includes/parser/Parser.php

   1 <?php
   2 /**
   3  * @defgroup Parser Parser
   4  *
   5  * @file
   6  * @ingroup Parser
   7  * File for Parser and related classes
   8  */
   9
  10
  11 /**
  12  * PHP Parser - Processes wiki markup (which uses a more user-friendly
  13  * syntax, such as "[[link]]" for making links), and provides a one-way
  14  * transformation of that wiki markup it into XHTML output / markup
  15  * (which in turn the browser understands, and can display).
  16  *
  17  * <pre>
  18  * There are five main entry points into the Parser class:
  19  * parse()
  20  *     produces HTML output
  21  * preSaveTransform().
  22  *     produces altered wiki markup.
  23  * preprocess()
  24  *     removes HTML comments and expands templates
  25  * cleanSig() / cleanSigInSig()
  26  *     Cleans a signature before saving it to preferences
  27  * getSection()
  28  *     Return the content of a section from an article for section editing
  29  * replaceSection()
  30  *     Replaces a section by number inside an article
  31  * getPreloadText()
  32  *     Removes <noinclude> sections, and <includeonly> tags.
  33  *
  34  * Globals used:
  35  *    objects:   $wgLang, $wgContLang
  36  *
  37  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  38  *
  39  * settings:
  40  *  $wgUseDynamicDates*, $wgInterwikiMagic*,
  41  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  42  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  43  *  $wgMaxArticleSize*
  44  *
  45  *  * only within ParserOptions
  46  * </pre>
  47  *
  48  * @ingroup Parser
  49  */
  50 class Parser {
  51         /**
  52          * Update this version number when the ParserOutput format
  53          * changes in an incompatible way, so the parser cache
  54          * can automatically discard old data.
  55          */
  56         const VERSION = '1.6.5';
  57
  58         /**
  59          * Update this version number when the output of serialiseHalfParsedText()
  60          * changes in an incompatible way
  61          */
  62         const HALF_PARSED_VERSION = 2;
  63
  64         # Flags for Parser::setFunctionHook
  65         # Also available as global constants from Defines.php
  66         const SFH_NO_HASH = 1;
  67         const SFH_OBJECT_ARGS = 2;
  68
  69         # Constants needed for external link processing
  70         # Everything except bracket, space, or control characters
  71         const EXT_LINK_URL_CLASS = '(?:[^\]\[<>"\\x00-\\x20\\x7F]|(?:\[\]))';
  72         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  73                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  74
  75         # State constants for the definition list colon extraction
  76         const COLON_STATE_TEXT = 0;
  77         const COLON_STATE_TAG = 1;
  78         const COLON_STATE_TAGSTART = 2;
  79         const COLON_STATE_CLOSETAG = 3;
  80         const COLON_STATE_TAGSLASH = 4;
  81         const COLON_STATE_COMMENT = 5;
  82         const COLON_STATE_COMMENTDASH = 6;
  83         const COLON_STATE_COMMENTDASHDASH = 7;
  84
  85         # Flags for preprocessToDom
  86         const PTD_FOR_INCLUSION = 1;
  87
  88         # Allowed values for $this->mOutputType
  89         # Parameter to startExternalParse().
  90         const OT_HTML = 1; # like parse()
  91         const OT_WIKI = 2; # like preSaveTransform()
  92         const OT_PREPROCESS = 3; # like preprocess()
  93         const OT_MSG = 3;
  94         const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
  95
  96         # Marker Suffix needs to be accessible staticly.
  97         const MARKER_SUFFIX = "-QINU\x7f";
  98
  99         # Persistent:
 100         var $mTagHooks = array();
 101         var $mTransparentTagHooks = array();
 102         var $mFunctionHooks = array();
 103         var $mFunctionSynonyms = array( 0 => array(), 1 => array() );
 104         var $mFunctionTagHooks = array();
 105         var $mStripList  = array();
 106         var $mDefaultStripList  = array();
 107         var $mVarCache = array();
 108         var $mImageParams = array();
 109         var $mImageParamsMagicArray = array();
 110         var $mMarkerIndex = 0;
 111         var $mFirstCall = true;
 112         var $mVariables, $mSubstWords; # Initialised by initialiseVariables()
 113         var $mConf, $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols; # Initialised in constructor
 114
 115         # Cleared with clearState():
 116         /**
 117          * @var ParserOutput
 118          */
 119         var $mOutput;
 120         var $mAutonumber, $mDTopen;
 121
 122         /**
 123          * @var StripState
 124          */
 125         var $mStripState;
 126
 127         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 128         var $mLinkHolders, $mLinkID;
 129         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 130         var $mTplExpandCache; # empty-frame expansion cache
 131         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 132         var $mExpensiveFunctionCount; # number of expensive parser function calls
 133
 134         /**
 135          * @var User
 136          */
 137         var $mUser; # User object; only used when doing pre-save transform
 138
 139         # Temporary
 140         # These are variables reset at least once per parse regardless of $clearState
 141
 142         /**
 143          * @var ParserOptions
 144          */
 145         var $mOptions;
 146
 147         /**
 148          * @var Title
 149          */
 150         var $mTitle;        # Title context, used for self-link rendering and similar things
 151         var $mOutputType;   # Output type, one of the OT_xxx constants
 152         var $ot;            # Shortcut alias, see setOutputType()
 153         var $mRevisionObject; # The revision object of the specified revision ID
 154         var $mRevisionId;   # ID to display in {{REVISIONID}} tags
 155         var $mRevisionTimestamp; # The timestamp of the specified revision ID
 156         var $mRevisionUser; # Userto display in {{REVISIONUSER}} tag
 157         var $mRevIdForTs;   # The revision ID which was used to fetch the timestamp
 158
 159         /**
 160          * Constructor
 161          */
 162         public function __construct( $conf = array() ) {
 163                 $this->mConf = $conf;
 164                 $this->mUrlProtocols = wfUrlProtocols();
 165                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 166                         '(?:[^\]\[<>"\x00-\x20\x7F]|\[\])+) *([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/S';
 167                 if ( isset( $conf['preprocessorClass'] ) ) {
 168                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 169                 } elseif ( extension_loaded( 'domxml' ) ) {
 170                         # PECL extension that conflicts with the core DOM extension (bug 13770)
 171                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 172                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 173                 } elseif ( extension_loaded( 'dom' ) ) {
 174                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 175                 } else {
 176                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 177                 }
 178         }
 179
 180         /**
 181          * Reduce memory usage to reduce the impact of circular references
 182          */
 183         function __destruct() {
 184                 if ( isset( $this->mLinkHolders ) ) {
 185                         unset( $this->mLinkHolders );
 186                 }
 187                 foreach ( $this as $name => $value ) {
 188                         unset( $this->$name );
 189                 }
 190         }
 191
 192         /**
 193          * Do various kinds of initialisation on the first call of the parser
 194          */
 195         function firstCallInit() {
 196                 if ( !$this->mFirstCall ) {
 197                         return;
 198                 }
 199                 $this->mFirstCall = false;
 200
 201                 wfProfileIn( __METHOD__ );
 202
 203                 CoreParserFunctions::register( $this );
 204                 CoreTagHooks::register( $this );
 205                 $this->initialiseVariables();
 206
 207                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 208                 wfProfileOut( __METHOD__ );
 209         }
 210
 211         /**
 212          * Clear Parser state
 213          *
 214          * @private
 215          */
 216         function clearState() {
 217                 wfProfileIn( __METHOD__ );
 218                 if ( $this->mFirstCall ) {
 219                         $this->firstCallInit();
 220                 }
 221                 $this->mOutput = new ParserOutput;
 222                 $this->mOptions->registerWatcher( array( $this->mOutput, 'recordOption' ) );
 223                 $this->mAutonumber = 0;
 224                 $this->mLastSection = '';
 225                 $this->mDTopen = false;
 226                 $this->mIncludeCount = array();
 227                 $this->mArgStack = false;
 228                 $this->mInPre = false;
 229                 $this->mLinkHolders = new LinkHolderArray( $this );
 230                 $this->mLinkID = 0;
 231                 $this->mRevisionObject = $this->mRevisionTimestamp =
 232                         $this->mRevisionId = $this->mRevisionUser = null;
 233                 $this->mVarCache = array();
 234                 $this->mUser = null;
 235
 236                 /**
 237                  * Prefix for temporary replacement strings for the multipass parser.
 238                  * \x07 should never appear in input as it's disallowed in XML.
 239                  * Using it at the front also gives us a little extra robustness
 240                  * since it shouldn't match when butted up against identifier-like
 241                  * string constructs.
 242                  *
 243                  * Must not consist of all title characters, or else it will change
 244                  * the behaviour of <nowiki> in a link.
 245                  */
 246                 # $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 247                 # Changed to \x7f to allow XML double-parsing -- TS
 248                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 249                 $this->mStripState = new StripState( $this->mUniqPrefix );
 250
 251
 252                 # Clear these on every parse, bug 4549
 253                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 254
 255                 $this->mShowToc = true;
 256                 $this->mForceTocPosition = false;
 257                 $this->mIncludeSizes = array(
 258                         'post-expand' => 0,
 259                         'arg' => 0,
 260                 );
 261                 $this->mPPNodeCount = 0;
 262                 $this->mDefaultSort = false;
 263                 $this->mHeadings = array();
 264                 $this->mDoubleUnderscores = array();
 265                 $this->mExpensiveFunctionCount = 0;
 266
 267                 # Fix cloning
 268                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 269                         $this->mPreprocessor = null;
 270                 }
 271
 272                 wfRunHooks( 'ParserClearState', array( &$this ) );
 273                 wfProfileOut( __METHOD__ );
 274         }
 275
 276         /**
 277          * Convert wikitext to HTML
 278          * Do not call this function recursively.
 279          *
 280          * @param $text String: text we want to parse
 281          * @param $title Title object
 282          * @param $options ParserOptions
 283          * @param $linestart boolean
 284          * @param $clearState boolean
 285          * @param $revid Int: number to pass in {{REVISIONID}}
 286          * @return ParserOutput a ParserOutput
 287          */
 288         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 289                 /**
 290                  * First pass--just handle <nowiki> sections, pass the rest off
 291                  * to internalParse() which does all the real work.
 292                  */
 293
 294                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang, $wgDisableLangConversion, $wgDisableTitleConversion;
 295                 $fname = __METHOD__.'-' . wfGetCaller();
 296                 wfProfileIn( __METHOD__ );
 297                 wfProfileIn( $fname );
 298
 299                 $this->startParse( $title, $options, self::OT_HTML, $clearState );
 300
 301                 $oldRevisionId = $this->mRevisionId;
 302                 $oldRevisionObject = $this->mRevisionObject;
 303                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 304                 $oldRevisionUser = $this->mRevisionUser;
 305                 if ( $revid !== null ) {
 306                         $this->mRevisionId = $revid;
 307                         $this->mRevisionObject = null;
 308                         $this->mRevisionTimestamp = null;
 309                         $this->mRevisionUser = null;
 310                 }
 311
 312                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 313                 # No more strip!
 314                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 315                 $text = $this->internalParse( $text );
 316
 317                 $text = $this->mStripState->unstripGeneral( $text );
 318
 319                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 320                 $fixtags = array(
 321                         # french spaces, last one Guillemet-left
 322                         # only if there is something before the space
 323                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;\\2',
 324                         # french spaces, Guillemet-right
 325                         '/(\\302\\253) /' => '\\1&#160;',
 326                         '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
 327                 );
 328                 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
 329
 330                 $text = $this->doBlockLevels( $text, $linestart );
 331
 332                 $this->replaceLinkHolders( $text );
 333
 334                 /**
 335                  * The page doesn't get language converted if
 336                  * a) It's disabled
 337                  * b) Content isn't converted
 338                  * c) It's a conversion table
 339                  */
 340                 if ( !( $wgDisableLangConversion
 341                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 342                                 || $this->mTitle->isConversionTable() ) ) {
 343
 344                         # The position of the convert() call should not be changed. it
 345                         # assumes that the links are all replaced and the only thing left
 346                         # is the <nowiki> mark.
 347
 348                         $text = $wgContLang->convert( $text );
 349                 }
 350
 351                 /**
 352                  * A converted title will be provided in the output object if title and
 353                  * content conversion are enabled, the article text does not contain
 354                  * a conversion-suppressing double-underscore tag, and no
 355                  * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
 356                  * automatic link conversion.
 357                  */
 358                 if ( !( $wgDisableLangConversion
 359                                 || $wgDisableTitleConversion
 360                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 361                                 || isset( $this->mDoubleUnderscores['notitleconvert'] )
 362                                 || $this->mOutput->getDisplayTitle() !== false ) )
 363                 {
 364                         $convruletitle = $wgContLang->getConvRuleTitle();
 365                         if ( $convruletitle ) {
 366                                 $this->mOutput->setTitleText( $convruletitle );
 367                         } else {
 368                                 $titleText = $wgContLang->convertTitle( $title );
 369                                 $this->mOutput->setTitleText( $titleText );
 370                         }
 371                 }
 372
 373                 $text = $this->mStripState->unstripNoWiki( $text );
 374
 375                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 376
 377                 $text = $this->replaceTransparentTags( $text );
 378                 $text = $this->mStripState->unstripGeneral( $text );
 379
 380                 $text = Sanitizer::normalizeCharReferences( $text );
 381
 382                 if ( ( $wgUseTidy && $this->mOptions->getTidy() ) || $wgAlwaysUseTidy ) {
 383                         $text = MWTidy::tidy( $text );
 384                 } else {
 385                         # attempt to sanitize at least some nesting problems
 386                         # (bug #2702 and quite a few others)
 387                         $tidyregs = array(
 388                                 # ''Something [http://www.cool.com cool''] -->
 389                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 390                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 391                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 392                                 # fix up an anchor inside another anchor, only
 393                                 # at least for a single single nested link (bug 3695)
 394                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 395                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 396                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 397                                 # contains a div, so fix it up here; replace
 398                                 # div with escaped text
 399                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 400                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 401                                 # remove empty italic or bold tag pairs, some
 402                                 # introduced by rules above
 403                                 '/<([bi])><\/\\1>/' => '',
 404                         );
 405
 406                         $text = preg_replace(
 407                                 array_keys( $tidyregs ),
 408                                 array_values( $tidyregs ),
 409                                 $text );
 410                 }
 411                 global $wgExpensiveParserFunctionLimit;
 412                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 413                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 414                 }
 415
 416                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 417
 418                 # Information on include size limits, for the benefit of users who try to skirt them
 419                 if ( $this->mOptions->getEnableLimitReport() ) {
 420                         $max = $this->mOptions->getMaxIncludeSize();
 421                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 422                         $limitReport =
 423                                 "NewPP limit report\n" .
 424                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->getMaxPPNodeCount()}\n" .
 425                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 426                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 427                                 $PFreport;
 428                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 429                         $text .= "\n<!-- \n$limitReport-->\n";
 430                 }
 431                 $this->mOutput->setText( $text );
 432
 433                 $this->mRevisionId = $oldRevisionId;
 434                 $this->mRevisionObject = $oldRevisionObject;
 435                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 436                 $this->mRevisionUser = $oldRevisionUser;
 437                 wfProfileOut( $fname );
 438                 wfProfileOut( __METHOD__ );
 439
 440                 return $this->mOutput;
 441         }
 442
 443         /**
 444          * Recursive parser entry point that can be called from an extension tag
 445          * hook.
 446          *
 447          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 448          *
 449          * @param $text String: text extension wants to have parsed
 450          * @param $frame PPFrame: The frame to use for expanding any template variables
 451          */
 452         function recursiveTagParse( $text, $frame=false ) {
 453                 wfProfileIn( __METHOD__ );
 454                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 455                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 456                 $text = $this->internalParse( $text, false, $frame );
 457                 wfProfileOut( __METHOD__ );
 458                 return $text;
 459         }
 460
 461         /**
 462          * Expand templates and variables in the text, producing valid, static wikitext.
 463          * Also removes comments.
 464          */
 465         function preprocess( $text, Title $title, ParserOptions $options, $revid = null ) {
 466                 wfProfileIn( __METHOD__ );
 467                 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
 468                 if ( $revid !== null ) {
 469                         $this->mRevisionId = $revid;
 470                 }
 471                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 472                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 473                 $text = $this->replaceVariables( $text );
 474                 $text = $this->mStripState->unstripBoth( $text );
 475                 wfProfileOut( __METHOD__ );
 476                 return $text;
 477         }
 478
 479         /**
 480          * Process the wikitext for the ?preload= feature. (bug 5210)
 481          *
 482          * <noinclude>, <includeonly> etc. are parsed as for template transclusion,
 483          * comments, templates, arguments, tags hooks and parser functions are untouched.
 484          */
 485         public function getPreloadText( $text, Title $title, ParserOptions $options ) {
 486                 # Parser (re)initialisation
 487                 $this->startParse( $title, $options, self::OT_PLAIN, true );
 488
 489                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 490                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
 491                 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
 492                 $text = $this->mStripState->unstripBoth( $text );
 493                 return $text;
 494         }
 495
 496         /**
 497          * Get a random string
 498          */
 499         static public function getRandomString() {
 500                 return dechex( mt_rand( 0, 0x7fffffff ) ) . dechex( mt_rand( 0, 0x7fffffff ) );
 501         }
 502
 503         /**
 504          * Set the current user.
 505          * Should only be used when doing pre-save transform.
 506          *
 507          * @param $user Mixed: User object or null (to reset)
 508          */
 509         function setUser( $user ) {
 510                 $this->mUser = $user;
 511         }
 512
 513         /**
 514          * Accessor for mUniqPrefix.
 515          *
 516          * @return String
 517          */
 518         public function uniqPrefix() {
 519                 if ( !isset( $this->mUniqPrefix ) ) {
 520                         # @todo Fixme: this is probably *horribly wrong*
 521                         # LanguageConverter seems to want $wgParser's uniqPrefix, however
 522                         # if this is called for a parser cache hit, the parser may not
 523                         # have ever been initialized in the first place.
 524                         # Not really sure what the heck is supposed to be going on here.
 525                         return '';
 526                         # throw new MWException( "Accessing uninitialized mUniqPrefix" );
 527                 }
 528                 return $this->mUniqPrefix;
 529         }
 530
 531         /**
 532          * Set the context title
 533          */
 534         function setTitle( $t ) {
 535                 if ( !$t || $t instanceof FakeTitle ) {
 536                         $t = Title::newFromText( 'NO TITLE' );
 537                 }
 538
 539                 if ( strval( $t->getFragment() ) !== '' ) {
 540                         # Strip the fragment to avoid various odd effects
 541                         $this->mTitle = clone $t;
 542                         $this->mTitle->setFragment( '' );
 543                 } else {
 544                         $this->mTitle = $t;
 545                 }
 546         }
 547
 548         /**
 549          * Accessor for the Title object
 550          *
 551          * @return Title object
 552          */
 553         function getTitle() {
 554                 return $this->mTitle;
 555         }
 556
 557         /**
 558          * Accessor/mutator for the Title object
 559          *
 560          * @param $x New Title object or null to just get the current one
 561          * @return Title object
 562          */
 563         function Title( $x = null ) {
 564                 return wfSetVar( $this->mTitle, $x );
 565         }
 566
 567         /**
 568          * Set the output type
 569          *
 570          * @param $ot Integer: new value
 571          */
 572         function setOutputType( $ot ) {
 573                 $this->mOutputType = $ot;
 574                 # Shortcut alias
 575                 $this->ot = array(
 576                         'html' => $ot == self::OT_HTML,
 577                         'wiki' => $ot == self::OT_WIKI,
 578                         'pre' => $ot == self::OT_PREPROCESS,
 579                         'plain' => $ot == self::OT_PLAIN,
 580                 );
 581         }
 582
 583         /**
 584          * Accessor/mutator for the output type
 585          *
 586          * @param $x New value or null to just get the current one
 587          * @return Integer
 588          */
 589         function OutputType( $x = null ) {
 590                 return wfSetVar( $this->mOutputType, $x );
 591         }
 592
 593         /**
 594          * Get the ParserOutput object
 595          *
 596          * @return ParserOutput object
 597          */
 598         function getOutput() {
 599                 return $this->mOutput;
 600         }
 601
 602         /**
 603          * Get the ParserOptions object
 604          *
 605          * @return ParserOptions object
 606          */
 607         function getOptions() {
 608                 return $this->mOptions;
 609         }
 610
 611         /**
 612          * Accessor/mutator for the ParserOptions object
 613          *
 614          * @param $x New value or null to just get the current one
 615          * @return Current ParserOptions object
 616          */
 617         function Options( $x = null ) {
 618                 return wfSetVar( $this->mOptions, $x );
 619         }
 620
 621         function nextLinkID() {
 622                 return $this->mLinkID++;
 623         }
 624
 625         function setLinkID( $id ) {
 626                 $this->mLinkID = $id;
 627         }
 628
 629         /**
 630          * @return Language
 631          */
 632         function getFunctionLang() {
 633                 global $wgLang, $wgContLang;
 634
 635                 $target = $this->mOptions->getTargetLanguage();
 636                 if ( $target !== null ) {
 637                         return $target;
 638                 } else {
 639                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 640                 }
 641         }
 642
 643         /**
 644          * Get a User object either from $this->mUser, if set, or from the
 645          * ParserOptions object otherwise
 646          *
 647          * @return User object
 648          */
 649         function getUser() {
 650                 if ( !is_null( $this->mUser ) ) {
 651                         return $this->mUser;
 652                 }
 653                 return $this->mOptions->getUser();
 654         }
 655
 656         /**
 657          * Get a preprocessor object
 658          *
 659          * @return Preprocessor instance
 660          */
 661         function getPreprocessor() {
 662                 if ( !isset( $this->mPreprocessor ) ) {
 663                         $class = $this->mPreprocessorClass;
 664                         $this->mPreprocessor = new $class( $this );
 665                 }
 666                 return $this->mPreprocessor;
 667         }
 668
 669         /**
 670          * Replaces all occurrences of HTML-style comments and the given tags
 671          * in the text with a random marker and returns the next text. The output
 672          * parameter $matches will be an associative array filled with data in
 673          * the form:
 674          *   'UNIQ-xxxxx' => array(
 675          *     'element',
 676          *     'tag content',
 677          *     array( 'param' => 'x' ),
 678          *     '<element param="x">tag content</element>' ) )
 679          *
 680          * @param $elements list of element names. Comments are always extracted.
 681          * @param $text Source text string.
 682          * @param $matches Out parameter, Array: extracted tags
 683          * @param $uniq_prefix
 684          * @return String: stripped text
 685          */
 686         public static function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
 687                 static $n = 1;
 688                 $stripped = '';
 689                 $matches = array();
 690
 691                 $taglist = implode( '|', $elements );
 692                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 693
 694                 while ( $text != '' ) {
 695                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 696                         $stripped .= $p[0];
 697                         if ( count( $p ) < 5 ) {
 698                                 break;
 699                         }
 700                         if ( count( $p ) > 5 ) {
 701                                 # comment
 702                                 $element    = $p[4];
 703                                 $attributes = '';
 704                                 $close      = '';
 705                                 $inside     = $p[5];
 706                         } else {
 707                                 # tag
 708                                 $element    = $p[1];
 709                                 $attributes = $p[2];
 710                                 $close      = $p[3];
 711                                 $inside     = $p[4];
 712                         }
 713
 714                         $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
 715                         $stripped .= $marker;
 716
 717                         if ( $close === '/>' ) {
 718                                 # Empty element tag, <tag />
 719                                 $content = null;
 720                                 $text = $inside;
 721                                 $tail = null;
 722                         } else {
 723                                 if ( $element === '!--' ) {
 724                                         $end = '/(-->)/';
 725                                 } else {
 726                                         $end = "/(<\\/$element\\s*>)/i";
 727                                 }
 728                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 729                                 $content = $q[0];
 730                                 if ( count( $q ) < 3 ) {
 731                                         # No end tag -- let it run out to the end of the text.
 732                                         $tail = '';
 733                                         $text = '';
 734                                 } else {
 735                                         $tail = $q[1];
 736                                         $text = $q[2];
 737                                 }
 738                         }
 739
 740                         $matches[$marker] = array( $element,
 741                                 $content,
 742                                 Sanitizer::decodeTagAttributes( $attributes ),
 743                                 "<$element$attributes$close$content$tail" );
 744                 }
 745                 return $stripped;
 746         }
 747
 748         /**
 749          * Get a list of strippable XML-like elements
 750          */
 751         function getStripList() {
 752                 return $this->mStripList;
 753         }
 754
 755         /**
 756          * @deprecated use replaceVariables
 757          */
 758         function strip( $text, $state, $stripcomments = false , $dontstrip = array() ) {
 759                 return $text;
 760         }
 761
 762         /**
 763          * Restores pre, math, and other extensions removed by strip()
 764          *
 765          * always call unstripNoWiki() after this one
 766          * @private
 767          * @deprecated use $this->mStripState->unstrip()
 768          */
 769         function unstrip( $text, $state ) {
 770                 return $state->unstripGeneral( $text );
 771         }
 772
 773         /**
 774          * Always call this after unstrip() to preserve the order
 775          *
 776          * @private
 777          * @deprecated use $this->mStripState->unstrip()
 778          */
 779         function unstripNoWiki( $text, $state ) {
 780                 return $state->unstripNoWiki( $text );
 781         }
 782
 783         /**
 784          * @deprecated use $this->mStripState->unstripBoth()
 785          */
 786         function unstripForHTML( $text ) {
 787                 return $this->mStripState->unstripBoth( $text );
 788         }
 789
 790         /**
 791          * Add an item to the strip state
 792          * Returns the unique tag which must be inserted into the stripped text
 793          * The tag will be replaced with the original text in unstrip()
 794          *
 795          * @private
 796          */
 797         function insertStripItem( $text ) {
 798                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 799                 $this->mMarkerIndex++;
 800                 $this->mStripState->addGeneral( $rnd, $text );
 801                 return $rnd;
 802         }
 803
 804         /**
 805          * parse the wiki syntax used to render tables
 806          *
 807          * @private
 808          */
 809         function doTableStuff( $text ) {
 810                 wfProfileIn( __METHOD__ );
 811
 812                 $lines = StringUtils::explode( "\n", $text );
 813                 $out = '';
 814                 $output =& $out;
 815
 816                 foreach ( $lines as $outLine ) {
 817                         $line = trim( $outLine );
 818
 819                         # empty line, go to next line,
 820                         # but only append \n if outside of table
 821                         if ( $line === '') {
 822                                 $output .= $outLine . "\n";
 823                                 continue;
 824                         }
 825                         $firstChars = $line[0];
 826                         if ( strlen( $line ) > 1 ) {
 827                                 $firstChars .= in_array( $line[1], array( '}', '+', '-' ) ) ? $line[1] : '';
 828                         }
 829                         $matches = array();
 830
 831                         if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line , $matches ) ) {
 832                                 $tables[] = array();
 833                                 $table =& $this->last( $tables );
 834                                 $table[0] = array(); // first row
 835                                 $currentRow =& $table[0];
 836                                 $table['indent'] = strlen( $matches[1] );
 837
 838                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 839                                 $attributes = Sanitizer::fixTagAttributes( $attributes , 'table' );
 840
 841                                 if ( $attributes !== '' ) {
 842                                         $table['attributes'] = $attributes;
 843                                 }
 844                         } else if ( !isset( $tables[0] ) ) {
 845                                 // we're outside the table
 846
 847                                 $out .= $outLine . "\n";
 848                         } else if ( $firstChars === '|}' ) {
 849                                 // trim the |} code from the line
 850                                 $line = substr ( $line , 2 );
 851
 852                                 // Shorthand for last row
 853                                 $lastRow =& $this->last( $table );
 854
 855                                 // a thead at the end becomes a tfoot, unless there is only one row
 856                                 // Do this before deleting empty last lines to allow headers at the bottom of tables
 857                                 if ( isset( $lastRow['type'] ) && $lastRow['type'] == 'thead' && isset( $table[1] ) ) {
 858                                         $lastRow['type'] = 'tfoot';
 859                                         for ( $i = 0; isset( $lastRow[$i] ); $i++ ) {
 860                                                 $lastRow[$i]['type'] = 'th';
 861                                         }
 862                                 }
 863
 864                                 // Delete empty last lines
 865                                 if ( empty( $lastRow ) ) {
 866                                         $lastRow = NULL;
 867                                 }
 868                                 $o = '';
 869                                 $curtable = array_pop( $tables );
 870
 871                                 #Add a line-ending before the table, but only if there isn't one already
 872                                 if ( substr( $out, -1 ) !== "\n" ) {
 873                                         $o .= "\n";
 874                                 }
 875                                 $o .= $this->generateTableHTML( $curtable ) . $line . "\n";
 876
 877                                 if ( count( $tables ) > 0 ) {
 878                                         $table =& $this->last( $tables );
 879                                         $currentRow =& $this->last( $table );
 880                                         $currentElement =& $this->last( $currentRow );
 881
 882                                         $output =& $currentElement['content'];
 883                                 } else {
 884                                         $output =& $out;
 885                                 }
 886
 887                                 $output .= $o;
 888
 889                         } else if ( $firstChars === '|-' ) {
 890                                 // start a new row element
 891                                 // but only when we haven't started one already
 892                                 if ( count( $currentRow ) != 0 ) {
 893                                         $table[] = array();
 894                                         $currentRow =& $this->last( $table );
 895                                 }
 896                                 // Get the attributes, there's nothing else useful in $line now
 897                                 $line = substr ( $line , 2 );
 898                                 $attributes = $this->mStripState->unstripBoth( $line );
 899                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 900                                 if ( $attributes !== '' ) {
 901                                         $currentRow['attributes'] = $attributes;
 902                                 }
 903
 904                         } else if ( $firstChars  === '|+' ) {
 905                                 // a table caption, but only proceed if there isn't one already
 906                                 if ( !isset ( $table['caption'] ) ) {
 907                                         $line = substr ( $line , 2 );
 908
 909                                         $c = $this->getCellAttr( $line , 'caption' );
 910                                         $table['caption'] = array();
 911                                         $table['caption']['content'] = $c[0];
 912                                         if ( isset( $c[1] ) ) $table['caption']['attributes'] = $c[1];
 913                                         unset( $c );
 914                                         $output =& $table['caption']['content'];
 915                                 }
 916                         } else if ( $firstChars === '|' || $firstChars === '!' || $firstChars === '!+' ) {
 917                                 // Which kind of cells are we dealing with
 918                                 $currentTag = 'td';
 919                                 $line = substr ( $line , 1 );
 920
 921                                 if ( $firstChars === '!'  || $firstChars === '!+' ) {
 922                                         $line = str_replace ( '!!' , '||' , $line );
 923                                         $currentTag = 'th';
 924                                 }
 925
 926                                 // Split up multiple cells on the same line.
 927                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 928                                 $line = ''; // save memory
 929
 930                                 // decide whether thead to tbody
 931                                 if ( !array_key_exists( 'type', $currentRow ) ) {
 932                                         $currentRow['type'] = ( $firstChars === '!' ) ? 'thead' : 'tbody' ;
 933                                 } else if ( $firstChars === '|' ) {
 934                                         $currentRow['type'] = 'tbody';
 935                                 }
 936
 937                                 // Loop through each table cell
 938                                 foreach ( $cells as $cell ) {
 939                                         // a new cell
 940                                         $currentRow[] = array();
 941                                         $currentElement =& $this->last( $currentRow );
 942
 943                                         $currentElement['type'] = $currentTag;
 944
 945                                         $c = $this->getCellAttr( $cell , $currentTag );
 946                                         $currentElement['content'] = $c[0];
 947                                         if ( isset( $c[1] ) ) $currentElement['attributes'] = $c[1];
 948                                         unset( $c );
 949                                 }
 950                                 $output =& $currentElement['content'];
 951
 952                         } else {
 953                                 $output .= "\n$outLine";
 954                         }
 955                 }
 956
 957                 # Remove trailing line-ending (b/c)
 958                 if ( substr( $out, -1 ) === "\n" ) {
 959                         $out = substr( $out, 0, -1 );
 960                 }
 961
 962                 # Close any unclosed tables
 963                 if ( isset( $tables ) && count( $tables ) > 0 ) {
 964                         for ( $i = 0; $i < count( $tables ); $i++ ) {
 965                                 $curtable = array_pop( $tables );
 966                                 $curtable = $this->generateTableHTML( $curtable );
 967                                 #Add a line-ending before the table, but only if there isn't one already
 968                                 if ( substr( $out, -1 ) !== "\n"  && $curtable !== "" ) {
 969                                         $out .= "\n";
 970                                 }
 971                                 $out .= $curtable;
 972                         }
 973                 }
 974
 975                 wfProfileOut( __METHOD__ );
 976
 977                 return $out;
 978         }
 979
 980         /**
 981          * Helper function for doTableStuff() separating the contents of cells from
 982          * attributes. Particularly useful as there's a possible bug and this action
 983          * is repeated twice.
 984          *
 985          * @private
 986          */
 987         function getCellAttr ( $cell, $tagName ) {
 988                 $content = null;
 989                 $attributes = null;
 990
 991                 $cell = trim ( $cell );
 992
 993                 // A cell could contain both parameters and data
 994                 $cellData = explode ( '|' , $cell , 2 );
 995
 996                 // Bug 553: Note that a '|' inside an invalid link should not
 997                 // be mistaken as delimiting cell parameters
 998                 if ( strpos( $cellData[0], '[[' ) !== false ) {
 999                         $content = trim ( $cell );
1000                 }
1001                 else if ( count ( $cellData ) == 1 ) {
1002                         $content = trim ( $cellData[0] );
1003                 }
1004                 else {
1005                         $attributes = $this->mStripState->unstripBoth( $cellData[0] );
1006                         $attributes = Sanitizer::fixTagAttributes( $attributes , $tagName );
1007
1008                         $content = trim ( $cellData[1] );
1009                 }
1010                 return array( $content, $attributes );
1011         }
1012
1013
1014         /**
1015          * Helper function for doTableStuff(). This converts the structured array into html.
1016          *
1017          * @private
1018          */
1019         function generateTableHTML ( &$table ) {
1020                 $return = "";
1021                 $return .= str_repeat( '<dl><dd>' , $table['indent'] );
1022                 $return .= '<table';
1023                 $return .= isset( $table['attributes'] ) ? $table['attributes'] : '';
1024                 $return .= '>';
1025                 unset( $table['attributes'] );
1026
1027                 if ( isset( $table['caption'] ) ) {
1028                         $return .= "\n<caption";
1029                         $return .= isset( $table['caption']['attributes'] ) ? $table['caption']['attributes'] : '';
1030                         $return .= '>';
1031                         $return .= $table['caption']['content'];
1032                         $return .= "\n</caption>";
1033                 }
1034                 $lastSection = '';
1035                 $empty = true;
1036                 $simple = true;
1037
1038                 // If we only have tbodies, mark table as simple
1039                 for ( $i = 0; isset( $table[$i] ); $i++ ) {
1040                         if ( !count( $table[$i] ) ) continue;
1041                         if ( !isset( $table[$i]['type'] ) ) $table[$i]['type'] = 'tbody';
1042                         if ( !$lastSection ) {
1043                                 $lastSection = $table[$i]['type'];
1044                         } else if ( $lastSection != $table[$i]['type'] ) {
1045                                 $simple = false;
1046                                 break;
1047                         }
1048                 }
1049                 $lastSection = '';
1050                 for ( $i = 0; isset( $table[$i] ); $i++ ) {
1051                         // Check for empty tables
1052                         if ( count( $table[$i] ) ) {
1053                                 $empty = false;
1054                         } else {
1055                                 continue;
1056                         }
1057                         if ( $table[$i]['type'] != $lastSection && !$simple ) {
1058                                 $return .= "\n<" . $table[$i]['type'] . '>';
1059                         }
1060
1061                         $return .= "\n<tr";
1062                         $return .= isset( $table[$i]['attributes'] ) ? $table[$i]['attributes'] : '';
1063                         $return .= '>';
1064                         for ( $j = 0; isset( $table[$i][$j] ); $j++ ) {
1065                                 if ( !isset( $table[$i][$j]['type'] ) ) $table[$i][$j]['type'] = 'td';
1066                                 $return .= "\n<" . $table[$i][$j]['type'];
1067                                 $return .= isset( $table[$i][$j]['attributes'] ) ? $table[$i][$j]['attributes'] : '';
1068                                 $return .= '>';
1069
1070                                 $return .= $table[$i][$j]['content'];
1071                                 if ( $table[$i][$j]['content'] != '' )
1072                                         $return .= "\n";
1073
1074                                 $return .= '</' . $table[$i][$j]['type'] . '>';
1075                                 unset( $table[$i][$j] );
1076                         }
1077                         $return .= "\n</tr>";
1078
1079                         if ( ( !isset( $table[$i + 1] ) && !$simple ) || ( isset( $table[$i + 1] ) && isset( $table[$i + 1]['type'] ) && $table[$i]['type'] != $table[$i + 1]['type'] ) ) {
1080                                 $return .= '</' . $table[$i]['type'] . '>';
1081                         }
1082                         $lastSection = $table[$i]['type'];
1083                         unset( $table[$i] );
1084                 }
1085                 if ( $empty ) {
1086                         if ( isset( $table['caption'] ) ) {
1087                                 $return .= "\n<tr><td></td></tr>";
1088                         } else {
1089                                 return '';
1090                         }
1091                 }
1092                 $return .= "\n</table>";
1093                 $return .= str_repeat( '</dd></dl>' , $table['indent'] );
1094
1095                 return $return;
1096         }
1097
1098         /**
1099          * like end() but only works on the numeric array index and php's internal pointers
1100          * returns a reference to the last element of an array much like "\$arr[-1]" in perl
1101          * ignores associative elements and will create a 0 key will a NULL value if there were
1102          * no numric elements and an array itself if not previously defined.
1103          *
1104          * @private
1105          */
1106         function &last ( &$arr ) {
1107                 for ( $i = count( $arr ); ( !isset( $arr[$i] ) && $i > 0 ); $i-- ) {  }
1108                 return $arr[$i];
1109         }
1110
1111         /**
1112          * Helper function for parse() that transforms wiki markup into
1113          * HTML. Only called for $mOutputType == self::OT_HTML.
1114          *
1115          * @private
1116          */
1117         function internalParse( $text, $isMain = true, $frame=false ) {
1118                 wfProfileIn( __METHOD__ );
1119
1120                 $origText = $text;
1121
1122                 # Hook to suspend the parser in this state
1123                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1124                         wfProfileOut( __METHOD__ );
1125                         return $text ;
1126                 }
1127
1128                 # if $frame is provided, then use $frame for replacing any variables
1129                 if ( $frame ) {
1130                         # use frame depth to infer how include/noinclude tags should be handled
1131                         # depth=0 means this is the top-level document; otherwise it's an included document
1132                         if ( !$frame->depth ) {
1133                                 $flag = 0;
1134                         } else {
1135                                 $flag = Parser::PTD_FOR_INCLUSION;
1136                         }
1137                         $dom = $this->preprocessToDom( $text, $flag );
1138                         $text = $frame->expand( $dom );
1139                 } else {
1140                         # if $frame is not provided, then use old-style replaceVariables
1141                         $text = $this->replaceVariables( $text );
1142                 }
1143
1144                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
1145                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1146
1147                 # Tables need to come after variable replacement for things to work
1148                 # properly; putting them before other transformations should keep
1149                 # exciting things like link expansions from showing up in surprising
1150                 # places.
1151                 $text = $this->doTableStuff( $text );
1152
1153                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1154
1155                 $text = $this->doDoubleUnderscore( $text );
1156
1157                 $text = $this->doHeadings( $text );
1158                 if ( $this->mOptions->getUseDynamicDates() ) {
1159                         $df = DateFormatter::getInstance();
1160                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1161                 }
1162                 $text = $this->replaceInternalLinks( $text );
1163                 $text = $this->doAllQuotes( $text );
1164                 $text = $this->replaceExternalLinks( $text );
1165
1166                 # replaceInternalLinks may sometimes leave behind
1167                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1168                 $text = str_replace( $this->mUniqPrefix.'NOPARSE', '', $text );
1169
1170                 $text = $this->doMagicLinks( $text );
1171                 $text = $this->formatHeadings( $text, $origText, $isMain );
1172
1173                 wfProfileOut( __METHOD__ );
1174                 return $text;
1175         }
1176
1177         /**
1178          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1179          * magic external links.
1180          *
1181          * DML
1182          * @private
1183          */
1184         function doMagicLinks( $text ) {
1185                 wfProfileIn( __METHOD__ );
1186                 $prots = $this->mUrlProtocols;
1187                 $urlChar = self::EXT_LINK_URL_CLASS;
1188                 $text = preg_replace_callback(
1189                         '!(?:                           # Start cases
1190                                 (<a[ \t\r\n>].*?</a>) |     # m[1]: Skip link text
1191                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1192                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
1193                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1194                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1195                                         (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1196                                         (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1197                                         [0-9Xx]                 # check digit
1198                                         \b)
1199                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1200                 wfProfileOut( __METHOD__ );
1201                 return $text;
1202         }
1203
1204         function magicLinkCallback( $m ) {
1205                 if ( isset( $m[1] ) && $m[1] !== '' ) {
1206                         # Skip anchor
1207                         return $m[0];
1208                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1209                         # Skip HTML element
1210                         return $m[0];
1211                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1212                         # Free external link
1213                         return $this->makeFreeExternalLink( $m[0] );
1214                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1215                         # RFC or PMID
1216                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1217                                 $keyword = 'RFC';
1218                                 $urlmsg = 'rfcurl';
1219                                 $CssClass = 'mw-magiclink-rfc';
1220                                 $id = $m[4];
1221                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1222                                 $keyword = 'PMID';
1223                                 $urlmsg = 'pubmedurl';
1224                                 $CssClass = 'mw-magiclink-pmid';
1225                                 $id = $m[4];
1226                         } else {
1227                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1228                                         substr( $m[0], 0, 20 ) . '"' );
1229                         }
1230                         $url = wfMsgForContent( $urlmsg, $id );
1231                         return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $CssClass );
1232                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1233                         # ISBN
1234                         $isbn = $m[5];
1235                         $num = strtr( $isbn, array(
1236                                 '-' => '',
1237                                 ' ' => '',
1238                                 'x' => 'X',
1239                         ));
1240                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1241                         return'<a href="' .
1242                                 $titleObj->escapeLocalUrl() .
1243                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1244                 } else {
1245                         return $m[0];
1246                 }
1247         }
1248
1249         /**
1250          * Make a free external link, given a user-supplied URL
1251          * @return HTML
1252          * @private
1253          */
1254         function makeFreeExternalLink( $url ) {
1255                 global $wgContLang;
1256                 wfProfileIn( __METHOD__ );
1257
1258                 $trail = '';
1259
1260                 # The characters '<' and '>' (which were escaped by
1261                 # removeHTMLtags()) should not be included in
1262                 # URLs, per RFC 2396.
1263                 $m2 = array();
1264                 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1265                         $trail = substr( $url, $m2[0][1] ) . $trail;
1266                         $url = substr( $url, 0, $m2[0][1] );
1267                 }
1268
1269                 # Move trailing punctuation to $trail
1270                 $sep = ',;\.:!?';
1271                 # If there is no left bracket, then consider right brackets fair game too
1272                 if ( strpos( $url, '(' ) === false ) {
1273                         $sep .= ')';
1274                 }
1275
1276                 $numSepChars = strspn( strrev( $url ), $sep );
1277                 if ( $numSepChars ) {
1278                         $trail = substr( $url, -$numSepChars ) . $trail;
1279                         $url = substr( $url, 0, -$numSepChars );
1280                 }
1281
1282                 $url = Sanitizer::cleanUrl( $url );
1283
1284                 # Is this an external image?
1285                 $text = $this->maybeMakeExternalImage( $url );
1286                 if ( $text === false ) {
1287                         # Not an image, make a link
1288                         $text = Linker::makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1289                                 $this->getExternalLinkAttribs( $url ) );
1290                         # Register it in the output object...
1291                         # Replace unnecessary URL escape codes with their equivalent characters
1292                         $pasteurized = self::replaceUnusualEscapes( $url );
1293                         $this->mOutput->addExternalLink( $pasteurized );
1294                 }
1295                 wfProfileOut( __METHOD__ );
1296                 return $text . $trail;
1297         }
1298
1299
1300         /**
1301          * Parse headers and return html
1302          *
1303          * @private
1304          */
1305         function doHeadings( $text ) {
1306                 wfProfileIn( __METHOD__ );
1307                 for ( $i = 6; $i >= 1; --$i ) {
1308                         $h = str_repeat( '=', $i );
1309                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1310                           "<h$i>\\1</h$i>", $text );
1311                 }
1312                 wfProfileOut( __METHOD__ );
1313                 return $text;
1314         }
1315
1316         /**
1317          * Replace single quotes with HTML markup
1318          * @private
1319          * @return string the altered text
1320          */
1321         function doAllQuotes( $text ) {
1322                 wfProfileIn( __METHOD__ );
1323                 $outtext = '';
1324                 $lines = StringUtils::explode( "\n", $text );
1325                 foreach ( $lines as $line ) {
1326                         $outtext .= $this->doQuotes( $line ) . "\n";
1327                 }
1328                 $outtext = substr( $outtext, 0,-1 );
1329                 wfProfileOut( __METHOD__ );
1330                 return $outtext;
1331         }
1332
1333         /**
1334          * Helper function for doAllQuotes()
1335          */
1336         public function doQuotes( $text ) {
1337                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1338                 if ( count( $arr ) == 1 ) {
1339                         return $text;
1340                 } else {
1341                         # First, do some preliminary work. This may shift some apostrophes from
1342                         # being mark-up to being text. It also counts the number of occurrences
1343                         # of bold and italics mark-ups.
1344                         $numbold = 0;
1345                         $numitalics = 0;
1346                         for ( $i = 0; $i < count( $arr ); $i++ ) {
1347                                 if ( ( $i % 2 ) == 1 ) {
1348                                         # If there are ever four apostrophes, assume the first is supposed to
1349                                         # be text, and the remaining three constitute mark-up for bold text.
1350                                         if ( strlen( $arr[$i] ) == 4 ) {
1351                                                 $arr[$i-1] .= "'";
1352                                                 $arr[$i] = "'''";
1353                                         } elseif ( strlen( $arr[$i] ) > 5 ) {
1354                                                 # If there are more than 5 apostrophes in a row, assume they're all
1355                                                 # text except for the last 5.
1356                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1357                                                 $arr[$i] = "'''''";
1358                                         }
1359                                         # Count the number of occurrences of bold and italics mark-ups.
1360                                         # We are not counting sequences of five apostrophes.
1361                                         if ( strlen( $arr[$i] ) == 2 ) {
1362                                                 $numitalics++;
1363                                         } elseif ( strlen( $arr[$i] ) == 3 ) {
1364                                                 $numbold++;
1365                                         } elseif ( strlen( $arr[$i] ) == 5 ) {
1366                                                 $numitalics++;
1367                                                 $numbold++;
1368                                         }
1369                                 }
1370                         }
1371
1372                         # If there is an odd number of both bold and italics, it is likely
1373                         # that one of the bold ones was meant to be an apostrophe followed
1374                         # by italics. Which one we cannot know for certain, but it is more
1375                         # likely to be one that has a single-letter word before it.
1376                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1377                                 $i = 0;
1378                                 $firstsingleletterword = -1;
1379                                 $firstmultiletterword = -1;
1380                                 $firstspace = -1;
1381                                 foreach ( $arr as $r ) {
1382                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1383                                                 $x1 = substr( $arr[$i-1], -1 );
1384                                                 $x2 = substr( $arr[$i-1], -2, 1 );
1385                                                 if ( $x1 === ' ' ) {
1386                                                         if ( $firstspace == -1 ) {
1387                                                                 $firstspace = $i;
1388                                                         }
1389                                                 } elseif ( $x2 === ' ') {
1390                                                         if ( $firstsingleletterword == -1 ) {
1391                                                                 $firstsingleletterword = $i;
1392                                                         }
1393                                                 } else {
1394                                                         if ( $firstmultiletterword == -1 ) {
1395                                                                 $firstmultiletterword = $i;
1396                                                         }
1397                                                 }
1398                                         }
1399                                         $i++;
1400                                 }
1401
1402                                 # If there is a single-letter word, use it!
1403                                 if ( $firstsingleletterword > -1 ) {
1404                                         $arr[$firstsingleletterword] = "''";
1405                                         $arr[$firstsingleletterword-1] .= "'";
1406                                 } elseif ( $firstmultiletterword > -1 ) {
1407                                         # If not, but there's a multi-letter word, use that one.
1408                                         $arr[$firstmultiletterword] = "''";
1409                                         $arr[$firstmultiletterword-1] .= "'";
1410                                 } elseif ( $firstspace > -1 ) {
1411                                         # ... otherwise use the first one that has neither.
1412                                         # (notice that it is possible for all three to be -1 if, for example,
1413                                         # there is only one pentuple-apostrophe in the line)
1414                                         $arr[$firstspace] = "''";
1415                                         $arr[$firstspace-1] .= "'";
1416                                 }
1417                         }
1418
1419                         # Now let's actually convert our apostrophic mush to HTML!
1420                         $output = '';
1421                         $buffer = '';
1422                         $state = '';
1423                         $i = 0;
1424                         foreach ( $arr as $r ) {
1425                                 if ( ( $i % 2 ) == 0 ) {
1426                                         if ( $state === 'both' ) {
1427                                                 $buffer .= $r;
1428                                         } else {
1429                                                 $output .= $r;
1430                                         }
1431                                 } else {
1432                                         if ( strlen( $r ) == 2 ) {
1433                                                 if ( $state === 'i' ) {
1434                                                         $output .= '</i>'; $state = '';
1435                                                 } elseif ( $state === 'bi' ) {
1436                                                         $output .= '</i>'; $state = 'b';
1437                                                 } elseif ( $state === 'ib' ) {
1438                                                         $output .= '</b></i><b>'; $state = 'b';
1439                                                 } elseif ( $state === 'both' ) {
1440                                                         $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1441                                                 } else { # $state can be 'b' or ''
1442                                                         $output .= '<i>'; $state .= 'i';
1443                                                 }
1444                                         } elseif ( strlen( $r ) == 3 ) {
1445                                                 if ( $state === 'b' ) {
1446                                                         $output .= '</b>'; $state = '';
1447                                                 } elseif ( $state === 'bi' ) {
1448                                                         $output .= '</i></b><i>'; $state = 'i';
1449                                                 } elseif ( $state === 'ib' ) {
1450                                                         $output .= '</b>'; $state = 'i';
1451                                                 } elseif ( $state === 'both' ) {
1452                                                         $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1453                                                 } else { # $state can be 'i' or ''
1454                                                         $output .= '<b>'; $state .= 'b';
1455                                                 }
1456                                         } elseif ( strlen( $r ) == 5 ) {
1457                                                 if ( $state === 'b' ) {
1458                                                         $output .= '</b><i>'; $state = 'i';
1459                                                 } elseif ( $state === 'i' ) {
1460                                                         $output .= '</i><b>'; $state = 'b';
1461                                                 } elseif ( $state === 'bi' ) {
1462                                                         $output .= '</i></b>'; $state = '';
1463                                                 } elseif ( $state === 'ib' ) {
1464                                                         $output .= '</b></i>'; $state = '';
1465                                                 } elseif ( $state === 'both' ) {
1466                                                         $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1467                                                 } else { # ($state == '')
1468                                                         $buffer = ''; $state = 'both';
1469                                                 }
1470                                         }
1471                                 }
1472                                 $i++;
1473                         }
1474                         # Now close all remaining tags.  Notice that the order is important.
1475                         if ( $state === 'b' || $state === 'ib' ) {
1476                                 $output .= '</b>';
1477                         }
1478                         if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1479                                 $output .= '</i>';
1480                         }
1481                         if ( $state === 'bi' ) {
1482                                 $output .= '</b>';
1483                         }
1484                         # There might be lonely ''''', so make sure we have a buffer
1485                         if ( $state === 'both' && $buffer ) {
1486                                 $output .= '<b><i>'.$buffer.'</i></b>';
1487                         }
1488                         return $output;
1489                 }
1490         }
1491
1492         /**
1493          * Replace external links (REL)
1494          *
1495          * Note: this is all very hackish and the order of execution matters a lot.
1496          * Make sure to run maintenance/parserTests.php if you change this code.
1497          *
1498          * @private
1499          */
1500         function replaceExternalLinks( $text ) {
1501                 global $wgContLang;
1502                 wfProfileIn( __METHOD__ );
1503
1504                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1505                 $s = array_shift( $bits );
1506
1507                 $i = 0;
1508                 while ( $i<count( $bits ) ) {
1509                         $url = $bits[$i++];
1510                         $protocol = $bits[$i++];
1511                         $text = $bits[$i++];
1512                         $trail = $bits[$i++];
1513
1514                         # The characters '<' and '>' (which were escaped by
1515                         # removeHTMLtags()) should not be included in
1516                         # URLs, per RFC 2396.
1517                         $m2 = array();
1518                         if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1519                                 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1520                                 $url = substr( $url, 0, $m2[0][1] );
1521                         }
1522
1523                         # If the link text is an image URL, replace it with an <img> tag
1524                         # This happened by accident in the original parser, but some people used it extensively
1525                         $img = $this->maybeMakeExternalImage( $text );
1526                         if ( $img !== false ) {
1527                                 $text = $img;
1528                         }
1529
1530                         $dtrail = '';
1531
1532                         # Set linktype for CSS - if URL==text, link is essentially free
1533                         $linktype = ( $text === $url ) ? 'free' : 'text';
1534
1535                         # No link text, e.g. [http://domain.tld/some.link]
1536                         if ( $text == '' ) {
1537                                 # Autonumber if allowed. See bug #5918
1538                                 if ( strpos( wfUrlProtocols(), substr( $protocol, 0, strpos( $protocol, ':' ) ) ) !== false ) {
1539                                         $langObj = $this->getFunctionLang();
1540                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1541                                         $linktype = 'autonumber';
1542                                 } else {
1543                                         # Otherwise just use the URL
1544                                         $text = htmlspecialchars( $url );
1545                                         $linktype = 'free';
1546                                 }
1547                         } else {
1548                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1549                                 # Check for trail
1550                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1551                         }
1552
1553                         $text = $wgContLang->markNoConversion( $text );
1554
1555                         $url = Sanitizer::cleanUrl( $url );
1556
1557                         # Use the encoded URL
1558                         # This means that users can paste URLs directly into the text
1559                         # Funny characters like ö aren't valid in URLs anyway
1560                         # This was changed in August 2004
1561                         $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1562                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1563
1564                         # Register link in the output object.
1565                         # Replace unnecessary URL escape codes with the referenced character
1566                         # This prevents spammers from hiding links from the filters
1567                         $pasteurized = self::replaceUnusualEscapes( $url );
1568                         $this->mOutput->addExternalLink( $pasteurized );
1569                 }
1570
1571                 wfProfileOut( __METHOD__ );
1572                 return $s;
1573         }
1574
1575         /**
1576          * Get an associative array of additional HTML attributes appropriate for a
1577          * particular external link.  This currently may include rel => nofollow
1578          * (depending on configuration, namespace, and the URL's domain) and/or a
1579          * target attribute (depending on configuration).
1580          *
1581          * @param $url String: optional URL, to extract the domain from for rel =>
1582          *   nofollow if appropriate
1583          * @return Array: associative array of HTML attributes
1584          */
1585         function getExternalLinkAttribs( $url = false ) {
1586                 $attribs = array();
1587                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1588                 $ns = $this->mTitle->getNamespace();
1589                 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) ) {
1590                         $attribs['rel'] = 'nofollow';
1591
1592                         global $wgNoFollowDomainExceptions;
1593                         if ( $wgNoFollowDomainExceptions ) {
1594                                 $bits = wfParseUrl( $url );
1595                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1596                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1597                                                 if ( substr( $bits['host'], -strlen( $domain ) ) == $domain ) {
1598                                                         unset( $attribs['rel'] );
1599                                                         break;
1600                                                 }
1601                                         }
1602                                 }
1603                         }
1604                 }
1605                 if ( $this->mOptions->getExternalLinkTarget() ) {
1606                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1607                 }
1608                 return $attribs;
1609         }
1610
1611         /**
1612          * Replace unusual URL escape codes with their equivalent characters
1613          *
1614          * @param $url String
1615          * @return String
1616          *
1617          * @todo  This can merge genuinely required bits in the path or query string,
1618          *        breaking legit URLs. A proper fix would treat the various parts of
1619          *        the URL differently; as a workaround, just use the output for
1620          *        statistical records, not for actual linking/output.
1621          */
1622         static function replaceUnusualEscapes( $url ) {
1623                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1624                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1625         }
1626
1627         /**
1628          * Callback function used in replaceUnusualEscapes().
1629          * Replaces unusual URL escape codes with their equivalent character
1630          */
1631         private static function replaceUnusualEscapesCallback( $matches ) {
1632                 $char = urldecode( $matches[0] );
1633                 $ord = ord( $char );
1634                 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1635                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1636                         # No, shouldn't be escaped
1637                         return $char;
1638                 } else {
1639                         # Yes, leave it escaped
1640                         return $matches[0];
1641                 }
1642         }
1643
1644         /**
1645          * make an image if it's allowed, either through the global
1646          * option, through the exception, or through the on-wiki whitelist
1647          * @private
1648          */
1649         function maybeMakeExternalImage( $url ) {
1650                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1651                 $imagesexception = !empty( $imagesfrom );
1652                 $text = false;
1653                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1654                 if ( $imagesexception && is_array( $imagesfrom ) ) {
1655                         $imagematch = false;
1656                         foreach ( $imagesfrom as $match ) {
1657                                 if ( strpos( $url, $match ) === 0 ) {
1658                                         $imagematch = true;
1659                                         break;
1660                                 }
1661                         }
1662                 } elseif ( $imagesexception ) {
1663                         $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1664                 } else {
1665                         $imagematch = false;
1666                 }
1667                 if ( $this->mOptions->getAllowExternalImages()
1668                          || ( $imagesexception && $imagematch ) ) {
1669                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1670                                 # Image found
1671                                 $text = Linker::makeExternalImage( $url );
1672                         }
1673                 }
1674                 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1675                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1676                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1677                         foreach ( $whitelist as $entry ) {
1678                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1679                                 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1680                                         continue;
1681                                 }
1682                                 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1683                                         # Image matches a whitelist entry
1684                                         $text = Linker::makeExternalImage( $url );
1685                                         break;
1686                                 }
1687                         }
1688                 }
1689                 return $text;
1690         }
1691
1692         /**
1693          * Process [[ ]] wikilinks
1694          * @return String: processed text
1695          *
1696          * @private
1697          */
1698         function replaceInternalLinks( $s ) {
1699                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1700                 return $s;
1701         }
1702
1703         /**
1704          * Process [[ ]] wikilinks (RIL)
1705          * @return LinkHolderArray
1706          *
1707          * @private
1708          */
1709         function replaceInternalLinks2( &$s ) {
1710                 global $wgContLang;
1711
1712                 wfProfileIn( __METHOD__ );
1713
1714                 wfProfileIn( __METHOD__.'-setup' );
1715                 static $tc = FALSE, $e1, $e1_img;
1716                 # the % is needed to support urlencoded titles as well
1717                 if ( !$tc ) {
1718                         $tc = Title::legalChars() . '#%';
1719                         # Match a link having the form [[namespace:link|alternate]]trail
1720                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1721                         # Match cases where there is no "]]", which might still be images
1722                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1723                 }
1724
1725                 $holders = new LinkHolderArray( $this );
1726
1727                 # split the entire text string on occurences of [[
1728                 $a = StringUtils::explode( '[[', ' ' . $s );
1729                 # get the first element (all text up to first [[), and remove the space we added
1730                 $s = $a->current();
1731                 $a->next();
1732                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1733                 $s = substr( $s, 1 );
1734
1735                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1736                 $e2 = null;
1737                 if ( $useLinkPrefixExtension ) {
1738                         # Match the end of a line for a word that's not followed by whitespace,
1739                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1740                         $e2 = wfMsgForContent( 'linkprefix' );
1741                 }
1742
1743                 if ( is_null( $this->mTitle ) ) {
1744                         wfProfileOut( __METHOD__.'-setup' );
1745                         wfProfileOut( __METHOD__ );
1746                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1747                 }
1748                 $nottalk = !$this->mTitle->isTalkPage();
1749
1750                 if ( $useLinkPrefixExtension ) {
1751                         $m = array();
1752                         if ( preg_match( $e2, $s, $m ) ) {
1753                                 $first_prefix = $m[2];
1754                         } else {
1755                                 $first_prefix = false;
1756                         }
1757                 } else {
1758                         $prefix = '';
1759                 }
1760
1761                 if ( $wgContLang->hasVariants() ) {
1762                         $selflink = $wgContLang->autoConvertToAllVariants( $this->mTitle->getPrefixedText() );
1763                 } else {
1764                         $selflink = array( $this->mTitle->getPrefixedText() );
1765                 }
1766                 $useSubpages = $this->areSubpagesAllowed();
1767                 wfProfileOut( __METHOD__.'-setup' );
1768
1769                 # Loop for each link
1770                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1771                         # Check for excessive memory usage
1772                         if ( $holders->isBig() ) {
1773                                 # Too big
1774                                 # Do the existence check, replace the link holders and clear the array
1775                                 $holders->replace( $s );
1776                                 $holders->clear();
1777                         }
1778
1779                         if ( $useLinkPrefixExtension ) {
1780                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1781                                 if ( preg_match( $e2, $s, $m ) ) {
1782                                         $prefix = $m[2];
1783                                         $s = $m[1];
1784                                 } else {
1785                                         $prefix='';
1786                                 }
1787                                 # first link
1788                                 if ( $first_prefix ) {
1789                                         $prefix = $first_prefix;
1790                                         $first_prefix = false;
1791                                 }
1792                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1793                         }
1794
1795                         $might_be_img = false;
1796
1797                         wfProfileIn( __METHOD__."-e1" );
1798                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1799                                 $text = $m[2];
1800                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1801                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1802                                 # the real problem is with the $e1 regex
1803                                 # See bug 1300.
1804                                 #
1805                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1806                                 # and no image is in sight. See bug 2095.
1807                                 #
1808                                 if ( $text !== '' &&
1809                                         substr( $m[3], 0, 1 ) === ']' &&
1810                                         strpos( $text, '[' ) !== false
1811                                 )
1812                                 {
1813                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1814                                         $m[3] = substr( $m[3], 1 );
1815                                 }
1816                                 # fix up urlencoded title texts
1817                                 if ( strpos( $m[1], '%' ) !== false ) {
1818                                         # Should anchors '#' also be rejected?
1819                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), rawurldecode( $m[1] ) );
1820                                 }
1821                                 $trail = $m[3];
1822                         } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1823                                 $might_be_img = true;
1824                                 $text = $m[2];
1825                                 if ( strpos( $m[1], '%' ) !== false ) {
1826                                         $m[1] = rawurldecode( $m[1] );
1827                                 }
1828                                 $trail = "";
1829                         } else { # Invalid form; output directly
1830                                 $s .= $prefix . '[[' . $line ;
1831                                 wfProfileOut( __METHOD__."-e1" );
1832                                 continue;
1833                         }
1834                         wfProfileOut( __METHOD__."-e1" );
1835                         wfProfileIn( __METHOD__."-misc" );
1836
1837                         # Don't allow internal links to pages containing
1838                         # PROTO: where PROTO is a valid URL protocol; these
1839                         # should be external links.
1840                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1841                                 $s .= $prefix . '[[' . $line ;
1842                                 wfProfileOut( __METHOD__."-misc" );
1843                                 continue;
1844                         }
1845
1846                         # Make subpage if necessary
1847                         if ( $useSubpages ) {
1848                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1849                         } else {
1850                                 $link = $m[1];
1851                         }
1852
1853                         $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1854                         if ( !$noforce ) {
1855                                 # Strip off leading ':'
1856                                 $link = substr( $link, 1 );
1857                         }
1858
1859                         wfProfileOut( __METHOD__."-misc" );
1860                         wfProfileIn( __METHOD__."-title" );
1861                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1862                         if ( $nt === null ) {
1863                                 $s .= $prefix . '[[' . $line;
1864                                 wfProfileOut( __METHOD__."-title" );
1865                                 continue;
1866                         }
1867
1868                         $ns = $nt->getNamespace();
1869                         $iw = $nt->getInterWiki();
1870                         wfProfileOut( __METHOD__."-title" );
1871
1872                         if ( $might_be_img ) { # if this is actually an invalid link
1873                                 wfProfileIn( __METHOD__."-might_be_img" );
1874                                 if ( $ns == NS_FILE && $noforce ) { # but might be an image
1875                                         $found = false;
1876                                         while ( true ) {
1877                                                 # look at the next 'line' to see if we can close it there
1878                                                 $a->next();
1879                                                 $next_line = $a->current();
1880                                                 if ( $next_line === false || $next_line === null ) {
1881                                                         break;
1882                                                 }
1883                                                 $m = explode( ']]', $next_line, 3 );
1884                                                 if ( count( $m ) == 3 ) {
1885                                                         # the first ]] closes the inner link, the second the image
1886                                                         $found = true;
1887                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1888                                                         $trail = $m[2];
1889                                                         break;
1890                                                 } elseif ( count( $m ) == 2 ) {
1891                                                         # if there's exactly one ]] that's fine, we'll keep looking
1892                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1893                                                 } else {
1894                                                         # if $next_line is invalid too, we need look no further
1895                                                         $text .= '[[' . $next_line;
1896                                                         break;
1897                                                 }
1898                                         }
1899                                         if ( !$found ) {
1900                                                 # we couldn't find the end of this imageLink, so output it raw
1901                                                 # but don't ignore what might be perfectly normal links in the text we've examined
1902                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1903                                                 $s .= "{$prefix}[[$link|$text";
1904                                                 # note: no $trail, because without an end, there *is* no trail
1905                                                 wfProfileOut( __METHOD__."-might_be_img" );
1906                                                 continue;
1907                                         }
1908                                 } else { # it's not an image, so output it raw
1909                                         $s .= "{$prefix}[[$link|$text";
1910                                         # note: no $trail, because without an end, there *is* no trail
1911                                         wfProfileOut( __METHOD__."-might_be_img" );
1912                                         continue;
1913                                 }
1914                                 wfProfileOut( __METHOD__."-might_be_img" );
1915                         }
1916
1917                         $wasblank = ( $text  == '' );
1918                         if ( $wasblank ) {
1919                                 $text = $link;
1920                         } else {
1921                                 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
1922                                 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
1923                                 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
1924                                 #    -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
1925                                 $text = $this->doQuotes( $text );
1926                         }
1927
1928                         # Link not escaped by : , create the various objects
1929                         if ( $noforce ) {
1930
1931                                 # Interwikis
1932                                 wfProfileIn( __METHOD__."-interwiki" );
1933                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1934                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1935                                         $s = rtrim( $s . $prefix );
1936                                         $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
1937                                         wfProfileOut( __METHOD__."-interwiki" );
1938                                         continue;
1939                                 }
1940                                 wfProfileOut( __METHOD__."-interwiki" );
1941
1942                                 if ( $ns == NS_FILE ) {
1943                                         wfProfileIn( __METHOD__."-image" );
1944                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1945                                                 if ( $wasblank ) {
1946                                                         # if no parameters were passed, $text
1947                                                         # becomes something like "File:Foo.png",
1948                                                         # which we don't want to pass on to the
1949                                                         # image generator
1950                                                         $text = '';
1951                                                 } else {
1952                                                         # recursively parse links inside the image caption
1953                                                         # actually, this will parse them in any other parameters, too,
1954                                                         # but it might be hard to fix that, and it doesn't matter ATM
1955                                                         $text = $this->replaceExternalLinks( $text );
1956                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1957                                                 }
1958                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1959                                                 $s .= $prefix . $this->armorLinks(
1960                                                         $this->makeImage( $nt, $text, $holders ) ) . $trail;
1961                                         } else {
1962                                                 $s .= $prefix . $trail;
1963                                         }
1964                                         wfProfileOut( __METHOD__."-image" );
1965                                         continue;
1966                                 }
1967
1968                                 if ( $ns == NS_CATEGORY ) {
1969                                         wfProfileIn( __METHOD__."-category" );
1970                                         $s = rtrim( $s . "\n" ); # bug 87
1971
1972                                         if ( $wasblank ) {
1973                                                 $sortkey = $this->getDefaultSort();
1974                                         } else {
1975                                                 $sortkey = $text;
1976                                         }
1977                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1978                                         $sortkey = str_replace( "\n", '', $sortkey );
1979                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1980                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1981
1982                                         /**
1983                                          * Strip the whitespace Category links produce, see bug 87
1984                                          * @todo We might want to use trim($tmp, "\n") here.
1985                                          */
1986                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
1987
1988                                         wfProfileOut( __METHOD__."-category" );
1989                                         continue;
1990                                 }
1991                         }
1992
1993                         # Self-link checking
1994                         if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1995                                 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1996                                         $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
1997                                         continue;
1998                                 }
1999                         }
2000
2001                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
2002                         # FIXME: Should do batch file existence checks, see comment below
2003                         if ( $ns == NS_MEDIA ) {
2004                                 wfProfileIn( __METHOD__."-media" );
2005                                 # Give extensions a chance to select the file revision for us
2006                                 $time = $sha1 = $descQuery = false;
2007                                 wfRunHooks( 'BeforeParserFetchFileAndTitle',
2008                                         array( $this, $nt, &$time, &$sha1, &$descQuery ) );
2009                                 # Fetch and register the file (file title may be different via hooks)
2010                                 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $time, $sha1 );
2011                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2012                                 $s .= $prefix . $this->armorLinks(
2013                                         Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2014                                 wfProfileOut( __METHOD__."-media" );
2015                                 continue;
2016                         }
2017
2018                         wfProfileIn( __METHOD__."-always_known" );
2019                         # Some titles, such as valid special pages or files in foreign repos, should
2020                         # be shown as bluelinks even though they're not included in the page table
2021                         #
2022                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2023                         # batch file existence checks for NS_FILE and NS_MEDIA
2024                         if ( $iw == '' && $nt->isAlwaysKnown() ) {
2025                                 $this->mOutput->addLink( $nt );
2026                                 $s .= $this->makeKnownLinkHolder( $nt, $text, array(), $trail, $prefix );
2027                         } else {
2028                                 # Links will be added to the output link list after checking
2029                                 $s .= $holders->makeHolder( $nt, $text, array(), $trail, $prefix );
2030                         }
2031                         wfProfileOut( __METHOD__."-always_known" );
2032                 }
2033                 wfProfileOut( __METHOD__ );
2034                 return $holders;
2035         }
2036
2037         /**
2038          * Make a link placeholder. The text returned can be later resolved to a real link with
2039          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
2040          * parsing of interwiki links, and secondly to allow all existence checks and
2041          * article length checks (for stub links) to be bundled into a single query.
2042          *
2043          * @deprecated
2044          */
2045         function makeLinkHolder( &$nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
2046                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
2047         }
2048
2049         /**
2050          * Render a forced-blue link inline; protect against double expansion of
2051          * URLs if we're in a mode that prepends full URL prefixes to internal links.
2052          * Since this little disaster has to split off the trail text to avoid
2053          * breaking URLs in the following text without breaking trails on the
2054          * wiki links, it's been made into a horrible function.
2055          *
2056          * @param $nt Title
2057          * @param $text String
2058          * @param $query Array or String
2059          * @param $trail String
2060          * @param $prefix String
2061          * @return String: HTML-wikitext mix oh yuck
2062          */
2063         function makeKnownLinkHolder( $nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
2064                 list( $inside, $trail ) = Linker::splitTrail( $trail );
2065
2066                 if ( is_string( $query ) ) {
2067                         $query = wfCgiToArray( $query );
2068                 }
2069                 if ( $text == '' ) {
2070                         $text = htmlspecialchars( $nt->getPrefixedText() );
2071                 }
2072
2073                 $link = Linker::linkKnown( $nt, "$prefix$text$inside", array(), $query );
2074
2075                 return $this->armorLinks( $link ) . $trail;
2076         }
2077
2078         /**
2079          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2080          * going to go through further parsing steps before inline URL expansion.
2081          *
2082          * Not needed quite as much as it used to be since free links are a bit
2083          * more sensible these days. But bracketed links are still an issue.
2084          *
2085          * @param $text String: more-or-less HTML
2086          * @return String: less-or-more HTML with NOPARSE bits
2087          */
2088         function armorLinks( $text ) {
2089                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
2090                         "{$this->mUniqPrefix}NOPARSE$1", $text );
2091         }
2092
2093         /**
2094          * Return true if subpage links should be expanded on this page.
2095          * @return Boolean
2096          */
2097         function areSubpagesAllowed() {
2098                 # Some namespaces don't allow subpages
2099                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2100         }
2101
2102         /**
2103          * Handle link to subpage if necessary
2104          *
2105          * @param $target String: the source of the link
2106          * @param &$text String: the link text, modified as necessary
2107          * @return string the full name of the link
2108          * @private
2109          */
2110         function maybeDoSubpageLink( $target, &$text ) {
2111                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2112         }
2113
2114         /**#@+
2115          * Used by doBlockLevels()
2116          * @private
2117          *
2118          * @return string
2119          */
2120         function closeParagraph() {
2121                 $result = '';
2122                 if ( $this->mLastSection != '' ) {
2123                         $result = '</' . $this->mLastSection  . ">\n";
2124                 }
2125                 $this->mInPre = false;
2126                 $this->mLastSection = '';
2127                 return $result;
2128         }
2129
2130         /**
2131          * getCommon() returns the length of the longest common substring
2132          * of both arguments, starting at the beginning of both.
2133          * @private
2134          */
2135         function getCommon( $st1, $st2 ) {
2136                 $fl = strlen( $st1 );
2137                 $shorter = strlen( $st2 );
2138                 if ( $fl < $shorter ) {
2139                         $shorter = $fl;
2140                 }
2141
2142                 for ( $i = 0; $i < $shorter; ++$i ) {
2143                         if ( $st1[$i] != $st2[$i] ) {
2144                                 break;
2145                         }
2146                 }
2147                 return $i;
2148         }
2149
2150         /**
2151          * These next three functions open, continue, and close the list
2152          * element appropriate to the prefix character passed into them.
2153          * @private
2154          *
2155          * @return string
2156          */
2157         function openList( $char ) {
2158                 $result = $this->closeParagraph();
2159
2160                 if ( '*' === $char ) {
2161                         $result .= '<ul><li>';
2162                 } elseif ( '#' === $char ) {
2163                         $result .= '<ol><li>';
2164                 } elseif ( ':' === $char ) {
2165                         $result .= '<dl><dd>';
2166                 } elseif ( ';' === $char ) {
2167                         $result .= '<dl><dt>';
2168                         $this->mDTopen = true;
2169                 } else {
2170                         $result = '<!-- ERR 1 -->';
2171                 }
2172
2173                 return $result;
2174         }
2175
2176         /**
2177          * TODO: document
2178          * @param $char String
2179          * @private
2180          *
2181          * @return string
2182          */
2183         function nextItem( $char ) {
2184                 if ( '*' === $char || '#' === $char ) {
2185                         return '</li><li>';
2186                 } elseif ( ':' === $char || ';' === $char ) {
2187                         $close = '</dd>';
2188                         if ( $this->mDTopen ) {
2189                                 $close = '</dt>';
2190                         }
2191                         if ( ';' === $char ) {
2192                                 $this->mDTopen = true;
2193                                 return $close . '<dt>';
2194                         } else {
2195                                 $this->mDTopen = false;
2196                                 return $close . '<dd>';
2197                         }
2198                 }
2199                 return '<!-- ERR 2 -->';
2200         }
2201
2202         /**
2203          * TODO: document
2204          * @param $char String
2205          * @private
2206          *
2207          * @return string
2208          */
2209         function closeList( $char ) {
2210                 if ( '*' === $char ) {
2211                         $text = '</li></ul>';
2212                 } elseif ( '#' === $char ) {
2213                         $text = '</li></ol>';
2214                 } elseif ( ':' === $char ) {
2215                         if ( $this->mDTopen ) {
2216                                 $this->mDTopen = false;
2217                                 $text = '</dt></dl>';
2218                         } else {
2219                                 $text = '</dd></dl>';
2220                         }
2221                 } else {
2222                         return '<!-- ERR 3 -->';
2223                 }
2224                 return $text."\n";
2225         }
2226         /**#@-*/
2227
2228         /**
2229          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2230          *
2231          * @param $text String
2232          * @param $linestart Boolean: whether or not this is at the start of a line.
2233          * @private
2234          * @return string the lists rendered as HTML
2235          */
2236         function doBlockLevels( $text, $linestart ) {
2237                 wfProfileIn( __METHOD__ );
2238
2239                 # Parsing through the text line by line.  The main thing
2240                 # happening here is handling of block-level elements p, pre,
2241                 # and making lists from lines starting with * # : etc.
2242                 #
2243                 $textLines = StringUtils::explode( "\n", $text );
2244
2245                 $lastPrefix = $output = '';
2246                 $this->mDTopen = $inBlockElem = false;
2247                 $prefixLength = 0;
2248                 $paragraphStack = false;
2249
2250                 foreach ( $textLines as $oLine ) {
2251                         # Fix up $linestart
2252                         if ( !$linestart ) {
2253                                 $output .= $oLine;
2254                                 $linestart = true;
2255                                 continue;
2256                         }
2257                         # * = ul
2258                         # # = ol
2259                         # ; = dt
2260                         # : = dd
2261
2262                         $lastPrefixLength = strlen( $lastPrefix );
2263                         $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2264                         $preOpenMatch = preg_match( '/<pre/i', $oLine );
2265                         # If not in a <pre> element, scan for and figure out what prefixes are there.
2266                         if ( !$this->mInPre ) {
2267                                 # Multiple prefixes may abut each other for nested lists.
2268                                 $prefixLength = strspn( $oLine, '*#:;' );
2269                                 $prefix = substr( $oLine, 0, $prefixLength );
2270
2271                                 # eh?
2272                                 # ; and : are both from definition-lists, so they're equivalent
2273                                 #  for the purposes of determining whether or not we need to open/close
2274                                 #  elements.
2275                                 $prefix2 = str_replace( ';', ':', $prefix );
2276                                 $t = substr( $oLine, $prefixLength );
2277                                 $this->mInPre = (bool)$preOpenMatch;
2278                         } else {
2279                                 # Don't interpret any other prefixes in preformatted text
2280                                 $prefixLength = 0;
2281                                 $prefix = $prefix2 = '';
2282                                 $t = $oLine;
2283                         }
2284
2285                         # List generation
2286                         if ( $prefixLength && $lastPrefix === $prefix2 ) {
2287                                 # Same as the last item, so no need to deal with nesting or opening stuff
2288                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2289                                 $paragraphStack = false;
2290
2291                                 if ( substr( $prefix, -1 ) === ';') {
2292                                         # The one nasty exception: definition lists work like this:
2293                                         # ; title : definition text
2294                                         # So we check for : in the remainder text to split up the
2295                                         # title and definition, without b0rking links.
2296                                         $term = $t2 = '';
2297                                         if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2298                                                 $t = $t2;
2299                                                 $output .= $term . $this->nextItem( ':' );
2300                                         }
2301                                 }
2302                         } elseif ( $prefixLength || $lastPrefixLength ) {
2303                                 # We need to open or close prefixes, or both.
2304
2305                                 # Either open or close a level...
2306                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2307                                 $paragraphStack = false;
2308
2309                                 # Close all the prefixes which aren't shared.
2310                                 while ( $commonPrefixLength < $lastPrefixLength ) {
2311                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2312                                         --$lastPrefixLength;
2313                                 }
2314
2315                                 # Continue the current prefix if appropriate.
2316                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2317                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2318                                 }
2319
2320                                 # Open prefixes where appropriate.
2321                                 while ( $prefixLength > $commonPrefixLength ) {
2322                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2323                                         $output .= $this->openList( $char );
2324
2325                                         if ( ';' === $char ) {
2326                                                 # FIXME: This is dupe of code above
2327                                                 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2328                                                         $t = $t2;
2329                                                         $output .= $term . $this->nextItem( ':' );
2330                                                 }
2331                                         }
2332                                         ++$commonPrefixLength;
2333                                 }
2334                                 $lastPrefix = $prefix2;
2335                         }
2336
2337                         # If we have no prefixes, go to paragraph mode.
2338                         if ( 0 == $prefixLength ) {
2339                                 wfProfileIn( __METHOD__."-paragraph" );
2340                                 # No prefix (not in list)--go to paragraph mode
2341                                 # XXX: use a stack for nestable elements like span, table and div
2342                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2343                                 $closematch = preg_match(
2344                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2345                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2346                                 if ( $openmatch or $closematch ) {
2347                                         $paragraphStack = false;
2348                                         # TODO bug 5718: paragraph closed
2349                                         $output .= $this->closeParagraph();
2350                                         if ( $preOpenMatch and !$preCloseMatch ) {
2351                                                 $this->mInPre = true;
2352                                         }
2353                                         $inBlockElem = !$closematch;
2354                                 } elseif ( !$inBlockElem && !$this->mInPre ) {
2355                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) ) {
2356                                                 # pre
2357                                                 if ( $this->mLastSection !== 'pre' ) {
2358                                                         $paragraphStack = false;
2359                                                         $output .= $this->closeParagraph().'<pre>';
2360                                                         $this->mLastSection = 'pre';
2361                                                 }
2362                                                 $t = substr( $t, 1 );
2363                                         } else {
2364                                                 # paragraph
2365                                                 if ( trim( $t ) === '' ) {
2366                                                         if ( $paragraphStack ) {
2367                                                                 $output .= $paragraphStack.'<br />';
2368                                                                 $paragraphStack = false;
2369                                                                 $this->mLastSection = 'p';
2370                                                         } else {
2371                                                                 if ( $this->mLastSection !== 'p' ) {
2372                                                                         $output .= $this->closeParagraph();
2373                                                                         $this->mLastSection = '';
2374                                                                         $paragraphStack = '<p>';
2375                                                                 } else {
2376                                                                         $paragraphStack = '</p><p>';
2377                                                                 }
2378                                                         }
2379                                                 } else {
2380                                                         if ( $paragraphStack ) {
2381                                                                 $output .= $paragraphStack;
2382                                                                 $paragraphStack = false;
2383                                                                 $this->mLastSection = 'p';
2384                                                         } elseif ( $this->mLastSection !== 'p' ) {
2385                                                                 $output .= $this->closeParagraph().'<p>';
2386                                                                 $this->mLastSection = 'p';
2387                                                         }
2388                                                 }
2389                                         }
2390                                 }
2391                                 wfProfileOut( __METHOD__."-paragraph" );
2392                         }
2393                         # somewhere above we forget to get out of pre block (bug 785)
2394                         if ( $preCloseMatch && $this->mInPre ) {
2395                                 $this->mInPre = false;
2396                         }
2397                         if ( $paragraphStack === false ) {
2398                                 $output .= $t."\n";
2399                         }
2400                 }
2401                 while ( $prefixLength ) {
2402                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2403                         --$prefixLength;
2404                 }
2405                 if ( $this->mLastSection != '' ) {
2406                         $output .= '</' . $this->mLastSection . '>';
2407                         $this->mLastSection = '';
2408                 }
2409
2410                 wfProfileOut( __METHOD__ );
2411                 return $output;
2412         }
2413
2414         /**
2415          * Split up a string on ':', ignoring any occurences inside tags
2416          * to prevent illegal overlapping.
2417          *
2418          * @param $str String: the string to split
2419          * @param &$before String: set to everything before the ':'
2420          * @param &$after String: set to everything after the ':'
2421          * return String: the position of the ':', or false if none found
2422          */
2423         function findColonNoLinks( $str, &$before, &$after ) {
2424                 wfProfileIn( __METHOD__ );
2425
2426                 $pos = strpos( $str, ':' );
2427                 if ( $pos === false ) {
2428                         # Nothing to find!
2429                         wfProfileOut( __METHOD__ );
2430                         return false;
2431                 }
2432
2433                 $lt = strpos( $str, '<' );
2434                 if ( $lt === false || $lt > $pos ) {
2435                         # Easy; no tag nesting to worry about
2436                         $before = substr( $str, 0, $pos );
2437                         $after = substr( $str, $pos+1 );
2438                         wfProfileOut( __METHOD__ );
2439                         return $pos;
2440                 }
2441
2442                 # Ugly state machine to walk through avoiding tags.
2443                 $state = self::COLON_STATE_TEXT;
2444                 $stack = 0;
2445                 $len = strlen( $str );
2446                 for( $i = 0; $i < $len; $i++ ) {
2447                         $c = $str[$i];
2448
2449                         switch( $state ) {
2450                         # (Using the number is a performance hack for common cases)
2451                         case 0: # self::COLON_STATE_TEXT:
2452                                 switch( $c ) {
2453                                 case "<":
2454                                         # Could be either a <start> tag or an </end> tag
2455                                         $state = self::COLON_STATE_TAGSTART;
2456                                         break;
2457                                 case ":":
2458                                         if ( $stack == 0 ) {
2459                                                 # We found it!
2460                                                 $before = substr( $str, 0, $i );
2461                                                 $after = substr( $str, $i + 1 );
2462                                                 wfProfileOut( __METHOD__ );
2463                                                 return $i;
2464                                         }
2465                                         # Embedded in a tag; don't break it.
2466                                         break;
2467                                 default:
2468                                         # Skip ahead looking for something interesting
2469                                         $colon = strpos( $str, ':', $i );
2470                                         if ( $colon === false ) {
2471                                                 # Nothing else interesting
2472                                                 wfProfileOut( __METHOD__ );
2473                                                 return false;
2474                                         }
2475                                         $lt = strpos( $str, '<', $i );
2476                                         if ( $stack === 0 ) {
2477                                                 if ( $lt === false || $colon < $lt ) {
2478                                                         # We found it!
2479                                                         $before = substr( $str, 0, $colon );
2480                                                         $after = substr( $str, $colon + 1 );
2481                                                         wfProfileOut( __METHOD__ );
2482                                                         return $i;
2483                                                 }
2484                                         }
2485                                         if ( $lt === false ) {
2486                                                 # Nothing else interesting to find; abort!
2487                                                 # We're nested, but there's no close tags left. Abort!
2488                                                 break 2;
2489                                         }
2490                                         # Skip ahead to next tag start
2491                                         $i = $lt;
2492                                         $state = self::COLON_STATE_TAGSTART;
2493                                 }
2494                                 break;
2495                         case 1: # self::COLON_STATE_TAG:
2496                                 # In a <tag>
2497                                 switch( $c ) {
2498                                 case ">":
2499                                         $stack++;
2500                                         $state = self::COLON_STATE_TEXT;
2501                                         break;
2502                                 case "/":
2503                                         # Slash may be followed by >?
2504                                         $state = self::COLON_STATE_TAGSLASH;
2505                                         break;
2506                                 default:
2507                                         # ignore
2508                                 }
2509                                 break;
2510                         case 2: # self::COLON_STATE_TAGSTART:
2511                                 switch( $c ) {
2512                                 case "/":
2513                                         $state = self::COLON_STATE_CLOSETAG;
2514                                         break;
2515                                 case "!":
2516                                         $state = self::COLON_STATE_COMMENT;
2517                                         break;
2518                                 case ">":
2519                                         # Illegal early close? This shouldn't happen D:
2520                                         $state = self::COLON_STATE_TEXT;
2521                                         break;
2522                                 default:
2523                                         $state = self::COLON_STATE_TAG;
2524                                 }
2525                                 break;
2526                         case 3: # self::COLON_STATE_CLOSETAG:
2527                                 # In a </tag>
2528                                 if ( $c === ">" ) {
2529                                         $stack--;
2530                                         if ( $stack < 0 ) {
2531                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2532                                                 wfProfileOut( __METHOD__ );
2533                                                 return false;
2534                                         }
2535                                         $state = self::COLON_STATE_TEXT;
2536                                 }
2537                                 break;
2538                         case self::COLON_STATE_TAGSLASH:
2539                                 if ( $c === ">" ) {
2540                                         # Yes, a self-closed tag <blah/>
2541                                         $state = self::COLON_STATE_TEXT;
2542                                 } else {
2543                                         # Probably we're jumping the gun, and this is an attribute
2544                                         $state = self::COLON_STATE_TAG;
2545                                 }
2546                                 break;
2547                         case 5: # self::COLON_STATE_COMMENT:
2548                                 if ( $c === "-" ) {
2549                                         $state = self::COLON_STATE_COMMENTDASH;
2550                                 }
2551                                 break;
2552                         case self::COLON_STATE_COMMENTDASH:
2553                                 if ( $c === "-" ) {
2554                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2555                                 } else {
2556                                         $state = self::COLON_STATE_COMMENT;
2557                                 }
2558                                 break;
2559                         case self::COLON_STATE_COMMENTDASHDASH:
2560                                 if ( $c === ">" ) {
2561                                         $state = self::COLON_STATE_TEXT;
2562                                 } else {
2563                                         $state = self::COLON_STATE_COMMENT;
2564                                 }
2565                                 break;
2566                         default:
2567                                 throw new MWException( "State machine error in " . __METHOD__ );
2568                         }
2569                 }
2570                 if ( $stack > 0 ) {
2571                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2572                         wfProfileOut( __METHOD__ );
2573                         return false;
2574                 }
2575                 wfProfileOut( __METHOD__ );
2576                 return false;
2577         }
2578
2579         /**
2580          * Return value of a magic variable (like PAGENAME)
2581          *
2582          * @private
2583          *
2584          * @param $index integer
2585          * @param $frame PPFrame
2586          */
2587         function getVariableValue( $index, $frame=false ) {
2588                 global $wgContLang, $wgSitename, $wgServer;
2589                 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2590
2591                 /**
2592                  * Some of these require message or data lookups and can be
2593                  * expensive to check many times.
2594                  */
2595                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2596                         if ( isset( $this->mVarCache[$index] ) ) {
2597                                 return $this->mVarCache[$index];
2598                         }
2599                 }
2600
2601                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2602                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2603
2604                 # Use the time zone
2605                 global $wgLocaltimezone;
2606                 if ( isset( $wgLocaltimezone ) ) {
2607                         $oldtz = date_default_timezone_get();
2608                         date_default_timezone_set( $wgLocaltimezone );
2609                 }
2610
2611                 $localTimestamp = date( 'YmdHis', $ts );
2612                 $localMonth = date( 'm', $ts );
2613                 $localMonth1 = date( 'n', $ts );
2614                 $localMonthName = date( 'n', $ts );
2615                 $localDay = date( 'j', $ts );
2616                 $localDay2 = date( 'd', $ts );
2617                 $localDayOfWeek = date( 'w', $ts );
2618                 $localWeek = date( 'W', $ts );
2619                 $localYear = date( 'Y', $ts );
2620                 $localHour = date( 'H', $ts );
2621                 if ( isset( $wgLocaltimezone ) ) {
2622                         date_default_timezone_set( $oldtz );
2623                 }
2624
2625                 switch ( $index ) {
2626                         case 'currentmonth':
2627                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2628                                 break;
2629                         case 'currentmonth1':
2630                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2631                                 break;
2632                         case 'currentmonthname':
2633                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2634                                 break;
2635                         case 'currentmonthnamegen':
2636                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2637                                 break;
2638                         case 'currentmonthabbrev':
2639                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2640                                 break;
2641                         case 'currentday':
2642                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2643                                 break;
2644                         case 'currentday2':
2645                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2646                                 break;
2647                         case 'localmonth':
2648                                 $value = $wgContLang->formatNum( $localMonth );
2649                                 break;
2650                         case 'localmonth1':
2651                                 $value = $wgContLang->formatNum( $localMonth1 );
2652                                 break;
2653                         case 'localmonthname':
2654                                 $value = $wgContLang->getMonthName( $localMonthName );
2655                                 break;
2656                         case 'localmonthnamegen':
2657                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2658                                 break;
2659                         case 'localmonthabbrev':
2660                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2661                                 break;
2662                         case 'localday':
2663                                 $value = $wgContLang->formatNum( $localDay );
2664                                 break;
2665                         case 'localday2':
2666                                 $value = $wgContLang->formatNum( $localDay2 );
2667                                 break;
2668                         case 'pagename':
2669                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2670                                 break;
2671                         case 'pagenamee':
2672                                 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2673                                 break;
2674                         case 'fullpagename':
2675                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2676                                 break;
2677                         case 'fullpagenamee':
2678                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2679                                 break;
2680                         case 'subpagename':
2681                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2682                                 break;
2683                         case 'subpagenamee':
2684                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2685                                 break;
2686                         case 'basepagename':
2687                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2688                                 break;
2689                         case 'basepagenamee':
2690                                 $value = wfEscapeWikiText( wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) ) );
2691                                 break;
2692                         case 'talkpagename':
2693                                 if ( $this->mTitle->canTalk() ) {
2694                                         $talkPage = $this->mTitle->getTalkPage();
2695                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2696                                 } else {
2697                                         $value = '';
2698                                 }
2699                                 break;
2700                         case 'talkpagenamee':
2701                                 if ( $this->mTitle->canTalk() ) {
2702                                         $talkPage = $this->mTitle->getTalkPage();
2703                                         $value = wfEscapeWikiText( $talkPage->getPrefixedUrl() );
2704                                 } else {
2705                                         $value = '';
2706                                 }
2707                                 break;
2708                         case 'subjectpagename':
2709                                 $subjPage = $this->mTitle->getSubjectPage();
2710                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2711                                 break;
2712                         case 'subjectpagenamee':
2713                                 $subjPage = $this->mTitle->getSubjectPage();
2714                                 $value = wfEscapeWikiText( $subjPage->getPrefixedUrl() );
2715                                 break;
2716                         case 'revisionid':
2717                                 # Let the edit saving system know we should parse the page
2718                                 # *after* a revision ID has been assigned.
2719                                 $this->mOutput->setFlag( 'vary-revision' );
2720                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2721                                 $value = $this->mRevisionId;
2722                                 break;
2723                         case 'revisionday':
2724                                 # Let the edit saving system know we should parse the page
2725                                 # *after* a revision ID has been assigned. This is for null edits.
2726                                 $this->mOutput->setFlag( 'vary-revision' );
2727                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2728                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2729                                 break;
2730                         case 'revisionday2':
2731                                 # Let the edit saving system know we should parse the page
2732                                 # *after* a revision ID has been assigned. This is for null edits.
2733                                 $this->mOutput->setFlag( 'vary-revision' );
2734                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2735                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2736                                 break;
2737                         case 'revisionmonth':
2738                                 # Let the edit saving system know we should parse the page
2739                                 # *after* a revision ID has been assigned. This is for null edits.
2740                                 $this->mOutput->setFlag( 'vary-revision' );
2741                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2742                                 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2743                                 break;
2744                         case 'revisionmonth1':
2745                                 # Let the edit saving system know we should parse the page
2746                                 # *after* a revision ID has been assigned. This is for null edits.
2747                                 $this->mOutput->setFlag( 'vary-revision' );
2748                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2749                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2750                                 break;
2751                         case 'revisionyear':
2752                                 # Let the edit saving system know we should parse the page
2753                                 # *after* a revision ID has been assigned. This is for null edits.
2754                                 $this->mOutput->setFlag( 'vary-revision' );
2755                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2756                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2757                                 break;
2758                         case 'revisiontimestamp':
2759                                 # Let the edit saving system know we should parse the page
2760                                 # *after* a revision ID has been assigned. This is for null edits.
2761                                 $this->mOutput->setFlag( 'vary-revision' );
2762                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2763                                 $value = $this->getRevisionTimestamp();
2764                                 break;
2765                         case 'revisionuser':
2766                                 # Let the edit saving system know we should parse the page
2767                                 # *after* a revision ID has been assigned. This is for null edits.
2768                                 $this->mOutput->setFlag( 'vary-revision' );
2769                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2770                                 $value = $this->getRevisionUser();
2771                                 break;
2772                         case 'namespace':
2773                                 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2774                                 break;
2775                         case 'namespacee':
2776                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2777                                 break;
2778                         case 'talkspace':
2779                                 $value = $this->mTitle->canTalk() ? str_replace( '_',' ',$this->mTitle->getTalkNsText() ) : '';
2780                                 break;
2781                         case 'talkspacee':
2782                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2783                                 break;
2784                         case 'subjectspace':
2785                                 $value = $this->mTitle->getSubjectNsText();
2786                                 break;
2787                         case 'subjectspacee':
2788                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2789                                 break;
2790                         case 'currentdayname':
2791                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2792                                 break;
2793                         case 'currentyear':
2794                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2795                                 break;
2796                         case 'currenttime':
2797                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2798                                 break;
2799                         case 'currenthour':
2800                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2801                                 break;
2802                         case 'currentweek':
2803                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2804                                 # int to remove the padding
2805                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2806                                 break;
2807                         case 'currentdow':
2808                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2809                                 break;
2810                         case 'localdayname':
2811                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2812                                 break;
2813                         case 'localyear':
2814                                 $value = $wgContLang->formatNum( $localYear, true );
2815                                 break;
2816                         case 'localtime':
2817                                 $value = $wgContLang->time( $localTimestamp, false, false );
2818                                 break;
2819                         case 'localhour':
2820                                 $value = $wgContLang->formatNum( $localHour, true );
2821                                 break;
2822                         case 'localweek':
2823                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2824                                 # int to remove the padding
2825                                 $value = $wgContLang->formatNum( (int)$localWeek );
2826                                 break;
2827                         case 'localdow':
2828                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2829                                 break;
2830                         case 'numberofarticles':
2831                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2832                                 break;
2833                         case 'numberoffiles':
2834                                 $value = $wgContLang->formatNum( SiteStats::images() );
2835                                 break;
2836                         case 'numberofusers':
2837                                 $value = $wgContLang->formatNum( SiteStats::users() );
2838                                 break;
2839                         case 'numberofactiveusers':
2840                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2841                                 break;
2842                         case 'numberofpages':
2843                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2844                                 break;
2845                         case 'numberofadmins':
2846                                 $value = $wgContLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2847                                 break;
2848                         case 'numberofedits':
2849                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2850                                 break;
2851                         case 'numberofviews':
2852                                 $value = $wgContLang->formatNum( SiteStats::views() );
2853                                 break;
2854                         case 'currenttimestamp':
2855                                 $value = wfTimestamp( TS_MW, $ts );
2856                                 break;
2857                         case 'localtimestamp':
2858                                 $value = $localTimestamp;
2859                                 break;
2860                         case 'currentversion':
2861                                 $value = SpecialVersion::getVersion();
2862                                 break;
2863                         case 'articlepath':
2864                                 return $wgArticlePath;
2865                         case 'sitename':
2866                                 return $wgSitename;
2867                         case 'server':
2868                                 return $wgServer;
2869                         case 'servername':
2870                                 wfSuppressWarnings(); # May give an E_WARNING in PHP < 5.3.3
2871                                 $serverName = parse_url( $wgServer, PHP_URL_HOST );
2872                                 wfRestoreWarnings();
2873                                 return $serverName ? $serverName : $wgServer;
2874                         case 'scriptpath':
2875                                 return $wgScriptPath;
2876                         case 'stylepath':
2877                                 return $wgStylePath;
2878                         case 'directionmark':
2879                                 return $wgContLang->getDirMark();
2880                         case 'contentlanguage':
2881                                 global $wgLanguageCode;
2882                                 return $wgLanguageCode;
2883                         default:
2884                                 $ret = null;
2885                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
2886                                         return $ret;
2887                                 } else {
2888                                         return null;
2889                                 }
2890                 }
2891
2892                 if ( $index )
2893                         $this->mVarCache[$index] = $value;
2894
2895                 return $value;
2896         }
2897
2898         /**
2899          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2900          *
2901          * @private
2902          */
2903         function initialiseVariables() {
2904                 wfProfileIn( __METHOD__ );
2905                 $variableIDs = MagicWord::getVariableIDs();
2906                 $substIDs = MagicWord::getSubstIDs();
2907
2908                 $this->mVariables = new MagicWordArray( $variableIDs );
2909                 $this->mSubstWords = new MagicWordArray( $substIDs );
2910                 wfProfileOut( __METHOD__ );
2911         }
2912
2913         /**
2914          * Preprocess some wikitext and return the document tree.
2915          * This is the ghost of replace_variables().
2916          *
2917          * @param $text String: The text to parse
2918          * @param $flags Integer: bitwise combination of:
2919          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2920          *                                     included. Default is to assume a direct page view.
2921          *
2922          * The generated DOM tree must depend only on the input text and the flags.
2923          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2924          *
2925          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2926          * change in the DOM tree for a given text, must be passed through the section identifier
2927          * in the section edit link and thus back to extractSections().
2928          *
2929          * The output of this function is currently only cached in process memory, but a persistent
2930          * cache may be implemented at a later date which takes further advantage of these strict
2931          * dependency requirements.
2932          *
2933          * @private
2934          *
2935          * @return PPNode
2936          */
2937         function preprocessToDom( $text, $flags = 0 ) {
2938                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2939                 return $dom;
2940         }
2941
2942         /**
2943          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2944          *
2945          * @return array
2946          */
2947         public static function splitWhitespace( $s ) {
2948                 $ltrimmed = ltrim( $s );
2949                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2950                 $trimmed = rtrim( $ltrimmed );
2951                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2952                 if ( $diff > 0 ) {
2953                         $w2 = substr( $ltrimmed, -$diff );
2954                 } else {
2955                         $w2 = '';
2956                 }
2957                 return array( $w1, $trimmed, $w2 );
2958         }
2959
2960         /**
2961          * Replace magic variables, templates, and template arguments
2962          * with the appropriate text. Templates are substituted recursively,
2963          * taking care to avoid infinite loops.
2964          *
2965          * Note that the substitution depends on value of $mOutputType:
2966          *  self::OT_WIKI: only {{subst:}} templates
2967          *  self::OT_PREPROCESS: templates but not extension tags
2968          *  self::OT_HTML: all templates and extension tags
2969          *
2970          * @param $text String: the text to transform
2971          * @param $frame PPFrame Object describing the arguments passed to the template.
2972          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2973          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2974          * @param $argsOnly Boolean: only do argument (triple-brace) expansion, not double-brace expansion
2975          * @private
2976          *
2977          * @return string
2978          */
2979         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2980                 # Is there any text? Also, Prevent too big inclusions!
2981                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2982                         return $text;
2983                 }
2984                 wfProfileIn( __METHOD__ );
2985
2986                 if ( $frame === false ) {
2987                         $frame = $this->getPreprocessor()->newFrame();
2988                 } elseif ( !( $frame instanceof PPFrame ) ) {
2989                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2990                         $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2991                 }
2992
2993                 $dom = $this->preprocessToDom( $text );
2994                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2995                 $text = $frame->expand( $dom, $flags );
2996
2997                 wfProfileOut( __METHOD__ );
2998                 return $text;
2999         }
3000
3001         /**
3002          * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3003          *
3004          * @return array
3005          */
3006         static function createAssocArgs( $args ) {
3007                 $assocArgs = array();
3008                 $index = 1;
3009                 foreach ( $args as $arg ) {
3010                         $eqpos = strpos( $arg, '=' );
3011                         if ( $eqpos === false ) {
3012                                 $assocArgs[$index++] = $arg;
3013                         } else {
3014                                 $name = trim( substr( $arg, 0, $eqpos ) );
3015                                 $value = trim( substr( $arg, $eqpos+1 ) );
3016                                 if ( $value === false ) {
3017                                         $value = '';
3018                                 }
3019                                 if ( $name !== false ) {
3020                                         $assocArgs[$name] = $value;
3021                                 }
3022                         }
3023                 }
3024
3025                 return $assocArgs;
3026         }
3027
3028         /**
3029          * Warn the user when a parser limitation is reached
3030          * Will warn at most once the user per limitation type
3031          *
3032          * @param $limitationType String: should be one of:
3033          *   'expensive-parserfunction' (corresponding messages:
3034          *       'expensive-parserfunction-warning',
3035          *       'expensive-parserfunction-category')
3036          *   'post-expand-template-argument' (corresponding messages:
3037          *       'post-expand-template-argument-warning',
3038          *       'post-expand-template-argument-category')
3039          *   'post-expand-template-inclusion' (corresponding messages:
3040          *       'post-expand-template-inclusion-warning',
3041          *       'post-expand-template-inclusion-category')
3042          * @param $current Current value
3043          * @param $max Maximum allowed, when an explicit limit has been
3044          *       exceeded, provide the values (optional)
3045          */
3046         function limitationWarn( $limitationType, $current=null, $max=null) {
3047                 # does no harm if $current and $max are present but are unnecessary for the message
3048                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
3049                 $this->mOutput->addWarning( $warning );
3050                 $this->addTrackingCategory( "$limitationType-category" );
3051         }
3052
3053         /**
3054          * Return the text of a template, after recursively
3055          * replacing any variables or templates within the template.
3056          *
3057          * @param $piece Array: the parts of the template
3058          *  $piece['title']: the title, i.e. the part before the |
3059          *  $piece['parts']: the parameter array
3060          *  $piece['lineStart']: whether the brace was at the start of a line
3061          * @param $frame PPFrame The current frame, contains template arguments
3062          * @return String: the text of the template
3063          * @private
3064          */
3065         function braceSubstitution( $piece, $frame ) {
3066                 global $wgContLang, $wgNonincludableNamespaces;
3067                 wfProfileIn( __METHOD__ );
3068                 wfProfileIn( __METHOD__.'-setup' );
3069
3070                 # Flags
3071                 $found = false;             # $text has been filled
3072                 $nowiki = false;            # wiki markup in $text should be escaped
3073                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
3074                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
3075                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
3076                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
3077
3078                 # Title object, where $text came from
3079                 $title = null;
3080
3081                 # $part1 is the bit before the first |, and must contain only title characters.
3082                 # Various prefixes will be stripped from it later.
3083                 $titleWithSpaces = $frame->expand( $piece['title'] );
3084                 $part1 = trim( $titleWithSpaces );
3085                 $titleText = false;
3086
3087                 # Original title text preserved for various purposes
3088                 $originalTitle = $part1;
3089
3090                 # $args is a list of argument nodes, starting from index 0, not including $part1
3091                 # *** FIXME if piece['parts'] is null then the call to getLength() below won't work b/c this $args isn't an object
3092                 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
3093                 wfProfileOut( __METHOD__.'-setup' );
3094
3095                 # SUBST
3096                 wfProfileIn( __METHOD__.'-modifiers' );
3097                 if ( !$found ) {
3098
3099                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3100
3101                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
3102                         # Decide whether to expand template or keep wikitext as-is.
3103                         if ( $this->ot['wiki'] ) {
3104                                 if ( $substMatch === false ) {
3105                                         $literal = true;  # literal when in PST with no prefix
3106                                 } else {
3107                                         $literal = false; # expand when in PST with subst: or safesubst:
3108                                 }
3109                         } else {
3110                                 if ( $substMatch == 'subst' ) {
3111                                         $literal = true;  # literal when not in PST with plain subst:
3112                                 } else {
3113                                         $literal = false; # expand when not in PST with safesubst: or no prefix
3114                                 }
3115                         }
3116                         if ( $literal ) {
3117                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3118                                 $isLocalObj = true;
3119                                 $found = true;
3120                         }
3121                 }
3122
3123                 # Variables
3124                 if ( !$found && $args->getLength() == 0 ) {
3125                         $id = $this->mVariables->matchStartToEnd( $part1 );
3126                         if ( $id !== false ) {
3127                                 $text = $this->getVariableValue( $id, $frame );
3128                                 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3129                                         $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3130                                 }
3131                                 $found = true;
3132                         }
3133                 }
3134
3135                 # MSG, MSGNW and RAW
3136                 if ( !$found ) {
3137                         # Check for MSGNW:
3138                         $mwMsgnw = MagicWord::get( 'msgnw' );
3139                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3140                                 $nowiki = true;
3141                         } else {
3142                                 # Remove obsolete MSG:
3143                                 $mwMsg = MagicWord::get( 'msg' );
3144                                 $mwMsg->matchStartAndRemove( $part1 );
3145                         }
3146
3147                         # Check for RAW:
3148                         $mwRaw = MagicWord::get( 'raw' );
3149                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3150                                 $forceRawInterwiki = true;
3151                         }
3152                 }
3153                 wfProfileOut( __METHOD__.'-modifiers' );
3154
3155                 # Parser functions
3156                 if ( !$found ) {
3157                         wfProfileIn( __METHOD__ . '-pfunc' );
3158
3159                         $colonPos = strpos( $part1, ':' );
3160                         if ( $colonPos !== false ) {
3161                                 # Case sensitive functions
3162                                 $function = substr( $part1, 0, $colonPos );
3163                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3164                                         $function = $this->mFunctionSynonyms[1][$function];
3165                                 } else {
3166                                         # Case insensitive functions
3167                                         $function = $wgContLang->lc( $function );
3168                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3169                                                 $function = $this->mFunctionSynonyms[0][$function];
3170                                         } else {
3171                                                 $function = false;
3172                                         }
3173                                 }
3174                                 if ( $function ) {
3175                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3176                                         $initialArgs = array( &$this );
3177                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3178                                         if ( $flags & SFH_OBJECT_ARGS ) {
3179                                                 # Add a frame parameter, and pass the arguments as an array
3180                                                 $allArgs = $initialArgs;
3181                                                 $allArgs[] = $frame;
3182                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3183                                                         $funcArgs[] = $args->item( $i );
3184                                                 }
3185                                                 $allArgs[] = $funcArgs;
3186                                         } else {
3187                                                 # Convert arguments to plain text
3188                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3189                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
3190                                                 }
3191                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3192                                         }
3193
3194                                         # Workaround for PHP bug 35229 and similar
3195                                         if ( !is_callable( $callback ) ) {
3196                                                 wfProfileOut( __METHOD__ . '-pfunc' );
3197                                                 wfProfileOut( __METHOD__ );
3198                                                 throw new MWException( "Tag hook for $function is not callable\n" );
3199                                         }
3200                                         $result = call_user_func_array( $callback, $allArgs );
3201                                         $found = true;
3202                                         $noparse = true;
3203                                         $preprocessFlags = 0;
3204
3205                                         if ( is_array( $result ) ) {
3206                                                 if ( isset( $result[0] ) ) {
3207                                                         $text = $result[0];
3208                                                         unset( $result[0] );
3209                                                 }
3210
3211                                                 # Extract flags into the local scope
3212                                                 # This allows callers to set flags such as nowiki, found, etc.
3213                                                 extract( $result );
3214                                         } else {
3215                                                 $text = $result;
3216                                         }
3217                                         if ( !$noparse ) {
3218                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
3219                                                 $isChildObj = true;
3220                                         }
3221                                 }
3222                         }
3223                         wfProfileOut( __METHOD__ . '-pfunc' );
3224                 }
3225
3226                 # Finish mangling title and then check for loops.
3227                 # Set $title to a Title object and $titleText to the PDBK
3228                 if ( !$found ) {
3229                         $ns = NS_TEMPLATE;
3230                         # Split the title into page and subpage
3231                         $subpage = '';
3232                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3233                         if ( $subpage !== '' ) {
3234                                 $ns = $this->mTitle->getNamespace();
3235                         }
3236                         $title = Title::newFromText( $part1, $ns );
3237                         if ( $title ) {
3238                                 $titleText = $title->getPrefixedText();
3239                                 # Check for language variants if the template is not found
3240                                 if ( $wgContLang->hasVariants() && $title->getArticleID() == 0 ) {
3241                                         $wgContLang->findVariantLink( $part1, $title, true );
3242                                 }
3243                                 # Do recursion depth check
3244                                 $limit = $this->mOptions->getMaxTemplateDepth();
3245                                 if ( $frame->depth >= $limit ) {
3246                                         $found = true;
3247                                         $text = '<span class="error">'
3248                                                 . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit )
3249                                                 . '</span>';
3250                                 }
3251                         }
3252                 }
3253
3254                 # Load from database
3255                 if ( !$found && $title ) {
3256                         wfProfileIn( __METHOD__ . '-loadtpl' );
3257                         if ( !$title->isExternal() ) {
3258                                 if ( $title->getNamespace() == NS_SPECIAL
3259                                         && $this->mOptions->getAllowSpecialInclusion()
3260                                         && $this->ot['html'] )
3261                                 {
3262                                         $text = SpecialPageFactory::capturePath( $title );
3263                                         if ( is_string( $text ) ) {
3264                                                 $found = true;
3265                                                 $isHTML = true;
3266                                                 $this->disableCache();
3267                                         }
3268                                 } elseif ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3269                                         $found = false; # access denied
3270                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3271                                 } else {
3272                                         list( $text, $title ) = $this->getTemplateDom( $title );
3273                                         if ( $text !== false ) {
3274                                                 $found = true;
3275                                                 $isChildObj = true;
3276                                         }
3277                                 }
3278
3279                                 # If the title is valid but undisplayable, make a link to it
3280                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3281                                         $text = "[[:$titleText]]";
3282                                         $found = true;
3283                                 }
3284                         } elseif ( $title->isTrans() ) {
3285                                 # Interwiki transclusion
3286                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3287                                         $text = $this->interwikiTransclude( $title, 'render' );
3288                                         $isHTML = true;
3289                                 } else {
3290                                         $text = $this->interwikiTransclude( $title, 'raw' );
3291                                         # Preprocess it like a template
3292                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3293                                         $isChildObj = true;
3294                                 }
3295                                 $found = true;
3296                         }
3297
3298                         # Do infinite loop check
3299                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3300                         if ( !$frame->loopCheck( $title ) ) {
3301                                 $found = true;
3302                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3303                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3304                         }
3305                         wfProfileOut( __METHOD__ . '-loadtpl' );
3306                 }
3307
3308                 # If we haven't found text to substitute by now, we're done
3309                 # Recover the source wikitext and return it
3310                 if ( !$found ) {
3311                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3312                         wfProfileOut( __METHOD__ );
3313                         return array( 'object' => $text );
3314                 }
3315
3316                 # Expand DOM-style return values in a child frame
3317                 if ( $isChildObj ) {
3318                         # Clean up argument array
3319                         $newFrame = $frame->newChild( $args, $title );
3320
3321                         if ( $nowiki ) {
3322                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3323                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3324                                 # Expansion is eligible for the empty-frame cache
3325                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3326                                         $text = $this->mTplExpandCache[$titleText];
3327                                 } else {
3328                                         $text = $newFrame->expand( $text );
3329                                         $this->mTplExpandCache[$titleText] = $text;
3330                                 }
3331                         } else {
3332                                 # Uncached expansion
3333                                 $text = $newFrame->expand( $text );
3334                         }
3335                 }
3336                 if ( $isLocalObj && $nowiki ) {
3337                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3338                         $isLocalObj = false;
3339                 }
3340
3341                 # Replace raw HTML by a placeholder
3342                 # Add a blank line preceding, to prevent it from mucking up
3343                 # immediately preceding headings
3344                 if ( $isHTML ) {
3345                         $text = "\n\n" . $this->insertStripItem( $text );
3346                 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3347                         # Escape nowiki-style return values
3348                         $text = wfEscapeWikiText( $text );
3349                 } elseif ( is_string( $text )
3350                         && !$piece['lineStart']
3351                         && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3352                 {
3353                         # Bug 529: if the template begins with a table or block-level
3354                         # element, it should be treated as beginning a new line.
3355                         # This behaviour is somewhat controversial.
3356                         $text = "\n" . $text;
3357                 }
3358
3359                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3360                         # Error, oversize inclusion
3361                         if ( $titleText !== false ) {
3362                                 # Make a working, properly escaped link if possible (bug 23588)
3363                                 $text = "[[:$titleText]]";
3364                         } else {
3365                                 # This will probably not be a working link, but at least it may
3366                                 # provide some hint of where the problem is
3367                                 preg_replace( '/^:/', '', $originalTitle );
3368                                 $text = "[[:$originalTitle]]";
3369                         }
3370                         $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3371                         $this->limitationWarn( 'post-expand-template-inclusion' );
3372                 }
3373
3374                 if ( $isLocalObj ) {
3375                         $ret = array( 'object' => $text );
3376                 } else {
3377                         $ret = array( 'text' => $text );
3378                 }
3379
3380                 wfProfileOut( __METHOD__ );
3381                 return $ret;
3382         }
3383
3384         /**
3385          * Get the semi-parsed DOM representation of a template with a given title,
3386          * and its redirect destination title. Cached.
3387          *
3388          * @return array
3389          */
3390         function getTemplateDom( $title ) {
3391                 $cacheTitle = $title;
3392                 $titleText = $title->getPrefixedDBkey();
3393
3394                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3395                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3396                         $title = Title::makeTitle( $ns, $dbk );
3397                         $titleText = $title->getPrefixedDBkey();
3398                 }
3399                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3400                         return array( $this->mTplDomCache[$titleText], $title );
3401                 }
3402
3403                 # Cache miss, go to the database
3404                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3405
3406                 if ( $text === false ) {
3407                         $this->mTplDomCache[$titleText] = false;
3408                         return array( false, $title );
3409                 }
3410
3411                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3412                 $this->mTplDomCache[ $titleText ] = $dom;
3413
3414                 if ( !$title->equals( $cacheTitle ) ) {
3415                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3416                                 array( $title->getNamespace(), $cdb = $title->getDBkey() );
3417                 }
3418
3419                 return array( $dom, $title );
3420         }
3421
3422         /**
3423          * Fetch the unparsed text of a template and register a reference to it.
3424          * @param Title $title
3425          * @return Array ( string or false, Title )
3426          */
3427         function fetchTemplateAndTitle( $title ) {
3428                 $templateCb = $this->mOptions->getTemplateCallback(); # Defaults to Parser::statelessFetchTemplate()
3429                 $stuff = call_user_func( $templateCb, $title, $this );
3430                 $text = $stuff['text'];
3431                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3432                 if ( isset( $stuff['deps'] ) ) {
3433                         foreach ( $stuff['deps'] as $dep ) {
3434                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3435                         }
3436                 }
3437                 return array( $text, $finalTitle );
3438         }
3439
3440         /**
3441          * Fetch the unparsed text of a template and register a reference to it.
3442          * @param Title $title
3443          * @return mixed string or false
3444          */
3445         function fetchTemplate( $title ) {
3446                 $rv = $this->fetchTemplateAndTitle( $title );
3447                 return $rv[0];
3448         }
3449
3450         /**
3451          * Static function to get a template
3452          * Can be overridden via ParserOptions::setTemplateCallback().
3453          *
3454          * @return array
3455          */
3456         static function statelessFetchTemplate( $title, $parser = false ) {
3457                 $text = $skip = false;
3458                 $finalTitle = $title;
3459                 $deps = array();
3460
3461                 # Loop to fetch the article, with up to 1 redirect
3462                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3463                         # Give extensions a chance to select the revision instead
3464                         $id = false; # Assume current
3465                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle',
3466                                 array( $parser, $title, &$skip, &$id ) );
3467
3468                         if ( $skip ) {
3469                                 $text = false;
3470                                 $deps[] = array(
3471                                         'title'         => $title,
3472                                         'page_id'       => $title->getArticleID(),
3473                                         'rev_id'        => null
3474                                 );
3475                                 break;
3476                         }
3477                         # Get the revision
3478                         $rev = $id
3479                                 ? Revision::newFromId( $id )
3480                                 : Revision::newFromTitle( $title );
3481                         $rev_id = $rev ? $rev->getId() : 0;
3482                         # If there is no current revision, there is no page
3483                         if ( $id === false && !$rev ) {
3484                                 $linkCache = LinkCache::singleton();
3485                                 $linkCache->addBadLinkObj( $title );
3486                         }
3487
3488                         $deps[] = array(
3489                                 'title'         => $title,
3490                                 'page_id'       => $title->getArticleID(),
3491                                 'rev_id'        => $rev_id );
3492                         if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3493                                 # We fetched a rev from a different title; register it too...
3494                                 $deps[] = array(
3495                                         'title'         => $rev->getTitle(),
3496                                         'page_id'       => $rev->getPage(),
3497                                         'rev_id'        => $rev_id );
3498                         }
3499
3500                         if ( $rev ) {
3501                                 $text = $rev->getText();
3502                         } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3503                                 global $wgContLang;
3504                                 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3505                                 if ( !$message->exists() ) {
3506                                         $text = false;
3507                                         break;
3508                                 }
3509                                 $text = $message->plain();
3510                         } else {
3511                                 break;
3512                         }
3513                         if ( $text === false ) {
3514                                 break;
3515                         }
3516                         # Redirect?
3517                         $finalTitle = $title;
3518                         $title = Title::newFromRedirect( $text );
3519                 }
3520                 return array(
3521                         'text' => $text,
3522                         'finalTitle' => $finalTitle,
3523                         'deps' => $deps );
3524         }
3525
3526         /**
3527          * Fetch a file and its title and register a reference to it.
3528          * @param Title $title
3529          * @param string $time MW timestamp
3530          * @param string $sha1 base 36 SHA-1
3531          * @return mixed File or false
3532          */
3533         function fetchFile( $title, $time = false, $sha1 = false ) {
3534                 $res = $this->fetchFileAndTitle( $title, $time, $sha1 );
3535                 return $res[0];
3536         }
3537
3538         /**
3539          * Fetch a file and its title and register a reference to it.
3540          * @param Title $title
3541          * @param string $time MW timestamp
3542          * @param string $sha1 base 36 SHA-1
3543          * @return Array ( File or false, Title of file )
3544          */
3545         function fetchFileAndTitle( $title, $time = false, $sha1 = false ) {
3546                 if ( $time === '0' ) {
3547                         $file = false; // broken thumbnail forced by hook
3548                 } elseif ( $sha1 ) { // get by (sha1,timestamp)
3549                         $file = RepoGroup::singleton()->findFileFromKey( $sha1, array( 'time' => $time ) );
3550                 } else { // get by (name,timestamp)
3551                         $file = wfFindFile( $title, array( 'time' => $time ) );
3552                 }
3553                 $time = $file ? $file->getTimestamp() : false;
3554                 $sha1 = $file ? $file->getSha1() : false;
3555                 # Register the file as a dependency...
3556                 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3557                 if ( $file && !$title->equals( $file->getTitle() ) ) {
3558                         # We fetched a rev from a different title; register it too...
3559                         $this->mOutput->addImage( $file->getTitle()->getDBkey(), $time, $sha1 );
3560                         # Update fetched file title
3561                         $title = $file->getTitle();
3562                 }
3563                 return array( $file, $title );
3564         }
3565
3566         /**
3567          * Transclude an interwiki link.
3568          *
3569          * @return string
3570          */
3571         function interwikiTransclude( $title, $action ) {
3572                 global $wgEnableScaryTranscluding;
3573
3574                 if ( !$wgEnableScaryTranscluding ) {
3575                         return wfMsgForContent('scarytranscludedisabled');
3576                 }
3577
3578                 $url = $title->getFullUrl( "action=$action" );
3579
3580                 if ( strlen( $url ) > 255 ) {
3581                         return wfMsgForContent( 'scarytranscludetoolong' );
3582                 }
3583                 return $this->fetchScaryTemplateMaybeFromCache( $url );
3584         }
3585
3586         /**
3587          * @param $url string
3588          * @return Mixed|String
3589          */
3590         function fetchScaryTemplateMaybeFromCache( $url ) {
3591                 global $wgTranscludeCacheExpiry;
3592                 $dbr = wfGetDB( DB_SLAVE );
3593                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3594                 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3595                                 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3596                 if ( $obj ) {
3597                         return $obj->tc_contents;
3598                 }
3599
3600                 $text = Http::get( $url );
3601                 if ( !$text ) {
3602                         return wfMsgForContent( 'scarytranscludefailed', $url );
3603                 }
3604
3605                 $dbw = wfGetDB( DB_MASTER );
3606                 $dbw->replace( 'transcache', array('tc_url'), array(
3607                         'tc_url' => $url,
3608                         'tc_time' => $dbw->timestamp( time() ),
3609                         'tc_contents' => $text)
3610                 );
3611                 return $text;
3612         }
3613
3614         /**
3615          * Triple brace replacement -- used for template arguments
3616          * @private
3617          *
3618          * @param $peice array
3619          * @param $frame PPFrame
3620          *
3621          * @return array
3622          */
3623         function argSubstitution( $piece, $frame ) {
3624                 wfProfileIn( __METHOD__ );
3625
3626                 $error = false;
3627                 $parts = $piece['parts'];
3628                 $nameWithSpaces = $frame->expand( $piece['title'] );
3629                 $argName = trim( $nameWithSpaces );
3630                 $object = false;
3631                 $text = $frame->getArgument( $argName );
3632                 if (  $text === false && $parts->getLength() > 0
3633                   && (
3634                         $this->ot['html']
3635                         || $this->ot['pre']
3636                         || ( $this->ot['wiki'] && $frame->isTemplate() )
3637                   )
3638                 ) {
3639                         # No match in frame, use the supplied default
3640                         $object = $parts->item( 0 )->getChildren();
3641                 }
3642                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3643                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3644                         $this->limitationWarn( 'post-expand-template-argument' );
3645                 }
3646
3647                 if ( $text === false && $object === false ) {
3648                         # No match anywhere
3649                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3650                 }
3651                 if ( $error !== false ) {
3652                         $text .= $error;
3653                 }
3654                 if ( $object !== false ) {
3655                         $ret = array( 'object' => $object );
3656                 } else {
3657                         $ret = array( 'text' => $text );
3658                 }
3659
3660                 wfProfileOut( __METHOD__ );
3661                 return $ret;
3662         }
3663
3664         /**
3665          * Return the text to be used for a given extension tag.
3666          * This is the ghost of strip().
3667          *
3668          * @param $params Associative array of parameters:
3669          *     name       PPNode for the tag name
3670          *     attr       PPNode for unparsed text where tag attributes are thought to be
3671          *     attributes Optional associative array of parsed attributes
3672          *     inner      Contents of extension element
3673          *     noClose    Original text did not have a close tag
3674          * @param $frame PPFrame
3675          *
3676          * @return string
3677          */
3678         function extensionSubstitution( $params, $frame ) {
3679                 $name = $frame->expand( $params['name'] );
3680                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3681                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3682                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3683
3684                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3685                         ( $this->ot['html'] || $this->ot['pre'] );
3686                 if ( $isFunctionTag ) {
3687                         $markerType = 'none';
3688                 } else {
3689                         $markerType = 'general';
3690                 }
3691                 if ( $this->ot['html'] || $isFunctionTag ) {
3692                         $name = strtolower( $name );
3693                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3694                         if ( isset( $params['attributes'] ) ) {
3695                                 $attributes = $attributes + $params['attributes'];
3696                         }
3697
3698                         if ( isset( $this->mTagHooks[$name] ) ) {
3699                                 # Workaround for PHP bug 35229 and similar
3700                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3701                                         throw new MWException( "Tag hook for $name is not callable\n" );
3702                                 }
3703                                 $output = call_user_func_array( $this->mTagHooks[$name],
3704                                         array( $content, $attributes, $this, $frame ) );
3705                         } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3706                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3707                                 if ( !is_callable( $callback ) ) {
3708                                         throw new MWException( "Tag hook for $name is not callable\n" );
3709                                 }
3710
3711                                 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3712                         } else {
3713                                 $output = '<span class="error">Invalid tag extension name: ' .
3714                                         htmlspecialchars( $name ) . '</span>';
3715                         }
3716
3717                         if ( is_array( $output ) ) {
3718                                 # Extract flags to local scope (to override $markerType)
3719                                 $flags = $output;
3720                                 $output = $flags[0];
3721                                 unset( $flags[0] );
3722                                 extract( $flags );
3723                         }
3724                 } else {
3725                         if ( is_null( $attrText ) ) {
3726                                 $attrText = '';
3727                         }
3728                         if ( isset( $params['attributes'] ) ) {
3729                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3730                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3731                                                 htmlspecialchars( $attrValue ) . '"';
3732                                 }
3733                         }
3734                         if ( $content === null ) {
3735                                 $output = "<$name$attrText/>";
3736                         } else {
3737                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3738                                 $output = "<$name$attrText>$content$close";
3739                         }
3740                 }
3741
3742                 if ( $markerType === 'none' ) {
3743                         return $output;
3744                 } elseif ( $markerType === 'nowiki' ) {
3745                         $this->mStripState->addNoWiki( $marker, $output );
3746                 } elseif ( $markerType === 'general' ) {
3747                         $this->mStripState->addGeneral( $marker, $output );
3748                 } else {
3749                         throw new MWException( __METHOD__.': invalid marker type' );
3750                 }
3751                 return $marker;
3752         }
3753
3754         /**
3755          * Increment an include size counter
3756          *
3757          * @param $type String: the type of expansion
3758          * @param $size Integer: the size of the text
3759          * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
3760          */
3761         function incrementIncludeSize( $type, $size ) {
3762                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3763                         return false;
3764                 } else {
3765                         $this->mIncludeSizes[$type] += $size;
3766                         return true;
3767                 }
3768         }
3769
3770         /**
3771          * Increment the expensive function count
3772          *
3773          * @return Boolean: false if the limit has been exceeded
3774          */
3775         function incrementExpensiveFunctionCount() {
3776                 global $wgExpensiveParserFunctionLimit;
3777                 $this->mExpensiveFunctionCount++;
3778                 if ( $this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit ) {
3779                         return true;
3780                 }
3781                 return false;
3782         }
3783
3784         /**
3785          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3786          * Fills $this->mDoubleUnderscores, returns the modified text
3787          */
3788         function doDoubleUnderscore( $text ) {
3789                 wfProfileIn( __METHOD__ );
3790
3791                 # The position of __TOC__ needs to be recorded
3792                 $mw = MagicWord::get( 'toc' );
3793                 if ( $mw->match( $text ) ) {
3794                         $this->mShowToc = true;
3795                         $this->mForceTocPosition = true;
3796
3797                         # Set a placeholder. At the end we'll fill it in with the TOC.
3798                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3799
3800                         # Only keep the first one.
3801                         $text = $mw->replace( '', $text );
3802                 }
3803
3804                 # Now match and remove the rest of them
3805                 $mwa = MagicWord::getDoubleUnderscoreArray();
3806                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3807
3808                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3809                         $this->mOutput->mNoGallery = true;
3810                 }
3811                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3812                         $this->mShowToc = false;
3813                 }
3814                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3815                         $this->addTrackingCategory( 'hidden-category-category' );
3816                 }
3817                 # (bug 8068) Allow control over whether robots index a page.
3818                 #
3819                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3820                 # is not desirable, the last one on the page should win.
3821                 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3822                         $this->mOutput->setIndexPolicy( 'noindex' );
3823                         $this->addTrackingCategory( 'noindex-category' );
3824                 }
3825                 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3826                         $this->mOutput->setIndexPolicy( 'index' );
3827                         $this->addTrackingCategory( 'index-category' );
3828                 }
3829
3830                 # Cache all double underscores in the database
3831                 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3832                         $this->mOutput->setProperty( $key, '' );
3833                 }
3834
3835                 wfProfileOut( __METHOD__ );
3836                 return $text;
3837         }
3838
3839         /**
3840          * Add a tracking category, getting the title from a system message,
3841          * or print a debug message if the title is invalid.
3842          *
3843          * @param $msg String: message key
3844          * @return Boolean: whether the addition was successful
3845          */
3846         protected function addTrackingCategory( $msg ) {
3847                 $cat = wfMsgForContent( $msg );
3848
3849                 # Allow tracking categories to be disabled by setting them to "-"
3850                 if ( $cat === '-' ) {
3851                         return false;
3852                 }
3853
3854                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3855                 if ( $containerCategory ) {
3856                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3857                         return true;
3858                 } else {
3859                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3860                         return false;
3861                 }
3862         }
3863
3864         /**
3865          * This function accomplishes several tasks:
3866          * 1) Auto-number headings if that option is enabled
3867          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3868          * 3) Add a Table of contents on the top for users who have enabled the option
3869          * 4) Auto-anchor headings
3870          *
3871          * It loops through all headlines, collects the necessary data, then splits up the
3872          * string and re-inserts the newly formatted headlines.
3873          *
3874          * @param $text String
3875          * @param $origText String: original, untouched wikitext
3876          * @param $isMain Boolean
3877          * @private
3878          */
3879         function formatHeadings( $text, $origText, $isMain=true ) {
3880                 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3881
3882                 # Inhibit editsection links if requested in the page
3883                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3884                         $showEditLink = 0;
3885                 } else {
3886                         $showEditLink = $this->mOptions->getEditSection();
3887                 }
3888                 if ( $showEditLink ) {
3889                         $this->mOutput->setEditSectionTokens( true );
3890                 }
3891
3892                 # Get all headlines for numbering them and adding funky stuff like [edit]
3893                 # links - this is for later, but we need the number of headlines right now
3894                 $matches = array();
3895                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3896
3897                 # if there are fewer than 4 headlines in the article, do not show TOC
3898                 # unless it's been explicitly enabled.
3899                 $enoughToc = $this->mShowToc &&
3900                         ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3901
3902                 # Allow user to stipulate that a page should have a "new section"
3903                 # link added via __NEWSECTIONLINK__
3904                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3905                         $this->mOutput->setNewSection( true );
3906                 }
3907
3908                 # Allow user to remove the "new section"
3909                 # link via __NONEWSECTIONLINK__
3910                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3911                         $this->mOutput->hideNewSection( true );
3912                 }
3913
3914                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3915                 # override above conditions and always show TOC above first header
3916                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3917                         $this->mShowToc = true;
3918                         $enoughToc = true;
3919                 }
3920
3921                 # headline counter
3922                 $headlineCount = 0;
3923                 $numVisible = 0;
3924
3925                 # Ugh .. the TOC should have neat indentation levels which can be
3926                 # passed to the skin functions. These are determined here
3927                 $toc = '';
3928                 $full = '';
3929                 $head = array();
3930                 $sublevelCount = array();
3931                 $levelCount = array();
3932                 $level = 0;
3933                 $prevlevel = 0;
3934                 $toclevel = 0;
3935                 $prevtoclevel = 0;
3936                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3937                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3938                 $oldType = $this->mOutputType;
3939                 $this->setOutputType( self::OT_WIKI );
3940                 $frame = $this->getPreprocessor()->newFrame();
3941                 $root = $this->preprocessToDom( $origText );
3942                 $node = $root->getFirstChild();
3943                 $byteOffset = 0;
3944                 $tocraw = array();
3945                 $refers = array();
3946
3947                 foreach ( $matches[3] as $headline ) {
3948                         $isTemplate = false;
3949                         $titleText = false;
3950                         $sectionIndex = false;
3951                         $numbering = '';
3952                         $markerMatches = array();
3953                         if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
3954                                 $serial = $markerMatches[1];
3955                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3956                                 $isTemplate = ( $titleText != $baseTitleText );
3957                                 $headline = preg_replace( "/^$markerRegex/", "", $headline );
3958                         }
3959
3960                         if ( $toclevel ) {
3961                                 $prevlevel = $level;
3962                         }
3963                         $level = $matches[1][$headlineCount];
3964
3965                         if ( $level > $prevlevel ) {
3966                                 # Increase TOC level
3967                                 $toclevel++;
3968                                 $sublevelCount[$toclevel] = 0;
3969                                 if ( $toclevel<$wgMaxTocLevel ) {
3970                                         $prevtoclevel = $toclevel;
3971                                         $toc .= Linker::tocIndent();
3972                                         $numVisible++;
3973                                 }
3974                         } elseif ( $level < $prevlevel && $toclevel > 1 ) {
3975                                 # Decrease TOC level, find level to jump to
3976
3977                                 for ( $i = $toclevel; $i > 0; $i-- ) {
3978                                         if ( $levelCount[$i] == $level ) {
3979                                                 # Found last matching level
3980                                                 $toclevel = $i;
3981                                                 break;
3982                                         } elseif ( $levelCount[$i] < $level ) {
3983                                                 # Found first matching level below current level
3984                                                 $toclevel = $i + 1;
3985                                                 break;
3986                                         }
3987                                 }
3988                                 if ( $i == 0 ) {
3989                                         $toclevel = 1;
3990                                 }
3991                                 if ( $toclevel<$wgMaxTocLevel ) {
3992                                         if ( $prevtoclevel < $wgMaxTocLevel ) {
3993                                                 # Unindent only if the previous toc level was shown :p
3994                                                 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
3995                                                 $prevtoclevel = $toclevel;
3996                                         } else {
3997                                                 $toc .= Linker::tocLineEnd();
3998                                         }
3999                                 }
4000                         } else {
4001                                 # No change in level, end TOC line
4002                                 if ( $toclevel<$wgMaxTocLevel ) {
4003                                         $toc .= Linker::tocLineEnd();
4004                                 }
4005                         }
4006
4007                         $levelCount[$toclevel] = $level;
4008
4009                         # count number of headlines for each level
4010                         @$sublevelCount[$toclevel]++;
4011                         $dot = 0;
4012                         for( $i = 1; $i <= $toclevel; $i++ ) {
4013                                 if ( !empty( $sublevelCount[$i] ) ) {
4014                                         if ( $dot ) {
4015                                                 $numbering .= '.';
4016                                         }
4017                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
4018                                         $dot = 1;
4019                                 }
4020                         }
4021
4022                         # The safe header is a version of the header text safe to use for links
4023                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4024                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
4025
4026                         # Remove link placeholders by the link text.
4027                         #     <!--LINK number-->
4028                         # turns into
4029                         #     link text with suffix
4030                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
4031
4032                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
4033                         $tocline = preg_replace(
4034                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
4035                                 array( '',                          '<$1>' ),
4036                                 $safeHeadline
4037                         );
4038                         $tocline = trim( $tocline );
4039
4040                         # For the anchor, strip out HTML-y stuff period
4041                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
4042                         $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4043
4044                         # Save headline for section edit hint before it's escaped
4045                         $headlineHint = $safeHeadline;
4046
4047                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
4048                                 # For reverse compatibility, provide an id that's
4049                                 # HTML4-compatible, like we used to.
4050                                 #
4051                                 # It may be worth noting, academically, that it's possible for
4052                                 # the legacy anchor to conflict with a non-legacy headline
4053                                 # anchor on the page.  In this case likely the "correct" thing
4054                                 # would be to either drop the legacy anchors or make sure
4055                                 # they're numbered first.  However, this would require people
4056                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4057                                 # manually, so let's not bother worrying about it.
4058                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
4059                                         array( 'noninitial', 'legacy' ) );
4060                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4061
4062                                 if ( $legacyHeadline == $safeHeadline ) {
4063                                         # No reason to have both (in fact, we can't)
4064                                         $legacyHeadline = false;
4065                                 }
4066                         } else {
4067                                 $legacyHeadline = false;
4068                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
4069                                         'noninitial' );
4070                         }
4071
4072                         # HTML names must be case-insensitively unique (bug 10721).
4073                         # This does not apply to Unicode characters per
4074                         # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
4075                         # FIXME: We may be changing them depending on the current locale.
4076                         $arrayKey = strtolower( $safeHeadline );
4077                         if ( $legacyHeadline === false ) {
4078                                 $legacyArrayKey = false;
4079                         } else {
4080                                 $legacyArrayKey = strtolower( $legacyHeadline );
4081                         }
4082
4083                         # count how many in assoc. array so we can track dupes in anchors
4084                         if ( isset( $refers[$arrayKey] ) ) {
4085                                 $refers[$arrayKey]++;
4086                         } else {
4087                                 $refers[$arrayKey] = 1;
4088                         }
4089                         if ( isset( $refers[$legacyArrayKey] ) ) {
4090                                 $refers[$legacyArrayKey]++;
4091                         } else {
4092                                 $refers[$legacyArrayKey] = 1;
4093                         }
4094
4095                         # Don't number the heading if it is the only one (looks silly)
4096                         if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4097                                 # the two are different if the line contains a link
4098                                 $headline = $numbering . ' ' . $headline;
4099                         }
4100
4101                         # Create the anchor for linking from the TOC to the section
4102                         $anchor = $safeHeadline;
4103                         $legacyAnchor = $legacyHeadline;
4104                         if ( $refers[$arrayKey] > 1 ) {
4105                                 $anchor .= '_' . $refers[$arrayKey];
4106                         }
4107                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
4108                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
4109                         }
4110                         if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4111                                 $toc .= Linker::tocLine( $anchor, $tocline,
4112                                         $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4113                         }
4114
4115                         # Add the section to the section tree
4116                         # Find the DOM node for this header
4117                         while ( $node && !$isTemplate ) {
4118                                 if ( $node->getName() === 'h' ) {
4119                                         $bits = $node->splitHeading();
4120                                         if ( $bits['i'] == $sectionIndex ) {
4121                                                 break;
4122                                         }
4123                                 }
4124                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4125                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4126                                 $node = $node->getNextSibling();
4127                         }
4128                         $tocraw[] = array(
4129                                 'toclevel' => $toclevel,
4130                                 'level' => $level,
4131                                 'line' => $tocline,
4132                                 'number' => $numbering,
4133                                 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4134                                 'fromtitle' => $titleText,
4135                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
4136                                 'anchor' => $anchor,
4137                         );
4138
4139                         # give headline the correct <h#> tag
4140                         if ( $showEditLink && $sectionIndex !== false ) {
4141                                 // Output edit section links as markers with styles that can be customized by skins
4142                                 if ( $isTemplate ) {
4143                                         # Put a T flag in the section identifier, to indicate to extractSections()
4144                                         # that sections inside <includeonly> should be counted.
4145                                         $editlinkArgs = array( $titleText, "T-$sectionIndex"/*, null */ );
4146                                 } else {
4147                                         $editlinkArgs = array( $this->mTitle->getPrefixedText(), $sectionIndex, $headlineHint );
4148                                 }
4149                                 // We use a bit of pesudo-xml for editsection markers. The language converter is run later on
4150                                 // Using a UNIQ style marker leads to the converter screwing up the tokens when it converts stuff
4151                                 // And trying to insert strip tags fails too. At this point all real inputted tags have already been escaped
4152                                 // so we don't have to worry about a user trying to input one of these markers directly.
4153                                 // We use a page and section attribute to stop the language converter from converting these important bits
4154                                 // of data, but put the headline hint inside a content block because the language converter is supposed to
4155                                 // be able to convert that piece of data.
4156                                 $editlink = '<mw:editsection page="' . htmlspecialchars($editlinkArgs[0]);
4157                                 $editlink .= '" section="' . htmlspecialchars($editlinkArgs[1]) .'"';
4158                                 if ( isset($editlinkArgs[2]) ) {
4159                                         $editlink .= '>' . $editlinkArgs[2] . '</mw:editsection>';
4160                                 } else {
4161                                         $editlink .= '/>';
4162                                 }
4163                         } else {
4164                                 $editlink = '';
4165                         }
4166                         $head[$headlineCount] = Linker::makeHeadline( $level,
4167                                 $matches['attrib'][$headlineCount], $anchor, $headline,
4168                                 $editlink, $legacyAnchor );
4169
4170                         $headlineCount++;
4171                 }
4172
4173                 $this->setOutputType( $oldType );
4174
4175                 # Never ever show TOC if no headers
4176                 if ( $numVisible < 1 ) {
4177                         $enoughToc = false;
4178                 }
4179
4180                 if ( $enoughToc ) {
4181                         if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4182                                 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4183                         }
4184                         $toc = Linker::tocList( $toc, $this->mOptions->getUserLang() );
4185                         $this->mOutput->setTOCHTML( $toc );
4186                 }
4187
4188                 if ( $isMain ) {
4189                         $this->mOutput->setSections( $tocraw );
4190                 }
4191
4192                 # split up and insert constructed headlines
4193
4194                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
4195                 $i = 0;
4196
4197                 foreach ( $blocks as $block ) {
4198                         if ( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
4199                                 # This is the [edit] link that appears for the top block of text when
4200                                 # section editing is enabled
4201
4202                                 # Disabled because it broke block formatting
4203                                 # For example, a bullet point in the top line
4204                                 # $full .= $sk->editSectionLink(0);
4205                         }
4206                         $full .= $block;
4207                         if ( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
4208                                 # Top anchor now in skin
4209                                 $full = $full.$toc;
4210                         }
4211
4212                         if ( !empty( $head[$i] ) ) {
4213                                 $full .= $head[$i];
4214                         }
4215                         $i++;
4216                 }
4217                 if ( $this->mForceTocPosition ) {
4218                         return str_replace( '<!--MWTOC-->', $toc, $full );
4219                 } else {
4220                         return $full;
4221                 }
4222         }
4223
4224         /**
4225          * Transform wiki markup when saving a page by doing \r\n -> \n
4226          * conversion, substitting signatures, {{subst:}} templates, etc.
4227          *
4228          * @param $text String: the text to transform
4229          * @param $title Title: the Title object for the current article
4230          * @param $user User: the User object describing the current user
4231          * @param $options ParserOptions: parsing options
4232          * @param $clearState Boolean: whether to clear the parser state first
4233          * @return String: the altered wiki markup
4234          */
4235         public function preSaveTransform( $text, Title $title, User $user, ParserOptions $options, $clearState = true ) {
4236                 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4237                 $this->setUser( $user );
4238
4239                 $pairs = array(
4240                         "\r\n" => "\n",
4241                 );
4242                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4243                 if( $options->getPreSaveTransform() ) {
4244                         $text = $this->pstPass2( $text, $user );
4245                 }
4246                 $text = $this->mStripState->unstripBoth( $text );
4247
4248                 $this->setUser( null ); #Reset
4249
4250                 return $text;
4251         }
4252
4253         /**
4254          * Pre-save transform helper function
4255          * @private
4256          */
4257         function pstPass2( $text, $user ) {
4258                 global $wgContLang, $wgLocaltimezone;
4259
4260                 # Note: This is the timestamp saved as hardcoded wikitext to
4261                 # the database, we use $wgContLang here in order to give
4262                 # everyone the same signature and use the default one rather
4263                 # than the one selected in each user's preferences.
4264                 # (see also bug 12815)
4265                 $ts = $this->mOptions->getTimestamp();
4266                 if ( isset( $wgLocaltimezone ) ) {
4267                         $tz = $wgLocaltimezone;
4268                 } else {
4269                         $tz = date_default_timezone_get();
4270                 }
4271
4272                 $unixts = wfTimestamp( TS_UNIX, $ts );
4273                 $oldtz = date_default_timezone_get();
4274                 date_default_timezone_set( $tz );
4275                 $ts = date( 'YmdHis', $unixts );
4276                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4277
4278                 # Allow translation of timezones through wiki. date() can return
4279                 # whatever crap the system uses, localised or not, so we cannot
4280                 # ship premade translations.
4281                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4282                 $msg = wfMessage( $key )->inContentLanguage();
4283                 if ( $msg->exists() ) {
4284                         $tzMsg = $msg->text();
4285                 }
4286
4287                 date_default_timezone_set( $oldtz );
4288
4289                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4290
4291                 # Variable replacement
4292                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4293                 $text = $this->replaceVariables( $text );
4294
4295                 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4296                 # which may corrupt this parser instance via its wfMsgExt( parsemag ) call-
4297
4298                 # Signatures
4299                 $sigText = $this->getUserSig( $user );
4300                 $text = strtr( $text, array(
4301                         '~~~~~' => $d,
4302                         '~~~~' => "$sigText $d",
4303                         '~~~' => $sigText
4304                 ) );
4305
4306                 # Context links: [[|name]] and [[name (context)|]]
4307                 global $wgLegalTitleChars;
4308                 $tc = "[$wgLegalTitleChars]";
4309                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4310
4311                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4312                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
4313                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4314                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4315
4316                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4317                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4318                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4319                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4320
4321                 $t = $this->mTitle->getText();
4322                 $m = array();
4323                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4324                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4325                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4326                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4327                 } else {
4328                         # if there's no context, don't bother duplicating the title
4329                         $text = preg_replace( $p2, '[[\\1]]', $text );
4330                 }
4331
4332                 # Trim trailing whitespace
4333                 $text = rtrim( $text );
4334
4335                 return $text;
4336         }
4337
4338         /**
4339          * Fetch the user's signature text, if any, and normalize to
4340          * validated, ready-to-insert wikitext.
4341          * If you have pre-fetched the nickname or the fancySig option, you can
4342          * specify them here to save a database query.
4343          * Do not reuse this parser instance after calling getUserSig(),
4344          * as it may have changed if it's the $wgParser.
4345          *
4346          * @param $user User
4347          * @param $nickname String: nickname to use or false to use user's default nickname
4348          * @param $fancySig Boolean: whether the nicknname is the complete signature
4349          *                  or null to use default value
4350          * @return string
4351          */
4352         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4353                 global $wgMaxSigChars;
4354
4355                 $username = $user->getName();
4356
4357                 # If not given, retrieve from the user object.
4358                 if ( $nickname === false )
4359                         $nickname = $user->getOption( 'nickname' );
4360
4361                 if ( is_null( $fancySig ) ) {
4362                         $fancySig = $user->getBoolOption( 'fancysig' );
4363                 }
4364
4365                 $nickname = $nickname == null ? $username : $nickname;
4366
4367                 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4368                         $nickname = $username;
4369                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4370                 } elseif ( $fancySig !== false ) {
4371                         # Sig. might contain markup; validate this
4372                         if ( $this->validateSig( $nickname ) !== false ) {
4373                                 # Validated; clean up (if needed) and return it
4374                                 return $this->cleanSig( $nickname, true );
4375                         } else {
4376                                 # Failed to validate; fall back to the default
4377                                 $nickname = $username;
4378                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4379                         }
4380                 }
4381
4382                 # Make sure nickname doesnt get a sig in a sig
4383                 $nickname = $this->cleanSigInSig( $nickname );
4384
4385                 # If we're still here, make it a link to the user page
4386                 $userText = wfEscapeWikiText( $username );
4387                 $nickText = wfEscapeWikiText( $nickname );
4388                 if ( $user->isAnon() )  {
4389                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4390                 } else {
4391                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4392                 }
4393         }
4394
4395         /**
4396          * Check that the user's signature contains no bad XML
4397          *
4398          * @param $text String
4399          * @return mixed An expanded string, or false if invalid.
4400          */
4401         function validateSig( $text ) {
4402                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4403         }
4404
4405         /**
4406          * Clean up signature text
4407          *
4408          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4409          * 2) Substitute all transclusions
4410          *
4411          * @param $text String
4412          * @param $parsing Whether we're cleaning (preferences save) or parsing
4413          * @return String: signature text
4414          */
4415         function cleanSig( $text, $parsing = false ) {
4416                 if ( !$parsing ) {
4417                         global $wgTitle;
4418                         $this->mOptions = new ParserOptions;
4419                         $this->clearState();
4420                         $this->setTitle( $wgTitle );
4421                         $this->setOutputType = self::OT_PREPROCESS;
4422                 }
4423
4424                 # Option to disable this feature
4425                 if ( !$this->mOptions->getCleanSignatures() ) {
4426                         return $text;
4427                 }
4428
4429                 # FIXME: regex doesn't respect extension tags or nowiki
4430                 #  => Move this logic to braceSubstitution()
4431                 $substWord = MagicWord::get( 'subst' );
4432                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4433                 $substText = '{{' . $substWord->getSynonym( 0 );
4434
4435                 $text = preg_replace( $substRegex, $substText, $text );
4436                 $text = $this->cleanSigInSig( $text );
4437                 $dom = $this->preprocessToDom( $text );
4438                 $frame = $this->getPreprocessor()->newFrame();
4439                 $text = $frame->expand( $dom );
4440
4441                 if ( !$parsing ) {
4442                         $text = $this->mStripState->unstripBoth( $text );
4443                 }
4444
4445                 return $text;
4446         }
4447
4448         /**
4449          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4450          *
4451          * @param $text String
4452          * @return String: signature text with /~{3,5}/ removed
4453          */
4454         function cleanSigInSig( $text ) {
4455                 $text = preg_replace( '/~{3,5}/', '', $text );
4456                 return $text;
4457         }
4458
4459         /**
4460          * Set up some variables which are usually set up in parse()
4461          * so that an external function can call some class members with confidence
4462          */
4463         public function startExternalParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4464                 $this->startParse( $title, $options, $outputType, $clearState );
4465         }
4466
4467         private function startParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4468                 $this->setTitle( $title );
4469                 $this->mOptions = $options;
4470                 $this->setOutputType( $outputType );
4471                 if ( $clearState ) {
4472                         $this->clearState();
4473                 }
4474         }
4475
4476         /**
4477          * Wrapper for preprocess()
4478          *
4479          * @param $text String: the text to preprocess
4480          * @param $options ParserOptions: options
4481          * @param $title Title object or null to use $wgTitle
4482          * @return String
4483          */
4484         public function transformMsg( $text, $options, $title = null ) {
4485                 static $executing = false;
4486
4487                 # Guard against infinite recursion
4488                 if ( $executing ) {
4489                         return $text;
4490                 }
4491                 $executing = true;
4492
4493                 wfProfileIn( __METHOD__ );
4494                 if ( !$title ) {
4495                         global $wgTitle;
4496                         $title = $wgTitle;
4497                 }
4498                 if ( !$title ) {
4499                         # It's not uncommon having a null $wgTitle in scripts. See r80898
4500                         # Create a ghost title in such case
4501                         $title = Title::newFromText( 'Dwimmerlaik' );
4502                 }
4503                 $text = $this->preprocess( $text, $title, $options );
4504
4505                 $executing = false;
4506                 wfProfileOut( __METHOD__ );
4507                 return $text;
4508         }
4509
4510         /**
4511          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4512          * The callback should have the following form:
4513          *    function myParserHook( $text, $params, $parser, $frame ) { ... }
4514          *
4515          * Transform and return $text. Use $parser for any required context, e.g. use
4516          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4517          *
4518          * Hooks may return extended information by returning an array, of which the
4519          * first numbered element (index 0) must be the return string, and all other
4520          * entries are extracted into local variables within an internal function
4521          * in the Parser class.
4522          *
4523          * This interface (introduced r61913) appears to be undocumented, but
4524          * 'markerName' is used by some core tag hooks to override which strip
4525          * array their results are placed in. **Use great caution if attempting
4526          * this interface, as it is not documented and injudicious use could smash
4527          * private variables.**
4528          *
4529          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4530          * @param $callback Mixed: the callback function (and object) to use for the tag
4531          * @return The old value of the mTagHooks array associated with the hook
4532          */
4533         public function setHook( $tag, $callback ) {
4534                 $tag = strtolower( $tag );
4535                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4536                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4537                 $this->mTagHooks[$tag] = $callback;
4538                 if ( !in_array( $tag, $this->mStripList ) ) {
4539                         $this->mStripList[] = $tag;
4540                 }
4541
4542                 return $oldVal;
4543         }
4544
4545         /**
4546          * As setHook(), but letting the contents be parsed.
4547          *
4548          * Transparent tag hooks are like regular XML-style tag hooks, except they
4549          * operate late in the transformation sequence, on HTML instead of wikitext.
4550          *
4551          * This is probably obsoleted by things dealing with parser frames?
4552          * The only extension currently using it is geoserver.
4553          *
4554          * @since 1.10
4555          * @todo better document or deprecate this
4556          *
4557          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4558          * @param $callback Mixed: the callback function (and object) to use for the tag
4559          * @return The old value of the mTagHooks array associated with the hook
4560          */
4561         function setTransparentTagHook( $tag, $callback ) {
4562                 $tag = strtolower( $tag );
4563                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4564                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4565                 $this->mTransparentTagHooks[$tag] = $callback;
4566
4567                 return $oldVal;
4568         }
4569
4570         /**
4571          * Remove all tag hooks
4572          */
4573         function clearTagHooks() {
4574                 $this->mTagHooks = array();
4575                 $this->mStripList = $this->mDefaultStripList;
4576         }
4577
4578         /**
4579          * Create a function, e.g. {{sum:1|2|3}}
4580          * The callback function should have the form:
4581          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4582          *
4583          * Or with SFH_OBJECT_ARGS:
4584          *    function myParserFunction( $parser, $frame, $args ) { ... }
4585          *
4586          * The callback may either return the text result of the function, or an array with the text
4587          * in element 0, and a number of flags in the other elements. The names of the flags are
4588          * specified in the keys. Valid flags are:
4589          *   found                     The text returned is valid, stop processing the template. This
4590          *                             is on by default.
4591          *   nowiki                    Wiki markup in the return value should be escaped
4592          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4593          *
4594          * @param $id String: The magic word ID
4595          * @param $callback Mixed: the callback function (and object) to use
4596          * @param $flags Integer: a combination of the following flags:
4597          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4598          *
4599          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4600          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4601          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4602          *     the arguments, and to control the way they are expanded.
4603          *
4604          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4605          *     arguments, for instance:
4606          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4607          *
4608          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4609          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4610          *     working if/when this is changed.
4611          *
4612          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4613          *     expansion.
4614          *
4615          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4616          *     about the methods available in PPFrame and PPNode.
4617          *
4618          * @return The old callback function for this name, if any
4619          */
4620         public function setFunctionHook( $id, $callback, $flags = 0 ) {
4621                 global $wgContLang;
4622
4623                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4624                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4625
4626                 # Add to function cache
4627                 $mw = MagicWord::get( $id );
4628                 if ( !$mw )
4629                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4630
4631                 $synonyms = $mw->getSynonyms();
4632                 $sensitive = intval( $mw->isCaseSensitive() );
4633
4634                 foreach ( $synonyms as $syn ) {
4635                         # Case
4636                         if ( !$sensitive ) {
4637                                 $syn = $wgContLang->lc( $syn );
4638                         }
4639                         # Add leading hash
4640                         if ( !( $flags & SFH_NO_HASH ) ) {
4641                                 $syn = '#' . $syn;
4642                         }
4643                         # Remove trailing colon
4644                         if ( substr( $syn, -1, 1 ) === ':' ) {
4645                                 $syn = substr( $syn, 0, -1 );
4646                         }
4647                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4648                 }
4649                 return $oldVal;
4650         }
4651
4652         /**
4653          * Get all registered function hook identifiers
4654          *
4655          * @return Array
4656          */
4657         function getFunctionHooks() {
4658                 return array_keys( $this->mFunctionHooks );
4659         }
4660
4661         /**
4662          * Create a tag function, e.g. <test>some stuff</test>.
4663          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4664          * Unlike parser functions, their content is not preprocessed.
4665          */
4666         function setFunctionTagHook( $tag, $callback, $flags ) {
4667                 $tag = strtolower( $tag );
4668                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4669                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4670                         $this->mFunctionTagHooks[$tag] : null;
4671                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4672
4673                 if ( !in_array( $tag, $this->mStripList ) ) {
4674                         $this->mStripList[] = $tag;
4675                 }
4676
4677                 return $old;
4678         }
4679
4680         /**
4681          * FIXME: update documentation. makeLinkObj() is deprecated.
4682          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4683          * Placeholders created in Skin::makeLinkObj()
4684          * Returns an array of link CSS classes, indexed by PDBK.
4685          */
4686         function replaceLinkHolders( &$text, $options = 0 ) {
4687                 return $this->mLinkHolders->replace( $text );
4688         }
4689
4690         /**
4691          * Replace <!--LINK--> link placeholders with plain text of links
4692          * (not HTML-formatted).
4693          *
4694          * @param $text String
4695          * @return String
4696          */
4697         function replaceLinkHoldersText( $text ) {
4698                 return $this->mLinkHolders->replaceText( $text );
4699         }
4700
4701         /**
4702          * Renders an image gallery from a text with one line per image.
4703          * text labels may be given by using |-style alternative text. E.g.
4704          *   Image:one.jpg|The number "1"
4705          *   Image:tree.jpg|A tree
4706          * given as text will return the HTML of a gallery with two images,
4707          * labeled 'The number "1"' and
4708          * 'A tree'.
4709          *
4710          * @param string $text
4711          * @param array $param
4712          * @return string HTML
4713          */
4714         function renderImageGallery( $text, $params ) {
4715                 $ig = new ImageGallery();
4716                 $ig->setContextTitle( $this->mTitle );
4717                 $ig->setShowBytes( false );
4718                 $ig->setShowFilename( false );
4719                 $ig->setParser( $this );
4720                 $ig->setHideBadImages();
4721                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4722
4723                 if ( isset( $params['showfilename'] ) ) {
4724                         $ig->setShowFilename( true );
4725                 } else {
4726                         $ig->setShowFilename( false );
4727                 }
4728                 if ( isset( $params['caption'] ) ) {
4729                         $caption = $params['caption'];
4730                         $caption = htmlspecialchars( $caption );
4731                         $caption = $this->replaceInternalLinks( $caption );
4732                         $ig->setCaptionHtml( $caption );
4733                 }
4734                 if ( isset( $params['perrow'] ) ) {
4735                         $ig->setPerRow( $params['perrow'] );
4736                 }
4737                 if ( isset( $params['widths'] ) ) {
4738                         $ig->setWidths( $params['widths'] );
4739                 }
4740                 if ( isset( $params['heights'] ) ) {
4741                         $ig->setHeights( $params['heights'] );
4742                 }
4743
4744                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4745
4746                 $lines = StringUtils::explode( "\n", $text );
4747                 foreach ( $lines as $line ) {
4748                         # match lines like these:
4749                         # Image:someimage.jpg|This is some image
4750                         $matches = array();
4751                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4752                         # Skip empty lines
4753                         if ( count( $matches ) == 0 ) {
4754                                 continue;
4755                         }
4756
4757                         if ( strpos( $matches[0], '%' ) !== false ) {
4758                                 $matches[1] = rawurldecode( $matches[1] );
4759                         }
4760                         $title = Title::newFromText( $matches[1], NS_FILE );
4761                         if ( is_null( $title ) ) {
4762                                 # Bogus title. Ignore these so we don't bomb out later.
4763                                 continue;
4764                         }
4765
4766                         $label = '';
4767                         $alt = '';
4768                         if ( isset( $matches[3] ) ) {
4769                                 // look for an |alt= definition while trying not to break existing
4770                                 // captions with multiple pipes (|) in it, until a more sensible grammar
4771                                 // is defined for images in galleries
4772
4773                                 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
4774                                 $altmatches = StringUtils::explode('|', $matches[3]);
4775                                 $magicWordAlt = MagicWord::get( 'img_alt' );
4776
4777                                 foreach ( $altmatches as $altmatch ) {
4778                                         $match = $magicWordAlt->matchVariableStartToEnd( $altmatch );
4779                                         if ( $match ) {
4780                                                 $alt = $this->stripAltText( $match, false );
4781                                         }
4782                                         else {
4783                                                 // concatenate all other pipes
4784                                                 $label .= '|' . $altmatch;
4785                                         }
4786                                 }
4787                                 // remove the first pipe
4788                                 $label = substr( $label, 1 );
4789                         }
4790
4791                         $ig->add( $title, $label, $alt );
4792                 }
4793                 return $ig->toHTML();
4794         }
4795
4796         function getImageParams( $handler ) {
4797                 if ( $handler ) {
4798                         $handlerClass = get_class( $handler );
4799                 } else {
4800                         $handlerClass = '';
4801                 }
4802                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4803                         # Initialise static lists
4804                         static $internalParamNames = array(
4805                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4806                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4807                                         'bottom', 'text-bottom' ),
4808                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4809                                         'upright', 'border', 'link', 'alt' ),
4810                         );
4811                         static $internalParamMap;
4812                         if ( !$internalParamMap ) {
4813                                 $internalParamMap = array();
4814                                 foreach ( $internalParamNames as $type => $names ) {
4815                                         foreach ( $names as $name ) {
4816                                                 $magicName = str_replace( '-', '_', "img_$name" );
4817                                                 $internalParamMap[$magicName] = array( $type, $name );
4818                                         }
4819                                 }
4820                         }
4821
4822                         # Add handler params
4823                         $paramMap = $internalParamMap;
4824                         if ( $handler ) {
4825                                 $handlerParamMap = $handler->getParamMap();
4826                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4827                                         $paramMap[$magic] = array( 'handler', $paramName );
4828                                 }
4829                         }
4830                         $this->mImageParams[$handlerClass] = $paramMap;
4831                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4832                 }
4833                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4834         }
4835
4836         /**
4837          * Parse image options text and use it to make an image
4838          *
4839          * @param $title Title
4840          * @param $options String
4841          * @param $holders LinkHolderArray
4842          * @return string HTML
4843          */
4844         function makeImage( $title, $options, $holders = false ) {
4845                 # Check if the options text is of the form "options|alt text"
4846                 # Options are:
4847                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4848                 #  * left       no resizing, just left align. label is used for alt= only
4849                 #  * right      same, but right aligned
4850                 #  * none       same, but not aligned
4851                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4852                 #  * center     center the image
4853                 #  * frame      Keep original image size, no magnify-button.
4854                 #  * framed     Same as "frame"
4855                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4856                 #  * upright    reduce width for upright images, rounded to full __0 px
4857                 #  * border     draw a 1px border around the image
4858                 #  * alt        Text for HTML alt attribute (defaults to empty)
4859                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4860                 # vertical-align values (no % or length right now):
4861                 #  * baseline
4862                 #  * sub
4863                 #  * super
4864                 #  * top
4865                 #  * text-top
4866                 #  * middle
4867                 #  * bottom
4868                 #  * text-bottom
4869
4870                 $parts = StringUtils::explode( "|", $options );
4871
4872                 # Give extensions a chance to select the file revision for us
4873                 $time = $sha1 = $descQuery = false;
4874                 wfRunHooks( 'BeforeParserFetchFileAndTitle',
4875                         array( $this, $title, &$time, &$sha1, &$descQuery ) );
4876                 # Fetch and register the file (file title may be different via hooks)
4877                 list( $file, $title ) = $this->fetchFileAndTitle( $title, $time, $sha1 );
4878
4879                 # Get parameter map
4880                 $handler = $file ? $file->getHandler() : false;
4881
4882                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4883
4884                 if ( !$file ) {
4885                         $this->addTrackingCategory( 'broken-file-category' );
4886                 }
4887
4888                 # Process the input parameters
4889                 $caption = '';
4890                 $params = array( 'frame' => array(), 'handler' => array(),
4891                         'horizAlign' => array(), 'vertAlign' => array() );
4892                 foreach ( $parts as $part ) {
4893                         $part = trim( $part );
4894                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4895                         $validated = false;
4896                         if ( isset( $paramMap[$magicName] ) ) {
4897                                 list( $type, $paramName ) = $paramMap[$magicName];
4898
4899                                 # Special case; width and height come in one variable together
4900                                 if ( $type === 'handler' && $paramName === 'width' ) {
4901                                         $m = array();
4902                                         # (bug 13500) In both cases (width/height and width only),
4903                                         # permit trailing "px" for backward compatibility.
4904                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4905                                                 $width = intval( $m[1] );
4906                                                 $height = intval( $m[2] );
4907                                                 if ( $handler->validateParam( 'width', $width ) ) {
4908                                                         $params[$type]['width'] = $width;
4909                                                         $validated = true;
4910                                                 }
4911                                                 if ( $handler->validateParam( 'height', $height ) ) {
4912                                                         $params[$type]['height'] = $height;
4913                                                         $validated = true;
4914                                                 }
4915                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4916                                                 $width = intval( $value );
4917                                                 if ( $handler->validateParam( 'width', $width ) ) {
4918                                                         $params[$type]['width'] = $width;
4919                                                         $validated = true;
4920                                                 }
4921                                         } # else no validation -- bug 13436
4922                                 } else {
4923                                         if ( $type === 'handler' ) {
4924                                                 # Validate handler parameter
4925                                                 $validated = $handler->validateParam( $paramName, $value );
4926                                         } else {
4927                                                 # Validate internal parameters
4928                                                 switch( $paramName ) {
4929                                                 case 'manualthumb':
4930                                                 case 'alt':
4931                                                         # @todo Fixme: possibly check validity here for
4932                                                         # manualthumb? downstream behavior seems odd with
4933                                                         # missing manual thumbs.
4934                                                         $validated = true;
4935                                                         $value = $this->stripAltText( $value, $holders );
4936                                                         break;
4937                                                 case 'link':
4938                                                         $chars = self::EXT_LINK_URL_CLASS;
4939                                                         $prots = $this->mUrlProtocols;
4940                                                         if ( $value === '' ) {
4941                                                                 $paramName = 'no-link';
4942                                                                 $value = true;
4943                                                                 $validated = true;
4944                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4945                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4946                                                                         $paramName = 'link-url';
4947                                                                         $this->mOutput->addExternalLink( $value );
4948                                                                         if ( $this->mOptions->getExternalLinkTarget() ) {
4949                                                                                 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
4950                                                                         }
4951                                                                         $validated = true;
4952                                                                 }
4953                                                         } else {
4954                                                                 $linkTitle = Title::newFromText( $value );
4955                                                                 if ( $linkTitle ) {
4956                                                                         $paramName = 'link-title';
4957                                                                         $value = $linkTitle;
4958                                                                         $this->mOutput->addLink( $linkTitle );
4959                                                                         $validated = true;
4960                                                                 }
4961                                                         }
4962                                                         break;
4963                                                 default:
4964                                                         # Most other things appear to be empty or numeric...
4965                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4966                                                 }
4967                                         }
4968
4969                                         if ( $validated ) {
4970                                                 $params[$type][$paramName] = $value;
4971                                         }
4972                                 }
4973                         }
4974                         if ( !$validated ) {
4975                                 $caption = $part;
4976                         }
4977                 }
4978
4979                 # Process alignment parameters
4980                 if ( $params['horizAlign'] ) {
4981                         $params['frame']['align'] = key( $params['horizAlign'] );
4982                 }
4983                 if ( $params['vertAlign'] ) {
4984                         $params['frame']['valign'] = key( $params['vertAlign'] );
4985                 }
4986
4987                 $params['frame']['caption'] = $caption;
4988
4989                 # Will the image be presented in a frame, with the caption below?
4990                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4991                                                  isset( $params['frame']['framed'] ) ||
4992                                                  isset( $params['frame']['thumbnail'] ) ||
4993                                                  isset( $params['frame']['manualthumb'] );
4994
4995                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4996                 # came to also set the caption, ordinary text after the image -- which
4997                 # makes no sense, because that just repeats the text multiple times in
4998                 # screen readers.  It *also* came to set the title attribute.
4999                 #
5000                 # Now that we have an alt attribute, we should not set the alt text to
5001                 # equal the caption: that's worse than useless, it just repeats the
5002                 # text.  This is the framed/thumbnail case.  If there's no caption, we
5003                 # use the unnamed parameter for alt text as well, just for the time be-
5004                 # ing, if the unnamed param is set and the alt param is not.
5005                 #
5006                 # For the future, we need to figure out if we want to tweak this more,
5007                 # e.g., introducing a title= parameter for the title; ignoring the un-
5008                 # named parameter entirely for images without a caption; adding an ex-
5009                 # plicit caption= parameter and preserving the old magic unnamed para-
5010                 # meter for BC; ...
5011                 if ( $imageIsFramed ) { # Framed image
5012                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5013                                 # No caption or alt text, add the filename as the alt text so
5014                                 # that screen readers at least get some description of the image
5015                                 $params['frame']['alt'] = $title->getText();
5016                         }
5017                         # Do not set $params['frame']['title'] because tooltips don't make sense
5018                         # for framed images
5019                 } else { # Inline image
5020                         if ( !isset( $params['frame']['alt'] ) ) {
5021                                 # No alt text, use the "caption" for the alt text
5022                                 if ( $caption !== '') {
5023                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5024                                 } else {
5025                                         # No caption, fall back to using the filename for the
5026                                         # alt text
5027                                         $params['frame']['alt'] = $title->getText();
5028                                 }
5029                         }
5030                         # Use the "caption" for the tooltip text
5031                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5032                 }
5033
5034                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
5035
5036                 # Linker does the rest
5037                 $ret = Linker::makeImageLink2( $title, $file, $params['frame'], $params['handler'],
5038                         $time, $descQuery, $this->mOptions->getThumbSize() );
5039
5040                 # Give the handler a chance to modify the parser object
5041                 if ( $handler ) {
5042                         $handler->parserTransformHook( $this, $file );
5043                 }
5044
5045                 return $ret;
5046         }
5047
5048         protected function stripAltText( $caption, $holders ) {
5049                 # Strip bad stuff out of the title (tooltip).  We can't just use
5050                 # replaceLinkHoldersText() here, because if this function is called
5051                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5052                 if ( $holders ) {
5053                         $tooltip = $holders->replaceText( $caption );
5054                 } else {
5055                         $tooltip = $this->replaceLinkHoldersText( $caption );
5056                 }
5057
5058                 # make sure there are no placeholders in thumbnail attributes
5059                 # that are later expanded to html- so expand them now and
5060                 # remove the tags
5061                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5062                 $tooltip = Sanitizer::stripAllTags( $tooltip );
5063
5064                 return $tooltip;
5065         }
5066
5067         /**
5068          * Set a flag in the output object indicating that the content is dynamic and
5069          * shouldn't be cached.
5070          */
5071         function disableCache() {
5072                 wfDebug( "Parser output marked as uncacheable.\n" );
5073                 $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
5074                 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5075         }
5076
5077         /**
5078          * Callback from the Sanitizer for expanding items found in HTML attribute
5079          * values, so they can be safely tested and escaped.
5080          *
5081          * @param $text String
5082          * @param $frame PPFrame
5083          * @return String
5084          */
5085         function attributeStripCallback( &$text, $frame = false ) {
5086                 $text = $this->replaceVariables( $text, $frame );
5087                 $text = $this->mStripState->unstripBoth( $text );
5088                 return $text;
5089         }
5090
5091         /**
5092          * Accessor
5093          *
5094          * @return array
5095          */
5096         function getTags() {
5097                 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ) );
5098         }
5099
5100         /**
5101          * Replace transparent tags in $text with the values given by the callbacks.
5102          *
5103          * Transparent tag hooks are like regular XML-style tag hooks, except they
5104          * operate late in the transformation sequence, on HTML instead of wikitext.
5105          */
5106         function replaceTransparentTags( $text ) {
5107                 $matches = array();
5108                 $elements = array_keys( $this->mTransparentTagHooks );
5109                 $text = self::extractTagsAndParams( $elements, $text, $matches, $this->mUniqPrefix );
5110
5111                 foreach ( $matches as $marker => $data ) {
5112                         list( $element, $content, $params, $tag ) = $data;
5113                         $tagName = strtolower( $element );
5114                         if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5115                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
5116                         } else {
5117                                 $output = $tag;
5118                         }
5119                         $this->mStripState->addGeneral( $marker, $output );
5120                 }
5121                 return $text;
5122         }
5123
5124         /**
5125          * Break wikitext input into sections, and either pull or replace
5126          * some particular section's text.
5127          *
5128          * External callers should use the getSection and replaceSection methods.
5129          *
5130          * @param $text String: Page wikitext
5131          * @param $section String: a section identifier string of the form:
5132          *   <flag1> - <flag2> - ... - <section number>
5133          *
5134          * Currently the only recognised flag is "T", which means the target section number
5135          * was derived during a template inclusion parse, in other words this is a template
5136          * section edit link. If no flags are given, it was an ordinary section edit link.
5137          * This flag is required to avoid a section numbering mismatch when a section is
5138          * enclosed by <includeonly> (bug 6563).
5139          *
5140          * The section number 0 pulls the text before the first heading; other numbers will
5141          * pull the given section along with its lower-level subsections. If the section is
5142          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5143          *
5144          * @param $mode String: one of "get" or "replace"
5145          * @param $newText String: replacement text for section data.
5146          * @return String: for "get", the extracted section text.
5147          *                 for "replace", the whole page with the section replaced.
5148          */
5149         private function extractSections( $text, $section, $mode, $newText='' ) {
5150                 global $wgTitle; # not generally used but removes an ugly failure mode
5151                 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5152                 $outText = '';
5153                 $frame = $this->getPreprocessor()->newFrame();
5154
5155                 # Process section extraction flags
5156                 $flags = 0;
5157                 $sectionParts = explode( '-', $section );
5158                 $sectionIndex = array_pop( $sectionParts );
5159                 foreach ( $sectionParts as $part ) {
5160                         if ( $part === 'T' ) {
5161                                 $flags |= self::PTD_FOR_INCLUSION;
5162                         }
5163                 }
5164                 # Preprocess the text
5165                 $root = $this->preprocessToDom( $text, $flags );
5166
5167                 # <h> nodes indicate section breaks
5168                 # They can only occur at the top level, so we can find them by iterating the root's children
5169                 $node = $root->getFirstChild();
5170
5171                 # Find the target section
5172                 if ( $sectionIndex == 0 ) {
5173                         # Section zero doesn't nest, level=big
5174                         $targetLevel = 1000;
5175                         if ( !$node ) {
5176                                 # The page definitely exists - we checked that earlier - so it must be blank: see bug #14005
5177                                 return $text;
5178                         }
5179                 } else {
5180                         while ( $node ) {
5181                                 if ( $node->getName() === 'h' ) {
5182                                         $bits = $node->splitHeading();
5183                                         if ( $bits['i'] == $sectionIndex ) {
5184                                                 $targetLevel = $bits['level'];
5185                                                 break;
5186                                         }
5187                                 }
5188                                 if ( $mode === 'replace' ) {
5189                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5190                                 }
5191                                 $node = $node->getNextSibling();
5192                         }
5193                 }
5194
5195                 if ( !$node ) {
5196                         # Not found
5197                         if ( $mode === 'get' ) {
5198                                 return $newText;
5199                         } else {
5200                                 return $text;
5201                         }
5202                 }
5203
5204                 # Find the end of the section, including nested sections
5205                 do {
5206                         if ( $node->getName() === 'h' ) {
5207                                 $bits = $node->splitHeading();
5208                                 $curLevel = $bits['level'];
5209                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5210                                         break;
5211                                 }
5212                         }
5213                         if ( $mode === 'get' ) {
5214                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5215                         }
5216                         $node = $node->getNextSibling();
5217                 } while ( $node );
5218
5219                 # Write out the remainder (in replace mode only)
5220                 if ( $mode === 'replace' ) {
5221                         # Output the replacement text
5222                         # Add two newlines on -- trailing whitespace in $newText is conventionally
5223                         # stripped by the editor, so we need both newlines to restore the paragraph gap
5224                         # Only add trailing whitespace if there is newText
5225                         if ( $newText != "" ) {
5226                                 $outText .= $newText . "\n\n";
5227                         }
5228
5229                         while ( $node ) {
5230                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5231                                 $node = $node->getNextSibling();
5232                         }
5233                 }
5234
5235                 if ( is_string( $outText ) ) {
5236                         # Re-insert stripped tags
5237                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5238                 }
5239
5240                 return $outText;
5241         }
5242
5243         /**
5244          * This function returns the text of a section, specified by a number ($section).
5245          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5246          * the first section before any such heading (section 0).
5247          *
5248          * If a section contains subsections, these are also returned.
5249          *
5250          * @param $text String: text to look in
5251          * @param $section String: section identifier
5252          * @param $deftext String: default to return if section is not found
5253          * @return string text of the requested section
5254          */
5255         public function getSection( $text, $section, $deftext='' ) {
5256                 return $this->extractSections( $text, $section, "get", $deftext );
5257         }
5258
5259         /**
5260          * This function returns $oldtext after the content of the section
5261          * specified by $section has been replaced with $text.
5262          *
5263          * @param $oldtext String: former text of the article
5264          * @param $section Numeric: section identifier
5265          * @param $text String: replacing text
5266          * @return String: modified text
5267          */
5268         public function replaceSection( $oldtext, $section, $text ) {
5269                 return $this->extractSections( $oldtext, $section, "replace", $text );
5270         }
5271
5272         /**
5273          * Get the ID of the revision we are parsing
5274          *
5275          * @return Mixed: integer or null
5276          */
5277         function getRevisionId() {
5278                 return $this->mRevisionId;
5279         }
5280
5281         /**
5282          * Get the revision object for $this->mRevisionId
5283          *
5284          * @return Revision|null either a Revision object or null
5285          */
5286         protected function getRevisionObject() {
5287                 if ( !is_null( $this->mRevisionObject ) ) {
5288                         return $this->mRevisionObject;
5289                 }
5290                 if ( is_null( $this->mRevisionId ) ) {
5291                         return null;
5292                 }
5293
5294                 $this->mRevisionObject = Revision::newFromId( $this->mRevisionId );
5295                 return $this->mRevisionObject;
5296         }
5297
5298         /**
5299          * Get the timestamp associated with the current revision, adjusted for
5300          * the default server-local timestamp
5301          */
5302         function getRevisionTimestamp() {
5303                 if ( is_null( $this->mRevisionTimestamp ) ) {
5304                         wfProfileIn( __METHOD__ );
5305
5306                         $revObject = $this->getRevisionObject();
5307                         $timestamp = $revObject ? $revObject->getTimestamp() : false;
5308
5309                         if( $timestamp !== false ) {
5310                                 global $wgContLang;
5311
5312                                 # The cryptic '' timezone parameter tells to use the site-default
5313                                 # timezone offset instead of the user settings.
5314                                 #
5315                                 # Since this value will be saved into the parser cache, served
5316                                 # to other users, and potentially even used inside links and such,
5317                                 # it needs to be consistent for all visitors.
5318                                 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5319                         }
5320
5321                         wfProfileOut( __METHOD__ );
5322                 }
5323                 return $this->mRevisionTimestamp;
5324         }
5325
5326         /**
5327          * Get the name of the user that edited the last revision
5328          *
5329          * @return String: user name
5330          */
5331         function getRevisionUser() {
5332                 if( is_null( $this->mRevisionUser ) ) {
5333                         $revObject = $this->getRevisionObject();
5334
5335                         # if this template is subst: the revision id will be blank,
5336                         # so just use the current user's name
5337                         if( $revObject ) {
5338                                 $this->mRevisionUser = $revObject->getUserText();
5339                         } elseif( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5340                                 $this->mRevisionUser = $this->getUser()->getName();
5341                         }
5342                 }
5343                 return $this->mRevisionUser;
5344         }
5345
5346         /**
5347          * Mutator for $mDefaultSort
5348          *
5349          * @param $sort New value
5350          */
5351         public function setDefaultSort( $sort ) {
5352                 $this->mDefaultSort = $sort;
5353                 $this->mOutput->setProperty( 'defaultsort', $sort );
5354         }
5355
5356         /**
5357          * Accessor for $mDefaultSort
5358          * Will use the empty string if none is set.
5359          *
5360          * This value is treated as a prefix, so the
5361          * empty string is equivalent to sorting by
5362          * page name.
5363          *
5364          * @return string
5365          */
5366         public function getDefaultSort() {
5367                 if ( $this->mDefaultSort !== false ) {
5368                         return $this->mDefaultSort;
5369                 } else {
5370                         return '';
5371                 }
5372         }
5373
5374         /**
5375          * Accessor for $mDefaultSort
5376          * Unlike getDefaultSort(), will return false if none is set
5377          *
5378          * @return string or false
5379          */
5380         public function getCustomDefaultSort() {
5381                 return $this->mDefaultSort;
5382         }
5383
5384         /**
5385          * Try to guess the section anchor name based on a wikitext fragment
5386          * presumably extracted from a heading, for example "Header" from
5387          * "== Header ==".
5388          */
5389         public function guessSectionNameFromWikiText( $text ) {
5390                 # Strip out wikitext links(they break the anchor)
5391                 $text = $this->stripSectionName( $text );
5392                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5393                 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5394         }
5395
5396         /**
5397          * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5398          * instead.  For use in redirects, since IE6 interprets Redirect: headers
5399          * as something other than UTF-8 (apparently?), resulting in breakage.
5400          *
5401          * @param $text String: The section name
5402          * @return string An anchor
5403          */
5404         public function guessLegacySectionNameFromWikiText( $text ) {
5405                 # Strip out wikitext links(they break the anchor)
5406                 $text = $this->stripSectionName( $text );
5407                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5408                 return '#' . Sanitizer::escapeId( $text, array( 'noninitial', 'legacy' ) );
5409         }
5410
5411         /**
5412          * Strips a text string of wikitext for use in a section anchor
5413          *
5414          * Accepts a text string and then removes all wikitext from the
5415          * string and leaves only the resultant text (i.e. the result of
5416          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5417          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5418          * to create valid section anchors by mimicing the output of the
5419          * parser when headings are parsed.
5420          *
5421          * @param $text String: text string to be stripped of wikitext
5422          * for use in a Section anchor
5423          * @return Filtered text string
5424          */
5425         public function stripSectionName( $text ) {
5426                 # Strip internal link markup
5427                 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5428                 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5429
5430                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5431                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5432                 # on how many empty links there are on the page - need to figure that out.
5433                 $text = preg_replace( '/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5434
5435                 # Parse wikitext quotes (italics & bold)
5436                 $text = $this->doQuotes( $text );
5437
5438                 # Strip HTML tags
5439                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5440                 return $text;
5441         }
5442
5443         /**
5444          * strip/replaceVariables/unstrip for preprocessor regression testing
5445          *
5446          * @return string
5447          */
5448         function testSrvus( $text, $title, ParserOptions $options, $outputType = self::OT_HTML ) {
5449                 if ( !$title instanceof Title ) {
5450                         $title = Title::newFromText( $title );
5451                 }
5452                 $this->startParse( $title, $options, $outputType, true );
5453
5454                 $text = $this->replaceVariables( $text );
5455                 $text = $this->mStripState->unstripBoth( $text );
5456                 $text = Sanitizer::removeHTMLtags( $text );
5457                 return $text;
5458         }
5459
5460         function testPst( $text, $title, $options ) {
5461                 global $wgUser;
5462                 if ( !$title instanceof Title ) {
5463                         $title = Title::newFromText( $title );
5464                 }
5465                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5466         }
5467
5468         function testPreprocess( $text, $title, $options ) {
5469                 if ( !$title instanceof Title ) {
5470                         $title = Title::newFromText( $title );
5471                 }
5472                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5473         }
5474
5475         /**
5476          * Call a callback function on all regions of the given text that are not
5477          * inside strip markers, and replace those regions with the return value
5478          * of the callback. For example, with input:
5479          *
5480          *  aaa<MARKER>bbb
5481          *
5482          * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5483          * two strings will be replaced with the value returned by the callback in
5484          * each case.
5485          *
5486          * @return string
5487          */
5488         function markerSkipCallback( $s, $callback ) {
5489                 $i = 0;
5490                 $out = '';
5491                 while ( $i < strlen( $s ) ) {
5492                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5493                         if ( $markerStart === false ) {
5494                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5495                                 break;
5496                         } else {
5497                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5498                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5499                                 if ( $markerEnd === false ) {
5500                                         $out .= substr( $s, $markerStart );
5501                                         break;
5502                                 } else {
5503                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5504                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5505                                         $i = $markerEnd;
5506                                 }
5507                         }
5508                 }
5509                 return $out;
5510         }
5511
5512         /**
5513          * Save the parser state required to convert the given half-parsed text to
5514          * HTML. "Half-parsed" in this context means the output of
5515          * recursiveTagParse() or internalParse(). This output has strip markers
5516          * from replaceVariables (extensionSubstitution() etc.), and link
5517          * placeholders from replaceLinkHolders().
5518          *
5519          * Returns an array which can be serialized and stored persistently. This
5520          * array can later be loaded into another parser instance with
5521          * unserializeHalfParsedText(). The text can then be safely incorporated into
5522          * the return value of a parser hook.
5523          *
5524          * @return array
5525          */
5526         function serializeHalfParsedText( $text ) {
5527                 wfProfileIn( __METHOD__ );
5528                 $data = array(
5529                         'text' => $text,
5530                         'version' => self::HALF_PARSED_VERSION,
5531                         'stripState' => $this->mStripState->getSubState( $text ),
5532                         'linkHolders' => $this->mLinkHolders->getSubArray( $text )
5533                 );
5534                 wfProfileOut( __METHOD__ );
5535                 return $data;
5536         }
5537
5538         /**
5539          * Load the parser state given in the $data array, which is assumed to
5540          * have been generated by serializeHalfParsedText(). The text contents is
5541          * extracted from the array, and its markers are transformed into markers
5542          * appropriate for the current Parser instance. This transformed text is
5543          * returned, and can be safely included in the return value of a parser
5544          * hook.
5545          *
5546          * If the $data array has been stored persistently, the caller should first
5547          * check whether it is still valid, by calling isValidHalfParsedText().
5548          *
5549          * @param $data Serialized data
5550          * @return String
5551          */
5552         function unserializeHalfParsedText( $data ) {
5553                 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
5554                         throw new MWException( __METHOD__.': invalid version' );
5555                 }
5556
5557                 # First, extract the strip state.
5558                 $texts = array( $data['text'] );
5559                 $texts = $this->mStripState->merge( $data['stripState'], $texts );
5560
5561                 # Now renumber links
5562                 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
5563
5564                 # Should be good to go.
5565                 return $texts[0];
5566         }
5567
5568         /**
5569          * Returns true if the given array, presumed to be generated by
5570          * serializeHalfParsedText(), is compatible with the current version of the
5571          * parser.
5572          *
5573          * @param $data Array
5574          *
5575          * @return bool
5576          */
5577         function isValidHalfParsedText( $data ) {
5578                 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
5579         }
5580 }