4523          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4524          *
4525          * Hooks may return extended information by returning an array, of which the
4526          * first numbered element (index 0) must be the return string, and all other
4527          * entries are extracted into local variables within an internal function
4528          * in the Parser class.
4529          *
4530          * This interface (introduced r61913) appears to be undocumented, but
4531          * 'markerName' is used by some core tag hooks to override which strip
4532          * array their results are placed in. **Use great caution if attempting
4533          * this interface, as it is not documented and injudicious use could smash
4534          * private variables.**
4535          *
4536          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4537          * @param $callback Mixed: the callback function (and object) to use for the tag
4538          * @return The old value of the mTagHooks array associated with the hook
4539          */
4540         public function setHook( $tag, $callback ) {
4541                 $tag = strtolower( $tag );
4542                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4543                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4544                 $this->mTagHooks[$tag] = $callback;
4545                 if ( !in_array( $tag, $this->mStripList ) ) {
4546                         $this->mStripList[] = $tag;
4547                 }
4548
4549                 return $oldVal;
4550         }
4551
4552         /**
4553          * As setHook(), but letting the contents be parsed.
4554          *
4555          * Transparent tag hooks are like regular XML-style tag hooks, except they
4556          * operate late in the transformation sequence, on HTML instead of wikitext.
4557          *
4558          * This is probably obsoleted by things dealing with parser frames?
4559          * The only extension currently using it is geoserver.
4560          *
4561          * @since 1.10
4562          * @todo better document or deprecate this
4563          *
4564          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4565          * @param $callback Mixed: the callback function (and object) to use for the tag
4566          * @return The old value of the mTagHooks array associated with the hook
4567          */
4568         function setTransparentTagHook( $tag, $callback ) {
4569                 $tag = strtolower( $tag );
4570                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4571                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4572                 $this->mTransparentTagHooks[$tag] = $callback;
4573
4574                 return $oldVal;
4575         }
4576
4577         /**
4578          * Remove all tag hooks
4579          */
4580         function clearTagHooks() {
4581                 $this->mTagHooks = array();
4582                 $this->mStripList = $this->mDefaultStripList;
4583         }
4584
4585         /**
4586          * Create a function, e.g. {{sum:1|2|3}}
4587          * The callback function should have the form:
4588          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4589          *
4590          * Or with SFH_OBJECT_ARGS:
4591          *    function myParserFunction( $parser, $frame, $args ) { ... }
4592          *
4593          * The callback may either return the text result of the function, or an array with the text
4594          * in element 0, and a number of flags in the other elements. The names of the flags are
4595          * specified in the keys. Valid flags are:
4596          *   found                     The text returned is valid, stop processing the template. This
4597          *                             is on by default.
4598          *   nowiki                    Wiki markup in the return value should be escaped
4599          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4600          *
4601          * @param $id String: The magic word ID
4602          * @param $callback Mixed: the callback function (and object) to use
4603          * @param $flags Integer: a combination of the following flags:
4604          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4605          *
4606          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4607          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4608          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4609          *     the arguments, and to control the way they are expanded.
4610          *
4611          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4612          *     arguments, for instance:
4613          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4614          *
4615          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4616          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4617          *     working if/when this is changed.
4618          *
4619          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4620          *     expansion.
4621          *
4622          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4623          *     about the methods available in PPFrame and PPNode.
4624          *
4625          * @return The old callback function for this name, if any
4626          */
4627         public function setFunctionHook( $id, $callback, $flags = 0 ) {
4628                 global $wgContLang;
4629
4630                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4631                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4632
4633                 # Add to function cache
4634                 $mw = MagicWord::get( $id );
4635                 if ( !$mw )
4636                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4637
4638                 $synonyms = $mw->getSynonyms();
4639                 $sensitive = intval( $mw->isCaseSensitive() );
4640
4641                 foreach ( $synonyms as $syn ) {
4642                         # Case
4643                         if ( !$sensitive ) {
4644                                 $syn = $wgContLang->lc( $syn );
4645                         }
4646                         # Add leading hash
4647                         if ( !( $flags & SFH_NO_HASH ) ) {
4648                                 $syn = '#' . $syn;
4649                         }
4650                         # Remove trailing colon
4651                         if ( substr( $syn, -1, 1 ) === ':' ) {
4652                                 $syn = substr( $syn, 0, -1 );
4653                         }
4654                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4655                 }
4656                 return $oldVal;
4657         }
4658
4659         /**
4660          * Get all registered function hook identifiers
4661          *
4662          * @return Array
4663          */
4664         function getFunctionHooks() {
4665                 return array_keys( $this->mFunctionHooks );
4666         }
4667
4668         /**
4669          * Create a tag function, e.g. <test>some stuff</test>.
4670          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4671          * Unlike parser functions, their content is not preprocessed.
4672          */
4673         function setFunctionTagHook( $tag, $callback, $flags ) {
4674                 $tag = strtolower( $tag );
4675                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4676                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4677                         $this->mFunctionTagHooks[$tag] : null;
4678                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4679
4680                 if ( !in_array( $tag, $this->mStripList ) ) {
4681                         $this->mStripList[] = $tag;
4682                 }
4683
4684                 return $old;
4685         }
4686
4687         /**
4688          * @todo FIXME: Update documentation. makeLinkObj() is deprecated.
4689          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4690          * Placeholders created in Skin::makeLinkObj()
4691          *
4692          * @param $text string
4693          * @param $options int
4694          *
4695          * @return array of link CSS classes, indexed by PDBK.
4696          */
4697         function replaceLinkHolders( &$text, $options = 0 ) {
4698                 return $this->mLinkHolders->replace( $text );
4699         }
4700
4701         /**
4702          * Replace <!--LINK--> link placeholders with plain text of links
4703          * (not HTML-formatted).
4704          *
4705          * @param $text String
4706          * @return String
4707          */
4708         function replaceLinkHoldersText( $text ) {
4709                 return $this->mLinkHolders->replaceText( $text );
4710         }
4711
4712         /**
4713          * Renders an image gallery from a text with one line per image.
4714          * text labels may be given by using |-style alternative text. E.g.
4715          *   Image:one.jpg|The number "1"
4716          *   Image:tree.jpg|A tree
4717          * given as text will return the HTML of a gallery with two images,
4718          * labeled 'The number "1"' and
4719          * 'A tree'.
4720          *
4721          * @param string $text
4722          * @param array $params
4723          * @return string HTML
4724          */
4725         function renderImageGallery( $text, $params ) {
4726                 $ig = new ImageGallery();
4727                 $ig->setContextTitle( $this->mTitle );
4728                 $ig->setShowBytes( false );
4729                 $ig->setShowFilename( false );
4730                 $ig->setParser( $this );
4731                 $ig->setHideBadImages();
4732                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4733
4734                 if ( isset( $params['showfilename'] ) ) {
4735                         $ig->setShowFilename( true );
4736                 } else {
4737                         $ig->setShowFilename( false );
4738                 }
4739                 if ( isset( $params['caption'] ) ) {
4740                         $caption = $params['caption'];
4741                         $caption = htmlspecialchars( $caption );
4742                         $caption = $this->replaceInternalLinks( $caption );
4743                         $ig->setCaptionHtml( $caption );
4744                 }
4745                 if ( isset( $params['perrow'] ) ) {
4746                         $ig->setPerRow( $params['perrow'] );
4747                 }
4748                 if ( isset( $params['widths'] ) ) {
4749                         $ig->setWidths( $params['widths'] );
4750                 }
4751                 if ( isset( $params['heights'] ) ) {
4752                         $ig->setHeights( $params['heights'] );
4753                 }
4754
4755                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4756
4757                 $lines = StringUtils::explode( "\n", $text );
4758                 foreach ( $lines as $line ) {
4759                         # match lines like these:
4760                         # Image:someimage.jpg|This is some image
4761                         $matches = array();
4762                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4763                         # Skip empty lines
4764                         if ( count( $matches ) == 0 ) {
4765                                 continue;
4766                         }
4767
4768                         if ( strpos( $matches[0], '%' ) !== false ) {
4769                                 $matches[1] = rawurldecode( $matches[1] );
4770                         }
4771                         $title = Title::newFromText( $matches[1], NS_FILE );
4772                         if ( is_null( $title ) ) {
4773                                 # Bogus title. Ignore these so we don't bomb out later.
4774                                 continue;
4775                         }
4776
4777                         $label = '';
4778                         $alt = '';
4779                         if ( isset( $matches[3] ) ) {
4780                                 // look for an |alt= definition while trying not to break existing
4781                                 // captions with multiple pipes (|) in it, until a more sensible grammar
4782                                 // is defined for images in galleries
4783
4784                                 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
4785                                 $altmatches = StringUtils::explode('|', $matches[3]);
4786                                 $magicWordAlt = MagicWord::get( 'img_alt' );
4787
4788                                 foreach ( $altmatches as $altmatch ) {
4789                                         $match = $magicWordAlt->matchVariableStartToEnd( $altmatch );
4790                                         if ( $match ) {
4791                                                 $alt = $this->stripAltText( $match, false );
4792                                         }
4793                                         else {
4794                                                 // concatenate all other pipes
4795                                                 $label .= '|' . $altmatch;
4796                                         }
4797                                 }
4798                                 // remove the first pipe
4799                                 $label = substr( $label, 1 );
4800                         }
4801
4802                         $ig->add( $title, $label, $alt );
4803                 }
4804                 return $ig->toHTML();
4805         }
4806
4807         /**
4808          * @param $handler
4809          * @return array
4810          */
4811         function getImageParams( $handler ) {
4812                 if ( $handler ) {
4813                         $handlerClass = get_class( $handler );
4814                 } else {
4815                         $handlerClass = '';
4816                 }
4817                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4818                         # Initialise static lists
4819                         static $internalParamNames = array(
4820                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4821                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4822                                         'bottom', 'text-bottom' ),
4823                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4824                                         'upright', 'border', 'link', 'alt' ),
4825                         );
4826                         static $internalParamMap;
4827                         if ( !$internalParamMap ) {
4828                                 $internalParamMap = array();
4829                                 foreach ( $internalParamNames as $type => $names ) {
4830                                         foreach ( $names as $name ) {
4831                                                 $magicName = str_replace( '-', '_', "img_$name" );
4832                                                 $internalParamMap[$magicName] = array( $type, $name );
4833                                         }
4834                                 }
4835                         }
4836
4837                         # Add handler params
4838                         $paramMap = $internalParamMap;
4839                         if ( $handler ) {
4840                                 $handlerParamMap = $handler->getParamMap();
4841                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4842                                         $paramMap[$magic] = array( 'handler', $paramName );
4843                                 }
4844                         }
4845                         $this->mImageParams[$handlerClass] = $paramMap;
4846                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4847                 }
4848                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4849         }
4850
4851         /**
4852          * Parse image options text and use it to make an image
4853          *
4854          * @param $title Title
4855          * @param $options String
4856          * @param $holders LinkHolderArray|false
4857          * @return string HTML
4858          */
4859         function makeImage( $title, $options, $holders = false ) {
4860                 # Check if the options text is of the form "options|alt text"
4861                 # Options are:
4862                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4863                 #  * left       no resizing, just left align. label is used for alt= only
4864                 #  * right      same, but right aligned
4865                 #  * none       same, but not aligned
4866                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4867                 #  * center     center the image
4868                 #  * frame      Keep original image size, no magnify-button.
4869                 #  * framed     Same as "frame"
4870                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4871                 #  * upright    reduce width for upright images, rounded to full __0 px
4872                 #  * border     draw a 1px border around the image
4873                 #  * alt        Text for HTML alt attribute (defaults to empty)
4874                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4875                 # vertical-align values (no % or length right now):
4876                 #  * baseline
4877                 #  * sub
4878                 #  * super
4879                 #  * top
4880                 #  * text-top
4881                 #  * middle
4882                 #  * bottom
4883                 #  * text-bottom
4884
4885                 $parts = StringUtils::explode( "|", $options );
4886
4887                 # Give extensions a chance to select the file revision for us
4888                 $options = array();
4889                 $descQuery = false;
4890                 wfRunHooks( 'BeforeParserFetchFileAndTitle',
4891                         array( $this, $title, &$options, &$descQuery ) );
4892                 # Fetch and register the file (file title may be different via hooks)
4893                 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
4894
4895                 # Get parameter map
4896                 $handler = $file ? $file->getHandler() : false;
4897
4898                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4899
4900                 if ( !$file ) {
4901                         $this->addTrackingCategory( 'broken-file-category' );
4902                 }
4903
4904                 # Process the input parameters
4905                 $caption = '';
4906                 $params = array( 'frame' => array(), 'handler' => array(),
4907                         'horizAlign' => array(), 'vertAlign' => array() );
4908                 foreach ( $parts as $part ) {
4909                         $part = trim( $part );
4910                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4911                         $validated = false;
4912                         if ( isset( $paramMap[$magicName] ) ) {
4913                                 list( $type, $paramName ) = $paramMap[$magicName];
4914
4915                                 # Special case; width and height come in one variable together
4916                                 if ( $type === 'handler' && $paramName === 'width' ) {
4917                                         $m = array();
4918                                         # (bug 13500) In both cases (width/height and width only),
4919                                         # permit trailing "px" for backward compatibility.
4920                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4921                                                 $width = intval( $m[1] );
4922                                                 $height = intval( $m[2] );
4923                                                 if ( $handler->validateParam( 'width', $width ) ) {
4924                                                         $params[$type]['width'] = $width;
4925                                                         $validated = true;
4926                                                 }
4927                                                 if ( $handler->validateParam( 'height', $height ) ) {
4928                                                         $params[$type]['height'] = $height;
4929                                                         $validated = true;
4930                                                 }
4931                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4932                                                 $width = intval( $value );
4933                                                 if ( $handler->validateParam( 'width', $width ) ) {
4934                                                         $params[$type]['width'] = $width;
4935                                                         $validated = true;
4936                                                 }
4937                                         } # else no validation -- bug 13436
4938                                 } else {
4939                                         if ( $type === 'handler' ) {
4940                                                 # Validate handler parameter
4941                                                 $validated = $handler->validateParam( $paramName, $value );
4942                                         } else {
4943                                                 # Validate internal parameters
4944                                                 switch( $paramName ) {
4945                                                 case 'manualthumb':
4946                                                 case 'alt':
4947                                                         # @todo FIXME: Possibly check validity here for
4948                                                         # manualthumb? downstream behavior seems odd with
4949                                                         # missing manual thumbs.
4950                                                         $validated = true;
4951                                                         $value = $this->stripAltText( $value, $holders );
4952                                                         break;
4953                                                 case 'link':
4954                                                         $chars = self::EXT_LINK_URL_CLASS;
4955                                                         $prots = $this->mUrlProtocols;
4956                                                         if ( $value === '' ) {
4957                                                                 $paramName = 'no-link';
4958                                                                 $value = true;
4959                                                                 $validated = true;
4960                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4961                                                                 if ( preg_match( "/^($prots)$chars+$/u", $value, $m ) ) {
4962                                                                         $paramName = 'link-url';
4963                                                                         $this->mOutput->addExternalLink( $value );
4964                                                                         if ( $this->mOptions->getExternalLinkTarget() ) {
4965                                                                                 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
4966                                                                         }
4967                                                                         $validated = true;
4968                                                                 }
4969                                                         } else {
4970                                                                 $linkTitle = Title::newFromText( $value );
4971                                                                 if ( $linkTitle ) {
4972                                                                         $paramName = 'link-title';
4973                                                                         $value = $linkTitle;
4974                                                                         $this->mOutput->addLink( $linkTitle );
4975                                                                         $validated = true;
4976                                                                 }
4977                                                         }
4978                                                         break;
4979                                                 default:
4980                                                         # Most other things appear to be empty or numeric...
4981                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4982                                                 }
4983                                         }
4984
4985                                         if ( $validated ) {
4986                                                 $params[$type][$paramName] = $value;
4987                                         }
4988                                 }
4989                         }
4990                         if ( !$validated ) {
4991                                 $caption = $part;
4992                         }
4993                 }
4994
4995                 # Process alignment parameters
4996                 if ( $params['horizAlign'] ) {
4997                         $params['frame']['align'] = key( $params['horizAlign'] );
4998                 }
4999                 if ( $params['vertAlign'] ) {
5000                         $params['frame']['valign'] = key( $params['vertAlign'] );
5001                 }
5002
5003                 $params['frame']['caption'] = $caption;
5004
5005                 # Will the image be presented in a frame, with the caption below?
5006                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
5007                                                  isset( $params['frame']['framed'] ) ||
5008                                                  isset( $params['frame']['thumbnail'] ) ||
5009                                                  isset( $params['frame']['manualthumb'] );
5010
5011                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
5012                 # came to also set the caption, ordinary text after the image -- which
5013                 # makes no sense, because that just repeats the text multiple times in
5014                 # screen readers.  It *also* came to set the title attribute.
5015                 #
5016                 # Now that we have an alt attribute, we should not set the alt text to
5017                 # equal the caption: that's worse than useless, it just repeats the
5018                 # text.  This is the framed/thumbnail case.  If there's no caption, we
5019                 # use the unnamed parameter for alt text as well, just for the time be-
5020                 # ing, if the unnamed param is set and the alt param is not.
5021                 #
5022                 # For the future, we need to figure out if we want to tweak this more,
5023                 # e.g., introducing a title= parameter for the title; ignoring the un-
5024                 # named parameter entirely for images without a caption; adding an ex-
5025                 # plicit caption= parameter and preserving the old magic unnamed para-
5026                 # meter for BC; ...
5027                 if ( $imageIsFramed ) { # Framed image
5028                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5029                                 # No caption or alt text, add the filename as the alt text so
5030                                 # that screen readers at least get some description of the image
5031                                 $params['frame']['alt'] = $title->getText();
5032                         }
5033                         # Do not set $params['frame']['title'] because tooltips don't make sense
5034                         # for framed images
5035                 } else { # Inline image
5036                         if ( !isset( $params['frame']['alt'] ) ) {
5037                                 # No alt text, use the "caption" for the alt text
5038                                 if ( $caption !== '') {
5039                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5040                                 } else {
5041                                         # No caption, fall back to using the filename for the
5042                                         # alt text
5043                                         $params['frame']['alt'] = $title->getText();
5044                                 }
5045                         }
5046                         # Use the "caption" for the tooltip text
5047                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5048                 }
5049
5050                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
5051
5052                 # Linker does the rest
5053                 $time = isset( $options['time'] ) ? $options['time'] : false;
5054                 $ret = Linker::makeImageLink2( $title, $file, $params['frame'], $params['handler'],
5055                         $time, $descQuery, $this->mOptions->getThumbSize() );
5056
5057                 # Give the handler a chance to modify the parser object
5058                 if ( $handler ) {
5059                         $handler->parserTransformHook( $this, $file );
5060                 }
5061
5062                 return $ret;
5063         }
5064
5065         /**
5066          * @param $caption
5067          * @param $holders LinkHolderArray
5068          * @return mixed|String
5069          */
5070         protected function stripAltText( $caption, $holders ) {
5071                 # Strip bad stuff out of the title (tooltip).  We can't just use
5072                 # replaceLinkHoldersText() here, because if this function is called
5073                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5074                 if ( $holders ) {
5075                         $tooltip = $holders->replaceText( $caption );
5076                 } else {
5077                         $tooltip = $this->replaceLinkHoldersText( $caption );
5078                 }
5079
5080                 # make sure there are no placeholders in thumbnail attributes
5081                 # that are later expanded to html- so expand them now and
5082                 # remove the tags
5083                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5084                 $tooltip = Sanitizer::stripAllTags( $tooltip );
5085
5086                 return $tooltip;
5087         }
5088
5089         /**
5090          * Set a flag in the output object indicating that the content is dynamic and
5091          * shouldn't be cached.
5092          */
5093         function disableCache() {
5094                 wfDebug( "Parser output marked as uncacheable.\n" );
5095                 $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
5096                 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5097         }
5098
5099         /**
5100          * Callback from the Sanitizer for expanding items found in HTML attribute
5101          * values, so they can be safely tested and escaped.
5102          *
5103          * @param $text String
5104          * @param $frame PPFrame
5105          * @return String
5106          */
5107         function attributeStripCallback( &$text, $frame = false ) {
5108                 $text = $this->replaceVariables( $text, $frame );
5109                 $text = $this->mStripState->unstripBoth( $text );
5110                 return $text;
5111         }
5112
5113         /**
5114          * Accessor
5115          *
5116          * @return array
5117          */
5118         function getTags() {
5119                 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ) );
5120         }
5121
5122         /**
5123          * Replace transparent tags in $text with the values given by the callbacks.
5124          *
5125          * Transparent tag hooks are like regular XML-style tag hooks, except they
5126          * operate late in the transformation sequence, on HTML instead of wikitext.
5127          *
5128          * @param $text string
5129          *
5130          * @return string
5131          */
5132         function replaceTransparentTags( $text ) {
5133                 $matches = array();
5134                 $elements = array_keys( $this->mTransparentTagHooks );
5135                 $text = self::extractTagsAndParams( $elements, $text, $matches, $this->mUniqPrefix );
5136                 $replacements = array();
5137
5138                 foreach ( $matches as $marker => $data ) {
5139                         list( $element, $content, $params, $tag ) = $data;
5140                         $tagName = strtolower( $element );
5141                         if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5142                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
5143                         } else {
5144                                 $output = $tag;
5145                         }
5146                         $replacements[$marker] = $output;
5147                 }
5148                 return strtr( $text, $replacements );
5149         }
5150
5151         /**
5152          * Break wikitext input into sections, and either pull or replace
5153          * some particular section's text.
5154          *
5155          * External callers should use the getSection and replaceSection methods.
5156          *
5157          * @param $text String: Page wikitext
5158          * @param $section String: a section identifier string of the form:
5159          *   <flag1> - <flag2> - ... - <section number>
5160          *
5161          * Currently the only recognised flag is "T", which means the target section number
5162          * was derived during a template inclusion parse, in other words this is a template
5163          * section edit link. If no flags are given, it was an ordinary section edit link.
5164          * This flag is required to avoid a section numbering mismatch when a section is
5165          * enclosed by <includeonly> (bug 6563).
5166          *
5167          * The section number 0 pulls the text before the first heading; other numbers will
5168          * pull the given section along with its lower-level subsections. If the section is
5169          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5170          *
5171          * Section 0 is always considered to exist, even if it only contains the empty
5172          * string. If $text is the empty string and section 0 is replaced, $newText is
5173          * returned.
5174          *
5175          * @param $mode String: one of "get" or "replace"
5176          * @param $newText String: replacement text for section data.
5177          * @return String: for "get", the extracted section text.
5178          *                 for "replace", the whole page with the section replaced.
5179          */
5180         private function extractSections( $text, $section, $mode, $newText='' ) {
5181                 global $wgTitle; # not generally used but removes an ugly failure mode
5182                 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5183                 $outText = '';
5184                 $frame = $this->getPreprocessor()->newFrame();
5185
5186                 # Process section extraction flags
5187                 $flags = 0;
5188                 $sectionParts = explode( '-', $section );
5189                 $sectionIndex = array_pop( $sectionParts );
5190                 foreach ( $sectionParts as $part ) {
5191                         if ( $part === 'T' ) {
5192                                 $flags |= self::PTD_FOR_INCLUSION;
5193                         }
5194                 }
5195
5196                 # Check for empty input
5197                 if ( strval( $text ) === '' ) {
5198                         # Only sections 0 and T-0 exist in an empty document
5199                         if ( $sectionIndex == 0 ) {
5200                                 if ( $mode === 'get' ) {
5201                                         return '';
5202                                 } else {
5203                                         return $newText;
5204                                 }
5205                         } else {
5206                                 if ( $mode === 'get' ) {
5207                                         return $newText;
5208                                 } else {
5209                                         return $text;
5210                                 }
5211                         }
5212                 }
5213
5214                 # Preprocess the text
5215                 $root = $this->preprocessToDom( $text, $flags );
5216
5217                 # <h> nodes indicate section breaks
5218                 # They can only occur at the top level, so we can find them by iterating the root's children
5219                 $node = $root->getFirstChild();
5220
5221                 # Find the target section
5222                 if ( $sectionIndex == 0 ) {
5223                         # Section zero doesn't nest, level=big
5224                         $targetLevel = 1000;
5225                 } else {
5226                         while ( $node ) {
5227                                 if ( $node->getName() === 'h' ) {
5228                                         $bits = $node->splitHeading();
5229                                         if ( $bits['i'] == $sectionIndex ) {
5230                                                 $targetLevel = $bits['level'];
5231                                                 break;
5232                                         }
5233                                 }
5234                                 if ( $mode === 'replace' ) {
5235                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5236                                 }
5237                                 $node = $node->getNextSibling();
5238                         }
5239                 }
5240
5241                 if ( !$node ) {
5242                         # Not found
5243                         if ( $mode === 'get' ) {
5244                                 return $newText;
5245                         } else {
5246                                 return $text;
5247                         }
5248                 }
5249
5250                 # Find the end of the section, including nested sections
5251                 do {
5252                         if ( $node->getName() === 'h' ) {
5253                                 $bits = $node->splitHeading();
5254                                 $curLevel = $bits['level'];
5255                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5256                                         break;
5257                                 }
5258                         }
5259                         if ( $mode === 'get' ) {
5260                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5261                         }
5262                         $node = $node->getNextSibling();
5263                 } while ( $node );
5264
5265                 # Write out the remainder (in replace mode only)
5266                 if ( $mode === 'replace' ) {
5267                         # Output the replacement text
5268                         # Add two newlines on -- trailing whitespace in $newText is conventionally
5269                         # stripped by the editor, so we need both newlines to restore the paragraph gap
5270                         # Only add trailing whitespace if there is newText
5271                         if ( $newText != "" ) {
5272                                 $outText .= $newText . "\n\n";
5273                         }
5274
5275                         while ( $node ) {
5276                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5277                                 $node = $node->getNextSibling();
5278                         }
5279                 }
5280
5281                 if ( is_string( $outText ) ) {
5282                         # Re-insert stripped tags
5283                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5284                 }
5285
5286                 return $outText;
5287         }
5288
5289         /**
5290          * This function returns the text of a section, specified by a number ($section).
5291          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5292          * the first section before any such heading (section 0).
5293          *
5294          * If a section contains subsections, these are also returned.
5295          *
5296          * @param $text String: text to look in
5297          * @param $section String: section identifier
5298          * @param $deftext String: default to return if section is not found
5299          * @return string text of the requested section
5300          */
5301         public function getSection( $text, $section, $deftext='' ) {
5302                 return $this->extractSections( $text, $section, "get", $deftext );
5303         }
5304
5305         /**
5306          * This function returns $oldtext after the content of the section
5307          * specified by $section has been replaced with $text. If the target
5308          * section does not exist, $oldtext is returned unchanged.
5309          *
5310          * @param $oldtext String: former text of the article
5311          * @param $section Numeric: section identifier
5312          * @param $text String: replacing text
5313          * @return String: modified text
5314          */
5315         public function replaceSection( $oldtext, $section, $text ) {
5316                 return $this->extractSections( $oldtext, $section, "replace", $text );
5317         }
5318
5319         /**
5320          * Get the ID of the revision we are parsing
5321          *
5322          * @return Mixed: integer or null
5323          */
5324         function getRevisionId() {
5325                 return $this->mRevisionId;
5326         }
5327
5328         /**
5329          * Get the revision object for $this->mRevisionId
5330          *
5331          * @return Revision|null either a Revision object or null
5332          */
5333         protected function getRevisionObject() {
5334                 if ( !is_null( $this->mRevisionObject ) ) {
5335                         return $this->mRevisionObject;
5336                 }
5337                 if ( is_null( $this->mRevisionId ) ) {
5338                         return null;
5339                 }
5340
5341                 $this->mRevisionObject = Revision::newFromId( $this->mRevisionId );
5342                 return $this->mRevisionObject;
5343         }
5344
5345         /**
5346          * Get the timestamp associated with the current revision, adjusted for
5347          * the default server-local timestamp
5348          */
5349         function getRevisionTimestamp() {
5350                 if ( is_null( $this->mRevisionTimestamp ) ) {
5351                         wfProfileIn( __METHOD__ );
5352
5353                         $revObject = $this->getRevisionObject();
5354                         $timestamp = $revObject ? $revObject->getTimestamp() : false;
5355
5356                         if( $timestamp !== false ) {
5357                                 global $wgContLang;
5358
5359                                 # The cryptic '' timezone parameter tells to use the site-default
5360                                 # timezone offset instead of the user settings.
5361                                 #
5362                                 # Since this value will be saved into the parser cache, served
5363                                 # to other users, and potentially even used inside links and such,
5364                                 # it needs to be consistent for all visitors.
5365                                 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5366                         }
5367
5368                         wfProfileOut( __METHOD__ );
5369                 }
5370                 return $this->mRevisionTimestamp;
5371         }
5372
5373         /**
5374          * Get the name of the user that edited the last revision
5375          *
5376          * @return String: user name
5377          */
5378         function getRevisionUser() {
5379                 if( is_null( $this->mRevisionUser ) ) {
5380                         $revObject = $this->getRevisionObject();
5381
5382                         # if this template is subst: the revision id will be blank,
5383                         # so just use the current user's name
5384                         if( $revObject ) {
5385                                 $this->mRevisionUser = $revObject->getUserText();
5386                         } elseif( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5387                                 $this->mRevisionUser = $this->getUser()->getName();
5388                         }
5389                 }
5390                 return $this->mRevisionUser;
5391         }
5392
5393         /**
5394          * Mutator for $mDefaultSort
5395          *
5396          * @param $sort New value
5397          */
5398         public function setDefaultSort( $sort ) {
5399                 $this->mDefaultSort = $sort;
5400                 $this->mOutput->setProperty( 'defaultsort', $sort );
5401         }
5402
5403         /**
5404          * Accessor for $mDefaultSort
5405          * Will use the empty string if none is set.
5406          *
5407          * This value is treated as a prefix, so the
5408          * empty string is equivalent to sorting by
5409          * page name.
5410          *
5411          * @return string
5412          */
5413         public function getDefaultSort() {
5414                 if ( $this->mDefaultSort !== false ) {
5415                         return $this->mDefaultSort;
5416                 } else {
5417                         return '';
5418                 }
5419         }
5420
5421         /**
5422          * Accessor for $mDefaultSort
5423          * Unlike getDefaultSort(), will return false if none is set
5424          *
5425          * @return string or false
5426          */
5427         public function getCustomDefaultSort() {
5428                 return $this->mDefaultSort;
5429         }
5430
5431         /**
5432          * Try to guess the section anchor name based on a wikitext fragment
5433          * presumably extracted from a heading, for example "Header" from
5434          * "== Header ==".
5435          *
5436          * @param $text string
5437          *
5438          * @return string
5439          */
5440         public function guessSectionNameFromWikiText( $text ) {
5441                 # Strip out wikitext links(they break the anchor)
5442                 $text = $this->stripSectionName( $text );
5443                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5444                 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5445         }
5446
5447         /**
5448          * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5449          * instead.  For use in redirects, since IE6 interprets Redirect: headers
5450          * as something other than UTF-8 (apparently?), resulting in breakage.
5451          *
5452          * @param $text String: The section name
5453          * @return string An anchor
5454          */
5455         public function guessLegacySectionNameFromWikiText( $text ) {
5456                 # Strip out wikitext links(they break the anchor)
5457                 $text = $this->stripSectionName( $text );
5458                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5459                 return '#' . Sanitizer::escapeId( $text, array( 'noninitial', 'legacy' ) );
5460         }
5461
5462         /**
5463          * Strips a text string of wikitext for use in a section anchor
5464          *
5465          * Accepts a text string and then removes all wikitext from the
5466          * string and leaves only the resultant text (i.e. the result of
5467          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5468          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5469          * to create valid section anchors by mimicing the output of the
5470          * parser when headings are parsed.
5471          *
5472          * @param $text String: text string to be stripped of wikitext
5473          * for use in a Section anchor
5474          * @return Filtered text string
5475          */
5476         public function stripSectionName( $text ) {
5477                 # Strip internal link markup
5478                 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5479                 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5480
5481                 # Strip external link markup
5482                 # @todo FIXME: Not tolerant to blank link text
5483                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5484                 # on how many empty links there are on the page - need to figure that out.
5485                 $text = preg_replace( '/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5486
5487                 # Parse wikitext quotes (italics & bold)
5488                 $text = $this->doQuotes( $text );
5489
5490                 # Strip HTML tags
5491                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5492                 return $text;
5493         }
5494
5495         /**
5496          * strip/replaceVariables/unstrip for preprocessor regression testing
5497          *
5498          * @param $text string
5499          * @param $title Title
5500          * @param $options ParserOptions
5501          * @param $outputType int
5502          *
5503          * @return string
5504          */
5505         function testSrvus( $text, Title $title, ParserOptions $options, $outputType = self::OT_HTML ) {
5506                 $this->startParse( $title, $options, $outputType, true );
5507
5508                 $text = $this->replaceVariables( $text );
5509                 $text = $this->mStripState->unstripBoth( $text );
5510                 $text = Sanitizer::removeHTMLtags( $text );
5511                 return $text;
5512         }
5513
5514         /**
5515          * @param $text string
5516          * @param $title Title
5517          * @param $options ParserOptions
5518          * @return string
5519          */
5520         function testPst( $text, Title $title, ParserOptions $options ) {
5521                 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
5522         }
5523
5524         /**
5525          * @param $text
5526          * @param $title Title
5527          * @param $options ParserOptions
5528          * @return string
5529          */
5530         function testPreprocess( $text, Title $title, ParserOptions $options ) {
5531                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5532         }
5533
5534         /**
5535          * Call a callback function on all regions of the given text that are not
5536          * inside strip markers, and replace those regions with the return value
5537          * of the callback. For example, with input:
5538          *
5539          *  aaa<MARKER>bbb
5540          *
5541          * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5542          * two strings will be replaced with the value returned by the callback in
5543          * each case.
5544          *
5545          * @param $s string
5546          * @param $callback
5547          *
5548          * @return string
5549          */
5550         function markerSkipCallback( $s, $callback ) {
5551                 $i = 0;
5552                 $out = '';
5553                 while ( $i < strlen( $s ) ) {
5554                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5555                         if ( $markerStart === false ) {
5556                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5557                                 break;
5558                         } else {
5559                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5560                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5561                                 if ( $markerEnd === false ) {
5562                                         $out .= substr( $s, $markerStart );
5563                                         break;
5564                                 } else {
5565                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5566                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5567                                         $i = $markerEnd;
5568                                 }
5569                         }
5570                 }
5571                 return $out;
5572         }
5573
5574         /**
5575          * Save the parser state required to convert the given half-parsed text to
5576          * HTML. "Half-parsed" in this context means the output of
5577          * recursiveTagParse() or internalParse(). This output has strip markers
5578          * from replaceVariables (extensionSubstitution() etc.), and link
5579          * placeholders from replaceLinkHolders().
5580          *
5581          * Returns an array which can be serialized and stored persistently. This
5582          * array can later be loaded into another parser instance with
5583          * unserializeHalfParsedText(). The text can then be safely incorporated into
5584          * the return value of a parser hook.
5585          *
5586          * @param $text string
5587          *
5588          * @return array
5589          */
5590         function serializeHalfParsedText( $text ) {
5591                 wfProfileIn( __METHOD__ );
5592                 $data = array(
5593                         'text' => $text,
5594                         'version' => self::HALF_PARSED_VERSION,
5595                         'stripState' => $this->mStripState->getSubState( $text ),
5596                         'linkHolders' => $this->mLinkHolders->getSubArray( $text )
5597                 );
5598                 wfProfileOut( __METHOD__ );
5599                 return $data;
5600         }
5601
5602         /**
5603          * Load the parser state given in the $data array, which is assumed to
5604          * have been generated by serializeHalfParsedText(). The text contents is
5605          * extracted from the array, and its markers are transformed into markers
5606          * appropriate for the current Parser instance. This transformed text is
5607          * returned, and can be safely included in the return value of a parser
5608          * hook.
5609          *
5610          * If the $data array has been stored persistently, the caller should first
5611          * check whether it is still valid, by calling isValidHalfParsedText().
5612          *
5613          * @param $data Serialized data
5614          * @return String
5615          */
5616         function unserializeHalfParsedText( $data ) {
5617                 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
5618                         throw new MWException( __METHOD__.': invalid version' );
5619                 }
5620
5621                 # First, extract the strip state.
5622                 $texts = array( $data['text'] );
5623                 $texts = $this->mStripState->merge( $data['stripState'], $texts );
5624
5625                 # Now renumber links
5626                 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
5627
5628                 # Should be good to go.
5629                 return $texts[0];
5630         }
5631
5632         /**
5633          * Returns true if the given array, presumed to be generated by
5634          * serializeHalfParsedText(), is compatible with the current version of the
5635          * parser.
5636          *
5637          * @param $data Array
5638          *
5639          * @return bool
5640          */
5641         function isValidHalfParsedText( $data ) {
5642                 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
5643         }
5644 }