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  *  $wgMaxArticleSize*
  44  *
  45  *  * only within ParserOptions
  46  * </pre>
  47  *
  48  * @ingroup Parser
  49  */
  50 class Parser {
  51         /**
  52          * Update this version number when the ParserOutput format
  53          * changes in an incompatible way, so the parser cache
  54          * can automatically discard old data.
  55          */
  56         const VERSION = '1.6.5';
  57
  58         /**
  59          * Update this version number when the output of serialiseHalfParsedText()
  60          * changes in an incompatible way
  61          */
  62         const HALF_PARSED_VERSION = 2;
  63
  64         # Flags for Parser::setFunctionHook
  65         # Also available as global constants from Defines.php
  66         const SFH_NO_HASH = 1;
  67         const SFH_OBJECT_ARGS = 2;
  68
  69         # Constants needed for external link processing
  70         # Everything except bracket, space, or control characters
  71         # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
  72         # as well as U+3000 is IDEOGRAPHIC SPACE for bug 19052
  73         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}]';
  74         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F\p{Zs}]+)
  75                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/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          * Process the wikitext for the ?preload= feature. (bug 5210)
 515          *
 516          * <noinclude>, <includeonly> etc. are parsed as for template transclusion,
 517          * comments, templates, arguments, tags hooks and parser functions are untouched.
 518          */
 519         public function getPreloadText( $text, Title $title, ParserOptions $options ) {
 520                 # Parser (re)initialisation
 521                 $this->startParse( $title, $options, self::OT_PLAIN, true );
 522
 523                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 524                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
 525                 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
 526                 $text = $this->mStripState->unstripBoth( $text );
 527                 return $text;
 528         }
 529
 530         /**
 531          * Get a random string
 532          *
 533          * @return string
 534          */
 535         static public function getRandomString() {
 536                 return dechex( mt_rand( 0, 0x7fffffff ) ) . dechex( mt_rand( 0, 0x7fffffff ) );
 537         }
 538
 539         /**
 540          * Set the current user.
 541          * Should only be used when doing pre-save transform.
 542          *
 543          * @param $user Mixed: User object or null (to reset)
 544          */
 545         function setUser( $user ) {
 546                 $this->mUser = $user;
 547         }
 548
 549         /**
 550          * Accessor for mUniqPrefix.
 551          *
 552          * @return String
 553          */
 554         public function uniqPrefix() {
 555                 if ( !isset( $this->mUniqPrefix ) ) {
 556                         # @todo FIXME: This is probably *horribly wrong*
 557                         # LanguageConverter seems to want $wgParser's uniqPrefix, however
 558                         # if this is called for a parser cache hit, the parser may not
 559                         # have ever been initialized in the first place.
 560                         # Not really sure what the heck is supposed to be going on here.
 561                         return '';
 562                         # throw new MWException( "Accessing uninitialized mUniqPrefix" );
 563                 }
 564                 return $this->mUniqPrefix;
 565         }
 566
 567         /**
 568          * Set the context title
 569          *
 570          * @param $t Title
 571          */
 572         function setTitle( $t ) {
 573                 if ( !$t || $t instanceof FakeTitle ) {
 574                         $t = Title::newFromText( 'NO TITLE' );
 575                 }
 576
 577                 if ( strval( $t->getFragment() ) !== '' ) {
 578                         # Strip the fragment to avoid various odd effects
 579                         $this->mTitle = clone $t;
 580                         $this->mTitle->setFragment( '' );
 581                 } else {
 582                         $this->mTitle = $t;
 583                 }
 584         }
 585
 586         /**
 587          * Accessor for the Title object
 588          *
 589          * @return Title object
 590          */
 591         function getTitle() {
 592                 return $this->mTitle;
 593         }
 594
 595         /**
 596          * Accessor/mutator for the Title object
 597          *
 598          * @param $x New Title object or null to just get the current one
 599          * @return Title object
 600          */
 601         function Title( $x = null ) {
 602                 return wfSetVar( $this->mTitle, $x );
 603         }
 604
 605         /**
 606          * Set the output type
 607          *
 608          * @param $ot Integer: new value
 609          */
 610         function setOutputType( $ot ) {
 611                 $this->mOutputType = $ot;
 612                 # Shortcut alias
 613                 $this->ot = array(
 614                         'html' => $ot == self::OT_HTML,
 615                         'wiki' => $ot == self::OT_WIKI,
 616                         'pre' => $ot == self::OT_PREPROCESS,
 617                         'plain' => $ot == self::OT_PLAIN,
 618                 );
 619         }
 620
 621         /**
 622          * Accessor/mutator for the output type
 623          *
 624          * @param $x New value or null to just get the current one
 625          * @return Integer
 626          */
 627         function OutputType( $x = null ) {
 628                 return wfSetVar( $this->mOutputType, $x );
 629         }
 630
 631         /**
 632          * Get the ParserOutput object
 633          *
 634          * @return ParserOutput object
 635          */
 636         function getOutput() {
 637                 return $this->mOutput;
 638         }
 639
 640         /**
 641          * Get the ParserOptions object
 642          *
 643          * @return ParserOptions object
 644          */
 645         function getOptions() {
 646                 return $this->mOptions;
 647         }
 648
 649         /**
 650          * Accessor/mutator for the ParserOptions object
 651          *
 652          * @param $x New value or null to just get the current one
 653          * @return Current ParserOptions object
 654          */
 655         function Options( $x = null ) {
 656                 return wfSetVar( $this->mOptions, $x );
 657         }
 658
 659         /**
 660          * @return int
 661          */
 662         function nextLinkID() {
 663                 return $this->mLinkID++;
 664         }
 665
 666         /**
 667          * @param $id int
 668          */
 669         function setLinkID( $id ) {
 670                 $this->mLinkID = $id;
 671         }
 672
 673         /**
 674          * @return Language
 675          */
 676         function getFunctionLang() {
 677                 global $wgLang;
 678
 679                 $target = $this->mOptions->getTargetLanguage();
 680                 if ( $target !== null ) {
 681                         return $target;
 682                 } else {
 683                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $this->mTitle->getPageLanguage();
 684                 }
 685         }
 686
 687         /**
 688          * Get a User object either from $this->mUser, if set, or from the
 689          * ParserOptions object otherwise
 690          *
 691          * @return User object
 692          */
 693         function getUser() {
 694                 if ( !is_null( $this->mUser ) ) {
 695                         return $this->mUser;
 696                 }
 697                 return $this->mOptions->getUser();
 698         }
 699
 700         /**
 701          * Get a preprocessor object
 702          *
 703          * @return Preprocessor instance
 704          */
 705         function getPreprocessor() {
 706                 if ( !isset( $this->mPreprocessor ) ) {
 707                         $class = $this->mPreprocessorClass;
 708                         $this->mPreprocessor = new $class( $this );
 709                 }
 710                 return $this->mPreprocessor;
 711         }
 712
 713         /**
 714          * Replaces all occurrences of HTML-style comments and the given tags
 715          * in the text with a random marker and returns the next text. The output
 716          * parameter $matches will be an associative array filled with data in
 717          * the form:
 718          *   'UNIQ-xxxxx' => array(
 719          *     'element',
 720          *     'tag content',
 721          *     array( 'param' => 'x' ),
 722          *     '<element param="x">tag content</element>' ) )
 723          *
 724          * @param $elements array list of element names. Comments are always extracted.
 725          * @param $text string Source text string.
 726          * @param $matches array Out parameter, Array: extracted tags
 727          * @param $uniq_prefix string
 728          * @return String: stripped text
 729          */
 730         public static function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
 731                 static $n = 1;
 732                 $stripped = '';
 733                 $matches = array();
 734
 735                 $taglist = implode( '|', $elements );
 736                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 737
 738                 while ( $text != '' ) {
 739                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 740                         $stripped .= $p[0];
 741                         if ( count( $p ) < 5 ) {
 742                                 break;
 743                         }
 744                         if ( count( $p ) > 5 ) {
 745                                 # comment
 746                                 $element    = $p[4];
 747                                 $attributes = '';
 748                                 $close      = '';
 749                                 $inside     = $p[5];
 750                         } else {
 751                                 # tag
 752                                 $element    = $p[1];
 753                                 $attributes = $p[2];
 754                                 $close      = $p[3];
 755                                 $inside     = $p[4];
 756                         }
 757
 758                         $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
 759                         $stripped .= $marker;
 760
 761                         if ( $close === '/>' ) {
 762                                 # Empty element tag, <tag />
 763                                 $content = null;
 764                                 $text = $inside;
 765                                 $tail = null;
 766                         } else {
 767                                 if ( $element === '!--' ) {
 768                                         $end = '/(-->)/';
 769                                 } else {
 770                                         $end = "/(<\\/$element\\s*>)/i";
 771                                 }
 772                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 773                                 $content = $q[0];
 774                                 if ( count( $q ) < 3 ) {
 775                                         # No end tag -- let it run out to the end of the text.
 776                                         $tail = '';
 777                                         $text = '';
 778                                 } else {
 779                                         $tail = $q[1];
 780                                         $text = $q[2];
 781                                 }
 782                         }
 783
 784                         $matches[$marker] = array( $element,
 785                                 $content,
 786                                 Sanitizer::decodeTagAttributes( $attributes ),
 787                                 "<$element$attributes$close$content$tail" );
 788                 }
 789                 return $stripped;
 790         }
 791
 792         /**
 793          * Get a list of strippable XML-like elements
 794          *
 795          * @return array
 796          */
 797         function getStripList() {
 798                 return $this->mStripList;
 799         }
 800
 801         /**
 802          * Add an item to the strip state
 803          * Returns the unique tag which must be inserted into the stripped text
 804          * The tag will be replaced with the original text in unstrip()
 805          *
 806          * @param $text string
 807          *
 808          * @return string
 809          */
 810         function insertStripItem( $text ) {
 811                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 812                 $this->mMarkerIndex++;
 813                 $this->mStripState->addGeneral( $rnd, $text );
 814                 return $rnd;
 815         }
 816
 817         /**
 818          * parse the wiki syntax used to render tables
 819          *
 820          * @private
 821          *
 822          * @param $text string
 823          *
 824          * @return string
 825          */
 826         function doTableStuff( $text ) {
 827                 wfProfileIn( __METHOD__ );
 828
 829                 $lines = StringUtils::explode( "\n", $text );
 830                 $out = '';
 831                 $output =& $out;
 832
 833                 foreach ( $lines as $outLine ) {
 834                         $line = trim( $outLine );
 835
 836                         # empty line, go to next line,
 837                         # but only append \n if outside of table
 838                         if ( $line === '') {
 839                                 $output .= $outLine . "\n";
 840                                 continue;
 841                         }
 842                         $firstChars = $line[0];
 843                         if ( strlen( $line ) > 1 ) {
 844                                 $firstChars .= in_array( $line[1], array( '}', '+', '-' ) ) ? $line[1] : '';
 845                         }
 846                         $matches = array();
 847
 848                         if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line , $matches ) ) {
 849                                 $tables[] = array();
 850                                 $table =& $this->last( $tables );
 851                                 $table[0] = array(); // first row
 852                                 $currentRow =& $table[0];
 853                                 $table['indent'] = strlen( $matches[1] );
 854
 855                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 856                                 $attributes = Sanitizer::fixTagAttributes( $attributes , 'table' );
 857
 858                                 if ( $attributes !== '' ) {
 859                                         $table['attributes'] = $attributes;
 860                                 }
 861                         } elseif ( !isset( $tables[0] ) ) {
 862                                 // we're outside the table
 863
 864                                 $out .= $outLine . "\n";
 865                         } elseif ( $firstChars === '|}' ) {
 866                                 // trim the |} code from the line
 867                                 $line = substr ( $line , 2 );
 868
 869                                 // Shorthand for last row
 870                                 $lastRow =& $this->last( $table );
 871
 872                                 // a thead at the end becomes a tfoot, unless there is only one row
 873                                 // Do this before deleting empty last lines to allow headers at the bottom of tables
 874                                 if ( isset( $lastRow['type'] ) && $lastRow['type'] == 'thead' && isset( $table[1] ) ) {
 875                                         $lastRow['type'] = 'tfoot';
 876                                         for ( $i = 0; isset( $lastRow[$i] ); $i++ ) {
 877                                                 $lastRow[$i]['type'] = 'th';
 878                                         }
 879                                 }
 880
 881                                 // Delete empty last lines
 882                                 if ( empty( $lastRow ) ) {
 883                                         $lastRow = NULL;
 884                                 }
 885                                 $o = '';
 886                                 $curtable = array_pop( $tables );
 887
 888                                 #Add a line-ending before the table, but only if there isn't one already
 889                                 if ( substr( $out, -1 ) !== "\n" ) {
 890                                         $o .= "\n";
 891                                 }
 892                                 $o .= $this->generateTableHTML( $curtable ) . $line . "\n";
 893
 894                                 if ( count( $tables ) > 0 ) {
 895                                         $table =& $this->last( $tables );
 896                                         $currentRow =& $this->last( $table );
 897                                         $currentElement =& $this->last( $currentRow );
 898
 899                                         $output =& $currentElement['content'];
 900                                 } else {
 901                                         $output =& $out;
 902                                 }
 903
 904                                 $output .= $o;
 905
 906                         } elseif ( $firstChars === '|-' ) {
 907                                 // start a new row element
 908                                 // but only when we haven't started one already
 909                                 if ( count( $currentRow ) != 0 ) {
 910                                         $table[] = array();
 911                                         $currentRow =& $this->last( $table );
 912                                 }
 913                                 // Get the attributes, there's nothing else useful in $line now
 914                                 $line = substr ( $line , 2 );
 915                                 $attributes = $this->mStripState->unstripBoth( $line );
 916                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 917                                 if ( $attributes !== '' ) {
 918                                         $currentRow['attributes'] = $attributes;
 919                                 }
 920
 921                         } elseif ( $firstChars  === '|+' ) {
 922                                 // a table caption, but only proceed if there isn't one already
 923                                 if ( !isset ( $table['caption'] ) ) {
 924                                         $line = substr ( $line , 2 );
 925
 926                                         $c = $this->getCellAttr( $line , 'caption' );
 927                                         $table['caption'] = array();
 928                                         $table['caption']['content'] = $c[0];
 929                                         if ( isset( $c[1] ) ) $table['caption']['attributes'] = $c[1];
 930                                         unset( $c );
 931                                         $output =& $table['caption']['content'];
 932                                 }
 933                         } elseif ( $firstChars === '|' || $firstChars === '!' || $firstChars === '!+' ) {
 934                                 // Which kind of cells are we dealing with
 935                                 $currentTag = 'td';
 936                                 $line = substr ( $line , 1 );
 937
 938                                 if ( $firstChars === '!'  || $firstChars === '!+' ) {
 939                                         $line = str_replace ( '!!' , '||' , $line );
 940                                         $currentTag = 'th';
 941                                 }
 942
 943                                 // Split up multiple cells on the same line.
 944                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 945                                 $line = ''; // save memory
 946
 947                                 // decide whether thead to tbody
 948                                 if ( !array_key_exists( 'type', $currentRow ) ) {
 949                                         $currentRow['type'] = ( $firstChars === '!' ) ? 'thead' : 'tbody' ;
 950                                 } elseif ( $firstChars === '|' ) {
 951                                         $currentRow['type'] = 'tbody';
 952                                 }
 953
 954                                 // Loop through each table cell
 955                                 foreach ( $cells as $cell ) {
 956                                         // a new cell
 957                                         $currentRow[] = array();
 958                                         $currentElement =& $this->last( $currentRow );
 959
 960                                         $currentElement['type'] = $currentTag;
 961
 962                                         $c = $this->getCellAttr( $cell , $currentTag );
 963                                         $currentElement['content'] = $c[0];
 964                                         if ( isset( $c[1] ) ) $currentElement['attributes'] = $c[1];
 965                                         unset( $c );
 966                                 }
 967                                 $output =& $currentElement['content'];
 968
 969                         } else {
 970                                 $output .= "\n$outLine";
 971                         }
 972                 }
 973
 974                 # Remove trailing line-ending (b/c)
 975                 if ( substr( $out, -1 ) === "\n" ) {
 976                         $out = substr( $out, 0, -1 );
 977                 }
 978
 979                 # Close any unclosed tables
 980                 if ( isset( $tables ) && count( $tables ) > 0 ) {
 981                         for ( $i = 0; $i < count( $tables ); $i++ ) {
 982                                 $curtable = array_pop( $tables );
 983                                 $curtable = $this->generateTableHTML( $curtable );
 984                                 #Add a line-ending before the table, but only if there isn't one already
 985                                 if ( substr( $out, -1 ) !== "\n"  && $curtable !== "" ) {
 986                                         $out .= "\n";
 987                                 }
 988                                 $out .= $curtable;
 989                         }
 990                 }
 991
 992                 wfProfileOut( __METHOD__ );
 993
 994                 return $out;
 995         }
 996
 997         /**
 998          * Helper function for doTableStuff() separating the contents of cells from
 999          * attributes. Particularly useful as there's a possible bug and this action
1000          * is repeated twice.
1001          *
1002          * @private
1003          * @param $cell
1004          * @param $tagName
1005          * @return array
1006          */
1007         function getCellAttr ( $cell, $tagName ) {
1008                 $attributes = null;
1009
1010                 $cell = trim ( $cell );
1011
1012                 // A cell could contain both parameters and data
1013                 $cellData = explode ( '|' , $cell , 2 );
1014
1015                 // Bug 553: Note that a '|' inside an invalid link should not
1016                 // be mistaken as delimiting cell parameters
1017                 if ( strpos( $cellData[0], '[[' ) !== false ) {
1018                         $content = trim ( $cell );
1019                 }
1020                 elseif ( count ( $cellData ) == 1 ) {
1021                         $content = trim ( $cellData[0] );
1022                 } else {
1023                         $attributes = $this->mStripState->unstripBoth( $cellData[0] );
1024                         $attributes = Sanitizer::fixTagAttributes( $attributes , $tagName );
1025
1026                         $content = trim ( $cellData[1] );
1027                 }
1028                 return array( $content, $attributes );
1029         }
1030
1031
1032         /**
1033          * Helper function for doTableStuff(). This converts the structured array into html.
1034          *
1035          * @private
1036          *
1037          * @param $table array
1038          *
1039          * @return string
1040          */
1041         function generateTableHTML( &$table ) {
1042                 $return = str_repeat( '<dl><dd>' , $table['indent'] );
1043                 $return .= '<table';
1044                 $return .= isset( $table['attributes'] ) ? $table['attributes'] : '';
1045                 $return .= '>';
1046                 unset( $table['attributes'] );
1047
1048                 if ( isset( $table['caption'] ) ) {
1049                         $return .= "\n<caption";
1050                         $return .= isset( $table['caption']['attributes'] ) ? $table['caption']['attributes'] : '';
1051                         $return .= '>';
1052                         $return .= $table['caption']['content'];
1053                         $return .= "\n</caption>";
1054                 }
1055                 $lastSection = '';
1056                 $empty = true;
1057                 $simple = true;
1058
1059                 // If we only have tbodies, mark table as simple
1060                 for ( $i = 0; isset( $table[$i] ); $i++ ) {
1061                         if ( !count( $table[$i] ) ) continue;
1062                         if ( !isset( $table[$i]['type'] ) ) {
1063                                 $table[$i]['type'] = 'tbody';
1064                         }
1065                         if ( !$lastSection ) {
1066                                 $lastSection = $table[$i]['type'];
1067                         } elseif ( $lastSection != $table[$i]['type'] ) {
1068                                 $simple = false;
1069                         }
1070                 }
1071                 $lastSection = '';
1072                 for ( $i = 0; isset( $table[$i] ); $i++ ) {
1073                         if ( !count( $table[$i] ) ) continue;
1074                         $empty = false; // check for empty tables
1075
1076                         if ( $table[$i]['type'] != $lastSection && !$simple ) {
1077                                 $return .= "\n<" . $table[$i]['type'] . '>';
1078                         }
1079
1080                         $return .= "\n<tr";
1081                         $return .= isset( $table[$i]['attributes'] ) ? $table[$i]['attributes'] : '';
1082                         $return .= '>';
1083                         for ( $j = 0; isset( $table[$i][$j] ); $j++ ) {
1084                                 if ( !isset( $table[$i][$j]['type'] ) ) $table[$i][$j]['type'] = 'td';
1085                                 $return .= "\n<" . $table[$i][$j]['type'];
1086                                 $return .= isset( $table[$i][$j]['attributes'] ) ? $table[$i][$j]['attributes'] : '';
1087                                 $return .= '>';
1088
1089                                 $return .= $table[$i][$j]['content'];
1090                                 if ( $table[$i][$j]['content'] != '' )
1091                                         $return .= "\n";
1092
1093                                 $return .= '</' . $table[$i][$j]['type'] . '>';
1094                                 unset( $table[$i][$j] );
1095                         }
1096                         $return .= "\n</tr>";
1097
1098                         if ( ( !isset( $table[$i + 1] ) && !$simple ) || ( isset( $table[$i + 1] ) && isset( $table[$i + 1]['type'] ) && $table[$i]['type'] != $table[$i + 1]['type'] ) ) {
1099                                 $return .= '</' . $table[$i]['type'] . '>';
1100                         }
1101                         $lastSection = $table[$i]['type'];
1102                         unset( $table[$i] );
1103                 }
1104                 if ( $empty ) {
1105                         if ( isset( $table['caption'] ) ) {
1106                                 $return .= "\n<tr><td></td></tr>";
1107                         } else {
1108                                 return '';
1109                         }
1110                 }
1111                 $return .= "\n</table>";
1112                 $return .= str_repeat( '</dd></dl>' , $table['indent'] );
1113
1114                 return $return;
1115         }
1116
1117         /**
1118          * like end() but only works on the numeric array index and php's internal pointers
1119          * returns a reference to the last element of an array much like "\$arr[-1]" in perl
1120          * ignores associative elements and will create a 0 key will a NULL value if there were
1121          * no numric elements and an array itself if not previously defined.
1122          *
1123          * @private
1124          *
1125          * @param $arr array
1126          */
1127         function &last ( &$arr ) {
1128                 for ( $i = count( $arr ); ( !isset( $arr[$i] ) && $i > 0 ); $i-- ) {  }
1129                 return $arr[$i];
1130         }
1131
1132         /**
1133          * Helper function for parse() that transforms wiki markup into
1134          * HTML. Only called for $mOutputType == self::OT_HTML.
1135          *
1136          * @private
1137          *
1138          * @param $text string
1139          * @param $isMain bool
1140          * @param $frame bool
1141          *
1142          * @return string
1143          */
1144         function internalParse( $text, $isMain = true, $frame = false ) {
1145                 wfProfileIn( __METHOD__ );
1146
1147                 $origText = $text;
1148
1149                 # Hook to suspend the parser in this state
1150                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1151                         wfProfileOut( __METHOD__ );
1152                         return $text ;
1153                 }
1154
1155                 # if $frame is provided, then use $frame for replacing any variables
1156                 if ( $frame ) {
1157                         # use frame depth to infer how include/noinclude tags should be handled
1158                         # depth=0 means this is the top-level document; otherwise it's an included document
1159                         if ( !$frame->depth ) {
1160                                 $flag = 0;
1161                         } else {
1162                                 $flag = Parser::PTD_FOR_INCLUSION;
1163                         }
1164                         $dom = $this->preprocessToDom( $text, $flag );
1165                         $text = $frame->expand( $dom );
1166                 } else {
1167                         # if $frame is not provided, then use old-style replaceVariables
1168                         $text = $this->replaceVariables( $text );
1169                 }
1170
1171                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
1172                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1173
1174                 # Tables need to come after variable replacement for things to work
1175                 # properly; putting them before other transformations should keep
1176                 # exciting things like link expansions from showing up in surprising
1177                 # places.
1178                 $text = $this->doTableStuff( $text );
1179
1180                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1181
1182                 $text = $this->doDoubleUnderscore( $text );
1183
1184                 $text = $this->doHeadings( $text );
1185                 if ( $this->mOptions->getUseDynamicDates() ) {
1186                         $df = DateFormatter::getInstance();
1187                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1188                 }
1189                 $text = $this->replaceInternalLinks( $text );
1190                 $text = $this->doAllQuotes( $text );
1191                 $text = $this->replaceExternalLinks( $text );
1192
1193                 # replaceInternalLinks may sometimes leave behind
1194                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1195                 $text = str_replace( $this->mUniqPrefix.'NOPARSE', '', $text );
1196
1197                 $text = $this->doMagicLinks( $text );
1198                 $text = $this->formatHeadings( $text, $origText, $isMain );
1199
1200                 wfProfileOut( __METHOD__ );
1201                 return $text;
1202         }
1203
1204         /**
1205          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1206          * magic external links.
1207          *
1208          * DML
1209          * @private
1210          *
1211          * @param $text string
1212          *
1213          * @return string
1214          */
1215         function doMagicLinks( $text ) {
1216                 wfProfileIn( __METHOD__ );
1217                 $prots = $this->mUrlProtocols;
1218                 $urlChar = self::EXT_LINK_URL_CLASS;
1219                 $text = preg_replace_callback(
1220                         '!(?:                           # Start cases
1221                                 (<a[ \t\r\n>].*?</a>) |     # m[1]: Skip link text
1222                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1223                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
1224                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1225                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1226                                         (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1227                                         (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1228                                         [0-9Xx]                 # check digit
1229                                         \b)
1230                         )!xu', array( &$this, 'magicLinkCallback' ), $text );
1231                 wfProfileOut( __METHOD__ );
1232                 return $text;
1233         }
1234
1235         /**
1236          * @throws MWException
1237          * @param $m array
1238          * @return HTML|string
1239          */
1240         function magicLinkCallback( $m ) {
1241                 if ( isset( $m[1] ) && $m[1] !== '' ) {
1242                         # Skip anchor
1243                         return $m[0];
1244                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1245                         # Skip HTML element
1246                         return $m[0];
1247                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1248                         # Free external link
1249                         return $this->makeFreeExternalLink( $m[0] );
1250                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1251                         # RFC or PMID
1252                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1253                                 $keyword = 'RFC';
1254                                 $urlmsg = 'rfcurl';
1255                                 $CssClass = 'mw-magiclink-rfc';
1256                                 $id = $m[4];
1257                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1258                                 $keyword = 'PMID';
1259                                 $urlmsg = 'pubmedurl';
1260                                 $CssClass = 'mw-magiclink-pmid';
1261                                 $id = $m[4];
1262                         } else {
1263                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1264                                         substr( $m[0], 0, 20 ) . '"' );
1265                         }
1266                         $url = wfMsgForContent( $urlmsg, $id );
1267                         return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $CssClass );
1268                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1269                         # ISBN
1270                         $isbn = $m[5];
1271                         $num = strtr( $isbn, array(
1272                                 '-' => '',
1273                                 ' ' => '',
1274                                 'x' => 'X',
1275                         ));
1276                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1277                         return'<a href="' .
1278                                 $titleObj->escapeLocalUrl() .
1279                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1280                 } else {
1281                         return $m[0];
1282                 }
1283         }
1284
1285         /**
1286          * Make a free external link, given a user-supplied URL
1287          *
1288          * @param $url string
1289          *
1290          * @return string HTML
1291          * @private
1292          */
1293         function makeFreeExternalLink( $url ) {
1294                 global $wgContLang;
1295                 wfProfileIn( __METHOD__ );
1296
1297                 $trail = '';
1298
1299                 # The characters '<' and '>' (which were escaped by
1300                 # removeHTMLtags()) should not be included in
1301                 # URLs, per RFC 2396.
1302                 $m2 = array();
1303                 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1304                         $trail = substr( $url, $m2[0][1] ) . $trail;
1305                         $url = substr( $url, 0, $m2[0][1] );
1306                 }
1307
1308                 # Move trailing punctuation to $trail
1309                 $sep = ',;\.:!?';
1310                 # If there is no left bracket, then consider right brackets fair game too
1311                 if ( strpos( $url, '(' ) === false ) {
1312                         $sep .= ')';
1313                 }
1314
1315                 $numSepChars = strspn( strrev( $url ), $sep );
1316                 if ( $numSepChars ) {
1317                         $trail = substr( $url, -$numSepChars ) . $trail;
1318                         $url = substr( $url, 0, -$numSepChars );
1319                 }
1320
1321                 $url = Sanitizer::cleanUrl( $url );
1322
1323                 # Is this an external image?
1324                 $text = $this->maybeMakeExternalImage( $url );
1325                 if ( $text === false ) {
1326                         # Not an image, make a link
1327                         $text = Linker::makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1328                                 $this->getExternalLinkAttribs( $url ) );
1329                         # Register it in the output object...
1330                         # Replace unnecessary URL escape codes with their equivalent characters
1331                         $pasteurized = self::replaceUnusualEscapes( $url );
1332                         $this->mOutput->addExternalLink( $pasteurized );
1333                 }
1334                 wfProfileOut( __METHOD__ );
1335                 return $text . $trail;
1336         }
1337
1338
1339         /**
1340          * Parse headers and return html
1341          *
1342          * @private
1343          *
1344          * @param $text string
1345          *
1346          * @return string
1347          */
1348         function doHeadings( $text ) {
1349                 wfProfileIn( __METHOD__ );
1350                 for ( $i = 6; $i >= 1; --$i ) {
1351                         $h = str_repeat( '=', $i );
1352                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1353                           "<h$i>\\1</h$i>", $text );
1354                 }
1355                 wfProfileOut( __METHOD__ );
1356                 return $text;
1357         }
1358
1359         /**
1360          * Replace single quotes with HTML markup
1361          * @private
1362          *
1363          * @param $text string
1364          *
1365          * @return string the altered text
1366          */
1367         function doAllQuotes( $text ) {
1368                 wfProfileIn( __METHOD__ );
1369                 $outtext = '';
1370                 $lines = StringUtils::explode( "\n", $text );
1371                 foreach ( $lines as $line ) {
1372                         $outtext .= $this->doQuotes( $line ) . "\n";
1373                 }
1374                 $outtext = substr( $outtext, 0,-1 );
1375                 wfProfileOut( __METHOD__ );
1376                 return $outtext;
1377         }
1378
1379         /**
1380          * Helper function for doAllQuotes()
1381          *
1382          * @param $text string
1383          *
1384          * @return string
1385          */
1386         public function doQuotes( $text ) {
1387                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1388                 if ( count( $arr ) == 1 ) {
1389                         return $text;
1390                 } else {
1391                         # First, do some preliminary work. This may shift some apostrophes from
1392                         # being mark-up to being text. It also counts the number of occurrences
1393                         # of bold and italics mark-ups.
1394                         $numbold = 0;
1395                         $numitalics = 0;
1396                         for ( $i = 0; $i < count( $arr ); $i++ ) {
1397                                 if ( ( $i % 2 ) == 1 ) {
1398                                         # If there are ever four apostrophes, assume the first is supposed to
1399                                         # be text, and the remaining three constitute mark-up for bold text.
1400                                         if ( strlen( $arr[$i] ) == 4 ) {
1401                                                 $arr[$i-1] .= "'";
1402                                                 $arr[$i] = "'''";
1403                                         } elseif ( strlen( $arr[$i] ) > 5 ) {
1404                                                 # If there are more than 5 apostrophes in a row, assume they're all
1405                                                 # text except for the last 5.
1406                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1407                                                 $arr[$i] = "'''''";
1408                                         }
1409                                         # Count the number of occurrences of bold and italics mark-ups.
1410                                         # We are not counting sequences of five apostrophes.
1411                                         if ( strlen( $arr[$i] ) == 2 ) {
1412                                                 $numitalics++;
1413                                         } elseif ( strlen( $arr[$i] ) == 3 ) {
1414                                                 $numbold++;
1415                                         } elseif ( strlen( $arr[$i] ) == 5 ) {
1416                                                 $numitalics++;
1417                                                 $numbold++;
1418                                         }
1419                                 }
1420                         }
1421
1422                         # If there is an odd number of both bold and italics, it is likely
1423                         # that one of the bold ones was meant to be an apostrophe followed
1424                         # by italics. Which one we cannot know for certain, but it is more
1425                         # likely to be one that has a single-letter word before it.
1426                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1427                                 $i = 0;
1428                                 $firstsingleletterword = -1;
1429                                 $firstmultiletterword = -1;
1430                                 $firstspace = -1;
1431                                 foreach ( $arr as $r ) {
1432                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1433                                                 $x1 = substr( $arr[$i-1], -1 );
1434                                                 $x2 = substr( $arr[$i-1], -2, 1 );
1435                                                 if ( $x1 === ' ' ) {
1436                                                         if ( $firstspace == -1 ) {
1437                                                                 $firstspace = $i;
1438                                                         }
1439                                                 } elseif ( $x2 === ' ') {
1440                                                         if ( $firstsingleletterword == -1 ) {
1441                                                                 $firstsingleletterword = $i;
1442                                                         }
1443                                                 } else {
1444                                                         if ( $firstmultiletterword == -1 ) {
1445                                                                 $firstmultiletterword = $i;
1446                                                         }
1447                                                 }
1448                                         }
1449                                         $i++;
1450                                 }
1451
1452                                 # If there is a single-letter word, use it!
1453                                 if ( $firstsingleletterword > -1 ) {
1454                                         $arr[$firstsingleletterword] = "''";
1455                                         $arr[$firstsingleletterword-1] .= "'";
1456                                 } elseif ( $firstmultiletterword > -1 ) {
1457                                         # If not, but there's a multi-letter word, use that one.
1458                                         $arr[$firstmultiletterword] = "''";
1459                                         $arr[$firstmultiletterword-1] .= "'";
1460                                 } elseif ( $firstspace > -1 ) {
1461                                         # ... otherwise use the first one that has neither.
1462                                         # (notice that it is possible for all three to be -1 if, for example,
1463                                         # there is only one pentuple-apostrophe in the line)
1464                                         $arr[$firstspace] = "''";
1465                                         $arr[$firstspace-1] .= "'";
1466                                 }
1467                         }
1468
1469                         # Now let's actually convert our apostrophic mush to HTML!
1470                         $output = '';
1471                         $buffer = '';
1472                         $state = '';
1473                         $i = 0;
1474                         foreach ( $arr as $r ) {
1475                                 if ( ( $i % 2 ) == 0 ) {
1476                                         if ( $state === 'both' ) {
1477                                                 $buffer .= $r;
1478                                         } else {
1479                                                 $output .= $r;
1480                                         }
1481                                 } else {
1482                                         if ( strlen( $r ) == 2 ) {
1483                                                 if ( $state === 'i' ) {
1484                                                         $output .= '</i>'; $state = '';
1485                                                 } elseif ( $state === 'bi' ) {
1486                                                         $output .= '</i>'; $state = 'b';
1487                                                 } elseif ( $state === 'ib' ) {
1488                                                         $output .= '</b></i><b>'; $state = 'b';
1489                                                 } elseif ( $state === 'both' ) {
1490                                                         $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1491                                                 } else { # $state can be 'b' or ''
1492                                                         $output .= '<i>'; $state .= 'i';
1493                                                 }
1494                                         } elseif ( strlen( $r ) == 3 ) {
1495                                                 if ( $state === 'b' ) {
1496                                                         $output .= '</b>'; $state = '';
1497                                                 } elseif ( $state === 'bi' ) {
1498                                                         $output .= '</i></b><i>'; $state = 'i';
1499                                                 } elseif ( $state === 'ib' ) {
1500                                                         $output .= '</b>'; $state = 'i';
1501                                                 } elseif ( $state === 'both' ) {
1502                                                         $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1503                                                 } else { # $state can be 'i' or ''
1504                                                         $output .= '<b>'; $state .= 'b';
1505                                                 }
1506                                         } elseif ( strlen( $r ) == 5 ) {
1507                                                 if ( $state === 'b' ) {
1508                                                         $output .= '</b><i>'; $state = 'i';
1509                                                 } elseif ( $state === 'i' ) {
1510                                                         $output .= '</i><b>'; $state = 'b';
1511                                                 } elseif ( $state === 'bi' ) {
1512                                                         $output .= '</i></b>'; $state = '';
1513                                                 } elseif ( $state === 'ib' ) {
1514                                                         $output .= '</b></i>'; $state = '';
1515                                                 } elseif ( $state === 'both' ) {
1516                                                         $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1517                                                 } else { # ($state == '')
1518                                                         $buffer = ''; $state = 'both';
1519                                                 }
1520                                         }
1521                                 }
1522                                 $i++;
1523                         }
1524                         # Now close all remaining tags.  Notice that the order is important.
1525                         if ( $state === 'b' || $state === 'ib' ) {
1526                                 $output .= '</b>';
1527                         }
1528                         if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1529                                 $output .= '</i>';
1530                         }
1531                         if ( $state === 'bi' ) {
1532                                 $output .= '</b>';
1533                         }
1534                         # There might be lonely ''''', so make sure we have a buffer
1535                         if ( $state === 'both' && $buffer ) {
1536                                 $output .= '<b><i>'.$buffer.'</i></b>';
1537                         }
1538                         return $output;
1539                 }
1540         }
1541
1542         /**
1543          * Replace external links (REL)
1544          *
1545          * Note: this is all very hackish and the order of execution matters a lot.
1546          * Make sure to run maintenance/parserTests.php if you change this code.
1547          *
1548          * @private
1549          *
1550          * @param $text string
1551          *
1552          * @return string
1553          */
1554         function replaceExternalLinks( $text ) {
1555                 global $wgContLang;
1556                 wfProfileIn( __METHOD__ );
1557
1558                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1559                 $s = array_shift( $bits );
1560
1561                 $i = 0;
1562                 while ( $i<count( $bits ) ) {
1563                         $url = $bits[$i++];
1564                         $protocol = $bits[$i++];
1565                         $text = $bits[$i++];
1566                         $trail = $bits[$i++];
1567
1568                         # The characters '<' and '>' (which were escaped by
1569                         # removeHTMLtags()) should not be included in
1570                         # URLs, per RFC 2396.
1571                         $m2 = array();
1572                         if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1573                                 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1574                                 $url = substr( $url, 0, $m2[0][1] );
1575                         }
1576
1577                         # If the link text is an image URL, replace it with an <img> tag
1578                         # This happened by accident in the original parser, but some people used it extensively
1579                         $img = $this->maybeMakeExternalImage( $text );
1580                         if ( $img !== false ) {
1581                                 $text = $img;
1582                         }
1583
1584                         $dtrail = '';
1585
1586                         # Set linktype for CSS - if URL==text, link is essentially free
1587                         $linktype = ( $text === $url ) ? 'free' : 'text';
1588
1589                         # No link text, e.g. [http://domain.tld/some.link]
1590                         if ( $text == '' ) {
1591                                 # Autonumber
1592                                 $langObj = $this->getFunctionLang();
1593                                 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1594                                 $linktype = 'autonumber';
1595                         } else {
1596                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1597                                 # Check for trail
1598                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1599                         }
1600
1601                         $text = $wgContLang->markNoConversion( $text );
1602
1603                         $url = Sanitizer::cleanUrl( $url );
1604
1605                         # Use the encoded URL
1606                         # This means that users can paste URLs directly into the text
1607                         # Funny characters like ö aren't valid in URLs anyway
1608                         # This was changed in August 2004
1609                         $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1610                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1611
1612                         # Register link in the output object.
1613                         # Replace unnecessary URL escape codes with the referenced character
1614                         # This prevents spammers from hiding links from the filters
1615                         $pasteurized = self::replaceUnusualEscapes( $url );
1616                         $this->mOutput->addExternalLink( $pasteurized );
1617                 }
1618
1619                 wfProfileOut( __METHOD__ );
1620                 return $s;
1621         }
1622
1623         /**
1624          * Get an associative array of additional HTML attributes appropriate for a
1625          * particular external link.  This currently may include rel => nofollow
1626          * (depending on configuration, namespace, and the URL's domain) and/or a
1627          * target attribute (depending on configuration).
1628          *
1629          * @param $url String|bool optional URL, to extract the domain from for rel =>
1630          *   nofollow if appropriate
1631          * @return Array associative array of HTML attributes
1632          */
1633         function getExternalLinkAttribs( $url = false ) {
1634                 $attribs = array();
1635                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1636                 $ns = $this->mTitle->getNamespace();
1637                 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) ) {
1638                         $attribs['rel'] = 'nofollow';
1639
1640                         global $wgNoFollowDomainExceptions;
1641                         if ( $wgNoFollowDomainExceptions ) {
1642                                 $bits = wfParseUrl( $url );
1643                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1644                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1645                                                 if ( substr( $bits['host'], -strlen( $domain ) ) == $domain ) {
1646                                                         unset( $attribs['rel'] );
1647                                                         break;
1648                                                 }
1649                                         }
1650                                 }
1651                         }
1652                 }
1653                 if ( $this->mOptions->getExternalLinkTarget() ) {
1654                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1655                 }
1656                 return $attribs;
1657         }
1658
1659         /**
1660          * Replace unusual URL escape codes with their equivalent characters
1661          *
1662          * @param $url String
1663          * @return String
1664          *
1665          * @todo  This can merge genuinely required bits in the path or query string,
1666          *        breaking legit URLs. A proper fix would treat the various parts of
1667          *        the URL differently; as a workaround, just use the output for
1668          *        statistical records, not for actual linking/output.
1669          */
1670         static function replaceUnusualEscapes( $url ) {
1671                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1672                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1673         }
1674
1675         /**
1676          * Callback function used in replaceUnusualEscapes().
1677          * Replaces unusual URL escape codes with their equivalent character
1678          *
1679          * @param $matches array
1680          *
1681          * @return string
1682          */
1683         private static function replaceUnusualEscapesCallback( $matches ) {
1684                 $char = urldecode( $matches[0] );
1685                 $ord = ord( $char );
1686                 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1687                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1688                         # No, shouldn't be escaped
1689                         return $char;
1690                 } else {
1691                         # Yes, leave it escaped
1692                         return $matches[0];
1693                 }
1694         }
1695
1696         /**
1697          * make an image if it's allowed, either through the global
1698          * option, through the exception, or through the on-wiki whitelist
1699          * @private
1700          *
1701          * $param $url string
1702          *
1703          * @return string
1704          */
1705         function maybeMakeExternalImage( $url ) {
1706                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1707                 $imagesexception = !empty( $imagesfrom );
1708                 $text = false;
1709                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1710                 if ( $imagesexception && is_array( $imagesfrom ) ) {
1711                         $imagematch = false;
1712                         foreach ( $imagesfrom as $match ) {
1713                                 if ( strpos( $url, $match ) === 0 ) {
1714                                         $imagematch = true;
1715                                         break;
1716                                 }
1717                         }
1718                 } elseif ( $imagesexception ) {
1719                         $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1720                 } else {
1721                         $imagematch = false;
1722                 }
1723                 if ( $this->mOptions->getAllowExternalImages()
1724                          || ( $imagesexception && $imagematch ) ) {
1725                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1726                                 # Image found
1727                                 $text = Linker::makeExternalImage( $url );
1728                         }
1729                 }
1730                 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1731                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1732                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1733                         foreach ( $whitelist as $entry ) {
1734                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1735                                 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1736                                         continue;
1737                                 }
1738                                 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1739                                         # Image matches a whitelist entry
1740                                         $text = Linker::makeExternalImage( $url );
1741                                         break;
1742                                 }
1743                         }
1744                 }
1745                 return $text;
1746         }
1747
1748         /**
1749          * Process [[ ]] wikilinks
1750          *
1751          * @param $s string
1752          *
1753          * @return String: processed text
1754          *
1755          * @private
1756          */
1757         function replaceInternalLinks( $s ) {
1758                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1759                 return $s;
1760         }
1761
1762         /**
1763          * Process [[ ]] wikilinks (RIL)
1764          * @return LinkHolderArray
1765          *
1766          * @private
1767          */
1768         function replaceInternalLinks2( &$s ) {
1769                 global $wgContLang;
1770
1771                 wfProfileIn( __METHOD__ );
1772
1773                 wfProfileIn( __METHOD__.'-setup' );
1774                 static $tc = FALSE, $e1, $e1_img;
1775                 # the % is needed to support urlencoded titles as well
1776                 if ( !$tc ) {
1777                         $tc = Title::legalChars() . '#%';
1778                         # Match a link having the form [[namespace:link|alternate]]trail
1779                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1780                         # Match cases where there is no "]]", which might still be images
1781                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1782                 }
1783
1784                 $holders = new LinkHolderArray( $this );
1785
1786                 # split the entire text string on occurences of [[
1787                 $a = StringUtils::explode( '[[', ' ' . $s );
1788                 # get the first element (all text up to first [[), and remove the space we added
1789                 $s = $a->current();
1790                 $a->next();
1791                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1792                 $s = substr( $s, 1 );
1793
1794                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1795                 $e2 = null;
1796                 if ( $useLinkPrefixExtension ) {
1797                         # Match the end of a line for a word that's not followed by whitespace,
1798                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1799                         $e2 = wfMsgForContent( 'linkprefix' );
1800                 }
1801
1802                 if ( is_null( $this->mTitle ) ) {
1803                         wfProfileOut( __METHOD__.'-setup' );
1804                         wfProfileOut( __METHOD__ );
1805                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1806                 }
1807                 $nottalk = !$this->mTitle->isTalkPage();
1808
1809                 if ( $useLinkPrefixExtension ) {
1810                         $m = array();
1811                         if ( preg_match( $e2, $s, $m ) ) {
1812                                 $first_prefix = $m[2];
1813                         } else {
1814                                 $first_prefix = false;
1815                         }
1816                 } else {
1817                         $prefix = '';
1818                 }
1819
1820                 if ( $wgContLang->hasVariants() ) {
1821                         $selflink = $wgContLang->autoConvertToAllVariants( $this->mTitle->getPrefixedText() );
1822                 } else {
1823                         $selflink = array( $this->mTitle->getPrefixedText() );
1824                 }
1825                 $useSubpages = $this->areSubpagesAllowed();
1826                 wfProfileOut( __METHOD__.'-setup' );
1827
1828                 # Loop for each link
1829                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1830                         # Check for excessive memory usage
1831                         if ( $holders->isBig() ) {
1832                                 # Too big
1833                                 # Do the existence check, replace the link holders and clear the array
1834                                 $holders->replace( $s );
1835                                 $holders->clear();
1836                         }
1837
1838                         if ( $useLinkPrefixExtension ) {
1839                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1840                                 if ( preg_match( $e2, $s, $m ) ) {
1841                                         $prefix = $m[2];
1842                                         $s = $m[1];
1843                                 } else {
1844                                         $prefix='';
1845                                 }
1846                                 # first link
1847                                 if ( $first_prefix ) {
1848                                         $prefix = $first_prefix;
1849                                         $first_prefix = false;
1850                                 }
1851                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1852                         }
1853
1854                         $might_be_img = false;
1855
1856                         wfProfileIn( __METHOD__."-e1" );
1857                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1858                                 $text = $m[2];
1859                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1860                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1861                                 # the real problem is with the $e1 regex
1862                                 # See bug 1300.
1863                                 #
1864                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1865                                 # and no image is in sight. See bug 2095.
1866                                 #
1867                                 if ( $text !== '' &&
1868                                         substr( $m[3], 0, 1 ) === ']' &&
1869                                         strpos( $text, '[' ) !== false
1870                                 )
1871                                 {
1872                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1873                                         $m[3] = substr( $m[3], 1 );
1874                                 }
1875                                 # fix up urlencoded title texts
1876                                 if ( strpos( $m[1], '%' ) !== false ) {
1877                                         # Should anchors '#' also be rejected?
1878                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), rawurldecode( $m[1] ) );
1879                                 }
1880                                 $trail = $m[3];
1881                         } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1882                                 $might_be_img = true;
1883                                 $text = $m[2];
1884                                 if ( strpos( $m[1], '%' ) !== false ) {
1885                                         $m[1] = rawurldecode( $m[1] );
1886                                 }
1887                                 $trail = "";
1888                         } else { # Invalid form; output directly
1889                                 $s .= $prefix . '[[' . $line ;
1890                                 wfProfileOut( __METHOD__."-e1" );
1891                                 continue;
1892                         }
1893                         wfProfileOut( __METHOD__."-e1" );
1894                         wfProfileIn( __METHOD__."-misc" );
1895
1896                         # Don't allow internal links to pages containing
1897                         # PROTO: where PROTO is a valid URL protocol; these
1898                         # should be external links.
1899                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1900                                 $s .= $prefix . '[[' . $line ;
1901                                 wfProfileOut( __METHOD__."-misc" );
1902                                 continue;
1903                         }
1904
1905                         # Make subpage if necessary
1906                         if ( $useSubpages ) {
1907                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1908                         } else {
1909                                 $link = $m[1];
1910                         }
1911
1912                         $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1913                         if ( !$noforce ) {
1914                                 # Strip off leading ':'
1915                                 $link = substr( $link, 1 );
1916                         }
1917
1918                         wfProfileOut( __METHOD__."-misc" );
1919                         wfProfileIn( __METHOD__."-title" );
1920                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1921                         if ( $nt === null ) {
1922                                 $s .= $prefix . '[[' . $line;
1923                                 wfProfileOut( __METHOD__."-title" );
1924                                 continue;
1925                         }
1926
1927                         $ns = $nt->getNamespace();
1928                         $iw = $nt->getInterWiki();
1929                         wfProfileOut( __METHOD__."-title" );
1930
1931                         if ( $might_be_img ) { # if this is actually an invalid link
1932                                 wfProfileIn( __METHOD__."-might_be_img" );
1933                                 if ( $ns == NS_FILE && $noforce ) { # but might be an image
1934                                         $found = false;
1935                                         while ( true ) {
1936                                                 # look at the next 'line' to see if we can close it there
1937                                                 $a->next();
1938                                                 $next_line = $a->current();
1939                                                 if ( $next_line === false || $next_line === null ) {
1940                                                         break;
1941                                                 }
1942                                                 $m = explode( ']]', $next_line, 3 );
1943                                                 if ( count( $m ) == 3 ) {
1944                                                         # the first ]] closes the inner link, the second the image
1945                                                         $found = true;
1946                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1947                                                         $trail = $m[2];
1948                                                         break;
1949                                                 } elseif ( count( $m ) == 2 ) {
1950                                                         # if there's exactly one ]] that's fine, we'll keep looking
1951                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1952                                                 } else {
1953                                                         # if $next_line is invalid too, we need look no further
1954                                                         $text .= '[[' . $next_line;
1955                                                         break;
1956                                                 }
1957                                         }
1958                                         if ( !$found ) {
1959                                                 # we couldn't find the end of this imageLink, so output it raw
1960                                                 # but don't ignore what might be perfectly normal links in the text we've examined
1961                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1962                                                 $s .= "{$prefix}[[$link|$text";
1963                                                 # note: no $trail, because without an end, there *is* no trail
1964                                                 wfProfileOut( __METHOD__."-might_be_img" );
1965                                                 continue;
1966                                         }
1967                                 } else { # it's not an image, so output it raw
1968                                         $s .= "{$prefix}[[$link|$text";
1969                                         # note: no $trail, because without an end, there *is* no trail
1970                                         wfProfileOut( __METHOD__."-might_be_img" );
1971                                         continue;
1972                                 }
1973                                 wfProfileOut( __METHOD__."-might_be_img" );
1974                         }
1975
1976                         $wasblank = ( $text  == '' );
1977                         if ( $wasblank ) {
1978                                 $text = $link;
1979                         } else {
1980                                 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
1981                                 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
1982                                 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
1983                                 #    -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
1984                                 $text = $this->doQuotes( $text );
1985                         }
1986
1987                         # Link not escaped by : , create the various objects
1988                         if ( $noforce ) {
1989
1990                                 # Interwikis
1991                                 wfProfileIn( __METHOD__."-interwiki" );
1992                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1993                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1994                                         $s = rtrim( $s . $prefix );
1995                                         $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
1996                                         wfProfileOut( __METHOD__."-interwiki" );
1997                                         continue;
1998                                 }
1999                                 wfProfileOut( __METHOD__."-interwiki" );
2000
2001                                 if ( $ns == NS_FILE ) {
2002                                         wfProfileIn( __METHOD__."-image" );
2003                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
2004                                                 if ( $wasblank ) {
2005                                                         # if no parameters were passed, $text
2006                                                         # becomes something like "File:Foo.png",
2007                                                         # which we don't want to pass on to the
2008                                                         # image generator
2009                                                         $text = '';
2010                                                 } else {
2011                                                         # recursively parse links inside the image caption
2012                                                         # actually, this will parse them in any other parameters, too,
2013                                                         # but it might be hard to fix that, and it doesn't matter ATM
2014                                                         $text = $this->replaceExternalLinks( $text );
2015                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
2016                                                 }
2017                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2018                                                 $s .= $prefix . $this->armorLinks(
2019                                                         $this->makeImage( $nt, $text, $holders ) ) . $trail;
2020                                         } else {
2021                                                 $s .= $prefix . $trail;
2022                                         }
2023                                         wfProfileOut( __METHOD__."-image" );
2024                                         continue;
2025                                 }
2026
2027                                 if ( $ns == NS_CATEGORY ) {
2028                                         wfProfileIn( __METHOD__."-category" );
2029                                         $s = rtrim( $s . "\n" ); # bug 87
2030
2031                                         if ( $wasblank ) {
2032                                                 $sortkey = $this->getDefaultSort();
2033                                         } else {
2034                                                 $sortkey = $text;
2035                                         }
2036                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
2037                                         $sortkey = str_replace( "\n", '', $sortkey );
2038                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
2039                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
2040
2041                                         /**
2042                                          * Strip the whitespace Category links produce, see bug 87
2043                                          * @todo We might want to use trim($tmp, "\n") here.
2044                                          */
2045                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
2046
2047                                         wfProfileOut( __METHOD__."-category" );
2048                                         continue;
2049                                 }
2050                         }
2051
2052                         # Self-link checking
2053                         if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
2054                                 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
2055                                         $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2056                                         continue;
2057                                 }
2058                         }
2059
2060                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
2061                         # @todo FIXME: Should do batch file existence checks, see comment below
2062                         if ( $ns == NS_MEDIA ) {
2063                                 wfProfileIn( __METHOD__."-media" );
2064                                 # Give extensions a chance to select the file revision for us
2065                                 $time = $sha1 = $descQuery = false;
2066                                 wfRunHooks( 'BeforeParserFetchFileAndTitle',
2067                                         array( $this, $nt, &$time, &$sha1, &$descQuery ) );
2068                                 # Fetch and register the file (file title may be different via hooks)
2069                                 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $time, $sha1 );
2070                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2071                                 $s .= $prefix . $this->armorLinks(
2072                                         Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2073                                 wfProfileOut( __METHOD__."-media" );
2074                                 continue;
2075                         }
2076
2077                         wfProfileIn( __METHOD__."-always_known" );
2078                         # Some titles, such as valid special pages or files in foreign repos, should
2079                         # be shown as bluelinks even though they're not included in the page table
2080                         #
2081                         # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2082                         # batch file existence checks for NS_FILE and NS_MEDIA
2083                         if ( $iw == '' && $nt->isAlwaysKnown() ) {
2084                                 $this->mOutput->addLink( $nt );
2085                                 $s .= $this->makeKnownLinkHolder( $nt, $text, array(), $trail, $prefix );
2086                         } else {
2087                                 # Links will be added to the output link list after checking
2088                                 $s .= $holders->makeHolder( $nt, $text, array(), $trail, $prefix );
2089                         }
2090                         wfProfileOut( __METHOD__."-always_known" );
2091                 }
2092                 wfProfileOut( __METHOD__ );
2093                 return $holders;
2094         }
2095
2096         /**
2097          * Render a forced-blue link inline; protect against double expansion of
2098          * URLs if we're in a mode that prepends full URL prefixes to internal links.
2099          * Since this little disaster has to split off the trail text to avoid
2100          * breaking URLs in the following text without breaking trails on the
2101          * wiki links, it's been made into a horrible function.
2102          *
2103          * @param $nt Title
2104          * @param $text String
2105          * @param $query Array or String
2106          * @param $trail String
2107          * @param $prefix String
2108          * @return String: HTML-wikitext mix oh yuck
2109          */
2110         function makeKnownLinkHolder( $nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
2111                 list( $inside, $trail ) = Linker::splitTrail( $trail );
2112
2113                 if ( is_string( $query ) ) {
2114                         $query = wfCgiToArray( $query );
2115                 }
2116                 if ( $text == '' ) {
2117                         $text = htmlspecialchars( $nt->getPrefixedText() );
2118                 }
2119
2120                 $link = Linker::linkKnown( $nt, "$prefix$text$inside", array(), $query );
2121
2122                 return $this->armorLinks( $link ) . $trail;
2123         }
2124
2125         /**
2126          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2127          * going to go through further parsing steps before inline URL expansion.
2128          *
2129          * Not needed quite as much as it used to be since free links are a bit
2130          * more sensible these days. But bracketed links are still an issue.
2131          *
2132          * @param $text String: more-or-less HTML
2133          * @return String: less-or-more HTML with NOPARSE bits
2134          */
2135         function armorLinks( $text ) {
2136                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
2137                         "{$this->mUniqPrefix}NOPARSE$1", $text );
2138         }
2139
2140         /**
2141          * Return true if subpage links should be expanded on this page.
2142          * @return Boolean
2143          */
2144         function areSubpagesAllowed() {
2145                 # Some namespaces don't allow subpages
2146                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2147         }
2148
2149         /**
2150          * Handle link to subpage if necessary
2151          *
2152          * @param $target String: the source of the link
2153          * @param &$text String: the link text, modified as necessary
2154          * @return string the full name of the link
2155          * @private
2156          */
2157         function maybeDoSubpageLink( $target, &$text ) {
2158                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2159         }
2160
2161         /**#@+
2162          * Used by doBlockLevels()
2163          * @private
2164          *
2165          * @return string
2166          */
2167         function closeParagraph() {
2168                 $result = '';
2169                 if ( $this->mLastSection != '' ) {
2170                         $result = '</' . $this->mLastSection  . ">\n";
2171                 }
2172                 $this->mInPre = false;
2173                 $this->mLastSection = '';
2174                 return $result;
2175         }
2176
2177         /**
2178          * getCommon() returns the length of the longest common substring
2179          * of both arguments, starting at the beginning of both.
2180          * @private
2181          *
2182          * @param $st1 string
2183          * @param $st2 string
2184          *
2185          * @return int
2186          */
2187         function getCommon( $st1, $st2 ) {
2188                 $fl = strlen( $st1 );
2189                 $shorter = strlen( $st2 );
2190                 if ( $fl < $shorter ) {
2191                         $shorter = $fl;
2192                 }
2193
2194                 for ( $i = 0; $i < $shorter; ++$i ) {
2195                         if ( $st1[$i] != $st2[$i] ) {
2196                                 break;
2197                         }
2198                 }
2199                 return $i;
2200         }
2201
2202         /**
2203          * These next three functions open, continue, and close the list
2204          * element appropriate to the prefix character passed into them.
2205          * @private
2206          *
2207          * @param $char char
2208          *
2209          * @return string
2210          */
2211         function openList( $char ) {
2212                 $result = $this->closeParagraph();
2213
2214                 if ( '*' === $char ) {
2215                         $result .= '<ul><li>';
2216                 } elseif ( '#' === $char ) {
2217                         $result .= '<ol><li>';
2218                 } elseif ( ':' === $char ) {
2219                         $result .= '<dl><dd>';
2220                 } elseif ( ';' === $char ) {
2221                         $result .= '<dl><dt>';
2222                         $this->mDTopen = true;
2223                 } else {
2224                         $result = '<!-- ERR 1 -->';
2225                 }
2226
2227                 return $result;
2228         }
2229
2230         /**
2231          * TODO: document
2232          * @param $char String
2233          * @private
2234          *
2235          * @return string
2236          */
2237         function nextItem( $char ) {
2238                 if ( '*' === $char || '#' === $char ) {
2239                         return '</li><li>';
2240                 } elseif ( ':' === $char || ';' === $char ) {
2241                         $close = '</dd>';
2242                         if ( $this->mDTopen ) {
2243                                 $close = '</dt>';
2244                         }
2245                         if ( ';' === $char ) {
2246                                 $this->mDTopen = true;
2247                                 return $close . '<dt>';
2248                         } else {
2249                                 $this->mDTopen = false;
2250                                 return $close . '<dd>';
2251                         }
2252                 }
2253                 return '<!-- ERR 2 -->';
2254         }
2255
2256         /**
2257          * TODO: document
2258          * @param $char String
2259          * @private
2260          *
2261          * @return string
2262          */
2263         function closeList( $char ) {
2264                 if ( '*' === $char ) {
2265                         $text = '</li></ul>';
2266                 } elseif ( '#' === $char ) {
2267                         $text = '</li></ol>';
2268                 } elseif ( ':' === $char ) {
2269                         if ( $this->mDTopen ) {
2270                                 $this->mDTopen = false;
2271                                 $text = '</dt></dl>';
2272                         } else {
2273                                 $text = '</dd></dl>';
2274                         }
2275                 } else {
2276                         return '<!-- ERR 3 -->';
2277                 }
2278                 return $text."\n";
2279         }
2280         /**#@-*/
2281
2282         /**
2283          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2284          *
2285          * @param $text String
2286          * @param $linestart Boolean: whether or not this is at the start of a line.
2287          * @private
2288          * @return string the lists rendered as HTML
2289          */
2290         function doBlockLevels( $text, $linestart ) {
2291                 wfProfileIn( __METHOD__ );
2292
2293                 # Parsing through the text line by line.  The main thing
2294                 # happening here is handling of block-level elements p, pre,
2295                 # and making lists from lines starting with * # : etc.
2296                 #
2297                 $textLines = StringUtils::explode( "\n", $text );
2298
2299                 $lastPrefix = $output = '';
2300                 $this->mDTopen = $inBlockElem = false;
2301                 $prefixLength = 0;
2302                 $paragraphStack = false;
2303
2304                 foreach ( $textLines as $oLine ) {
2305                         # Fix up $linestart
2306                         if ( !$linestart ) {
2307                                 $output .= $oLine;
2308                                 $linestart = true;
2309                                 continue;
2310                         }
2311                         # * = ul
2312                         # # = ol
2313                         # ; = dt
2314                         # : = dd
2315
2316                         $lastPrefixLength = strlen( $lastPrefix );
2317                         $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2318                         $preOpenMatch = preg_match( '/<pre/i', $oLine );
2319                         # If not in a <pre> element, scan for and figure out what prefixes are there.
2320                         if ( !$this->mInPre ) {
2321                                 # Multiple prefixes may abut each other for nested lists.
2322                                 $prefixLength = strspn( $oLine, '*#:;' );
2323                                 $prefix = substr( $oLine, 0, $prefixLength );
2324
2325                                 # eh?
2326                                 # ; and : are both from definition-lists, so they're equivalent
2327                                 #  for the purposes of determining whether or not we need to open/close
2328                                 #  elements.
2329                                 $prefix2 = str_replace( ';', ':', $prefix );
2330                                 $t = substr( $oLine, $prefixLength );
2331                                 $this->mInPre = (bool)$preOpenMatch;
2332                         } else {
2333                                 # Don't interpret any other prefixes in preformatted text
2334                                 $prefixLength = 0;
2335                                 $prefix = $prefix2 = '';
2336                                 $t = $oLine;
2337                         }
2338
2339                         # List generation
2340                         if ( $prefixLength && $lastPrefix === $prefix2 ) {
2341                                 # Same as the last item, so no need to deal with nesting or opening stuff
2342                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2343                                 $paragraphStack = false;
2344
2345                                 if ( substr( $prefix, -1 ) === ';') {
2346                                         # The one nasty exception: definition lists work like this:
2347                                         # ; title : definition text
2348                                         # So we check for : in the remainder text to split up the
2349                                         # title and definition, without b0rking links.
2350                                         $term = $t2 = '';
2351                                         if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2352                                                 $t = $t2;
2353                                                 $output .= $term . $this->nextItem( ':' );
2354                                         }
2355                                 }
2356                         } elseif ( $prefixLength || $lastPrefixLength ) {
2357                                 # We need to open or close prefixes, or both.
2358
2359                                 # Either open or close a level...
2360                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2361                                 $paragraphStack = false;
2362
2363                                 # Close all the prefixes which aren't shared.
2364                                 while ( $commonPrefixLength < $lastPrefixLength ) {
2365                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2366                                         --$lastPrefixLength;
2367                                 }
2368
2369                                 # Continue the current prefix if appropriate.
2370                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2371                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2372                                 }
2373
2374                                 # Open prefixes where appropriate.
2375                                 while ( $prefixLength > $commonPrefixLength ) {
2376                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2377                                         $output .= $this->openList( $char );
2378
2379                                         if ( ';' === $char ) {
2380                                                 # @todo FIXME: This is dupe of code above
2381                                                 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2382                                                         $t = $t2;
2383                                                         $output .= $term . $this->nextItem( ':' );
2384                                                 }
2385                                         }
2386                                         ++$commonPrefixLength;
2387                                 }
2388                                 $lastPrefix = $prefix2;
2389                         }
2390
2391                         # If we have no prefixes, go to paragraph mode.
2392                         if ( 0 == $prefixLength ) {
2393                                 wfProfileIn( __METHOD__."-paragraph" );
2394                                 # No prefix (not in list)--go to paragraph mode
2395                                 # XXX: use a stack for nestable elements like span, table and div
2396                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2397                                 $closematch = preg_match(
2398                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2399                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2400                                 if ( $openmatch or $closematch ) {
2401                                         $paragraphStack = false;
2402                                         # TODO bug 5718: paragraph closed
2403                                         $output .= $this->closeParagraph();
2404                                         if ( $preOpenMatch and !$preCloseMatch ) {
2405                                                 $this->mInPre = true;
2406                                         }
2407                                         $inBlockElem = !$closematch;
2408                                 } elseif ( !$inBlockElem && !$this->mInPre ) {
2409                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) ) {
2410                                                 # pre
2411                                                 if ( $this->mLastSection !== 'pre' ) {
2412                                                         $paragraphStack = false;
2413                                                         $output .= $this->closeParagraph().'<pre>';
2414                                                         $this->mLastSection = 'pre';
2415                                                 }
2416                                                 $t = substr( $t, 1 );
2417                                         } else {
2418                                                 # paragraph
2419                                                 if ( trim( $t ) === '' ) {
2420                                                         if ( $paragraphStack ) {
2421                                                                 $output .= $paragraphStack.'<br />';
2422                                                                 $paragraphStack = false;
2423                                                                 $this->mLastSection = 'p';
2424                                                         } else {
2425                                                                 if ( $this->mLastSection !== 'p' ) {
2426                                                                         $output .= $this->closeParagraph();
2427                                                                         $this->mLastSection = '';
2428                                                                         $paragraphStack = '<p>';
2429                                                                 } else {
2430                                                                         $paragraphStack = '</p><p>';
2431                                                                 }
2432                                                         }
2433                                                 } else {
2434                                                         if ( $paragraphStack ) {
2435                                                                 $output .= $paragraphStack;
2436                                                                 $paragraphStack = false;
2437                                                                 $this->mLastSection = 'p';
2438                                                         } elseif ( $this->mLastSection !== 'p' ) {
2439                                                                 $output .= $this->closeParagraph().'<p>';
2440                                                                 $this->mLastSection = 'p';
2441                                                         }
2442                                                 }
2443                                         }
2444                                 }
2445                                 wfProfileOut( __METHOD__."-paragraph" );
2446                         }
2447                         # somewhere above we forget to get out of pre block (bug 785)
2448                         if ( $preCloseMatch && $this->mInPre ) {
2449                                 $this->mInPre = false;
2450                         }
2451                         if ( $paragraphStack === false ) {
2452                                 $output .= $t."\n";
2453                         }
2454                 }
2455                 while ( $prefixLength ) {
2456                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2457                         --$prefixLength;
2458                 }
2459                 if ( $this->mLastSection != '' ) {
2460                         $output .= '</' . $this->mLastSection . '>';
2461                         $this->mLastSection = '';
2462                 }
2463
2464                 wfProfileOut( __METHOD__ );
2465                 return $output;
2466         }
2467
2468         /**
2469          * Split up a string on ':', ignoring any occurences inside tags
2470          * to prevent illegal overlapping.
2471          *
2472          * @param $str String the string to split
2473          * @param &$before String set to everything before the ':'
2474          * @param &$after String set to everything after the ':'
2475          * @return String the position of the ':', or false if none found
2476          */
2477         function findColonNoLinks( $str, &$before, &$after ) {
2478                 wfProfileIn( __METHOD__ );
2479
2480                 $pos = strpos( $str, ':' );
2481                 if ( $pos === false ) {
2482                         # Nothing to find!
2483                         wfProfileOut( __METHOD__ );
2484                         return false;
2485                 }
2486
2487                 $lt = strpos( $str, '<' );
2488                 if ( $lt === false || $lt > $pos ) {
2489                         # Easy; no tag nesting to worry about
2490                         $before = substr( $str, 0, $pos );
2491                         $after = substr( $str, $pos+1 );
2492                         wfProfileOut( __METHOD__ );
2493                         return $pos;
2494                 }
2495
2496                 # Ugly state machine to walk through avoiding tags.
2497                 $state = self::COLON_STATE_TEXT;
2498                 $stack = 0;
2499                 $len = strlen( $str );
2500                 for( $i = 0; $i < $len; $i++ ) {
2501                         $c = $str[$i];
2502
2503                         switch( $state ) {
2504                         # (Using the number is a performance hack for common cases)
2505                         case 0: # self::COLON_STATE_TEXT:
2506                                 switch( $c ) {
2507                                 case "<":
2508                                         # Could be either a <start> tag or an </end> tag
2509                                         $state = self::COLON_STATE_TAGSTART;
2510                                         break;
2511                                 case ":":
2512                                         if ( $stack == 0 ) {
2513                                                 # We found it!
2514                                                 $before = substr( $str, 0, $i );
2515                                                 $after = substr( $str, $i + 1 );
2516                                                 wfProfileOut( __METHOD__ );
2517                                                 return $i;
2518                                         }
2519                                         # Embedded in a tag; don't break it.
2520                                         break;
2521                                 default:
2522                                         # Skip ahead looking for something interesting
2523                                         $colon = strpos( $str, ':', $i );
2524                                         if ( $colon === false ) {
2525                                                 # Nothing else interesting
2526                                                 wfProfileOut( __METHOD__ );
2527                                                 return false;
2528                                         }
2529                                         $lt = strpos( $str, '<', $i );
2530                                         if ( $stack === 0 ) {
2531                                                 if ( $lt === false || $colon < $lt ) {
2532                                                         # We found it!
2533                                                         $before = substr( $str, 0, $colon );
2534                                                         $after = substr( $str, $colon + 1 );
2535                                                         wfProfileOut( __METHOD__ );
2536                                                         return $i;
2537                                                 }
2538                                         }
2539                                         if ( $lt === false ) {
2540                                                 # Nothing else interesting to find; abort!
2541                                                 # We're nested, but there's no close tags left. Abort!
2542                                                 break 2;
2543                                         }
2544                                         # Skip ahead to next tag start
2545                                         $i = $lt;
2546                                         $state = self::COLON_STATE_TAGSTART;
2547                                 }
2548                                 break;
2549                         case 1: # self::COLON_STATE_TAG:
2550                                 # In a <tag>
2551                                 switch( $c ) {
2552                                 case ">":
2553                                         $stack++;
2554                                         $state = self::COLON_STATE_TEXT;
2555                                         break;
2556                                 case "/":
2557                                         # Slash may be followed by >?
2558                                         $state = self::COLON_STATE_TAGSLASH;
2559                                         break;
2560                                 default:
2561                                         # ignore
2562                                 }
2563                                 break;
2564                         case 2: # self::COLON_STATE_TAGSTART:
2565                                 switch( $c ) {
2566                                 case "/":
2567                                         $state = self::COLON_STATE_CLOSETAG;
2568                                         break;
2569                                 case "!":
2570                                         $state = self::COLON_STATE_COMMENT;
2571                                         break;
2572                                 case ">":
2573                                         # Illegal early close? This shouldn't happen D:
2574                                         $state = self::COLON_STATE_TEXT;
2575                                         break;
2576                                 default:
2577                                         $state = self::COLON_STATE_TAG;
2578                                 }
2579                                 break;
2580                         case 3: # self::COLON_STATE_CLOSETAG:
2581                                 # In a </tag>
2582                                 if ( $c === ">" ) {
2583                                         $stack--;
2584                                         if ( $stack < 0 ) {
2585                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2586                                                 wfProfileOut( __METHOD__ );
2587                                                 return false;
2588                                         }
2589                                         $state = self::COLON_STATE_TEXT;
2590                                 }
2591                                 break;
2592                         case self::COLON_STATE_TAGSLASH:
2593                                 if ( $c === ">" ) {
2594                                         # Yes, a self-closed tag <blah/>
2595                                         $state = self::COLON_STATE_TEXT;
2596                                 } else {
2597                                         # Probably we're jumping the gun, and this is an attribute
2598                                         $state = self::COLON_STATE_TAG;
2599                                 }
2600                                 break;
2601                         case 5: # self::COLON_STATE_COMMENT:
2602                                 if ( $c === "-" ) {
2603                                         $state = self::COLON_STATE_COMMENTDASH;
2604                                 }
2605                                 break;
2606                         case self::COLON_STATE_COMMENTDASH:
2607                                 if ( $c === "-" ) {
2608                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2609                                 } else {
2610                                         $state = self::COLON_STATE_COMMENT;
2611                                 }
2612                                 break;
2613                         case self::COLON_STATE_COMMENTDASHDASH:
2614                                 if ( $c === ">" ) {
2615                                         $state = self::COLON_STATE_TEXT;
2616                                 } else {
2617                                         $state = self::COLON_STATE_COMMENT;
2618                                 }
2619                                 break;
2620                         default:
2621                                 throw new MWException( "State machine error in " . __METHOD__ );
2622                         }
2623                 }
2624                 if ( $stack > 0 ) {
2625                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2626                         wfProfileOut( __METHOD__ );
2627                         return false;
2628                 }
2629                 wfProfileOut( __METHOD__ );
2630                 return false;
2631         }
2632
2633         /**
2634          * Return value of a magic variable (like PAGENAME)
2635          *
2636          * @private
2637          *
2638          * @param $index integer
2639          * @param $frame PPFrame
2640          *
2641          * @return string
2642          */
2643         function getVariableValue( $index, $frame = false ) {
2644                 global $wgContLang, $wgSitename, $wgServer;
2645                 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2646
2647                 /**
2648                  * Some of these require message or data lookups and can be
2649                  * expensive to check many times.
2650                  */
2651                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2652                         if ( isset( $this->mVarCache[$index] ) ) {
2653                                 return $this->mVarCache[$index];
2654                         }
2655                 }
2656
2657                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2658                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2659
2660                 # Use the time zone
2661                 global $wgLocaltimezone;
2662                 if ( isset( $wgLocaltimezone ) ) {
2663                         $oldtz = date_default_timezone_get();
2664                         date_default_timezone_set( $wgLocaltimezone );
2665                 }
2666
2667                 $localTimestamp = date( 'YmdHis', $ts );
2668                 $localMonth = date( 'm', $ts );
2669                 $localMonth1 = date( 'n', $ts );
2670                 $localMonthName = date( 'n', $ts );
2671                 $localDay = date( 'j', $ts );
2672                 $localDay2 = date( 'd', $ts );
2673                 $localDayOfWeek = date( 'w', $ts );
2674                 $localWeek = date( 'W', $ts );
2675                 $localYear = date( 'Y', $ts );
2676                 $localHour = date( 'H', $ts );
2677                 if ( isset( $wgLocaltimezone ) ) {
2678                         date_default_timezone_set( $oldtz );
2679                 }
2680
2681                 switch ( $index ) {
2682                         case 'currentmonth':
2683                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2684                                 break;
2685                         case 'currentmonth1':
2686                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2687                                 break;
2688                         case 'currentmonthname':
2689                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2690                                 break;
2691                         case 'currentmonthnamegen':
2692                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2693                                 break;
2694                         case 'currentmonthabbrev':
2695                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2696                                 break;
2697                         case 'currentday':
2698                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2699                                 break;
2700                         case 'currentday2':
2701                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2702                                 break;
2703                         case 'localmonth':
2704                                 $value = $wgContLang->formatNum( $localMonth );
2705                                 break;
2706                         case 'localmonth1':
2707                                 $value = $wgContLang->formatNum( $localMonth1 );
2708                                 break;
2709                         case 'localmonthname':
2710                                 $value = $wgContLang->getMonthName( $localMonthName );
2711                                 break;
2712                         case 'localmonthnamegen':
2713                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2714                                 break;
2715                         case 'localmonthabbrev':
2716                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2717                                 break;
2718                         case 'localday':
2719                                 $value = $wgContLang->formatNum( $localDay );
2720                                 break;
2721                         case 'localday2':
2722                                 $value = $wgContLang->formatNum( $localDay2 );
2723                                 break;
2724                         case 'pagename':
2725                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2726                                 break;
2727                         case 'pagenamee':
2728                                 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2729                                 break;
2730                         case 'fullpagename':
2731                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2732                                 break;
2733                         case 'fullpagenamee':
2734                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2735                                 break;
2736                         case 'subpagename':
2737                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2738                                 break;
2739                         case 'subpagenamee':
2740                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2741                                 break;
2742                         case 'basepagename':
2743                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2744                                 break;
2745                         case 'basepagenamee':
2746                                 $value = wfEscapeWikiText( wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) ) );
2747                                 break;
2748                         case 'talkpagename':
2749                                 if ( $this->mTitle->canTalk() ) {
2750                                         $talkPage = $this->mTitle->getTalkPage();
2751                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2752                                 } else {
2753                                         $value = '';
2754                                 }
2755                                 break;
2756                         case 'talkpagenamee':
2757                                 if ( $this->mTitle->canTalk() ) {
2758                                         $talkPage = $this->mTitle->getTalkPage();
2759                                         $value = wfEscapeWikiText( $talkPage->getPrefixedUrl() );
2760                                 } else {
2761                                         $value = '';
2762                                 }
2763                                 break;
2764                         case 'subjectpagename':
2765                                 $subjPage = $this->mTitle->getSubjectPage();
2766                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2767                                 break;
2768                         case 'subjectpagenamee':
2769                                 $subjPage = $this->mTitle->getSubjectPage();
2770                                 $value = wfEscapeWikiText( $subjPage->getPrefixedUrl() );
2771                                 break;
2772                         case 'revisionid':
2773                                 # Let the edit saving system know we should parse the page
2774                                 # *after* a revision ID has been assigned.
2775                                 $this->mOutput->setFlag( 'vary-revision' );
2776                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2777                                 $value = $this->mRevisionId;
2778                                 break;
2779                         case 'revisionday':
2780                                 # Let the edit saving system know we should parse the page
2781                                 # *after* a revision ID has been assigned. This is for null edits.
2782                                 $this->mOutput->setFlag( 'vary-revision' );
2783                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2784                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2785                                 break;
2786                         case 'revisionday2':
2787                                 # Let the edit saving system know we should parse the page
2788                                 # *after* a revision ID has been assigned. This is for null edits.
2789                                 $this->mOutput->setFlag( 'vary-revision' );
2790                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2791                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2792                                 break;
2793                         case 'revisionmonth':
2794                                 # Let the edit saving system know we should parse the page
2795                                 # *after* a revision ID has been assigned. This is for null edits.
2796                                 $this->mOutput->setFlag( 'vary-revision' );
2797                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2798                                 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2799                                 break;
2800                         case 'revisionmonth1':
2801                                 # Let the edit saving system know we should parse the page
2802                                 # *after* a revision ID has been assigned. This is for null edits.
2803                                 $this->mOutput->setFlag( 'vary-revision' );
2804                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2805                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2806                                 break;
2807                         case 'revisionyear':
2808                                 # Let the edit saving system know we should parse the page
2809                                 # *after* a revision ID has been assigned. This is for null edits.
2810                                 $this->mOutput->setFlag( 'vary-revision' );
2811                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2812                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2813                                 break;
2814                         case 'revisiontimestamp':
2815                                 # Let the edit saving system know we should parse the page
2816                                 # *after* a revision ID has been assigned. This is for null edits.
2817                                 $this->mOutput->setFlag( 'vary-revision' );
2818                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2819                                 $value = $this->getRevisionTimestamp();
2820                                 break;
2821                         case 'revisionuser':
2822                                 # Let the edit saving system know we should parse the page
2823                                 # *after* a revision ID has been assigned. This is for null edits.
2824                                 $this->mOutput->setFlag( 'vary-revision' );
2825                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2826                                 $value = $this->getRevisionUser();
2827                                 break;
2828                         case 'namespace':
2829                                 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2830                                 break;
2831                         case 'namespacee':
2832                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2833                                 break;
2834                         case 'talkspace':
2835                                 $value = $this->mTitle->canTalk() ? str_replace( '_',' ',$this->mTitle->getTalkNsText() ) : '';
2836                                 break;
2837                         case 'talkspacee':
2838                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2839                                 break;
2840                         case 'subjectspace':
2841                                 $value = $this->mTitle->getSubjectNsText();
2842                                 break;
2843                         case 'subjectspacee':
2844                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2845                                 break;
2846                         case 'currentdayname':
2847                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2848                                 break;
2849                         case 'currentyear':
2850                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2851                                 break;
2852                         case 'currenttime':
2853                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2854                                 break;
2855                         case 'currenthour':
2856                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2857                                 break;
2858                         case 'currentweek':
2859                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2860                                 # int to remove the padding
2861                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2862                                 break;
2863                         case 'currentdow':
2864                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2865                                 break;
2866                         case 'localdayname':
2867                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2868                                 break;
2869                         case 'localyear':
2870                                 $value = $wgContLang->formatNum( $localYear, true );
2871                                 break;
2872                         case 'localtime':
2873                                 $value = $wgContLang->time( $localTimestamp, false, false );
2874                                 break;
2875                         case 'localhour':
2876                                 $value = $wgContLang->formatNum( $localHour, true );
2877                                 break;
2878                         case 'localweek':
2879                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2880                                 # int to remove the padding
2881                                 $value = $wgContLang->formatNum( (int)$localWeek );
2882                                 break;
2883                         case 'localdow':
2884                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2885                                 break;
2886                         case 'numberofarticles':
2887                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2888                                 break;
2889                         case 'numberoffiles':
2890                                 $value = $wgContLang->formatNum( SiteStats::images() );
2891                                 break;
2892                         case 'numberofusers':
2893                                 $value = $wgContLang->formatNum( SiteStats::users() );
2894                                 break;
2895                         case 'numberofactiveusers':
2896                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2897                                 break;
2898                         case 'numberofpages':
2899                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2900                                 break;
2901                         case 'numberofadmins':
2902                                 $value = $wgContLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2903                                 break;
2904                         case 'numberofedits':
2905                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2906                                 break;
2907                         case 'numberofviews':
2908                                 $value = $wgContLang->formatNum( SiteStats::views() );
2909                                 break;
2910                         case 'currenttimestamp':
2911                                 $value = wfTimestamp( TS_MW, $ts );
2912                                 break;
2913                         case 'localtimestamp':
2914                                 $value = $localTimestamp;
2915                                 break;
2916                         case 'currentversion':
2917                                 $value = SpecialVersion::getVersion();
2918                                 break;
2919                         case 'articlepath':
2920                                 return $wgArticlePath;
2921                         case 'sitename':
2922                                 return $wgSitename;
2923                         case 'server':
2924                                 return $wgServer;
2925                         case 'servername':
2926                                 wfSuppressWarnings(); # May give an E_WARNING in PHP < 5.3.3
2927                                 $serverName = parse_url( $wgServer, PHP_URL_HOST );
2928                                 wfRestoreWarnings();
2929                                 return $serverName ? $serverName : $wgServer;
2930                         case 'scriptpath':
2931                                 return $wgScriptPath;
2932                         case 'stylepath':
2933                                 return $wgStylePath;
2934                         case 'directionmark':
2935                                 return $wgContLang->getDirMark();
2936                         case 'contentlanguage':
2937                                 global $wgLanguageCode;
2938                                 return $wgLanguageCode;
2939                         default:
2940                                 $ret = null;
2941                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
2942                                         return $ret;
2943                                 } else {
2944                                         return null;
2945                                 }
2946                 }
2947
2948                 if ( $index ) {
2949                         $this->mVarCache[$index] = $value;
2950                 }
2951
2952                 return $value;
2953         }
2954
2955         /**
2956          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2957          *
2958          * @private
2959          */
2960         function initialiseVariables() {
2961                 wfProfileIn( __METHOD__ );
2962                 $variableIDs = MagicWord::getVariableIDs();
2963                 $substIDs = MagicWord::getSubstIDs();
2964
2965                 $this->mVariables = new MagicWordArray( $variableIDs );
2966                 $this->mSubstWords = new MagicWordArray( $substIDs );
2967                 wfProfileOut( __METHOD__ );
2968         }
2969
2970         /**
2971          * Preprocess some wikitext and return the document tree.
2972          * This is the ghost of replace_variables().
2973          *
2974          * @param $text String: The text to parse
2975          * @param $flags Integer: bitwise combination of:
2976          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2977          *                                     included. Default is to assume a direct page view.
2978          *
2979          * The generated DOM tree must depend only on the input text and the flags.
2980          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2981          *
2982          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2983          * change in the DOM tree for a given text, must be passed through the section identifier
2984          * in the section edit link and thus back to extractSections().
2985          *
2986          * The output of this function is currently only cached in process memory, but a persistent
2987          * cache may be implemented at a later date which takes further advantage of these strict
2988          * dependency requirements.
2989          *
2990          * @private
2991          *
2992          * @return PPNode
2993          */
2994         function preprocessToDom( $text, $flags = 0 ) {
2995                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2996                 return $dom;
2997         }
2998
2999         /**
3000          * Return a three-element array: leading whitespace, string contents, trailing whitespace
3001          *
3002          * @param $s string
3003          *
3004          * @return array
3005          */
3006         public static function splitWhitespace( $s ) {
3007                 $ltrimmed = ltrim( $s );
3008                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
3009                 $trimmed = rtrim( $ltrimmed );
3010                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
3011                 if ( $diff > 0 ) {
3012                         $w2 = substr( $ltrimmed, -$diff );
3013                 } else {
3014                         $w2 = '';
3015                 }
3016                 return array( $w1, $trimmed, $w2 );
3017         }
3018
3019         /**
3020          * Replace magic variables, templates, and template arguments
3021          * with the appropriate text. Templates are substituted recursively,
3022          * taking care to avoid infinite loops.
3023          *
3024          * Note that the substitution depends on value of $mOutputType:
3025          *  self::OT_WIKI: only {{subst:}} templates
3026          *  self::OT_PREPROCESS: templates but not extension tags
3027          *  self::OT_HTML: all templates and extension tags
3028          *
3029          * @param $text String the text to transform
3030          * @param $frame PPFrame Object describing the arguments passed to the template.
3031          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
3032          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
3033          * @param $argsOnly Boolean only do argument (triple-brace) expansion, not double-brace expansion
3034          * @private
3035          *
3036          * @return string
3037          */
3038         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3039                 # Is there any text? Also, Prevent too big inclusions!
3040                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
3041                         return $text;
3042                 }
3043                 wfProfileIn( __METHOD__ );
3044
3045                 if ( $frame === false ) {
3046                         $frame = $this->getPreprocessor()->newFrame();
3047                 } elseif ( !( $frame instanceof PPFrame ) ) {
3048                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
3049                         $frame = $this->getPreprocessor()->newCustomFrame( $frame );
3050                 }
3051
3052                 $dom = $this->preprocessToDom( $text );
3053                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
3054                 $text = $frame->expand( $dom, $flags );
3055
3056                 wfProfileOut( __METHOD__ );
3057                 return $text;
3058         }
3059
3060         /**
3061          * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3062          *
3063          * @param $args array
3064          *
3065          * @return array
3066          */
3067         static function createAssocArgs( $args ) {
3068                 $assocArgs = array();
3069                 $index = 1;
3070                 foreach ( $args as $arg ) {
3071                         $eqpos = strpos( $arg, '=' );
3072                         if ( $eqpos === false ) {
3073                                 $assocArgs[$index++] = $arg;
3074                         } else {
3075                                 $name = trim( substr( $arg, 0, $eqpos ) );
3076                                 $value = trim( substr( $arg, $eqpos+1 ) );
3077                                 if ( $value === false ) {
3078                                         $value = '';
3079                                 }
3080                                 if ( $name !== false ) {
3081                                         $assocArgs[$name] = $value;
3082                                 }
3083                         }
3084                 }
3085
3086                 return $assocArgs;
3087         }
3088
3089         /**
3090          * Warn the user when a parser limitation is reached
3091          * Will warn at most once the user per limitation type
3092          *
3093          * @param $limitationType String: should be one of:
3094          *   'expensive-parserfunction' (corresponding messages:
3095          *       'expensive-parserfunction-warning',
3096          *       'expensive-parserfunction-category')
3097          *   'post-expand-template-argument' (corresponding messages:
3098          *       'post-expand-template-argument-warning',
3099          *       'post-expand-template-argument-category')
3100          *   'post-expand-template-inclusion' (corresponding messages:
3101          *       'post-expand-template-inclusion-warning',
3102          *       'post-expand-template-inclusion-category')
3103          * @param $current Current value
3104          * @param $max Maximum allowed, when an explicit limit has been
3105          *       exceeded, provide the values (optional)
3106          */
3107         function limitationWarn( $limitationType, $current=null, $max=null) {
3108                 # does no harm if $current and $max are present but are unnecessary for the message
3109                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
3110                 $this->mOutput->addWarning( $warning );
3111                 $this->addTrackingCategory( "$limitationType-category" );
3112         }
3113
3114         /**
3115          * Return the text of a template, after recursively
3116          * replacing any variables or templates within the template.
3117          *
3118          * @param $piece Array: the parts of the template
3119          *  $piece['title']: the title, i.e. the part before the |
3120          *  $piece['parts']: the parameter array
3121          *  $piece['lineStart']: whether the brace was at the start of a line
3122          * @param $frame PPFrame The current frame, contains template arguments
3123          * @return String: the text of the template
3124          * @private
3125          */
3126         function braceSubstitution( $piece, $frame ) {
3127                 global $wgContLang, $wgNonincludableNamespaces;
3128                 wfProfileIn( __METHOD__ );
3129                 wfProfileIn( __METHOD__.'-setup' );
3130
3131                 # Flags
3132                 $found = false;             # $text has been filled
3133                 $nowiki = false;            # wiki markup in $text should be escaped
3134                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
3135                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
3136                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
3137                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
3138
3139                 # Title object, where $text came from
3140                 $title = null;
3141
3142                 # $part1 is the bit before the first |, and must contain only title characters.
3143                 # Various prefixes will be stripped from it later.
3144                 $titleWithSpaces = $frame->expand( $piece['title'] );
3145                 $part1 = trim( $titleWithSpaces );
3146                 $titleText = false;
3147
3148                 # Original title text preserved for various purposes
3149                 $originalTitle = $part1;
3150
3151                 # $args is a list of argument nodes, starting from index 0, not including $part1
3152                 # @todo FIXME: If piece['parts'] is null then the call to getLength() below won't work b/c this $args isn't an object
3153                 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
3154                 wfProfileOut( __METHOD__.'-setup' );
3155                 wfProfileIn( __METHOD__."-title-$originalTitle" );
3156
3157                 # SUBST
3158                 wfProfileIn( __METHOD__.'-modifiers' );
3159                 if ( !$found ) {
3160
3161                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3162
3163                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
3164                         # Decide whether to expand template or keep wikitext as-is.
3165                         if ( $this->ot['wiki'] ) {
3166                                 if ( $substMatch === false ) {
3167                                         $literal = true;  # literal when in PST with no prefix
3168                                 } else {
3169                                         $literal = false; # expand when in PST with subst: or safesubst:
3170                                 }
3171                         } else {
3172                                 if ( $substMatch == 'subst' ) {
3173                                         $literal = true;  # literal when not in PST with plain subst:
3174                                 } else {
3175                                         $literal = false; # expand when not in PST with safesubst: or no prefix
3176                                 }
3177                         }
3178                         if ( $literal ) {
3179                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3180                                 $isLocalObj = true;
3181                                 $found = true;
3182                         }
3183                 }
3184
3185                 # Variables
3186                 if ( !$found && $args->getLength() == 0 ) {
3187                         $id = $this->mVariables->matchStartToEnd( $part1 );
3188                         if ( $id !== false ) {
3189                                 $text = $this->getVariableValue( $id, $frame );
3190                                 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3191                                         $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3192                                 }
3193                                 $found = true;
3194                         }
3195                 }
3196
3197                 # MSG, MSGNW and RAW
3198                 if ( !$found ) {
3199                         # Check for MSGNW:
3200                         $mwMsgnw = MagicWord::get( 'msgnw' );
3201                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3202                                 $nowiki = true;
3203                         } else {
3204                                 # Remove obsolete MSG:
3205                                 $mwMsg = MagicWord::get( 'msg' );
3206                                 $mwMsg->matchStartAndRemove( $part1 );
3207                         }
3208
3209                         # Check for RAW:
3210                         $mwRaw = MagicWord::get( 'raw' );
3211                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3212                                 $forceRawInterwiki = true;
3213                         }
3214                 }
3215                 wfProfileOut( __METHOD__.'-modifiers' );
3216
3217                 # Parser functions
3218                 if ( !$found ) {
3219                         wfProfileIn( __METHOD__ . '-pfunc' );
3220
3221                         $colonPos = strpos( $part1, ':' );
3222                         if ( $colonPos !== false ) {
3223                                 # Case sensitive functions
3224                                 $function = substr( $part1, 0, $colonPos );
3225                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3226                                         $function = $this->mFunctionSynonyms[1][$function];
3227                                 } else {
3228                                         # Case insensitive functions
3229                                         $function = $wgContLang->lc( $function );
3230                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3231                                                 $function = $this->mFunctionSynonyms[0][$function];
3232                                         } else {
3233                                                 $function = false;
3234                                         }
3235                                 }
3236                                 if ( $function ) {
3237                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3238                                         $initialArgs = array( &$this );
3239                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3240                                         if ( $flags & SFH_OBJECT_ARGS ) {
3241                                                 # Add a frame parameter, and pass the arguments as an array
3242                                                 $allArgs = $initialArgs;
3243                                                 $allArgs[] = $frame;
3244                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3245                                                         $funcArgs[] = $args->item( $i );
3246                                                 }
3247                                                 $allArgs[] = $funcArgs;
3248                                         } else {
3249                                                 # Convert arguments to plain text
3250                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3251                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
3252                                                 }
3253                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3254                                         }
3255
3256                                         # Workaround for PHP bug 35229 and similar
3257                                         if ( !is_callable( $callback ) ) {
3258                                                 wfProfileOut( __METHOD__ . '-pfunc' );
3259                                                 wfProfileOut( __METHOD__ );
3260                                                 throw new MWException( "Tag hook for $function is not callable\n" );
3261                                         }
3262                                         $result = call_user_func_array( $callback, $allArgs );
3263                                         $found = true;
3264                                         $noparse = true;
3265                                         $preprocessFlags = 0;
3266
3267                                         if ( is_array( $result ) ) {
3268                                                 if ( isset( $result[0] ) ) {
3269                                                         $text = $result[0];
3270                                                         unset( $result[0] );
3271                                                 }
3272
3273                                                 # Extract flags into the local scope
3274                                                 # This allows callers to set flags such as nowiki, found, etc.
3275                                                 extract( $result );
3276                                         } else {
3277                                                 $text = $result;
3278                                         }
3279                                         if ( !$noparse ) {
3280                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
3281                                                 $isChildObj = true;
3282                                         }
3283                                 }
3284                         }
3285                         wfProfileOut( __METHOD__ . '-pfunc' );
3286                 }
3287
3288                 # Finish mangling title and then check for loops.
3289                 # Set $title to a Title object and $titleText to the PDBK
3290                 if ( !$found ) {
3291                         $ns = NS_TEMPLATE;
3292                         # Split the title into page and subpage
3293                         $subpage = '';
3294                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3295                         if ( $subpage !== '' ) {
3296                                 $ns = $this->mTitle->getNamespace();
3297                         }
3298                         $title = Title::newFromText( $part1, $ns );
3299                         if ( $title ) {
3300                                 $titleText = $title->getPrefixedText();
3301                                 # Check for language variants if the template is not found
3302                                 if ( $wgContLang->hasVariants() && $title->getArticleID() == 0 ) {
3303                                         $wgContLang->findVariantLink( $part1, $title, true );
3304                                 }
3305                                 # Do recursion depth check
3306                                 $limit = $this->mOptions->getMaxTemplateDepth();
3307                                 if ( $frame->depth >= $limit ) {
3308                                         $found = true;
3309                                         $text = '<span class="error">'
3310                                                 . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit )
3311                                                 . '</span>';
3312                                 }
3313                         }
3314                 }
3315
3316                 # Load from database
3317                 if ( !$found && $title ) {
3318                         wfProfileIn( __METHOD__ . '-loadtpl' );
3319                         if ( !$title->isExternal() ) {
3320                                 if ( $title->getNamespace() == NS_SPECIAL
3321                                         && $this->mOptions->getAllowSpecialInclusion()
3322                                         && $this->ot['html'] )
3323                                 {
3324                                         $pageArgs = array();
3325                                         for ( $i = 0; $i < $args->getLength(); $i++ ) {
3326                                                 $bits = $args->item( $i )->splitArg();
3327                                                 if ( strval( $bits['index'] ) === '' ) {
3328                                                         $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3329                                                         $value = trim( $frame->expand( $bits['value'] ) );
3330                                                         $pageArgs[$name] = $value;
3331                                                 }
3332                                         }
3333                                         $context = new RequestContext;
3334                                         $context->setTitle( $title );
3335                                         $context->setRequest( new FauxRequest( $pageArgs ) );
3336                                         $context->setUser( $this->getUser() );
3337                                         $context->setLang( Language::factory( $this->mOptions->getUserLang() ) );
3338                                         $ret = SpecialPageFactory::capturePath( $title, $context );
3339                                         if ( $ret ) {
3340                                                 $text = $context->getOutput()->getHTML();
3341                                                 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3342                                                 $found = true;
3343                                                 $isHTML = true;
3344                                                 $this->disableCache();
3345                                         }
3346                                 } elseif ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3347                                         $found = false; # access denied
3348                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3349                                 } else {
3350                                         list( $text, $title ) = $this->getTemplateDom( $title );
3351                                         if ( $text !== false ) {
3352                                                 $found = true;
3353                                                 $isChildObj = true;
3354                                         }
3355                                 }
3356
3357                                 # If the title is valid but undisplayable, make a link to it
3358                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3359                                         $text = "[[:$titleText]]";
3360                                         $found = true;
3361                                 }
3362                         } elseif ( $title->isTrans() ) {
3363                                 # Interwiki transclusion
3364                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3365                                         $text = $this->interwikiTransclude( $title, 'render' );
3366                                         $isHTML = true;
3367                                 } else {
3368                                         $text = $this->interwikiTransclude( $title, 'raw' );
3369                                         # Preprocess it like a template
3370                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3371                                         $isChildObj = true;
3372                                 }
3373                                 $found = true;
3374                         }
3375
3376                         # Do infinite loop check
3377                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3378                         if ( !$frame->loopCheck( $title ) ) {
3379                                 $found = true;
3380                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3381                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3382                         }
3383                         wfProfileOut( __METHOD__ . '-loadtpl' );
3384                 }
3385
3386                 # If we haven't found text to substitute by now, we're done
3387                 # Recover the source wikitext and return it
3388                 if ( !$found ) {
3389                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3390                         wfProfileOut( __METHOD__."-title-$originalTitle" );
3391                         wfProfileOut( __METHOD__ );
3392                         return array( 'object' => $text );
3393                 }
3394
3395                 # Expand DOM-style return values in a child frame
3396                 if ( $isChildObj ) {
3397                         # Clean up argument array
3398                         $newFrame = $frame->newChild( $args, $title );
3399
3400                         if ( $nowiki ) {
3401                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3402                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3403                                 # Expansion is eligible for the empty-frame cache
3404                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3405                                         $text = $this->mTplExpandCache[$titleText];
3406                                 } else {
3407                                         $text = $newFrame->expand( $text );
3408                                         $this->mTplExpandCache[$titleText] = $text;
3409                                 }
3410                         } else {
3411                                 # Uncached expansion
3412                                 $text = $newFrame->expand( $text );
3413                         }
3414                 }
3415                 if ( $isLocalObj && $nowiki ) {
3416                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3417                         $isLocalObj = false;
3418                 }
3419
3420                 # Replace raw HTML by a placeholder
3421                 # Add a blank line preceding, to prevent it from mucking up
3422                 # immediately preceding headings
3423                 if ( $isHTML ) {
3424                         $text = "\n\n" . $this->insertStripItem( $text );
3425                 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3426                         # Escape nowiki-style return values
3427                         $text = wfEscapeWikiText( $text );
3428                 } elseif ( is_string( $text )
3429                         && !$piece['lineStart']
3430                         && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3431                 {
3432                         # Bug 529: if the template begins with a table or block-level
3433                         # element, it should be treated as beginning a new line.
3434                         # This behaviour is somewhat controversial.
3435                         $text = "\n" . $text;
3436                 }
3437
3438                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3439                         # Error, oversize inclusion
3440                         if ( $titleText !== false ) {
3441                                 # Make a working, properly escaped link if possible (bug 23588)
3442                                 $text = "[[:$titleText]]";
3443                         } else {
3444                                 # This will probably not be a working link, but at least it may
3445                                 # provide some hint of where the problem is
3446                                 preg_replace( '/^:/', '', $originalTitle );
3447                                 $text = "[[:$originalTitle]]";
3448                         }
3449                         $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3450                         $this->limitationWarn( 'post-expand-template-inclusion' );
3451                 }
3452
3453                 if ( $isLocalObj ) {
3454                         $ret = array( 'object' => $text );
3455                 } else {
3456                         $ret = array( 'text' => $text );
3457                 }
3458
3459                 wfProfileOut( __METHOD__."-title-$originalTitle" );
3460                 wfProfileOut( __METHOD__ );
3461                 return $ret;
3462         }
3463
3464         /**
3465          * Get the semi-parsed DOM representation of a template with a given title,
3466          * and its redirect destination title. Cached.
3467          *
3468          * @param $title Title
3469          *
3470          * @return array
3471          */
3472         function getTemplateDom( $title ) {
3473                 $cacheTitle = $title;
3474                 $titleText = $title->getPrefixedDBkey();
3475
3476                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3477                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3478                         $title = Title::makeTitle( $ns, $dbk );
3479                         $titleText = $title->getPrefixedDBkey();
3480                 }
3481                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3482                         return array( $this->mTplDomCache[$titleText], $title );
3483                 }
3484
3485                 # Cache miss, go to the database
3486                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3487
3488                 if ( $text === false ) {
3489                         $this->mTplDomCache[$titleText] = false;
3490                         return array( false, $title );
3491                 }
3492
3493                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3494                 $this->mTplDomCache[ $titleText ] = $dom;
3495
3496                 if ( !$title->equals( $cacheTitle ) ) {
3497                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3498                                 array( $title->getNamespace(), $cdb = $title->getDBkey() );
3499                 }
3500
3501                 return array( $dom, $title );
3502         }
3503
3504         /**
3505          * Fetch the unparsed text of a template and register a reference to it.
3506          * @param Title $title
3507          * @return Array ( string or false, Title )
3508          */
3509         function fetchTemplateAndTitle( $title ) {
3510                 $templateCb = $this->mOptions->getTemplateCallback(); # Defaults to Parser::statelessFetchTemplate()
3511                 $stuff = call_user_func( $templateCb, $title, $this );
3512                 $text = $stuff['text'];
3513                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3514                 if ( isset( $stuff['deps'] ) ) {
3515                         foreach ( $stuff['deps'] as $dep ) {
3516                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3517                         }
3518                 }
3519                 return array( $text, $finalTitle );
3520         }
3521
3522         /**
3523          * Fetch the unparsed text of a template and register a reference to it.
3524          * @param Title $title
3525          * @return mixed string or false
3526          */
3527         function fetchTemplate( $title ) {
3528                 $rv = $this->fetchTemplateAndTitle( $title );
3529                 return $rv[0];
3530         }
3531
3532         /**
3533          * Static function to get a template
3534          * Can be overridden via ParserOptions::setTemplateCallback().
3535          *
3536          * @parma $title Title
3537          * @param $parser Parser
3538          *
3539          * @return array
3540          */
3541         static function statelessFetchTemplate( $title, $parser = false ) {
3542                 $text = $skip = false;
3543                 $finalTitle = $title;
3544                 $deps = array();
3545
3546                 # Loop to fetch the article, with up to 1 redirect
3547                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3548                         # Give extensions a chance to select the revision instead
3549                         $id = false; # Assume current
3550                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle',
3551                                 array( $parser, $title, &$skip, &$id ) );
3552
3553                         if ( $skip ) {
3554                                 $text = false;
3555                                 $deps[] = array(
3556                                         'title'         => $title,
3557                                         'page_id'       => $title->getArticleID(),
3558                                         'rev_id'        => null
3559                                 );
3560                                 break;
3561                         }
3562                         # Get the revision
3563                         $rev = $id
3564                                 ? Revision::newFromId( $id )
3565                                 : Revision::newFromTitle( $title );
3566                         $rev_id = $rev ? $rev->getId() : 0;
3567                         # If there is no current revision, there is no page
3568                         if ( $id === false && !$rev ) {
3569                                 $linkCache = LinkCache::singleton();
3570                                 $linkCache->addBadLinkObj( $title );
3571                         }
3572
3573                         $deps[] = array(
3574                                 'title'         => $title,
3575                                 'page_id'       => $title->getArticleID(),
3576                                 'rev_id'        => $rev_id );
3577                         if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3578                                 # We fetched a rev from a different title; register it too...
3579                                 $deps[] = array(
3580                                         'title'         => $rev->getTitle(),
3581                                         'page_id'       => $rev->getPage(),
3582                                         'rev_id'        => $rev_id );
3583                         }
3584
3585                         if ( $rev ) {
3586                                 $text = $rev->getText();
3587                         } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3588                                 global $wgContLang;
3589                                 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3590                                 if ( !$message->exists() ) {
3591                                         $text = false;
3592                                         break;
3593                                 }
3594                                 $text = $message->plain();
3595                         } else {
3596                                 break;
3597                         }
3598                         if ( $text === false ) {
3599                                 break;
3600                         }
3601                         # Redirect?
3602                         $finalTitle = $title;
3603                         $title = Title::newFromRedirect( $text );
3604                 }
3605                 return array(
3606                         'text' => $text,
3607                         'finalTitle' => $finalTitle,
3608                         'deps' => $deps );
3609         }
3610
3611         /**
3612          * Fetch a file and its title and register a reference to it.
3613          * @param Title $title
3614          * @param string $time MW timestamp
3615          * @param string $sha1 base 36 SHA-1
3616          * @return File|false
3617          */
3618         function fetchFile( $title, $time = false, $sha1 = false ) {
3619                 $res = $this->fetchFileAndTitle( $title, $time, $sha1 );
3620                 return $res[0];
3621         }
3622
3623         /**
3624          * Fetch a file and its title and register a reference to it.
3625          * @param Title $title
3626          * @param string $time MW timestamp
3627          * @param string $sha1 base 36 SHA-1
3628          * @return Array ( File or false, Title of file )
3629          */
3630         function fetchFileAndTitle( $title, $time = false, $sha1 = false ) {
3631                 if ( $time === '0' ) {
3632                         $file = false; // broken thumbnail forced by hook
3633                 } elseif ( $sha1 ) { // get by (sha1,timestamp)
3634                         $file = RepoGroup::singleton()->findFileFromKey( $sha1, array( 'time' => $time ) );
3635                 } else { // get by (name,timestamp)
3636                         $file = wfFindFile( $title, array( 'time' => $time ) );
3637                 }
3638                 $time = $file ? $file->getTimestamp() : false;
3639                 $sha1 = $file ? $file->getSha1() : false;
3640                 # Register the file as a dependency...
3641                 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3642                 if ( $file && !$title->equals( $file->getTitle() ) ) {
3643                         # Update fetched file title
3644                         $title = $file->getTitle();
3645                         if ( is_null( $file->getRedirectedTitle() ) ) {
3646                                 # This file was not a redirect, but the title does not match.
3647                                 # Register under the new name because otherwise the link will
3648                                 # get lost.
3649                                 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3650                         }
3651                 }
3652                 return array( $file, $title );
3653         }
3654
3655         /**
3656          * Transclude an interwiki link.
3657          *
3658          * @param $title Title
3659          * @param $action
3660          *
3661          * @return string
3662          */
3663         function interwikiTransclude( $title, $action ) {
3664                 global $wgEnableScaryTranscluding;
3665
3666                 if ( !$wgEnableScaryTranscluding ) {
3667                         return wfMsgForContent('scarytranscludedisabled');
3668                 }
3669
3670                 $url = $title->getFullUrl( "action=$action" );
3671
3672                 if ( strlen( $url ) > 255 ) {
3673                         return wfMsgForContent( 'scarytranscludetoolong' );
3674                 }
3675                 return $this->fetchScaryTemplateMaybeFromCache( $url );
3676         }
3677
3678         /**
3679          * @param $url string
3680          * @return Mixed|String
3681          */
3682         function fetchScaryTemplateMaybeFromCache( $url ) {
3683                 global $wgTranscludeCacheExpiry;
3684                 $dbr = wfGetDB( DB_SLAVE );
3685                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3686                 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3687                                 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3688                 if ( $obj ) {
3689                         return $obj->tc_contents;
3690                 }
3691
3692                 $text = Http::get( $url );
3693                 if ( !$text ) {
3694                         return wfMsgForContent( 'scarytranscludefailed', $url );
3695                 }
3696
3697                 $dbw = wfGetDB( DB_MASTER );
3698                 $dbw->replace( 'transcache', array('tc_url'), array(
3699                         'tc_url' => $url,
3700                         'tc_time' => $dbw->timestamp( time() ),
3701                         'tc_contents' => $text)
3702                 );
3703                 return $text;
3704         }
3705
3706         /**
3707          * Triple brace replacement -- used for template arguments
3708          * @private
3709          *
3710          * @param $peice array
3711          * @param $frame PPFrame
3712          *
3713          * @return array
3714          */
3715         function argSubstitution( $piece, $frame ) {
3716                 wfProfileIn( __METHOD__ );
3717
3718                 $error = false;
3719                 $parts = $piece['parts'];
3720                 $nameWithSpaces = $frame->expand( $piece['title'] );
3721                 $argName = trim( $nameWithSpaces );
3722                 $object = false;
3723                 $text = $frame->getArgument( $argName );
3724                 if (  $text === false && $parts->getLength() > 0
3725                   && (
3726                         $this->ot['html']
3727                         || $this->ot['pre']
3728                         || ( $this->ot['wiki'] && $frame->isTemplate() )
3729                   )
3730                 ) {
3731                         # No match in frame, use the supplied default
3732                         $object = $parts->item( 0 )->getChildren();
3733                 }
3734                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3735                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3736                         $this->limitationWarn( 'post-expand-template-argument' );
3737                 }
3738
3739                 if ( $text === false && $object === false ) {
3740                         # No match anywhere
3741                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3742                 }
3743                 if ( $error !== false ) {
3744                         $text .= $error;
3745                 }
3746                 if ( $object !== false ) {
3747                         $ret = array( 'object' => $object );
3748                 } else {
3749                         $ret = array( 'text' => $text );
3750                 }
3751
3752                 wfProfileOut( __METHOD__ );
3753                 return $ret;
3754         }
3755
3756         /**
3757          * Return the text to be used for a given extension tag.
3758          * This is the ghost of strip().
3759          *
3760          * @param $params Associative array of parameters:
3761          *     name       PPNode for the tag name
3762          *     attr       PPNode for unparsed text where tag attributes are thought to be
3763          *     attributes Optional associative array of parsed attributes
3764          *     inner      Contents of extension element
3765          *     noClose    Original text did not have a close tag
3766          * @param $frame PPFrame
3767          *
3768          * @return string
3769          */
3770         function extensionSubstitution( $params, $frame ) {
3771                 $name = $frame->expand( $params['name'] );
3772                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3773                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3774                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3775
3776                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3777                         ( $this->ot['html'] || $this->ot['pre'] );
3778                 if ( $isFunctionTag ) {
3779                         $markerType = 'none';
3780                 } else {
3781                         $markerType = 'general';
3782                 }
3783                 if ( $this->ot['html'] || $isFunctionTag ) {
3784                         $name = strtolower( $name );
3785                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3786                         if ( isset( $params['attributes'] ) ) {
3787                                 $attributes = $attributes + $params['attributes'];
3788                         }
3789
3790                         if ( isset( $this->mTagHooks[$name] ) ) {
3791                                 # Workaround for PHP bug 35229 and similar
3792                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3793                                         throw new MWException( "Tag hook for $name is not callable\n" );
3794                                 }
3795                                 $output = call_user_func_array( $this->mTagHooks[$name],
3796                                         array( $content, $attributes, $this, $frame ) );
3797                         } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3798                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3799                                 if ( !is_callable( $callback ) ) {
3800                                         throw new MWException( "Tag hook for $name is not callable\n" );
3801                                 }
3802
3803                                 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3804                         } else {
3805                                 $output = '<span class="error">Invalid tag extension name: ' .
3806                                         htmlspecialchars( $name ) . '</span>';
3807                         }
3808
3809                         if ( is_array( $output ) ) {
3810                                 # Extract flags to local scope (to override $markerType)
3811                                 $flags = $output;
3812                                 $output = $flags[0];
3813                                 unset( $flags[0] );
3814                                 extract( $flags );
3815                         }
3816                 } else {
3817                         if ( is_null( $attrText ) ) {
3818                                 $attrText = '';
3819                         }
3820                         if ( isset( $params['attributes'] ) ) {
3821                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3822                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3823                                                 htmlspecialchars( $attrValue ) . '"';
3824                                 }
3825                         }
3826                         if ( $content === null ) {
3827                                 $output = "<$name$attrText/>";
3828                         } else {
3829                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3830                                 $output = "<$name$attrText>$content$close";
3831                         }
3832                 }
3833
3834                 if ( $markerType === 'none' ) {
3835                         return $output;
3836                 } elseif ( $markerType === 'nowiki' ) {
3837                         $this->mStripState->addNoWiki( $marker, $output );
3838                 } elseif ( $markerType === 'general' ) {
3839                         $this->mStripState->addGeneral( $marker, $output );
3840                 } else {
3841                         throw new MWException( __METHOD__.': invalid marker type' );
3842                 }
3843                 return $marker;
3844         }
3845
3846         /**
3847          * Increment an include size counter
3848          *
3849          * @param $type String: the type of expansion
3850          * @param $size Integer: the size of the text
3851          * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
3852          */
3853         function incrementIncludeSize( $type, $size ) {
3854                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3855                         return false;
3856                 } else {
3857                         $this->mIncludeSizes[$type] += $size;
3858                         return true;
3859                 }
3860         }
3861
3862         /**
3863          * Increment the expensive function count
3864          *
3865          * @return Boolean: false if the limit has been exceeded
3866          */
3867         function incrementExpensiveFunctionCount() {
3868                 global $wgExpensiveParserFunctionLimit;
3869                 $this->mExpensiveFunctionCount++;
3870                 if ( $this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit ) {
3871                         return true;
3872                 }
3873                 return false;
3874         }
3875
3876         /**
3877          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3878          * Fills $this->mDoubleUnderscores, returns the modified text
3879          *
3880          * @param $text string
3881          *
3882          * @return string
3883          */
3884         function doDoubleUnderscore( $text ) {
3885                 wfProfileIn( __METHOD__ );
3886
3887                 # The position of __TOC__ needs to be recorded
3888                 $mw = MagicWord::get( 'toc' );
3889                 if ( $mw->match( $text ) ) {
3890                         $this->mShowToc = true;
3891                         $this->mForceTocPosition = true;
3892
3893                         # Set a placeholder. At the end we'll fill it in with the TOC.
3894                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3895
3896                         # Only keep the first one.
3897                         $text = $mw->replace( '', $text );
3898                 }
3899
3900                 # Now match and remove the rest of them
3901                 $mwa = MagicWord::getDoubleUnderscoreArray();
3902                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3903
3904                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3905                         $this->mOutput->mNoGallery = true;
3906                 }
3907                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3908                         $this->mShowToc = false;
3909                 }
3910                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3911                         $this->addTrackingCategory( 'hidden-category-category' );
3912                 }
3913                 # (bug 8068) Allow control over whether robots index a page.
3914                 #
3915                 # @todo FIXME: Bug 14899: __INDEX__ always overrides __NOINDEX__ here!  This
3916                 # is not desirable, the last one on the page should win.
3917                 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3918                         $this->mOutput->setIndexPolicy( 'noindex' );
3919                         $this->addTrackingCategory( 'noindex-category' );
3920                 }
3921                 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3922                         $this->mOutput->setIndexPolicy( 'index' );
3923                         $this->addTrackingCategory( 'index-category' );
3924                 }
3925
3926                 # Cache all double underscores in the database
3927                 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3928                         $this->mOutput->setProperty( $key, '' );
3929                 }
3930
3931                 wfProfileOut( __METHOD__ );
3932                 return $text;
3933         }
3934
3935         /**
3936          * Add a tracking category, getting the title from a system message,
3937          * or print a debug message if the title is invalid.
3938          *
3939          * @param $msg String: message key
3940          * @return Boolean: whether the addition was successful
3941          */
3942         protected function addTrackingCategory( $msg ) {
3943                 $cat = wfMsgForContent( $msg );
3944
3945                 # Allow tracking categories to be disabled by setting them to "-"
3946                 if ( $cat === '-' ) {
3947                         return false;
3948                 }
3949
3950                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3951                 if ( $containerCategory ) {
3952                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3953                         return true;
3954                 } else {
3955                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3956                         return false;
3957                 }
3958         }
3959
3960         /**
3961          * This function accomplishes several tasks:
3962          * 1) Auto-number headings if that option is enabled
3963          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3964          * 3) Add a Table of contents on the top for users who have enabled the option
3965          * 4) Auto-anchor headings
3966          *
3967          * It loops through all headlines, collects the necessary data, then splits up the
3968          * string and re-inserts the newly formatted headlines.
3969          *
3970          * @param $text String
3971          * @param $origText String: original, untouched wikitext
3972          * @param $isMain Boolean
3973          * @private
3974          */
3975         function formatHeadings( $text, $origText, $isMain=true ) {
3976                 global $wgMaxTocLevel, $wgHtml5, $wgExperimentalHtmlIds;
3977
3978                 # Inhibit editsection links if requested in the page
3979                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3980                         $showEditLink = 0;
3981                 } else {
3982                         $showEditLink = $this->mOptions->getEditSection();
3983                 }
3984                 if ( $showEditLink ) {
3985                         $this->mOutput->setEditSectionTokens( true );
3986                 }
3987
3988                 # Get all headlines for numbering them and adding funky stuff like [edit]
3989                 # links - this is for later, but we need the number of headlines right now
3990                 $matches = array();
3991                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3992
3993                 # if there are fewer than 4 headlines in the article, do not show TOC
3994                 # unless it's been explicitly enabled.
3995                 $enoughToc = $this->mShowToc &&
3996                         ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3997
3998                 # Allow user to stipulate that a page should have a "new section"
3999                 # link added via __NEWSECTIONLINK__
4000                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
4001                         $this->mOutput->setNewSection( true );
4002                 }
4003
4004                 # Allow user to remove the "new section"
4005                 # link via __NONEWSECTIONLINK__
4006                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
4007                         $this->mOutput->hideNewSection( true );
4008                 }
4009
4010                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4011                 # override above conditions and always show TOC above first header
4012                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4013                         $this->mShowToc = true;
4014                         $enoughToc = true;
4015                 }
4016
4017                 # headline counter
4018                 $headlineCount = 0;
4019                 $numVisible = 0;
4020
4021                 # Ugh .. the TOC should have neat indentation levels which can be
4022                 # passed to the skin functions. These are determined here
4023                 $toc = '';
4024                 $full = '';
4025                 $head = array();
4026                 $sublevelCount = array();
4027                 $levelCount = array();
4028                 $level = 0;
4029                 $prevlevel = 0;
4030                 $toclevel = 0;
4031                 $prevtoclevel = 0;
4032                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
4033                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4034                 $oldType = $this->mOutputType;
4035                 $this->setOutputType( self::OT_WIKI );
4036                 $frame = $this->getPreprocessor()->newFrame();
4037                 $root = $this->preprocessToDom( $origText );
4038                 $node = $root->getFirstChild();
4039                 $byteOffset = 0;
4040                 $tocraw = array();
4041                 $refers = array();
4042
4043                 foreach ( $matches[3] as $headline ) {
4044                         $isTemplate = false;
4045                         $titleText = false;
4046                         $sectionIndex = false;
4047                         $numbering = '';
4048                         $markerMatches = array();
4049                         if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
4050                                 $serial = $markerMatches[1];
4051                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4052                                 $isTemplate = ( $titleText != $baseTitleText );
4053                                 $headline = preg_replace( "/^$markerRegex/", "", $headline );
4054                         }
4055
4056                         if ( $toclevel ) {
4057                                 $prevlevel = $level;
4058                         }
4059                         $level = $matches[1][$headlineCount];
4060
4061                         if ( $level > $prevlevel ) {
4062                                 # Increase TOC level
4063                                 $toclevel++;
4064                                 $sublevelCount[$toclevel] = 0;
4065                                 if ( $toclevel<$wgMaxTocLevel ) {
4066                                         $prevtoclevel = $toclevel;
4067                                         $toc .= Linker::tocIndent();
4068                                         $numVisible++;
4069                                 }
4070                         } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4071                                 # Decrease TOC level, find level to jump to
4072
4073                                 for ( $i = $toclevel; $i > 0; $i-- ) {
4074                                         if ( $levelCount[$i] == $level ) {
4075                                                 # Found last matching level
4076                                                 $toclevel = $i;
4077                                                 break;
4078                                         } elseif ( $levelCount[$i] < $level ) {
4079                                                 # Found first matching level below current level
4080                                                 $toclevel = $i + 1;
4081                                                 break;
4082                                         }
4083                                 }
4084                                 if ( $i == 0 ) {
4085                                         $toclevel = 1;
4086                                 }
4087                                 if ( $toclevel<$wgMaxTocLevel ) {
4088                                         if ( $prevtoclevel < $wgMaxTocLevel ) {
4089                                                 # Unindent only if the previous toc level was shown :p
4090                                                 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4091                                                 $prevtoclevel = $toclevel;
4092                                         } else {
4093                                                 $toc .= Linker::tocLineEnd();
4094                                         }
4095                                 }
4096                         } else {
4097                                 # No change in level, end TOC line
4098                                 if ( $toclevel<$wgMaxTocLevel ) {
4099                                         $toc .= Linker::tocLineEnd();
4100                                 }
4101                         }
4102
4103                         $levelCount[$toclevel] = $level;
4104
4105                         # count number of headlines for each level
4106                         @$sublevelCount[$toclevel]++;
4107                         $dot = 0;
4108                         for( $i = 1; $i <= $toclevel; $i++ ) {
4109                                 if ( !empty( $sublevelCount[$i] ) ) {
4110                                         if ( $dot ) {
4111                                                 $numbering .= '.';
4112                                         }
4113                                         $numbering .= $this->getFunctionLang()->formatNum( $sublevelCount[$i] );
4114                                         $dot = 1;
4115                                 }
4116                         }
4117
4118                         # The safe header is a version of the header text safe to use for links
4119                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4120                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
4121
4122                         # Remove link placeholders by the link text.
4123                         #     <!--LINK number-->
4124                         # turns into
4125                         #     link text with suffix
4126                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
4127
4128                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
4129                         $tocline = preg_replace(
4130                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
4131                                 array( '',                          '<$1>' ),
4132                                 $safeHeadline
4133                         );
4134                         $tocline = trim( $tocline );
4135
4136                         # For the anchor, strip out HTML-y stuff period
4137                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
4138                         $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4139
4140                         # Save headline for section edit hint before it's escaped
4141                         $headlineHint = $safeHeadline;
4142
4143                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
4144                                 # For reverse compatibility, provide an id that's
4145                                 # HTML4-compatible, like we used to.
4146                                 #
4147                                 # It may be worth noting, academically, that it's possible for
4148                                 # the legacy anchor to conflict with a non-legacy headline
4149                                 # anchor on the page.  In this case likely the "correct" thing
4150                                 # would be to either drop the legacy anchors or make sure
4151                                 # they're numbered first.  However, this would require people
4152                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4153                                 # manually, so let's not bother worrying about it.
4154                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
4155                                         array( 'noninitial', 'legacy' ) );
4156                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4157
4158                                 if ( $legacyHeadline == $safeHeadline ) {
4159                                         # No reason to have both (in fact, we can't)
4160                                         $legacyHeadline = false;
4161                                 }
4162                         } else {
4163                                 $legacyHeadline = false;
4164                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
4165                                         'noninitial' );
4166                         }
4167
4168                         # HTML names must be case-insensitively unique (bug 10721).
4169                         # This does not apply to Unicode characters per
4170                         # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
4171                         # @todo FIXME: We may be changing them depending on the current locale.
4172                         $arrayKey = strtolower( $safeHeadline );
4173                         if ( $legacyHeadline === false ) {
4174                                 $legacyArrayKey = false;
4175                         } else {
4176                                 $legacyArrayKey = strtolower( $legacyHeadline );
4177                         }
4178
4179                         # count how many in assoc. array so we can track dupes in anchors
4180                         if ( isset( $refers[$arrayKey] ) ) {
4181                                 $refers[$arrayKey]++;
4182                         } else {
4183                                 $refers[$arrayKey] = 1;
4184                         }
4185                         if ( isset( $refers[$legacyArrayKey] ) ) {
4186                                 $refers[$legacyArrayKey]++;
4187                         } else {
4188                                 $refers[$legacyArrayKey] = 1;
4189                         }
4190
4191                         # Don't number the heading if it is the only one (looks silly)
4192                         if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4193                                 # the two are different if the line contains a link
4194                                 $headline = $numbering . ' ' . $headline;
4195                         }
4196
4197                         # Create the anchor for linking from the TOC to the section
4198                         $anchor = $safeHeadline;
4199                         $legacyAnchor = $legacyHeadline;
4200                         if ( $refers[$arrayKey] > 1 ) {
4201                                 $anchor .= '_' . $refers[$arrayKey];
4202                         }
4203                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
4204                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
4205                         }
4206                         if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4207                                 $toc .= Linker::tocLine( $anchor, $tocline,
4208                                         $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4209                         }
4210
4211                         # Add the section to the section tree
4212                         # Find the DOM node for this header
4213                         while ( $node && !$isTemplate ) {
4214                                 if ( $node->getName() === 'h' ) {
4215                                         $bits = $node->splitHeading();
4216                                         if ( $bits['i'] == $sectionIndex ) {
4217                                                 break;
4218                                         }
4219                                 }
4220                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4221                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4222                                 $node = $node->getNextSibling();
4223                         }
4224                         $tocraw[] = array(
4225                                 'toclevel' => $toclevel,
4226                                 'level' => $level,
4227                                 'line' => $tocline,
4228                                 'number' => $numbering,
4229                                 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4230                                 'fromtitle' => $titleText,
4231                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
4232                                 'anchor' => $anchor,
4233                         );
4234
4235                         # give headline the correct <h#> tag
4236                         if ( $showEditLink && $sectionIndex !== false ) {
4237                                 // Output edit section links as markers with styles that can be customized by skins
4238                                 if ( $isTemplate ) {
4239                                         # Put a T flag in the section identifier, to indicate to extractSections()
4240                                         # that sections inside <includeonly> should be counted.
4241                                         $editlinkArgs = array( $titleText, "T-$sectionIndex"/*, null */ );
4242                                 } else {
4243                                         $editlinkArgs = array( $this->mTitle->getPrefixedText(), $sectionIndex, $headlineHint );
4244                                 }
4245                                 // We use a bit of pesudo-xml for editsection markers. The language converter is run later on
4246                                 // Using a UNIQ style marker leads to the converter screwing up the tokens when it converts stuff
4247                                 // And trying to insert strip tags fails too. At this point all real inputted tags have already been escaped
4248                                 // so we don't have to worry about a user trying to input one of these markers directly.
4249                                 // We use a page and section attribute to stop the language converter from converting these important bits
4250                                 // of data, but put the headline hint inside a content block because the language converter is supposed to
4251                                 // be able to convert that piece of data.
4252                                 $editlink = '<mw:editsection page="' . htmlspecialchars($editlinkArgs[0]);
4253                                 $editlink .= '" section="' . htmlspecialchars($editlinkArgs[1]) .'"';
4254                                 if ( isset($editlinkArgs[2]) ) {
4255                                         $editlink .= '>' . $editlinkArgs[2] . '</mw:editsection>';
4256                                 } else {
4257                                         $editlink .= '/>';
4258                                 }
4259                         } else {
4260                                 $editlink = '';
4261                         }
4262                         $head[$headlineCount] = Linker::makeHeadline( $level,
4263                                 $matches['attrib'][$headlineCount], $anchor, $headline,
4264                                 $editlink, $legacyAnchor );
4265
4266                         $headlineCount++;
4267                 }
4268
4269                 $this->setOutputType( $oldType );
4270
4271                 # Never ever show TOC if no headers
4272                 if ( $numVisible < 1 ) {
4273                         $enoughToc = false;
4274                 }
4275
4276                 if ( $enoughToc ) {
4277                         if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4278                                 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4279                         }
4280                         $toc = Linker::tocList( $toc, $this->mOptions->getUserLang() );
4281                         $this->mOutput->setTOCHTML( $toc );
4282                 }
4283
4284                 if ( $isMain ) {
4285                         $this->mOutput->setSections( $tocraw );
4286                 }
4287
4288                 # split up and insert constructed headlines
4289                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
4290                 $i = 0;
4291
4292                 // build an array of document sections
4293                 $sections = array();
4294                 foreach ( $blocks as $block ) {
4295                         // $head is zero-based, sections aren't.
4296                         if ( empty( $head[$i - 1] ) ) {
4297                                 $sections[$i] = $block;
4298                         } else {
4299                                 $sections[$i] = $head[$i - 1] . $block;
4300                         }
4301
4302                         /**
4303                          * Send a hook, one per section.
4304                          * The idea here is to be able to make section-level DIVs, but to do so in a
4305                          * lower-impact, more correct way than r50769
4306                          *
4307                          * $this : caller
4308                          * $section : the section number
4309                          * &$sectionContent : ref to the content of the section
4310                          * $showEditLinks : boolean describing whether this section has an edit link
4311                          */
4312                         wfRunHooks( 'ParserSectionCreate', array( $this, $i, &$sections[$i], $showEditLink ) );
4313
4314                         $i++;
4315                 }
4316
4317                 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4318                         // append the TOC at the beginning
4319                         // Top anchor now in skin
4320                         $sections[0] = $sections[0] . $toc . "\n";
4321                 }
4322
4323                 $full .= join( '', $sections );
4324
4325                 if ( $this->mForceTocPosition ) {
4326                         return str_replace( '<!--MWTOC-->', $toc, $full );
4327                 } else {
4328                         return $full;
4329                 }
4330         }
4331
4332         /**
4333          * Transform wiki markup when saving a page by doing \r\n -> \n
4334          * conversion, substitting signatures, {{subst:}} templates, etc.
4335          *
4336          * @param $text String: the text to transform
4337          * @param $title Title: the Title object for the current article
4338          * @param $user User: the User object describing the current user
4339          * @param $options ParserOptions: parsing options
4340          * @param $clearState Boolean: whether to clear the parser state first
4341          * @return String: the altered wiki markup
4342          */
4343         public function preSaveTransform( $text, Title $title, User $user, ParserOptions $options, $clearState = true ) {
4344                 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4345                 $this->setUser( $user );
4346
4347                 $pairs = array(
4348                         "\r\n" => "\n",
4349                 );
4350                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4351                 if( $options->getPreSaveTransform() ) {
4352                         $text = $this->pstPass2( $text, $user );
4353                 }
4354                 $text = $this->mStripState->unstripBoth( $text );
4355
4356                 $this->setUser( null ); #Reset
4357
4358                 return $text;
4359         }
4360
4361         /**
4362          * Pre-save transform helper function
4363          * @private
4364          *
4365          * @param $text string
4366          * @param $user User
4367          *
4368          * @return string
4369          */
4370         function pstPass2( $text, $user ) {
4371                 global $wgContLang, $wgLocaltimezone;
4372
4373                 # Note: This is the timestamp saved as hardcoded wikitext to
4374                 # the database, we use $wgContLang here in order to give
4375                 # everyone the same signature and use the default one rather
4376                 # than the one selected in each user's preferences.
4377                 # (see also bug 12815)
4378                 $ts = $this->mOptions->getTimestamp();
4379                 if ( isset( $wgLocaltimezone ) ) {
4380                         $tz = $wgLocaltimezone;
4381                 } else {
4382                         $tz = date_default_timezone_get();
4383                 }
4384
4385                 $unixts = wfTimestamp( TS_UNIX, $ts );
4386                 $oldtz = date_default_timezone_get();
4387                 date_default_timezone_set( $tz );
4388                 $ts = date( 'YmdHis', $unixts );
4389                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4390
4391                 # Allow translation of timezones through wiki. date() can return
4392                 # whatever crap the system uses, localised or not, so we cannot
4393                 # ship premade translations.
4394                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4395                 $msg = wfMessage( $key )->inContentLanguage();
4396                 if ( $msg->exists() ) {
4397                         $tzMsg = $msg->text();
4398                 }
4399
4400                 date_default_timezone_set( $oldtz );
4401
4402                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4403
4404                 # Variable replacement
4405                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4406                 $text = $this->replaceVariables( $text );
4407
4408                 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4409                 # which may corrupt this parser instance via its wfMsgExt( parsemag ) call-
4410
4411                 # Signatures
4412                 $sigText = $this->getUserSig( $user );
4413                 $text = strtr( $text, array(
4414                         '~~~~~' => $d,
4415                         '~~~~' => "$sigText $d",
4416                         '~~~' => $sigText
4417                 ) );
4418
4419                 # Context links: [[|name]] and [[name (context)|]]
4420                 global $wgLegalTitleChars;
4421                 $tc = "[$wgLegalTitleChars]";
4422                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4423
4424                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";           # [[ns:page (context)|]]
4425                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";           # [[ns:page（context）|]]
4426                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)(, $tc+|)\\|]]/"; # [[ns:page (context), context|]]
4427                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4428
4429                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4430                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4431                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4432                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4433
4434                 $t = $this->mTitle->getText();
4435                 $m = array();
4436                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4437                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4438                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4439                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4440                 } else {
4441                         # if there's no context, don't bother duplicating the title
4442                         $text = preg_replace( $p2, '[[\\1]]', $text );
4443                 }
4444
4445                 # Trim trailing whitespace
4446                 $text = rtrim( $text );
4447
4448                 return $text;
4449         }
4450
4451         /**
4452          * Fetch the user's signature text, if any, and normalize to
4453          * validated, ready-to-insert wikitext.
4454          * If you have pre-fetched the nickname or the fancySig option, you can
4455          * specify them here to save a database query.
4456          * Do not reuse this parser instance after calling getUserSig(),
4457          * as it may have changed if it's the $wgParser.
4458          *
4459          * @param $user User
4460          * @param $nickname String|bool nickname to use or false to use user's default nickname
4461          * @param $fancySig Boolean|null whether the nicknname is the complete signature
4462          *                  or null to use default value
4463          * @return string
4464          */
4465         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4466                 global $wgMaxSigChars;
4467
4468                 $username = $user->getName();
4469
4470                 # If not given, retrieve from the user object.
4471                 if ( $nickname === false )
4472                         $nickname = $user->getOption( 'nickname' );
4473
4474                 if ( is_null( $fancySig ) ) {
4475                         $fancySig = $user->getBoolOption( 'fancysig' );
4476                 }
4477
4478                 $nickname = $nickname == null ? $username : $nickname;
4479
4480                 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4481                         $nickname = $username;
4482                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4483                 } elseif ( $fancySig !== false ) {
4484                         # Sig. might contain markup; validate this
4485                         if ( $this->validateSig( $nickname ) !== false ) {
4486                                 # Validated; clean up (if needed) and return it
4487                                 return $this->cleanSig( $nickname, true );
4488                         } else {
4489                                 # Failed to validate; fall back to the default
4490                                 $nickname = $username;
4491                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4492                         }
4493                 }
4494
4495                 # Make sure nickname doesnt get a sig in a sig
4496                 $nickname = $this->cleanSigInSig( $nickname );
4497
4498                 # If we're still here, make it a link to the user page
4499                 $userText = wfEscapeWikiText( $username );
4500                 $nickText = wfEscapeWikiText( $nickname );
4501                 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4502
4503                 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()->title( $this->getTitle() )->text();
4504         }
4505
4506         /**
4507          * Check that the user's signature contains no bad XML
4508          *
4509          * @param $text String
4510          * @return mixed An expanded string, or false if invalid.
4511          */
4512         function validateSig( $text ) {
4513                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4514         }
4515
4516         /**
4517          * Clean up signature text
4518          *
4519          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4520          * 2) Substitute all transclusions
4521          *
4522          * @param $text String
4523          * @param $parsing bool Whether we're cleaning (preferences save) or parsing
4524          * @return String: signature text
4525          */
4526         function cleanSig( $text, $parsing = false ) {
4527                 if ( !$parsing ) {
4528                         global $wgTitle;
4529                         $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4530                 }
4531
4532                 # Option to disable this feature
4533                 if ( !$this->mOptions->getCleanSignatures() ) {
4534                         return $text;
4535                 }
4536
4537                 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4538                 #  => Move this logic to braceSubstitution()
4539                 $substWord = MagicWord::get( 'subst' );
4540                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4541                 $substText = '{{' . $substWord->getSynonym( 0 );
4542
4543                 $text = preg_replace( $substRegex, $substText, $text );
4544                 $text = $this->cleanSigInSig( $text );
4545                 $dom = $this->preprocessToDom( $text );
4546                 $frame = $this->getPreprocessor()->newFrame();
4547                 $text = $frame->expand( $dom );
4548
4549                 if ( !$parsing ) {
4550                         $text = $this->mStripState->unstripBoth( $text );
4551                 }
4552
4553                 return $text;
4554         }
4555
4556         /**
4557          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4558          *
4559          * @param $text String
4560          * @return String: signature text with /~{3,5}/ removed
4561          */
4562         function cleanSigInSig( $text ) {
4563                 $text = preg_replace( '/~{3,5}/', '', $text );
4564                 return $text;
4565         }
4566
4567         /**
4568          * Set up some variables which are usually set up in parse()
4569          * so that an external function can call some class members with confidence
4570          *
4571          * @param $title Title|null
4572          * @param $options ParserOptions
4573          * @param $outputType
4574          * @param $clearState bool
4575          */
4576         public function startExternalParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4577                 $this->startParse( $title, $options, $outputType, $clearState );
4578         }
4579
4580         /**
4581          * @param $title Title|null
4582          * @param $options ParserOptions
4583          * @param $outputType
4584          * @param $clearState bool
4585          */
4586         private function startParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4587                 $this->setTitle( $title );
4588                 $this->mOptions = $options;
4589                 $this->setOutputType( $outputType );
4590                 if ( $clearState ) {
4591                         $this->clearState();
4592                 }
4593         }
4594
4595         /**
4596          * Wrapper for preprocess()
4597          *
4598          * @param $text String: the text to preprocess
4599          * @param $options ParserOptions: options
4600          * @param $title Title object or null to use $wgTitle
4601          * @return String
4602          */
4603         public function transformMsg( $text, $options, $title = null ) {
4604                 static $executing = false;
4605
4606                 # Guard against infinite recursion
4607                 if ( $executing ) {
4608                         return $text;
4609                 }
4610                 $executing = true;
4611
4612                 wfProfileIn( __METHOD__ );
4613                 if ( !$title ) {
4614                         global $wgTitle;
4615                         $title = $wgTitle;
4616                 }
4617                 if ( !$title ) {
4618                         # It's not uncommon having a null $wgTitle in scripts. See r80898
4619                         # Create a ghost title in such case
4620                         $title = Title::newFromText( 'Dwimmerlaik' );
4621                 }
4622                 $text = $this->preprocess( $text, $title, $options );
4623
4624                 $executing = false;
4625                 wfProfileOut( __METHOD__ );
4626                 return $text;
4627         }
4628
4629         /**
4630          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4631          * The callback should have the following form:
4632          *    function myParserHook( $text, $params, $parser, $frame ) { ... }
4633          *
4634          * Transform and return $text. Use $parser for any required context, e.g. use
4635          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4636          *
4637          * Hooks may return extended information by returning an array, of which the
4638          * first numbered element (index 0) must be the return string, and all other
4639          * entries are extracted into local variables within an internal function
4640          * in the Parser class.
4641          *
4642          * This interface (introduced r61913) appears to be undocumented, but
4643          * 'markerName' is used by some core tag hooks to override which strip
4644          * array their results are placed in. **Use great caution if attempting
4645          * this interface, as it is not documented and injudicious use could smash
4646          * private variables.**
4647          *
4648          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4649          * @param $callback Mixed: the callback function (and object) to use for the tag
4650          * @return The old value of the mTagHooks array associated with the hook
4651          */
4652         public function setHook( $tag, $callback ) {
4653                 $tag = strtolower( $tag );
4654                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4655                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4656                 $this->mTagHooks[$tag] = $callback;
4657                 if ( !in_array( $tag, $this->mStripList ) ) {
4658                         $this->mStripList[] = $tag;
4659                 }
4660
4661                 return $oldVal;
4662         }
4663
4664         /**
4665          * As setHook(), but letting the contents be parsed.
4666          *
4667          * Transparent tag hooks are like regular XML-style tag hooks, except they
4668          * operate late in the transformation sequence, on HTML instead of wikitext.
4669          *
4670          * This is probably obsoleted by things dealing with parser frames?
4671          * The only extension currently using it is geoserver.
4672          *
4673          * @since 1.10
4674          * @todo better document or deprecate this
4675          *
4676          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4677          * @param $callback Mixed: the callback function (and object) to use for the tag
4678          * @return The old value of the mTagHooks array associated with the hook
4679          */
4680         function setTransparentTagHook( $tag, $callback ) {
4681                 $tag = strtolower( $tag );
4682                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4683                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4684                 $this->mTransparentTagHooks[$tag] = $callback;
4685
4686                 return $oldVal;
4687         }
4688
4689         /**
4690          * Remove all tag hooks
4691          */
4692         function clearTagHooks() {
4693                 $this->mTagHooks = array();
4694                 $this->mStripList = $this->mDefaultStripList;
4695         }
4696
4697         /**
4698          * Remove a specific tag hook. Should not be called on $wgParser.
4699          * Does not change the strip list.
4700          *
4701          * @param string $tag
4702          * @return void
4703          */
4704         function clearTagHook( $tag ) {
4705                 if ( isset( $this->mTagHooks[$tag] ) ) {
4706                         unset( $this->mTagHooks[$tag] );
4707                 }
4708         }
4709
4710         /**
4711          * Create a function, e.g. {{sum:1|2|3}}
4712          * The callback function should have the form:
4713          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4714          *
4715          * Or with SFH_OBJECT_ARGS:
4716          *    function myParserFunction( $parser, $frame, $args ) { ... }
4717          *
4718          * The callback may either return the text result of the function, or an array with the text
4719          * in element 0, and a number of flags in the other elements. The names of the flags are
4720          * specified in the keys. Valid flags are:
4721          *   found                     The text returned is valid, stop processing the template. This
4722          *                             is on by default.
4723          *   nowiki                    Wiki markup in the return value should be escaped
4724          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4725          *
4726          * @param $id String: The magic word ID
4727          * @param $callback Mixed: the callback function (and object) to use
4728          * @param $flags Integer: a combination of the following flags:
4729          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4730          *
4731          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4732          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4733          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4734          *     the arguments, and to control the way they are expanded.
4735          *
4736          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4737          *     arguments, for instance:
4738          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4739          *
4740          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4741          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4742          *     working if/when this is changed.
4743          *
4744          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4745          *     expansion.
4746          *
4747          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4748          *     about the methods available in PPFrame and PPNode.
4749          *
4750          * @return The old callback function for this name, if any
4751          */
4752         public function setFunctionHook( $id, $callback, $flags = 0 ) {
4753                 global $wgContLang;
4754
4755                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4756                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4757
4758                 # Add to function cache
4759                 $mw = MagicWord::get( $id );
4760                 if ( !$mw )
4761                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4762
4763                 $synonyms = $mw->getSynonyms();
4764                 $sensitive = intval( $mw->isCaseSensitive() );
4765
4766                 foreach ( $synonyms as $syn ) {
4767                         # Case
4768                         if ( !$sensitive ) {
4769                                 $syn = $wgContLang->lc( $syn );
4770                         }
4771                         # Add leading hash
4772                         if ( !( $flags & SFH_NO_HASH ) ) {
4773                                 $syn = '#' . $syn;
4774                         }
4775                         # Remove trailing colon
4776                         if ( substr( $syn, -1, 1 ) === ':' ) {
4777                                 $syn = substr( $syn, 0, -1 );
4778                         }
4779                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4780                 }
4781                 return $oldVal;
4782         }
4783
4784         /**
4785          * Get all registered function hook identifiers
4786          *
4787          * @return Array
4788          */
4789         function getFunctionHooks() {
4790                 return array_keys( $this->mFunctionHooks );
4791         }
4792
4793         /**
4794          * Create a tag function, e.g. <test>some stuff</test>.
4795          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4796          * Unlike parser functions, their content is not preprocessed.
4797          */
4798         function setFunctionTagHook( $tag, $callback, $flags ) {
4799                 $tag = strtolower( $tag );
4800                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4801                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4802                         $this->mFunctionTagHooks[$tag] : null;
4803                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4804
4805                 if ( !in_array( $tag, $this->mStripList ) ) {
4806                         $this->mStripList[] = $tag;
4807                 }
4808
4809                 return $old;
4810         }
4811
4812         /**
4813          * @todo FIXME: Update documentation. makeLinkObj() is deprecated.
4814          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4815          * Placeholders created in Skin::makeLinkObj()
4816          *
4817          * @param $text string
4818          * @param $options int
4819          *
4820          * @return array of link CSS classes, indexed by PDBK.
4821          */
4822         function replaceLinkHolders( &$text, $options = 0 ) {
4823                 return $this->mLinkHolders->replace( $text );
4824         }
4825
4826         /**
4827          * Replace <!--LINK--> link placeholders with plain text of links
4828          * (not HTML-formatted).
4829          *
4830          * @param $text String
4831          * @return String
4832          */
4833         function replaceLinkHoldersText( $text ) {
4834                 return $this->mLinkHolders->replaceText( $text );
4835         }
4836
4837         /**
4838          * Renders an image gallery from a text with one line per image.
4839          * text labels may be given by using |-style alternative text. E.g.
4840          *   Image:one.jpg|The number "1"
4841          *   Image:tree.jpg|A tree
4842          * given as text will return the HTML of a gallery with two images,
4843          * labeled 'The number "1"' and
4844          * 'A tree'.
4845          *
4846          * @param string $text
4847          * @param array $params
4848          * @return string HTML
4849          */
4850         function renderImageGallery( $text, $params ) {
4851                 $ig = new ImageGallery();
4852                 $ig->setContextTitle( $this->mTitle );
4853                 $ig->setShowBytes( false );
4854                 $ig->setShowFilename( false );
4855                 $ig->setParser( $this );
4856                 $ig->setHideBadImages();
4857                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4858
4859                 if ( isset( $params['showfilename'] ) ) {
4860                         $ig->setShowFilename( true );
4861                 } else {
4862                         $ig->setShowFilename( false );
4863                 }
4864                 if ( isset( $params['caption'] ) ) {
4865                         $caption = $params['caption'];
4866                         $caption = htmlspecialchars( $caption );
4867                         $caption = $this->replaceInternalLinks( $caption );
4868                         $ig->setCaptionHtml( $caption );
4869                 }
4870                 if ( isset( $params['perrow'] ) ) {
4871                         $ig->setPerRow( $params['perrow'] );
4872                 }
4873                 if ( isset( $params['widths'] ) ) {
4874                         $ig->setWidths( $params['widths'] );
4875                 }
4876                 if ( isset( $params['heights'] ) ) {
4877                         $ig->setHeights( $params['heights'] );
4878                 }
4879
4880                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4881
4882                 $lines = StringUtils::explode( "\n", $text );
4883                 foreach ( $lines as $line ) {
4884                         # match lines like these:
4885                         # Image:someimage.jpg|This is some image
4886                         $matches = array();
4887                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4888                         # Skip empty lines
4889                         if ( count( $matches ) == 0 ) {
4890                                 continue;
4891                         }
4892
4893                         if ( strpos( $matches[0], '%' ) !== false ) {
4894                                 $matches[1] = rawurldecode( $matches[1] );
4895                         }
4896                         $title = Title::newFromText( $matches[1], NS_FILE );
4897                         if ( is_null( $title ) ) {
4898                                 # Bogus title. Ignore these so we don't bomb out later.
4899                                 continue;
4900                         }
4901
4902                         $label = '';
4903                         $alt = '';
4904                         if ( isset( $matches[3] ) ) {
4905                                 // look for an |alt= definition while trying not to break existing
4906                                 // captions with multiple pipes (|) in it, until a more sensible grammar
4907                                 // is defined for images in galleries
4908
4909                                 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
4910                                 $altmatches = StringUtils::explode('|', $matches[3]);
4911                                 $magicWordAlt = MagicWord::get( 'img_alt' );
4912
4913                                 foreach ( $altmatches as $altmatch ) {
4914                                         $match = $magicWordAlt->matchVariableStartToEnd( $altmatch );
4915                                         if ( $match ) {
4916                                                 $alt = $this->stripAltText( $match, false );
4917                                         }
4918                                         else {
4919                                                 // concatenate all other pipes
4920                                                 $label .= '|' . $altmatch;
4921                                         }
4922                                 }
4923                                 // remove the first pipe
4924                                 $label = substr( $label, 1 );
4925                         }
4926
4927                         $ig->add( $title, $label, $alt );
4928                 }
4929                 return $ig->toHTML();
4930         }
4931
4932         /**
4933          * @param $handler
4934          * @return array
4935          */
4936         function getImageParams( $handler ) {
4937                 if ( $handler ) {
4938                         $handlerClass = get_class( $handler );
4939                 } else {
4940                         $handlerClass = '';
4941                 }
4942                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4943                         # Initialise static lists
4944                         static $internalParamNames = array(
4945                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4946                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4947                                         'bottom', 'text-bottom' ),
4948                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4949                                         'upright', 'border', 'link', 'alt' ),
4950                         );
4951                         static $internalParamMap;
4952                         if ( !$internalParamMap ) {
4953                                 $internalParamMap = array();
4954                                 foreach ( $internalParamNames as $type => $names ) {
4955                                         foreach ( $names as $name ) {
4956                                                 $magicName = str_replace( '-', '_', "img_$name" );
4957                                                 $internalParamMap[$magicName] = array( $type, $name );
4958                                         }
4959                                 }
4960                         }
4961
4962                         # Add handler params
4963                         $paramMap = $internalParamMap;
4964                         if ( $handler ) {
4965                                 $handlerParamMap = $handler->getParamMap();
4966                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4967                                         $paramMap[$magic] = array( 'handler', $paramName );
4968                                 }
4969                         }
4970                         $this->mImageParams[$handlerClass] = $paramMap;
4971                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4972                 }
4973                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4974         }
4975
4976         /**
4977          * Parse image options text and use it to make an image
4978          *
4979          * @param $title Title
4980          * @param $options String
4981          * @param $holders LinkHolderArray|false
4982          * @return string HTML
4983          */
4984         function makeImage( $title, $options, $holders = false ) {
4985                 # Check if the options text is of the form "options|alt text"
4986                 # Options are:
4987                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4988                 #  * left       no resizing, just left align. label is used for alt= only
4989                 #  * right      same, but right aligned
4990                 #  * none       same, but not aligned
4991                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4992                 #  * center     center the image
4993                 #  * frame      Keep original image size, no magnify-button.
4994                 #  * framed     Same as "frame"
4995                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4996                 #  * upright    reduce width for upright images, rounded to full __0 px
4997                 #  * border     draw a 1px border around the image
4998                 #  * alt        Text for HTML alt attribute (defaults to empty)
4999                 #  * link       Set the target of the image link. Can be external, interwiki, or local
5000                 # vertical-align values (no % or length right now):
5001                 #  * baseline
5002                 #  * sub
5003                 #  * super
5004                 #  * top
5005                 #  * text-top
5006                 #  * middle
5007                 #  * bottom
5008                 #  * text-bottom
5009
5010                 $parts = StringUtils::explode( "|", $options );
5011
5012                 # Give extensions a chance to select the file revision for us
5013                 $time = $sha1 = $descQuery = false;
5014                 wfRunHooks( 'BeforeParserFetchFileAndTitle',
5015                         array( $this, $title, &$time, &$sha1, &$descQuery ) );
5016                 # Fetch and register the file (file title may be different via hooks)
5017                 list( $file, $title ) = $this->fetchFileAndTitle( $title, $time, $sha1 );
5018
5019                 # Get parameter map
5020                 $handler = $file ? $file->getHandler() : false;
5021
5022                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5023
5024                 if ( !$file ) {
5025                         $this->addTrackingCategory( 'broken-file-category' );
5026                 }
5027
5028                 # Process the input parameters
5029                 $caption = '';
5030                 $params = array( 'frame' => array(), 'handler' => array(),
5031                         'horizAlign' => array(), 'vertAlign' => array() );
5032                 foreach ( $parts as $part ) {
5033                         $part = trim( $part );
5034                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5035                         $validated = false;
5036                         if ( isset( $paramMap[$magicName] ) ) {
5037                                 list( $type, $paramName ) = $paramMap[$magicName];
5038
5039                                 # Special case; width and height come in one variable together
5040                                 if ( $type === 'handler' && $paramName === 'width' ) {
5041                                         $m = array();
5042                                         # (bug 13500) In both cases (width/height and width only),
5043                                         # permit trailing "px" for backward compatibility.
5044                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
5045                                                 $width = intval( $m[1] );
5046                                                 $height = intval( $m[2] );
5047                                                 if ( $handler->validateParam( 'width', $width ) ) {
5048                                                         $params[$type]['width'] = $width;
5049                                                         $validated = true;
5050                                                 }
5051                                                 if ( $handler->validateParam( 'height', $height ) ) {
5052                                                         $params[$type]['height'] = $height;
5053                                                         $validated = true;
5054                                                 }
5055                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
5056                                                 $width = intval( $value );
5057                                                 if ( $handler->validateParam( 'width', $width ) ) {
5058                                                         $params[$type]['width'] = $width;
5059                                                         $validated = true;
5060                                                 }
5061                                         } # else no validation -- bug 13436
5062                                 } else {
5063                                         if ( $type === 'handler' ) {
5064                                                 # Validate handler parameter
5065                                                 $validated = $handler->validateParam( $paramName, $value );
5066                                         } else {
5067                                                 # Validate internal parameters
5068                                                 switch( $paramName ) {
5069                                                 case 'manualthumb':
5070                                                 case 'alt':
5071                                                         # @todo FIXME: Possibly check validity here for
5072                                                         # manualthumb? downstream behavior seems odd with
5073                                                         # missing manual thumbs.
5074                                                         $validated = true;
5075                                                         $value = $this->stripAltText( $value, $holders );
5076                                                         break;
5077                                                 case 'link':
5078                                                         $chars = self::EXT_LINK_URL_CLASS;
5079                                                         $prots = $this->mUrlProtocols;
5080                                                         if ( $value === '' ) {
5081                                                                 $paramName = 'no-link';
5082                                                                 $value = true;
5083                                                                 $validated = true;
5084                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
5085                                                                 if ( preg_match( "/^($prots)$chars+$/u", $value, $m ) ) {
5086                                                                         $paramName = 'link-url';
5087                                                                         $this->mOutput->addExternalLink( $value );
5088                                                                         if ( $this->mOptions->getExternalLinkTarget() ) {
5089                                                                                 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5090                                                                         }
5091                                                                         $validated = true;
5092                                                                 }
5093                                                         } else {
5094                                                                 $linkTitle = Title::newFromText( $value );
5095                                                                 if ( $linkTitle ) {
5096                                                                         $paramName = 'link-title';
5097                                                                         $value = $linkTitle;
5098                                                                         $this->mOutput->addLink( $linkTitle );
5099                                                                         $validated = true;
5100                                                                 }
5101                                                         }
5102                                                         break;
5103                                                 default:
5104                                                         # Most other things appear to be empty or numeric...
5105                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
5106                                                 }
5107                                         }
5108
5109                                         if ( $validated ) {
5110                                                 $params[$type][$paramName] = $value;
5111                                         }
5112                                 }
5113                         }
5114                         if ( !$validated ) {
5115                                 $caption = $part;
5116                         }
5117                 }
5118
5119                 # Process alignment parameters
5120                 if ( $params['horizAlign'] ) {
5121                         $params['frame']['align'] = key( $params['horizAlign'] );
5122                 }
5123                 if ( $params['vertAlign'] ) {
5124                         $params['frame']['valign'] = key( $params['vertAlign'] );
5125                 }
5126
5127                 $params['frame']['caption'] = $caption;
5128
5129                 # Will the image be presented in a frame, with the caption below?
5130                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
5131                                                  isset( $params['frame']['framed'] ) ||
5132                                                  isset( $params['frame']['thumbnail'] ) ||
5133                                                  isset( $params['frame']['manualthumb'] );
5134
5135                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
5136                 # came to also set the caption, ordinary text after the image -- which
5137                 # makes no sense, because that just repeats the text multiple times in
5138                 # screen readers.  It *also* came to set the title attribute.
5139                 #
5140                 # Now that we have an alt attribute, we should not set the alt text to
5141                 # equal the caption: that's worse than useless, it just repeats the
5142                 # text.  This is the framed/thumbnail case.  If there's no caption, we
5143                 # use the unnamed parameter for alt text as well, just for the time be-
5144                 # ing, if the unnamed param is set and the alt param is not.
5145                 #
5146                 # For the future, we need to figure out if we want to tweak this more,
5147                 # e.g., introducing a title= parameter for the title; ignoring the un-
5148                 # named parameter entirely for images without a caption; adding an ex-
5149                 # plicit caption= parameter and preserving the old magic unnamed para-
5150                 # meter for BC; ...
5151                 if ( $imageIsFramed ) { # Framed image
5152                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5153                                 # No caption or alt text, add the filename as the alt text so
5154                                 # that screen readers at least get some description of the image
5155                                 $params['frame']['alt'] = $title->getText();
5156                         }
5157                         # Do not set $params['frame']['title'] because tooltips don't make sense
5158                         # for framed images
5159                 } else { # Inline image
5160                         if ( !isset( $params['frame']['alt'] ) ) {
5161                                 # No alt text, use the "caption" for the alt text
5162                                 if ( $caption !== '') {
5163                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5164                                 } else {
5165                                         # No caption, fall back to using the filename for the
5166                                         # alt text
5167                                         $params['frame']['alt'] = $title->getText();
5168                                 }
5169                         }
5170                         # Use the "caption" for the tooltip text
5171                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5172                 }
5173
5174                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
5175
5176                 # Linker does the rest
5177                 $ret = Linker::makeImageLink2( $title, $file, $params['frame'], $params['handler'],
5178                         $time, $descQuery, $this->mOptions->getThumbSize() );
5179
5180                 # Give the handler a chance to modify the parser object
5181                 if ( $handler ) {
5182                         $handler->parserTransformHook( $this, $file );
5183                 }
5184
5185                 return $ret;
5186         }
5187
5188         /**
5189          * @param $caption
5190          * @param $holders LinkHolderArray
5191          * @return mixed|String
5192          */
5193         protected function stripAltText( $caption, $holders ) {
5194                 # Strip bad stuff out of the title (tooltip).  We can't just use
5195                 # replaceLinkHoldersText() here, because if this function is called
5196                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5197                 if ( $holders ) {
5198                         $tooltip = $holders->replaceText( $caption );
5199                 } else {
5200                         $tooltip = $this->replaceLinkHoldersText( $caption );
5201                 }
5202
5203                 # make sure there are no placeholders in thumbnail attributes
5204                 # that are later expanded to html- so expand them now and
5205                 # remove the tags
5206                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5207                 $tooltip = Sanitizer::stripAllTags( $tooltip );
5208
5209                 return $tooltip;
5210         }
5211
5212         /**
5213          * Set a flag in the output object indicating that the content is dynamic and
5214          * shouldn't be cached.
5215          */
5216         function disableCache() {
5217                 wfDebug( "Parser output marked as uncacheable.\n" );
5218                 $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
5219                 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5220         }
5221
5222         /**
5223          * Callback from the Sanitizer for expanding items found in HTML attribute
5224          * values, so they can be safely tested and escaped.
5225          *
5226          * @param $text String
5227          * @param $frame PPFrame
5228          * @return String
5229          */
5230         function attributeStripCallback( &$text, $frame = false ) {
5231                 $text = $this->replaceVariables( $text, $frame );
5232                 $text = $this->mStripState->unstripBoth( $text );
5233                 return $text;
5234         }
5235
5236         /**
5237          * Accessor
5238          *
5239          * @return array
5240          */
5241         function getTags() {
5242                 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ) );
5243         }
5244
5245         /**
5246          * Replace transparent tags in $text with the values given by the callbacks.
5247          *
5248          * Transparent tag hooks are like regular XML-style tag hooks, except they
5249          * operate late in the transformation sequence, on HTML instead of wikitext.
5250          *
5251          * @param $text string
5252          *
5253          * @return string
5254          */
5255         function replaceTransparentTags( $text ) {
5256                 $matches = array();
5257                 $elements = array_keys( $this->mTransparentTagHooks );
5258                 $text = self::extractTagsAndParams( $elements, $text, $matches, $this->mUniqPrefix );
5259                 $replacements = array();
5260
5261                 foreach ( $matches as $marker => $data ) {
5262                         list( $element, $content, $params, $tag ) = $data;
5263                         $tagName = strtolower( $element );
5264                         if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5265                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
5266                         } else {
5267                                 $output = $tag;
5268                         }
5269                         $replacements[$marker] = $output;
5270                 }
5271                 return strtr( $text, $replacements );
5272         }
5273
5274         /**
5275          * Break wikitext input into sections, and either pull or replace
5276          * some particular section's text.
5277          *
5278          * External callers should use the getSection and replaceSection methods.
5279          *
5280          * @param $text String: Page wikitext
5281          * @param $section String: a section identifier string of the form:
5282          *   <flag1> - <flag2> - ... - <section number>
5283          *
5284          * Currently the only recognised flag is "T", which means the target section number
5285          * was derived during a template inclusion parse, in other words this is a template
5286          * section edit link. If no flags are given, it was an ordinary section edit link.
5287          * This flag is required to avoid a section numbering mismatch when a section is
5288          * enclosed by <includeonly> (bug 6563).
5289          *
5290          * The section number 0 pulls the text before the first heading; other numbers will
5291          * pull the given section along with its lower-level subsections. If the section is
5292          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5293          *
5294          * @param $mode String: one of "get" or "replace"
5295          * @param $newText String: replacement text for section data.
5296          * @return String: for "get", the extracted section text.
5297          *                 for "replace", the whole page with the section replaced.
5298          *                 If the page is empty and section 0 is requested, $text (as '')
5299          *                  is returned
5300          */
5301         private function extractSections( $text, $section, $mode, $newText='' ) {
5302                 global $wgTitle; # not generally used but removes an ugly failure mode
5303                 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5304                 $outText = '';
5305                 $frame = $this->getPreprocessor()->newFrame();
5306
5307                 # Process section extraction flags
5308                 $flags = 0;
5309                 $sectionParts = explode( '-', $section );
5310                 $sectionIndex = array_pop( $sectionParts );
5311                 foreach ( $sectionParts as $part ) {
5312                         if ( $part === 'T' ) {
5313                                 $flags |= self::PTD_FOR_INCLUSION;
5314                         }
5315                 }
5316                 # Preprocess the text
5317                 $root = $this->preprocessToDom( $text, $flags );
5318
5319                 # <h> nodes indicate section breaks
5320                 # They can only occur at the top level, so we can find them by iterating the root's children
5321                 $node = $root->getFirstChild();
5322
5323                 # Find the target section
5324                 if ( $sectionIndex == 0 ) {
5325                         # Section zero doesn't nest, level=big
5326                         $targetLevel = 1000;
5327                         if ( !$node ) {
5328                                 # The page definitely exists - we checked that earlier - so it must be blank: see bug #14005
5329                                 return $text;
5330                         }
5331                 } else {
5332                         while ( $node ) {
5333                                 if ( $node->getName() === 'h' ) {
5334                                         $bits = $node->splitHeading();
5335                                         if ( $bits['i'] == $sectionIndex ) {
5336                                                 $targetLevel = $bits['level'];
5337                                                 break;
5338                                         }
5339                                 }
5340                                 if ( $mode === 'replace' ) {
5341                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5342                                 }
5343                                 $node = $node->getNextSibling();
5344                         }
5345                 }
5346
5347                 if ( !$node ) {
5348                         # Not found
5349                         if ( $mode === 'get' ) {
5350                                 return $newText;
5351                         } else {
5352                                 return $text;
5353                         }
5354                 }
5355
5356                 # Find the end of the section, including nested sections
5357                 do {
5358                         if ( $node->getName() === 'h' ) {
5359                                 $bits = $node->splitHeading();
5360                                 $curLevel = $bits['level'];
5361                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5362                                         break;
5363                                 }
5364                         }
5365                         if ( $mode === 'get' ) {
5366                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5367                         }
5368                         $node = $node->getNextSibling();
5369                 } while ( $node );
5370
5371                 # Write out the remainder (in replace mode only)
5372                 if ( $mode === 'replace' ) {
5373                         # Output the replacement text
5374                         # Add two newlines on -- trailing whitespace in $newText is conventionally
5375                         # stripped by the editor, so we need both newlines to restore the paragraph gap
5376                         # Only add trailing whitespace if there is newText
5377                         if ( $newText != "" ) {
5378                                 $outText .= $newText . "\n\n";
5379                         }
5380
5381                         while ( $node ) {
5382                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5383                                 $node = $node->getNextSibling();
5384                         }
5385                 }
5386
5387                 if ( is_string( $outText ) ) {
5388                         # Re-insert stripped tags
5389                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5390                 }
5391
5392                 return $outText;
5393         }
5394
5395         /**
5396          * This function returns the text of a section, specified by a number ($section).
5397          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5398          * the first section before any such heading (section 0).
5399          *
5400          * If a section contains subsections, these are also returned.
5401          *
5402          * @param $text String: text to look in
5403          * @param $section String: section identifier
5404          * @param $deftext String: default to return if section is not found
5405          * @return string text of the requested section
5406          */
5407         public function getSection( $text, $section, $deftext='' ) {
5408                 return $this->extractSections( $text, $section, "get", $deftext );
5409         }
5410
5411         /**
5412          * This function returns $oldtext after the content of the section
5413          * specified by $section has been replaced with $text.
5414          *
5415          * @param $oldtext String: former text of the article
5416          * @param $section Numeric: section identifier
5417          * @param $text String: replacing text
5418          * @return String: modified text
5419          */
5420         public function replaceSection( $oldtext, $section, $text ) {
5421                 return $this->extractSections( $oldtext, $section, "replace", $text );
5422         }
5423
5424         /**
5425          * Get the ID of the revision we are parsing
5426          *
5427          * @return Mixed: integer or null
5428          */
5429         function getRevisionId() {
5430                 return $this->mRevisionId;
5431         }
5432
5433         /**
5434          * Get the revision object for $this->mRevisionId
5435          *
5436          * @return Revision|null either a Revision object or null
5437          */
5438         protected function getRevisionObject() {
5439                 if ( !is_null( $this->mRevisionObject ) ) {
5440                         return $this->mRevisionObject;
5441                 }
5442                 if ( is_null( $this->mRevisionId ) ) {
5443                         return null;
5444                 }
5445
5446                 $this->mRevisionObject = Revision::newFromId( $this->mRevisionId );
5447                 return $this->mRevisionObject;
5448         }
5449
5450         /**
5451          * Get the timestamp associated with the current revision, adjusted for
5452          * the default server-local timestamp
5453          */
5454         function getRevisionTimestamp() {
5455                 if ( is_null( $this->mRevisionTimestamp ) ) {
5456                         wfProfileIn( __METHOD__ );
5457
5458                         $revObject = $this->getRevisionObject();
5459                         $timestamp = $revObject ? $revObject->getTimestamp() : false;
5460
5461                         if( $timestamp !== false ) {
5462                                 global $wgContLang;
5463
5464                                 # The cryptic '' timezone parameter tells to use the site-default
5465                                 # timezone offset instead of the user settings.
5466                                 #
5467                                 # Since this value will be saved into the parser cache, served
5468                                 # to other users, and potentially even used inside links and such,
5469                                 # it needs to be consistent for all visitors.
5470                                 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5471                         }
5472
5473                         wfProfileOut( __METHOD__ );
5474                 }
5475                 return $this->mRevisionTimestamp;
5476         }
5477
5478         /**
5479          * Get the name of the user that edited the last revision
5480          *
5481          * @return String: user name
5482          */
5483         function getRevisionUser() {
5484                 if( is_null( $this->mRevisionUser ) ) {
5485                         $revObject = $this->getRevisionObject();
5486
5487                         # if this template is subst: the revision id will be blank,
5488                         # so just use the current user's name
5489                         if( $revObject ) {
5490                                 $this->mRevisionUser = $revObject->getUserText();
5491                         } elseif( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5492                                 $this->mRevisionUser = $this->getUser()->getName();
5493                         }
5494                 }
5495                 return $this->mRevisionUser;
5496         }
5497
5498         /**
5499          * Mutator for $mDefaultSort
5500          *
5501          * @param $sort New value
5502          */
5503         public function setDefaultSort( $sort ) {
5504                 $this->mDefaultSort = $sort;
5505                 $this->mOutput->setProperty( 'defaultsort', $sort );
5506         }
5507
5508         /**
5509          * Accessor for $mDefaultSort
5510          * Will use the empty string if none is set.
5511          *
5512          * This value is treated as a prefix, so the
5513          * empty string is equivalent to sorting by
5514          * page name.
5515          *
5516          * @return string
5517          */
5518         public function getDefaultSort() {
5519                 if ( $this->mDefaultSort !== false ) {
5520                         return $this->mDefaultSort;
5521                 } else {
5522                         return '';
5523                 }
5524         }
5525
5526         /**
5527          * Accessor for $mDefaultSort
5528          * Unlike getDefaultSort(), will return false if none is set
5529          *
5530          * @return string or false
5531          */
5532         public function getCustomDefaultSort() {
5533                 return $this->mDefaultSort;
5534         }
5535
5536         /**
5537          * Try to guess the section anchor name based on a wikitext fragment
5538          * presumably extracted from a heading, for example "Header" from
5539          * "== Header ==".
5540          *
5541          * @param $text string
5542          *
5543          * @return string
5544          */
5545         public function guessSectionNameFromWikiText( $text ) {
5546                 # Strip out wikitext links(they break the anchor)
5547                 $text = $this->stripSectionName( $text );
5548                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5549                 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5550         }
5551
5552         /**
5553          * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5554          * instead.  For use in redirects, since IE6 interprets Redirect: headers
5555          * as something other than UTF-8 (apparently?), resulting in breakage.
5556          *
5557          * @param $text String: The section name
5558          * @return string An anchor
5559          */
5560         public function guessLegacySectionNameFromWikiText( $text ) {
5561                 # Strip out wikitext links(they break the anchor)
5562                 $text = $this->stripSectionName( $text );
5563                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5564                 return '#' . Sanitizer::escapeId( $text, array( 'noninitial', 'legacy' ) );
5565         }
5566
5567         /**
5568          * Strips a text string of wikitext for use in a section anchor
5569          *
5570          * Accepts a text string and then removes all wikitext from the
5571          * string and leaves only the resultant text (i.e. the result of
5572          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5573          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5574          * to create valid section anchors by mimicing the output of the
5575          * parser when headings are parsed.
5576          *
5577          * @param $text String: text string to be stripped of wikitext
5578          * for use in a Section anchor
5579          * @return Filtered text string
5580          */
5581         public function stripSectionName( $text ) {
5582                 # Strip internal link markup
5583                 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5584                 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5585
5586                 # Strip external link markup
5587                 # @todo FIXME: Not tolerant to blank link text
5588                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5589                 # on how many empty links there are on the page - need to figure that out.
5590                 $text = preg_replace( '/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5591
5592                 # Parse wikitext quotes (italics & bold)
5593                 $text = $this->doQuotes( $text );
5594
5595                 # Strip HTML tags
5596                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5597                 return $text;
5598         }
5599
5600         /**
5601          * strip/replaceVariables/unstrip for preprocessor regression testing
5602          *
5603          * @param $text string
5604          * @param $title Title
5605          * @param $options ParserOptions
5606          * @param $outputType int
5607          *
5608          * @return string
5609          */
5610         function testSrvus( $text, Title $title, ParserOptions $options, $outputType = self::OT_HTML ) {
5611                 $this->startParse( $title, $options, $outputType, true );
5612
5613                 $text = $this->replaceVariables( $text );
5614                 $text = $this->mStripState->unstripBoth( $text );
5615                 $text = Sanitizer::removeHTMLtags( $text );
5616                 return $text;
5617         }
5618
5619         /**
5620          * @param $text string
5621          * @param $title Title
5622          * @param $options ParserOptions
5623          * @return string
5624          */
5625         function testPst( $text, Title $title, ParserOptions $options ) {
5626                 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
5627         }
5628
5629         /**
5630          * @param $text
5631          * @param $title Title
5632          * @param $options ParserOptions
5633          * @return string
5634          */
5635         function testPreprocess( $text, Title $title, ParserOptions $options ) {
5636                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5637         }
5638
5639         /**
5640          * Call a callback function on all regions of the given text that are not
5641          * inside strip markers, and replace those regions with the return value
5642          * of the callback. For example, with input:
5643          *
5644          *  aaa<MARKER>bbb
5645          *
5646          * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5647          * two strings will be replaced with the value returned by the callback in
5648          * each case.
5649          *
5650          * @param $s string
5651          * @param $callback
5652          *
5653          * @return string
5654          */
5655         function markerSkipCallback( $s, $callback ) {
5656                 $i = 0;
5657                 $out = '';
5658                 while ( $i < strlen( $s ) ) {
5659                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5660                         if ( $markerStart === false ) {
5661                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5662                                 break;
5663                         } else {
5664                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5665                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5666                                 if ( $markerEnd === false ) {
5667                                         $out .= substr( $s, $markerStart );
5668                                         break;
5669                                 } else {
5670                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5671                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5672                                         $i = $markerEnd;
5673                                 }
5674                         }
5675                 }
5676                 return $out;
5677         }
5678
5679         /**
5680          * Save the parser state required to convert the given half-parsed text to
5681          * HTML. "Half-parsed" in this context means the output of
5682          * recursiveTagParse() or internalParse(). This output has strip markers
5683          * from replaceVariables (extensionSubstitution() etc.), and link
5684          * placeholders from replaceLinkHolders().
5685          *
5686          * Returns an array which can be serialized and stored persistently. This
5687          * array can later be loaded into another parser instance with
5688          * unserializeHalfParsedText(). The text can then be safely incorporated into
5689          * the return value of a parser hook.
5690          *
5691          * @param $text string
5692          *
5693          * @return array
5694          */
5695         function serializeHalfParsedText( $text ) {
5696                 wfProfileIn( __METHOD__ );
5697                 $data = array(
5698                         'text' => $text,
5699                         'version' => self::HALF_PARSED_VERSION,
5700                         'stripState' => $this->mStripState->getSubState( $text ),
5701                         'linkHolders' => $this->mLinkHolders->getSubArray( $text )
5702                 );
5703                 wfProfileOut( __METHOD__ );
5704                 return $data;
5705         }
5706
5707         /**
5708          * Load the parser state given in the $data array, which is assumed to
5709          * have been generated by serializeHalfParsedText(). The text contents is
5710          * extracted from the array, and its markers are transformed into markers
5711          * appropriate for the current Parser instance. This transformed text is
5712          * returned, and can be safely included in the return value of a parser
5713          * hook.
5714          *
5715          * If the $data array has been stored persistently, the caller should first
5716          * check whether it is still valid, by calling isValidHalfParsedText().
5717          *
5718          * @param $data Serialized data
5719          * @return String
5720          */
5721         function unserializeHalfParsedText( $data ) {
5722                 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
5723                         throw new MWException( __METHOD__.': invalid version' );
5724                 }
5725
5726                 # First, extract the strip state.
5727                 $texts = array( $data['text'] );
5728                 $texts = $this->mStripState->merge( $data['stripState'], $texts );
5729
5730                 # Now renumber links
5731                 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
5732
5733                 # Should be good to go.
5734                 return $texts[0];
5735         }
5736
5737         /**
5738          * Returns true if the given array, presumed to be generated by
5739          * serializeHalfParsedText(), is compatible with the current version of the
5740          * parser.
5741          *
5742          * @param $data Array
5743          *
5744          * @return bool
5745          */
5746         function isValidHalfParsedText( $data ) {
5747                 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
5748         }
5749 }