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          * @static
 687          */
 688         public function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
 689                 static $n = 1;
 690                 $stripped = '';
 691                 $matches = array();
 692
 693                 $taglist = implode( '|', $elements );
 694                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 695
 696                 while ( $text != '' ) {
 697                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 698                         $stripped .= $p[0];
 699                         if ( count( $p ) < 5 ) {
 700                                 break;
 701                         }
 702                         if ( count( $p ) > 5 ) {
 703                                 # comment
 704                                 $element    = $p[4];
 705                                 $attributes = '';
 706                                 $close      = '';
 707                                 $inside     = $p[5];
 708                         } else {
 709                                 # tag
 710                                 $element    = $p[1];
 711                                 $attributes = $p[2];
 712                                 $close      = $p[3];
 713                                 $inside     = $p[4];
 714                         }
 715
 716                         $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
 717                         $stripped .= $marker;
 718
 719                         if ( $close === '/>' ) {
 720                                 # Empty element tag, <tag />
 721                                 $content = null;
 722                                 $text = $inside;
 723                                 $tail = null;
 724                         } else {
 725                                 if ( $element === '!--' ) {
 726                                         $end = '/(-->)/';
 727                                 } else {
 728                                         $end = "/(<\\/$element\\s*>)/i";
 729                                 }
 730                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 731                                 $content = $q[0];
 732                                 if ( count( $q ) < 3 ) {
 733                                         # No end tag -- let it run out to the end of the text.
 734                                         $tail = '';
 735                                         $text = '';
 736                                 } else {
 737                                         $tail = $q[1];
 738                                         $text = $q[2];
 739                                 }
 740                         }
 741
 742                         $matches[$marker] = array( $element,
 743                                 $content,
 744                                 Sanitizer::decodeTagAttributes( $attributes ),
 745                                 "<$element$attributes$close$content$tail" );
 746                 }
 747                 return $stripped;
 748         }
 749
 750         /**
 751          * Get a list of strippable XML-like elements
 752          */
 753         function getStripList() {
 754                 return $this->mStripList;
 755         }
 756
 757         /**
 758          * @deprecated use replaceVariables
 759          */
 760         function strip( $text, $state, $stripcomments = false , $dontstrip = array() ) {
 761                 return $text;
 762         }
 763
 764         /**
 765          * Restores pre, math, and other extensions removed by strip()
 766          *
 767          * always call unstripNoWiki() after this one
 768          * @private
 769          * @deprecated use $this->mStripState->unstrip()
 770          */
 771         function unstrip( $text, $state ) {
 772                 return $state->unstripGeneral( $text );
 773         }
 774
 775         /**
 776          * Always call this after unstrip() to preserve the order
 777          *
 778          * @private
 779          * @deprecated use $this->mStripState->unstrip()
 780          */
 781         function unstripNoWiki( $text, $state ) {
 782                 return $state->unstripNoWiki( $text );
 783         }
 784
 785         /**
 786          * @deprecated use $this->mStripState->unstripBoth()
 787          */
 788         function unstripForHTML( $text ) {
 789                 return $this->mStripState->unstripBoth( $text );
 790         }
 791
 792         /**
 793          * Add an item to the strip state
 794          * Returns the unique tag which must be inserted into the stripped text
 795          * The tag will be replaced with the original text in unstrip()
 796          *
 797          * @private
 798          */
 799         function insertStripItem( $text ) {
 800                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 801                 $this->mMarkerIndex++;
 802                 $this->mStripState->addGeneral( $rnd, $text );
 803                 return $rnd;
 804         }
 805
 806         /**
 807          * Interface with html tidy
 808          * @deprecated Use MWTidy::tidy()
 809          */
 810         public static function tidy( $text ) {
 811                 wfDeprecated( __METHOD__ );
 812                 return MWTidy::tidy( $text );
 813         }
 814
 815         /**
 816          * parse the wiki syntax used to render tables
 817          *
 818          * @private
 819          */
 820         function doTableStuff( $text ) {
 821                 wfProfileIn( __METHOD__ );
 822
 823                 $lines = StringUtils::explode( "\n", $text );
 824                 $out = '';
 825                 $output =& $out;
 826
 827                 foreach ( $lines as $outLine ) {
 828                         $line = trim( $outLine );
 829
 830                         # empty line, go to next line,
 831                         # but only append \n if outside of table
 832                         if ( $line === '') {
 833                                 $output .= $outLine . "\n";
 834                                 continue;
 835                         }
 836                         $firstChars = $line[0];
 837                         if ( strlen( $line ) > 1 ) {
 838                                 $firstChars .= in_array( $line[1], array( '}', '+', '-' ) ) ? $line[1] : '';
 839                         }
 840                         $matches = array();
 841
 842                         if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line , $matches ) ) {
 843                                 $tables[] = array();
 844                                 $table =& $this->last( $tables );
 845                                 $table[0] = array(); // first row
 846                                 $currentRow =& $table[0];
 847                                 $table['indent'] = strlen( $matches[1] );
 848
 849                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 850                                 $attributes = Sanitizer::fixTagAttributes( $attributes , 'table' );
 851
 852                                 if ( $attributes !== '' ) {
 853                                         $table['attributes'] = $attributes;
 854                                 }
 855                         } else if ( !isset( $tables[0] ) ) {
 856                                 // we're outside the table
 857
 858                                 $out .= $outLine . "\n";
 859                         } else if ( $firstChars === '|}' ) {
 860                                 // trim the |} code from the line
 861                                 $line = substr ( $line , 2 );
 862
 863                                 // Shorthand for last row
 864                                 $lastRow =& $this->last( $table );
 865
 866                                 // a thead at the end becomes a tfoot, unless there is only one row
 867                                 // Do this before deleting empty last lines to allow headers at the bottom of tables
 868                                 if ( isset( $lastRow['type'] ) && $lastRow['type'] == 'thead' && isset( $table[1] ) ) {
 869                                         $lastRow['type'] = 'tfoot';
 870                                         for ( $i = 0; isset( $lastRow[$i] ); $i++ ) {
 871                                                 $lastRow[$i]['type'] = 'th';
 872                                         }
 873                                 }
 874
 875                                 // Delete empty last lines
 876                                 if ( empty( $lastRow ) ) {
 877                                         $lastRow = NULL;
 878                                 }
 879                                 $o = '';
 880                                 $curtable = array_pop( $tables );
 881
 882                                 #Add a line-ending before the table, but only if there isn't one already
 883                                 if ( substr( $out, -1 ) !== "\n" ) {
 884                                         $o .= "\n";
 885                                 }
 886                                 $o .= $this->generateTableHTML( $curtable ) . $line . "\n";
 887
 888                                 if ( count( $tables ) > 0 ) {
 889                                         $table =& $this->last( $tables );
 890                                         $currentRow =& $this->last( $table );
 891                                         $currentElement =& $this->last( $currentRow );
 892
 893                                         $output =& $currentElement['content'];
 894                                 } else {
 895                                         $output =& $out;
 896                                 }
 897
 898                                 $output .= $o;
 899
 900                         } else if ( $firstChars === '|-' ) {
 901                                 // start a new row element
 902                                 // but only when we haven't started one already
 903                                 if ( count( $currentRow ) != 0 ) {
 904                                         $table[] = array();
 905                                         $currentRow =& $this->last( $table );
 906                                 }
 907                                 // Get the attributes, there's nothing else useful in $line now
 908                                 $line = substr ( $line , 2 );
 909                                 $attributes = $this->mStripState->unstripBoth( $line );
 910                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 911                                 if ( $attributes !== '' ) {
 912                                         $currentRow['attributes'] = $attributes;
 913                                 }
 914
 915                         } else if ( $firstChars  === '|+' ) {
 916                                 // a table caption, but only proceed if there isn't one already
 917                                 if ( !isset ( $table['caption'] ) ) {
 918                                         $line = substr ( $line , 2 );
 919
 920                                         $c = $this->getCellAttr( $line , 'caption' );
 921                                         $table['caption'] = array();
 922                                         $table['caption']['content'] = $c[0];
 923                                         if ( isset( $c[1] ) ) $table['caption']['attributes'] = $c[1];
 924                                         unset( $c );
 925                                         $output =& $table['caption']['content'];
 926                                 }
 927                         } else if ( $firstChars === '|' || $firstChars === '!' || $firstChars === '!+' ) {
 928                                 // Which kind of cells are we dealing with
 929                                 $currentTag = 'td';
 930                                 $line = substr ( $line , 1 );
 931
 932                                 if ( $firstChars === '!'  || $firstChars === '!+' ) {
 933                                         $line = str_replace ( '!!' , '||' , $line );
 934                                         $currentTag = 'th';
 935                                 }
 936
 937                                 // Split up multiple cells on the same line.
 938                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 939                                 $line = ''; // save memory
 940
 941                                 // decide whether thead to tbody
 942                                 if ( !array_key_exists( 'type', $currentRow ) ) {
 943                                         $currentRow['type'] = ( $firstChars === '!' ) ? 'thead' : 'tbody' ;
 944                                 } else if ( $firstChars === '|' ) {
 945                                         $currentRow['type'] = 'tbody';
 946                                 }
 947
 948                                 // Loop through each table cell
 949                                 foreach ( $cells as $cell ) {
 950                                         // a new cell
 951                                         $currentRow[] = array();
 952                                         $currentElement =& $this->last( $currentRow );
 953
 954                                         $currentElement['type'] = $currentTag;
 955
 956                                         $c = $this->getCellAttr( $cell , $currentTag );
 957                                         $currentElement['content'] = $c[0];
 958                                         if ( isset( $c[1] ) ) $currentElement['attributes'] = $c[1];
 959                                         unset( $c );
 960                                 }
 961                                 $output =& $currentElement['content'];
 962
 963                         } else {
 964                                 $output .= "\n$outLine";
 965                         }
 966                 }
 967
 968                 # Remove trailing line-ending (b/c)
 969                 if ( substr( $out, -1 ) === "\n" ) {
 970                         $out = substr( $out, 0, -1 );
 971                 }
 972
 973                 # Close any unclosed tables
 974                 if ( isset( $tables ) && count( $tables ) > 0 ) {
 975                         for ( $i = 0; $i < count( $tables ); $i++ ) {
 976                                 $curtable = array_pop( $tables );
 977                                 $curtable = $this->generateTableHTML( $curtable );
 978                                 #Add a line-ending before the table, but only if there isn't one already
 979                                 if ( substr( $out, -1 ) !== "\n"  && $curtable !== "" ) {
 980                                         $out .= "\n";
 981                                 }
 982                                 $out .= $curtable;
 983                         }
 984                 }
 985
 986                 wfProfileOut( __METHOD__ );
 987
 988                 return $out;
 989         }
 990
 991         /**
 992          * Helper function for doTableStuff() separating the contents of cells from
 993          * attributes. Particularly useful as there's a possible bug and this action
 994          * is repeated twice.
 995          *
 996          * @private
 997          */
 998         function getCellAttr ( $cell, $tagName ) {
 999                 $content = null;
1000                 $attributes = null;
1001
1002                 $cell = trim ( $cell );
1003
1004                 // A cell could contain both parameters and data
1005                 $cellData = explode ( '|' , $cell , 2 );
1006
1007                 // Bug 553: Note that a '|' inside an invalid link should not
1008                 // be mistaken as delimiting cell parameters
1009                 if ( strpos( $cellData[0], '[[' ) !== false ) {
1010                         $content = trim ( $cell );
1011                 }
1012                 else if ( count ( $cellData ) == 1 ) {
1013                         $content = trim ( $cellData[0] );
1014                 }
1015                 else {
1016                         $attributes = $this->mStripState->unstripBoth( $cellData[0] );
1017                         $attributes = Sanitizer::fixTagAttributes( $attributes , $tagName );
1018
1019                         $content = trim ( $cellData[1] );
1020                 }
1021                 return array( $content, $attributes );
1022         }
1023
1024
1025         /**
1026          * Helper function for doTableStuff(). This converts the structured array into html.
1027          *
1028          * @private
1029          */
1030         function generateTableHTML ( &$table ) {
1031                 $return = "";
1032                 $return .= str_repeat( '<dl><dd>' , $table['indent'] );
1033                 $return .= '<table';
1034                 $return .= isset( $table['attributes'] ) ? $table['attributes'] : '';
1035                 $return .= '>';
1036                 unset( $table['attributes'] );
1037
1038                 if ( isset( $table['caption'] ) ) {
1039                         $return .= "\n<caption";
1040                         $return .= isset( $table['caption']['attributes'] ) ? $table['caption']['attributes'] : '';
1041                         $return .= '>';
1042                         $return .= $table['caption']['content'];
1043                         $return .= "\n</caption>";
1044                 }
1045                 $lastSection = '';
1046                 $empty = true;
1047                 $simple = true;
1048
1049                 // If we only have tbodies, mark table as simple
1050                 for ( $i = 0; isset( $table[$i] ); $i++ ) {
1051                         if ( !count( $table[$i] ) ) continue;
1052                         if ( !isset( $table[$i]['type'] ) ) $table[$i]['type'] = 'tbody';
1053                         if ( !$lastSection ) {
1054                                 $lastSection = $table[$i]['type'];
1055                         } else if ( $lastSection != $table[$i]['type'] ) {
1056                                 $simple = false;
1057                                 break;
1058                         }
1059                 }
1060                 $lastSection = '';
1061                 for ( $i = 0; isset( $table[$i] ); $i++ ) {
1062                         // Check for empty tables
1063                         if ( count( $table[$i] ) ) {
1064                                 $empty = false;
1065                         } else {
1066                                 continue;
1067                         }
1068                         if ( $table[$i]['type'] != $lastSection && !$simple ) {
1069                                 $return .= "\n<" . $table[$i]['type'] . '>';
1070                         }
1071
1072                         $return .= "\n<tr";
1073                         $return .= isset( $table[$i]['attributes'] ) ? $table[$i]['attributes'] : '';
1074                         $return .= '>';
1075                         for ( $j = 0; isset( $table[$i][$j] ); $j++ ) {
1076                                 if ( !isset( $table[$i][$j]['type'] ) ) $table[$i][$j]['type'] = 'td';
1077                                 $return .= "\n<" . $table[$i][$j]['type'];
1078                                 $return .= isset( $table[$i][$j]['attributes'] ) ? $table[$i][$j]['attributes'] : '';
1079                                 $return .= '>';
1080
1081                                 $return .= $table[$i][$j]['content'];
1082                                 if ( $table[$i][$j]['content'] != '' )
1083                                         $return .= "\n";
1084
1085                                 $return .= '</' . $table[$i][$j]['type'] . '>';
1086                                 unset( $table[$i][$j] );
1087                         }
1088                         $return .= "\n</tr>";
1089
1090                         if ( ( !isset( $table[$i + 1] ) && !$simple ) || ( isset( $table[$i + 1] ) && ( $table[$i]['type'] != $table[$i + 1]['type'] ) ) ) {
1091                                 $return .= '</' . $table[$i]['type'] . '>';
1092                         }
1093                         $lastSection = $table[$i]['type'];
1094                         unset( $table[$i] );
1095                 }
1096                 if ( $empty ) {
1097                         if ( isset( $table['caption'] ) ) {
1098                                 $return .= "\n<tr><td></td></tr>";
1099                         } else {
1100                                 return '';
1101                         }
1102                 }
1103                 $return .= "\n</table>";
1104                 $return .= str_repeat( '</dd></dl>' , $table['indent'] );
1105
1106                 return $return;
1107         }
1108
1109         /**
1110          * like end() but only works on the numeric array index and php's internal pointers
1111          * returns a reference to the last element of an array much like "\$arr[-1]" in perl
1112          * ignores associative elements and will create a 0 key will a NULL value if there were
1113          * no numric elements and an array itself if not previously defined.
1114          *
1115          * @private
1116          */
1117         function &last ( &$arr ) {
1118                 for ( $i = count( $arr ); ( !isset( $arr[$i] ) && $i > 0 ); $i-- ) {  }
1119                 return $arr[$i];
1120         }
1121
1122         /**
1123          * Helper function for parse() that transforms wiki markup into
1124          * HTML. Only called for $mOutputType == self::OT_HTML.
1125          *
1126          * @private
1127          */
1128         function internalParse( $text, $isMain = true, $frame=false ) {
1129                 wfProfileIn( __METHOD__ );
1130
1131                 $origText = $text;
1132
1133                 # Hook to suspend the parser in this state
1134                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1135                         wfProfileOut( __METHOD__ );
1136                         return $text ;
1137                 }
1138
1139                 # if $frame is provided, then use $frame for replacing any variables
1140                 if ( $frame ) {
1141                         # use frame depth to infer how include/noinclude tags should be handled
1142                         # depth=0 means this is the top-level document; otherwise it's an included document
1143                         if ( !$frame->depth ) {
1144                                 $flag = 0;
1145                         } else {
1146                                 $flag = Parser::PTD_FOR_INCLUSION;
1147                         }
1148                         $dom = $this->preprocessToDom( $text, $flag );
1149                         $text = $frame->expand( $dom );
1150                 } else {
1151                         # if $frame is not provided, then use old-style replaceVariables
1152                         $text = $this->replaceVariables( $text );
1153                 }
1154
1155                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
1156                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1157
1158                 # Tables need to come after variable replacement for things to work
1159                 # properly; putting them before other transformations should keep
1160                 # exciting things like link expansions from showing up in surprising
1161                 # places.
1162                 $text = $this->doTableStuff( $text );
1163
1164                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1165
1166                 $text = $this->doDoubleUnderscore( $text );
1167
1168                 $text = $this->doHeadings( $text );
1169                 if ( $this->mOptions->getUseDynamicDates() ) {
1170                         $df = DateFormatter::getInstance();
1171                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1172                 }
1173                 $text = $this->replaceInternalLinks( $text );
1174                 $text = $this->doAllQuotes( $text );
1175                 $text = $this->replaceExternalLinks( $text );
1176
1177                 # replaceInternalLinks may sometimes leave behind
1178                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1179                 $text = str_replace( $this->mUniqPrefix.'NOPARSE', '', $text );
1180
1181                 $text = $this->doMagicLinks( $text );
1182                 $text = $this->formatHeadings( $text, $origText, $isMain );
1183
1184                 wfProfileOut( __METHOD__ );
1185                 return $text;
1186         }
1187
1188         /**
1189          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1190          * magic external links.
1191          *
1192          * DML
1193          * @private
1194          */
1195         function doMagicLinks( $text ) {
1196                 wfProfileIn( __METHOD__ );
1197                 $prots = $this->mUrlProtocols;
1198                 $urlChar = self::EXT_LINK_URL_CLASS;
1199                 $text = preg_replace_callback(
1200                         '!(?:                           # Start cases
1201                                 (<a[ \t\r\n>].*?</a>) |     # m[1]: Skip link text
1202                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1203                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
1204                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1205                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1206                                         (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1207                                         (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1208                                         [0-9Xx]                 # check digit
1209                                         \b)
1210                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1211                 wfProfileOut( __METHOD__ );
1212                 return $text;
1213         }
1214
1215         function magicLinkCallback( $m ) {
1216                 if ( isset( $m[1] ) && $m[1] !== '' ) {
1217                         # Skip anchor
1218                         return $m[0];
1219                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1220                         # Skip HTML element
1221                         return $m[0];
1222                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1223                         # Free external link
1224                         return $this->makeFreeExternalLink( $m[0] );
1225                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1226                         # RFC or PMID
1227                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1228                                 $keyword = 'RFC';
1229                                 $urlmsg = 'rfcurl';
1230                                 $CssClass = 'mw-magiclink-rfc';
1231                                 $id = $m[4];
1232                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1233                                 $keyword = 'PMID';
1234                                 $urlmsg = 'pubmedurl';
1235                                 $CssClass = 'mw-magiclink-pmid';
1236                                 $id = $m[4];
1237                         } else {
1238                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1239                                         substr( $m[0], 0, 20 ) . '"' );
1240                         }
1241                         $url = wfMsgForContent( $urlmsg, $id );
1242                         return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $CssClass );
1243                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1244                         # ISBN
1245                         $isbn = $m[5];
1246                         $num = strtr( $isbn, array(
1247                                 '-' => '',
1248                                 ' ' => '',
1249                                 'x' => 'X',
1250                         ));
1251                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1252                         return'<a href="' .
1253                                 $titleObj->escapeLocalUrl() .
1254                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1255                 } else {
1256                         return $m[0];
1257                 }
1258         }
1259
1260         /**
1261          * Make a free external link, given a user-supplied URL
1262          * @return HTML
1263          * @private
1264          */
1265         function makeFreeExternalLink( $url ) {
1266                 global $wgContLang;
1267                 wfProfileIn( __METHOD__ );
1268
1269                 $trail = '';
1270
1271                 # The characters '<' and '>' (which were escaped by
1272                 # removeHTMLtags()) should not be included in
1273                 # URLs, per RFC 2396.
1274                 $m2 = array();
1275                 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1276                         $trail = substr( $url, $m2[0][1] ) . $trail;
1277                         $url = substr( $url, 0, $m2[0][1] );
1278                 }
1279
1280                 # Move trailing punctuation to $trail
1281                 $sep = ',;\.:!?';
1282                 # If there is no left bracket, then consider right brackets fair game too
1283                 if ( strpos( $url, '(' ) === false ) {
1284                         $sep .= ')';
1285                 }
1286
1287                 $numSepChars = strspn( strrev( $url ), $sep );
1288                 if ( $numSepChars ) {
1289                         $trail = substr( $url, -$numSepChars ) . $trail;
1290                         $url = substr( $url, 0, -$numSepChars );
1291                 }
1292
1293                 $url = Sanitizer::cleanUrl( $url );
1294
1295                 # Is this an external image?
1296                 $text = $this->maybeMakeExternalImage( $url );
1297                 if ( $text === false ) {
1298                         # Not an image, make a link
1299                         $text = Linker::makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1300                                 $this->getExternalLinkAttribs( $url ) );
1301                         # Register it in the output object...
1302                         # Replace unnecessary URL escape codes with their equivalent characters
1303                         $pasteurized = self::replaceUnusualEscapes( $url );
1304                         $this->mOutput->addExternalLink( $pasteurized );
1305                 }
1306                 wfProfileOut( __METHOD__ );
1307                 return $text . $trail;
1308         }
1309
1310
1311         /**
1312          * Parse headers and return html
1313          *
1314          * @private
1315          */
1316         function doHeadings( $text ) {
1317                 wfProfileIn( __METHOD__ );
1318                 for ( $i = 6; $i >= 1; --$i ) {
1319                         $h = str_repeat( '=', $i );
1320                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1321                           "<h$i>\\1</h$i>", $text );
1322                 }
1323                 wfProfileOut( __METHOD__ );
1324                 return $text;
1325         }
1326
1327         /**
1328          * Replace single quotes with HTML markup
1329          * @private
1330          * @return string the altered text
1331          */
1332         function doAllQuotes( $text ) {
1333                 wfProfileIn( __METHOD__ );
1334                 $outtext = '';
1335                 $lines = StringUtils::explode( "\n", $text );
1336                 foreach ( $lines as $line ) {
1337                         $outtext .= $this->doQuotes( $line ) . "\n";
1338                 }
1339                 $outtext = substr( $outtext, 0,-1 );
1340                 wfProfileOut( __METHOD__ );
1341                 return $outtext;
1342         }
1343
1344         /**
1345          * Helper function for doAllQuotes()
1346          */
1347         public function doQuotes( $text ) {
1348                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1349                 if ( count( $arr ) == 1 ) {
1350                         return $text;
1351                 } else {
1352                         # First, do some preliminary work. This may shift some apostrophes from
1353                         # being mark-up to being text. It also counts the number of occurrences
1354                         # of bold and italics mark-ups.
1355                         $numbold = 0;
1356                         $numitalics = 0;
1357                         for ( $i = 0; $i < count( $arr ); $i++ ) {
1358                                 if ( ( $i % 2 ) == 1 ) {
1359                                         # If there are ever four apostrophes, assume the first is supposed to
1360                                         # be text, and the remaining three constitute mark-up for bold text.
1361                                         if ( strlen( $arr[$i] ) == 4 ) {
1362                                                 $arr[$i-1] .= "'";
1363                                                 $arr[$i] = "'''";
1364                                         } elseif ( strlen( $arr[$i] ) > 5 ) {
1365                                                 # If there are more than 5 apostrophes in a row, assume they're all
1366                                                 # text except for the last 5.
1367                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1368                                                 $arr[$i] = "'''''";
1369                                         }
1370                                         # Count the number of occurrences of bold and italics mark-ups.
1371                                         # We are not counting sequences of five apostrophes.
1372                                         if ( strlen( $arr[$i] ) == 2 ) {
1373                                                 $numitalics++;
1374                                         } elseif ( strlen( $arr[$i] ) == 3 ) {
1375                                                 $numbold++;
1376                                         } elseif ( strlen( $arr[$i] ) == 5 ) {
1377                                                 $numitalics++;
1378                                                 $numbold++;
1379                                         }
1380                                 }
1381                         }
1382
1383                         # If there is an odd number of both bold and italics, it is likely
1384                         # that one of the bold ones was meant to be an apostrophe followed
1385                         # by italics. Which one we cannot know for certain, but it is more
1386                         # likely to be one that has a single-letter word before it.
1387                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1388                                 $i = 0;
1389                                 $firstsingleletterword = -1;
1390                                 $firstmultiletterword = -1;
1391                                 $firstspace = -1;
1392                                 foreach ( $arr as $r ) {
1393                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1394                                                 $x1 = substr( $arr[$i-1], -1 );
1395                                                 $x2 = substr( $arr[$i-1], -2, 1 );
1396                                                 if ( $x1 === ' ' ) {
1397                                                         if ( $firstspace == -1 ) {
1398                                                                 $firstspace = $i;
1399                                                         }
1400                                                 } elseif ( $x2 === ' ') {
1401                                                         if ( $firstsingleletterword == -1 ) {
1402                                                                 $firstsingleletterword = $i;
1403                                                         }
1404                                                 } else {
1405                                                         if ( $firstmultiletterword == -1 ) {
1406                                                                 $firstmultiletterword = $i;
1407                                                         }
1408                                                 }
1409                                         }
1410                                         $i++;
1411                                 }
1412
1413                                 # If there is a single-letter word, use it!
1414                                 if ( $firstsingleletterword > -1 ) {
1415                                         $arr[$firstsingleletterword] = "''";
1416                                         $arr[$firstsingleletterword-1] .= "'";
1417                                 } elseif ( $firstmultiletterword > -1 ) {
1418                                         # If not, but there's a multi-letter word, use that one.
1419                                         $arr[$firstmultiletterword] = "''";
1420                                         $arr[$firstmultiletterword-1] .= "'";
1421                                 } elseif ( $firstspace > -1 ) {
1422                                         # ... otherwise use the first one that has neither.
1423                                         # (notice that it is possible for all three to be -1 if, for example,
1424                                         # there is only one pentuple-apostrophe in the line)
1425                                         $arr[$firstspace] = "''";
1426                                         $arr[$firstspace-1] .= "'";
1427                                 }
1428                         }
1429
1430                         # Now let's actually convert our apostrophic mush to HTML!
1431                         $output = '';
1432                         $buffer = '';
1433                         $state = '';
1434                         $i = 0;
1435                         foreach ( $arr as $r ) {
1436                                 if ( ( $i % 2 ) == 0 ) {
1437                                         if ( $state === 'both' ) {
1438                                                 $buffer .= $r;
1439                                         } else {
1440                                                 $output .= $r;
1441                                         }
1442                                 } else {
1443                                         if ( strlen( $r ) == 2 ) {
1444                                                 if ( $state === 'i' ) {
1445                                                         $output .= '</i>'; $state = '';
1446                                                 } elseif ( $state === 'bi' ) {
1447                                                         $output .= '</i>'; $state = 'b';
1448                                                 } elseif ( $state === 'ib' ) {
1449                                                         $output .= '</b></i><b>'; $state = 'b';
1450                                                 } elseif ( $state === 'both' ) {
1451                                                         $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1452                                                 } else { # $state can be 'b' or ''
1453                                                         $output .= '<i>'; $state .= 'i';
1454                                                 }
1455                                         } elseif ( strlen( $r ) == 3 ) {
1456                                                 if ( $state === 'b' ) {
1457                                                         $output .= '</b>'; $state = '';
1458                                                 } elseif ( $state === 'bi' ) {
1459                                                         $output .= '</i></b><i>'; $state = 'i';
1460                                                 } elseif ( $state === 'ib' ) {
1461                                                         $output .= '</b>'; $state = 'i';
1462                                                 } elseif ( $state === 'both' ) {
1463                                                         $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1464                                                 } else { # $state can be 'i' or ''
1465                                                         $output .= '<b>'; $state .= 'b';
1466                                                 }
1467                                         } elseif ( strlen( $r ) == 5 ) {
1468                                                 if ( $state === 'b' ) {
1469                                                         $output .= '</b><i>'; $state = 'i';
1470                                                 } elseif ( $state === 'i' ) {
1471                                                         $output .= '</i><b>'; $state = 'b';
1472                                                 } elseif ( $state === 'bi' ) {
1473                                                         $output .= '</i></b>'; $state = '';
1474                                                 } elseif ( $state === 'ib' ) {
1475                                                         $output .= '</b></i>'; $state = '';
1476                                                 } elseif ( $state === 'both' ) {
1477                                                         $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1478                                                 } else { # ($state == '')
1479                                                         $buffer = ''; $state = 'both';
1480                                                 }
1481                                         }
1482                                 }
1483                                 $i++;
1484                         }
1485                         # Now close all remaining tags.  Notice that the order is important.
1486                         if ( $state === 'b' || $state === 'ib' ) {
1487                                 $output .= '</b>';
1488                         }
1489                         if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1490                                 $output .= '</i>';
1491                         }
1492                         if ( $state === 'bi' ) {
1493                                 $output .= '</b>';
1494                         }
1495                         # There might be lonely ''''', so make sure we have a buffer
1496                         if ( $state === 'both' && $buffer ) {
1497                                 $output .= '<b><i>'.$buffer.'</i></b>';
1498                         }
1499                         return $output;
1500                 }
1501         }
1502
1503         /**
1504          * Replace external links (REL)
1505          *
1506          * Note: this is all very hackish and the order of execution matters a lot.
1507          * Make sure to run maintenance/parserTests.php if you change this code.
1508          *
1509          * @private
1510          */
1511         function replaceExternalLinks( $text ) {
1512                 global $wgContLang;
1513                 wfProfileIn( __METHOD__ );
1514
1515                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1516                 $s = array_shift( $bits );
1517
1518                 $i = 0;
1519                 while ( $i<count( $bits ) ) {
1520                         $url = $bits[$i++];
1521                         $protocol = $bits[$i++];
1522                         $text = $bits[$i++];
1523                         $trail = $bits[$i++];
1524
1525                         # The characters '<' and '>' (which were escaped by
1526                         # removeHTMLtags()) should not be included in
1527                         # URLs, per RFC 2396.
1528                         $m2 = array();
1529                         if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1530                                 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1531                                 $url = substr( $url, 0, $m2[0][1] );
1532                         }
1533
1534                         # If the link text is an image URL, replace it with an <img> tag
1535                         # This happened by accident in the original parser, but some people used it extensively
1536                         $img = $this->maybeMakeExternalImage( $text );
1537                         if ( $img !== false ) {
1538                                 $text = $img;
1539                         }
1540
1541                         $dtrail = '';
1542
1543                         # Set linktype for CSS - if URL==text, link is essentially free
1544                         $linktype = ( $text === $url ) ? 'free' : 'text';
1545
1546                         # No link text, e.g. [http://domain.tld/some.link]
1547                         if ( $text == '' ) {
1548                                 # Autonumber if allowed. See bug #5918
1549                                 if ( strpos( wfUrlProtocols(), substr( $protocol, 0, strpos( $protocol, ':' ) ) ) !== false ) {
1550                                         $langObj = $this->getFunctionLang();
1551                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1552                                         $linktype = 'autonumber';
1553                                 } else {
1554                                         # Otherwise just use the URL
1555                                         $text = htmlspecialchars( $url );
1556                                         $linktype = 'free';
1557                                 }
1558                         } else {
1559                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1560                                 # Check for trail
1561                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1562                         }
1563
1564                         $text = $wgContLang->markNoConversion( $text );
1565
1566                         $url = Sanitizer::cleanUrl( $url );
1567
1568                         # Use the encoded URL
1569                         # This means that users can paste URLs directly into the text
1570                         # Funny characters like ö aren't valid in URLs anyway
1571                         # This was changed in August 2004
1572                         $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1573                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1574
1575                         # Register link in the output object.
1576                         # Replace unnecessary URL escape codes with the referenced character
1577                         # This prevents spammers from hiding links from the filters
1578                         $pasteurized = self::replaceUnusualEscapes( $url );
1579                         $this->mOutput->addExternalLink( $pasteurized );
1580                 }
1581
1582                 wfProfileOut( __METHOD__ );
1583                 return $s;
1584         }
1585
1586         /**
1587          * Get an associative array of additional HTML attributes appropriate for a
1588          * particular external link.  This currently may include rel => nofollow
1589          * (depending on configuration, namespace, and the URL's domain) and/or a
1590          * target attribute (depending on configuration).
1591          *
1592          * @param $url String: optional URL, to extract the domain from for rel =>
1593          *   nofollow if appropriate
1594          * @return Array: associative array of HTML attributes
1595          */
1596         function getExternalLinkAttribs( $url = false ) {
1597                 $attribs = array();
1598                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1599                 $ns = $this->mTitle->getNamespace();
1600                 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) ) {
1601                         $attribs['rel'] = 'nofollow';
1602
1603                         global $wgNoFollowDomainExceptions;
1604                         if ( $wgNoFollowDomainExceptions ) {
1605                                 $bits = wfParseUrl( $url );
1606                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1607                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1608                                                 if ( substr( $bits['host'], -strlen( $domain ) ) == $domain ) {
1609                                                         unset( $attribs['rel'] );
1610                                                         break;
1611                                                 }
1612                                         }
1613                                 }
1614                         }
1615                 }
1616                 if ( $this->mOptions->getExternalLinkTarget() ) {
1617                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1618                 }
1619                 return $attribs;
1620         }
1621
1622         /**
1623          * Replace unusual URL escape codes with their equivalent characters
1624          *
1625          * @param $url String
1626          * @return String
1627          *
1628          * @todo  This can merge genuinely required bits in the path or query string,
1629          *        breaking legit URLs. A proper fix would treat the various parts of
1630          *        the URL differently; as a workaround, just use the output for
1631          *        statistical records, not for actual linking/output.
1632          */
1633         static function replaceUnusualEscapes( $url ) {
1634                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1635                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1636         }
1637
1638         /**
1639          * Callback function used in replaceUnusualEscapes().
1640          * Replaces unusual URL escape codes with their equivalent character
1641          */
1642         private static function replaceUnusualEscapesCallback( $matches ) {
1643                 $char = urldecode( $matches[0] );
1644                 $ord = ord( $char );
1645                 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1646                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1647                         # No, shouldn't be escaped
1648                         return $char;
1649                 } else {
1650                         # Yes, leave it escaped
1651                         return $matches[0];
1652                 }
1653         }
1654
1655         /**
1656          * make an image if it's allowed, either through the global
1657          * option, through the exception, or through the on-wiki whitelist
1658          * @private
1659          */
1660         function maybeMakeExternalImage( $url ) {
1661                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1662                 $imagesexception = !empty( $imagesfrom );
1663                 $text = false;
1664                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1665                 if ( $imagesexception && is_array( $imagesfrom ) ) {
1666                         $imagematch = false;
1667                         foreach ( $imagesfrom as $match ) {
1668                                 if ( strpos( $url, $match ) === 0 ) {
1669                                         $imagematch = true;
1670                                         break;
1671                                 }
1672                         }
1673                 } elseif ( $imagesexception ) {
1674                         $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1675                 } else {
1676                         $imagematch = false;
1677                 }
1678                 if ( $this->mOptions->getAllowExternalImages()
1679                          || ( $imagesexception && $imagematch ) ) {
1680                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1681                                 # Image found
1682                                 $text = Linker::makeExternalImage( $url );
1683                         }
1684                 }
1685                 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1686                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1687                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1688                         foreach ( $whitelist as $entry ) {
1689                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1690                                 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1691                                         continue;
1692                                 }
1693                                 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1694                                         # Image matches a whitelist entry
1695                                         $text = Linker::makeExternalImage( $url );
1696                                         break;
1697                                 }
1698                         }
1699                 }
1700                 return $text;
1701         }
1702
1703         /**
1704          * Process [[ ]] wikilinks
1705          * @return String: processed text
1706          *
1707          * @private
1708          */
1709         function replaceInternalLinks( $s ) {
1710                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1711                 return $s;
1712         }
1713
1714         /**
1715          * Process [[ ]] wikilinks (RIL)
1716          * @return LinkHolderArray
1717          *
1718          * @private
1719          */
1720         function replaceInternalLinks2( &$s ) {
1721                 global $wgContLang;
1722
1723                 wfProfileIn( __METHOD__ );
1724
1725                 wfProfileIn( __METHOD__.'-setup' );
1726                 static $tc = FALSE, $e1, $e1_img;
1727                 # the % is needed to support urlencoded titles as well
1728                 if ( !$tc ) {
1729                         $tc = Title::legalChars() . '#%';
1730                         # Match a link having the form [[namespace:link|alternate]]trail
1731                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1732                         # Match cases where there is no "]]", which might still be images
1733                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1734                 }
1735
1736                 $holders = new LinkHolderArray( $this );
1737
1738                 # split the entire text string on occurences of [[
1739                 $a = StringUtils::explode( '[[', ' ' . $s );
1740                 # get the first element (all text up to first [[), and remove the space we added
1741                 $s = $a->current();
1742                 $a->next();
1743                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1744                 $s = substr( $s, 1 );
1745
1746                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1747                 $e2 = null;
1748                 if ( $useLinkPrefixExtension ) {
1749                         # Match the end of a line for a word that's not followed by whitespace,
1750                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1751                         $e2 = wfMsgForContent( 'linkprefix' );
1752                 }
1753
1754                 if ( is_null( $this->mTitle ) ) {
1755                         wfProfileOut( __METHOD__.'-setup' );
1756                         wfProfileOut( __METHOD__ );
1757                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1758                 }
1759                 $nottalk = !$this->mTitle->isTalkPage();
1760
1761                 if ( $useLinkPrefixExtension ) {
1762                         $m = array();
1763                         if ( preg_match( $e2, $s, $m ) ) {
1764                                 $first_prefix = $m[2];
1765                         } else {
1766                                 $first_prefix = false;
1767                         }
1768                 } else {
1769                         $prefix = '';
1770                 }
1771
1772                 if ( $wgContLang->hasVariants() ) {
1773                         $selflink = $wgContLang->autoConvertToAllVariants( $this->mTitle->getPrefixedText() );
1774                 } else {
1775                         $selflink = array( $this->mTitle->getPrefixedText() );
1776                 }
1777                 $useSubpages = $this->areSubpagesAllowed();
1778                 wfProfileOut( __METHOD__.'-setup' );
1779
1780                 # Loop for each link
1781                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1782                         # Check for excessive memory usage
1783                         if ( $holders->isBig() ) {
1784                                 # Too big
1785                                 # Do the existence check, replace the link holders and clear the array
1786                                 $holders->replace( $s );
1787                                 $holders->clear();
1788                         }
1789
1790                         if ( $useLinkPrefixExtension ) {
1791                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1792                                 if ( preg_match( $e2, $s, $m ) ) {
1793                                         $prefix = $m[2];
1794                                         $s = $m[1];
1795                                 } else {
1796                                         $prefix='';
1797                                 }
1798                                 # first link
1799                                 if ( $first_prefix ) {
1800                                         $prefix = $first_prefix;
1801                                         $first_prefix = false;
1802                                 }
1803                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1804                         }
1805
1806                         $might_be_img = false;
1807
1808                         wfProfileIn( __METHOD__."-e1" );
1809                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1810                                 $text = $m[2];
1811                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1812                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1813                                 # the real problem is with the $e1 regex
1814                                 # See bug 1300.
1815                                 #
1816                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1817                                 # and no image is in sight. See bug 2095.
1818                                 #
1819                                 if ( $text !== '' &&
1820                                         substr( $m[3], 0, 1 ) === ']' &&
1821                                         strpos( $text, '[' ) !== false
1822                                 )
1823                                 {
1824                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1825                                         $m[3] = substr( $m[3], 1 );
1826                                 }
1827                                 # fix up urlencoded title texts
1828                                 if ( strpos( $m[1], '%' ) !== false ) {
1829                                         # Should anchors '#' also be rejected?
1830                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), rawurldecode( $m[1] ) );
1831                                 }
1832                                 $trail = $m[3];
1833                         } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1834                                 $might_be_img = true;
1835                                 $text = $m[2];
1836                                 if ( strpos( $m[1], '%' ) !== false ) {
1837                                         $m[1] = rawurldecode( $m[1] );
1838                                 }
1839                                 $trail = "";
1840                         } else { # Invalid form; output directly
1841                                 $s .= $prefix . '[[' . $line ;
1842                                 wfProfileOut( __METHOD__."-e1" );
1843                                 continue;
1844                         }
1845                         wfProfileOut( __METHOD__."-e1" );
1846                         wfProfileIn( __METHOD__."-misc" );
1847
1848                         # Don't allow internal links to pages containing
1849                         # PROTO: where PROTO is a valid URL protocol; these
1850                         # should be external links.
1851                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1852                                 $s .= $prefix . '[[' . $line ;
1853                                 wfProfileOut( __METHOD__."-misc" );
1854                                 continue;
1855                         }
1856
1857                         # Make subpage if necessary
1858                         if ( $useSubpages ) {
1859                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1860                         } else {
1861                                 $link = $m[1];
1862                         }
1863
1864                         $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1865                         if ( !$noforce ) {
1866                                 # Strip off leading ':'
1867                                 $link = substr( $link, 1 );
1868                         }
1869
1870                         wfProfileOut( __METHOD__."-misc" );
1871                         wfProfileIn( __METHOD__."-title" );
1872                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1873                         if ( $nt === null ) {
1874                                 $s .= $prefix . '[[' . $line;
1875                                 wfProfileOut( __METHOD__."-title" );
1876                                 continue;
1877                         }
1878
1879                         $ns = $nt->getNamespace();
1880                         $iw = $nt->getInterWiki();
1881                         wfProfileOut( __METHOD__."-title" );
1882
1883                         if ( $might_be_img ) { # if this is actually an invalid link
1884                                 wfProfileIn( __METHOD__."-might_be_img" );
1885                                 if ( $ns == NS_FILE && $noforce ) { # but might be an image
1886                                         $found = false;
1887                                         while ( true ) {
1888                                                 # look at the next 'line' to see if we can close it there
1889                                                 $a->next();
1890                                                 $next_line = $a->current();
1891                                                 if ( $next_line === false || $next_line === null ) {
1892                                                         break;
1893                                                 }
1894                                                 $m = explode( ']]', $next_line, 3 );
1895                                                 if ( count( $m ) == 3 ) {
1896                                                         # the first ]] closes the inner link, the second the image
1897                                                         $found = true;
1898                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1899                                                         $trail = $m[2];
1900                                                         break;
1901                                                 } elseif ( count( $m ) == 2 ) {
1902                                                         # if there's exactly one ]] that's fine, we'll keep looking
1903                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1904                                                 } else {
1905                                                         # if $next_line is invalid too, we need look no further
1906                                                         $text .= '[[' . $next_line;
1907                                                         break;
1908                                                 }
1909                                         }
1910                                         if ( !$found ) {
1911                                                 # we couldn't find the end of this imageLink, so output it raw
1912                                                 # but don't ignore what might be perfectly normal links in the text we've examined
1913                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1914                                                 $s .= "{$prefix}[[$link|$text";
1915                                                 # note: no $trail, because without an end, there *is* no trail
1916                                                 wfProfileOut( __METHOD__."-might_be_img" );
1917                                                 continue;
1918                                         }
1919                                 } else { # it's not an image, so output it raw
1920                                         $s .= "{$prefix}[[$link|$text";
1921                                         # note: no $trail, because without an end, there *is* no trail
1922                                         wfProfileOut( __METHOD__."-might_be_img" );
1923                                         continue;
1924                                 }
1925                                 wfProfileOut( __METHOD__."-might_be_img" );
1926                         }
1927
1928                         $wasblank = ( $text  == '' );
1929                         if ( $wasblank ) {
1930                                 $text = $link;
1931                         } else {
1932                                 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
1933                                 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
1934                                 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
1935                                 #    -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
1936                                 $text = $this->doQuotes( $text );
1937                         }
1938
1939                         # Link not escaped by : , create the various objects
1940                         if ( $noforce ) {
1941
1942                                 # Interwikis
1943                                 wfProfileIn( __METHOD__."-interwiki" );
1944                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1945                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1946                                         $s = rtrim( $s . $prefix );
1947                                         $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
1948                                         wfProfileOut( __METHOD__."-interwiki" );
1949                                         continue;
1950                                 }
1951                                 wfProfileOut( __METHOD__."-interwiki" );
1952
1953                                 if ( $ns == NS_FILE ) {
1954                                         wfProfileIn( __METHOD__."-image" );
1955                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1956                                                 if ( $wasblank ) {
1957                                                         # if no parameters were passed, $text
1958                                                         # becomes something like "File:Foo.png",
1959                                                         # which we don't want to pass on to the
1960                                                         # image generator
1961                                                         $text = '';
1962                                                 } else {
1963                                                         # recursively parse links inside the image caption
1964                                                         # actually, this will parse them in any other parameters, too,
1965                                                         # but it might be hard to fix that, and it doesn't matter ATM
1966                                                         $text = $this->replaceExternalLinks( $text );
1967                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1968                                                 }
1969                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1970                                                 $s .= $prefix . $this->armorLinks(
1971                                                         $this->makeImage( $nt, $text, $holders ) ) . $trail;
1972                                         } else {
1973                                                 $s .= $prefix . $trail;
1974                                         }
1975                                         wfProfileOut( __METHOD__."-image" );
1976                                         continue;
1977                                 }
1978
1979                                 if ( $ns == NS_CATEGORY ) {
1980                                         wfProfileIn( __METHOD__."-category" );
1981                                         $s = rtrim( $s . "\n" ); # bug 87
1982
1983                                         if ( $wasblank ) {
1984                                                 $sortkey = $this->getDefaultSort();
1985                                         } else {
1986                                                 $sortkey = $text;
1987                                         }
1988                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1989                                         $sortkey = str_replace( "\n", '', $sortkey );
1990                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1991                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1992
1993                                         /**
1994                                          * Strip the whitespace Category links produce, see bug 87
1995                                          * @todo We might want to use trim($tmp, "\n") here.
1996                                          */
1997                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
1998
1999                                         wfProfileOut( __METHOD__."-category" );
2000                                         continue;
2001                                 }
2002                         }
2003
2004                         # Self-link checking
2005                         if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
2006                                 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
2007                                         $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2008                                         continue;
2009                                 }
2010                         }
2011
2012                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
2013                         # FIXME: Should do batch file existence checks, see comment below
2014                         if ( $ns == NS_MEDIA ) {
2015                                 wfProfileIn( __METHOD__."-media" );
2016                                 # Give extensions a chance to select the file revision for us
2017                                 $time = $sha1 = $descQuery = false;
2018                                 wfRunHooks( 'BeforeParserFetchFileAndTitle',
2019                                         array( $this, $nt, &$time, &$sha1, &$descQuery ) );
2020                                 # Fetch and register the file (file title may be different via hooks)
2021                                 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $time, $sha1 );
2022                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2023                                 $s .= $prefix . $this->armorLinks(
2024                                         Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2025                                 wfProfileOut( __METHOD__."-media" );
2026                                 continue;
2027                         }
2028
2029                         wfProfileIn( __METHOD__."-always_known" );
2030                         # Some titles, such as valid special pages or files in foreign repos, should
2031                         # be shown as bluelinks even though they're not included in the page table
2032                         #
2033                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2034                         # batch file existence checks for NS_FILE and NS_MEDIA
2035                         if ( $iw == '' && $nt->isAlwaysKnown() ) {
2036                                 $this->mOutput->addLink( $nt );
2037                                 $s .= $this->makeKnownLinkHolder( $nt, $text, array(), $trail, $prefix );
2038                         } else {
2039                                 # Links will be added to the output link list after checking
2040                                 $s .= $holders->makeHolder( $nt, $text, array(), $trail, $prefix );
2041                         }
2042                         wfProfileOut( __METHOD__."-always_known" );
2043                 }
2044                 wfProfileOut( __METHOD__ );
2045                 return $holders;
2046         }
2047
2048         /**
2049          * Make a link placeholder. The text returned can be later resolved to a real link with
2050          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
2051          * parsing of interwiki links, and secondly to allow all existence checks and
2052          * article length checks (for stub links) to be bundled into a single query.
2053          *
2054          * @deprecated
2055          */
2056         function makeLinkHolder( &$nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
2057                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
2058         }
2059
2060         /**
2061          * Render a forced-blue link inline; protect against double expansion of
2062          * URLs if we're in a mode that prepends full URL prefixes to internal links.
2063          * Since this little disaster has to split off the trail text to avoid
2064          * breaking URLs in the following text without breaking trails on the
2065          * wiki links, it's been made into a horrible function.
2066          *
2067          * @param $nt Title
2068          * @param $text String
2069          * @param $query Array or String
2070          * @param $trail String
2071          * @param $prefix String
2072          * @return String: HTML-wikitext mix oh yuck
2073          */
2074         function makeKnownLinkHolder( $nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
2075                 list( $inside, $trail ) = Linker::splitTrail( $trail );
2076
2077                 if ( is_string( $query ) ) {
2078                         $query = wfCgiToArray( $query );
2079                 }
2080                 if ( $text == '' ) {
2081                         $text = htmlspecialchars( $nt->getPrefixedText() );
2082                 }
2083
2084                 $link = Linker::linkKnown( $nt, "$prefix$text$inside", array(), $query );
2085
2086                 return $this->armorLinks( $link ) . $trail;
2087         }
2088
2089         /**
2090          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2091          * going to go through further parsing steps before inline URL expansion.
2092          *
2093          * Not needed quite as much as it used to be since free links are a bit
2094          * more sensible these days. But bracketed links are still an issue.
2095          *
2096          * @param $text String: more-or-less HTML
2097          * @return String: less-or-more HTML with NOPARSE bits
2098          */
2099         function armorLinks( $text ) {
2100                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
2101                         "{$this->mUniqPrefix}NOPARSE$1", $text );
2102         }
2103
2104         /**
2105          * Return true if subpage links should be expanded on this page.
2106          * @return Boolean
2107          */
2108         function areSubpagesAllowed() {
2109                 # Some namespaces don't allow subpages
2110                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2111         }
2112
2113         /**
2114          * Handle link to subpage if necessary
2115          *
2116          * @param $target String: the source of the link
2117          * @param &$text String: the link text, modified as necessary
2118          * @return string the full name of the link
2119          * @private
2120          */
2121         function maybeDoSubpageLink( $target, &$text ) {
2122                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2123         }
2124
2125         /**#@+
2126          * Used by doBlockLevels()
2127          * @private
2128          *
2129          * @return string
2130          */
2131         function closeParagraph() {
2132                 $result = '';
2133                 if ( $this->mLastSection != '' ) {
2134                         $result = '</' . $this->mLastSection  . ">\n";
2135                 }
2136                 $this->mInPre = false;
2137                 $this->mLastSection = '';
2138                 return $result;
2139         }
2140
2141         /**
2142          * getCommon() returns the length of the longest common substring
2143          * of both arguments, starting at the beginning of both.
2144          * @private
2145          */
2146         function getCommon( $st1, $st2 ) {
2147                 $fl = strlen( $st1 );
2148                 $shorter = strlen( $st2 );
2149                 if ( $fl < $shorter ) {
2150                         $shorter = $fl;
2151                 }
2152
2153                 for ( $i = 0; $i < $shorter; ++$i ) {
2154                         if ( $st1[$i] != $st2[$i] ) {
2155                                 break;
2156                         }
2157                 }
2158                 return $i;
2159         }
2160
2161         /**
2162          * These next three functions open, continue, and close the list
2163          * element appropriate to the prefix character passed into them.
2164          * @private
2165          *
2166          * @return string
2167          */
2168         function openList( $char ) {
2169                 $result = $this->closeParagraph();
2170
2171                 if ( '*' === $char ) {
2172                         $result .= '<ul><li>';
2173                 } elseif ( '#' === $char ) {
2174                         $result .= '<ol><li>';
2175                 } elseif ( ':' === $char ) {
2176                         $result .= '<dl><dd>';
2177                 } elseif ( ';' === $char ) {
2178                         $result .= '<dl><dt>';
2179                         $this->mDTopen = true;
2180                 } else {
2181                         $result = '<!-- ERR 1 -->';
2182                 }
2183
2184                 return $result;
2185         }
2186
2187         /**
2188          * TODO: document
2189          * @param $char String
2190          * @private
2191          *
2192          * @return string
2193          */
2194         function nextItem( $char ) {
2195                 if ( '*' === $char || '#' === $char ) {
2196                         return '</li><li>';
2197                 } elseif ( ':' === $char || ';' === $char ) {
2198                         $close = '</dd>';
2199                         if ( $this->mDTopen ) {
2200                                 $close = '</dt>';
2201                         }
2202                         if ( ';' === $char ) {
2203                                 $this->mDTopen = true;
2204                                 return $close . '<dt>';
2205                         } else {
2206                                 $this->mDTopen = false;
2207                                 return $close . '<dd>';
2208                         }
2209                 }
2210                 return '<!-- ERR 2 -->';
2211         }
2212
2213         /**
2214          * TODO: document
2215          * @param $char String
2216          * @private
2217          *
2218          * @return string
2219          */
2220         function closeList( $char ) {
2221                 if ( '*' === $char ) {
2222                         $text = '</li></ul>';
2223                 } elseif ( '#' === $char ) {
2224                         $text = '</li></ol>';
2225                 } elseif ( ':' === $char ) {
2226                         if ( $this->mDTopen ) {
2227                                 $this->mDTopen = false;
2228                                 $text = '</dt></dl>';
2229                         } else {
2230                                 $text = '</dd></dl>';
2231                         }
2232                 } else {
2233                         return '<!-- ERR 3 -->';
2234                 }
2235                 return $text."\n";
2236         }
2237         /**#@-*/
2238
2239         /**
2240          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2241          *
2242          * @param $text String
2243          * @param $linestart Boolean: whether or not this is at the start of a line.
2244          * @private
2245          * @return string the lists rendered as HTML
2246          */
2247         function doBlockLevels( $text, $linestart ) {
2248                 wfProfileIn( __METHOD__ );
2249
2250                 # Parsing through the text line by line.  The main thing
2251                 # happening here is handling of block-level elements p, pre,
2252                 # and making lists from lines starting with * # : etc.
2253                 #
2254                 $textLines = StringUtils::explode( "\n", $text );
2255
2256                 $lastPrefix = $output = '';
2257                 $this->mDTopen = $inBlockElem = false;
2258                 $prefixLength = 0;
2259                 $paragraphStack = false;
2260
2261                 foreach ( $textLines as $oLine ) {
2262                         # Fix up $linestart
2263                         if ( !$linestart ) {
2264                                 $output .= $oLine;
2265                                 $linestart = true;
2266                                 continue;
2267                         }
2268                         # * = ul
2269                         # # = ol
2270                         # ; = dt
2271                         # : = dd
2272
2273                         $lastPrefixLength = strlen( $lastPrefix );
2274                         $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2275                         $preOpenMatch = preg_match( '/<pre/i', $oLine );
2276                         # If not in a <pre> element, scan for and figure out what prefixes are there.
2277                         if ( !$this->mInPre ) {
2278                                 # Multiple prefixes may abut each other for nested lists.
2279                                 $prefixLength = strspn( $oLine, '*#:;' );
2280                                 $prefix = substr( $oLine, 0, $prefixLength );
2281
2282                                 # eh?
2283                                 # ; and : are both from definition-lists, so they're equivalent
2284                                 #  for the purposes of determining whether or not we need to open/close
2285                                 #  elements.
2286                                 $prefix2 = str_replace( ';', ':', $prefix );
2287                                 $t = substr( $oLine, $prefixLength );
2288                                 $this->mInPre = (bool)$preOpenMatch;
2289                         } else {
2290                                 # Don't interpret any other prefixes in preformatted text
2291                                 $prefixLength = 0;
2292                                 $prefix = $prefix2 = '';
2293                                 $t = $oLine;
2294                         }
2295
2296                         # List generation
2297                         if ( $prefixLength && $lastPrefix === $prefix2 ) {
2298                                 # Same as the last item, so no need to deal with nesting or opening stuff
2299                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2300                                 $paragraphStack = false;
2301
2302                                 if ( substr( $prefix, -1 ) === ';') {
2303                                         # The one nasty exception: definition lists work like this:
2304                                         # ; title : definition text
2305                                         # So we check for : in the remainder text to split up the
2306                                         # title and definition, without b0rking links.
2307                                         $term = $t2 = '';
2308                                         if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2309                                                 $t = $t2;
2310                                                 $output .= $term . $this->nextItem( ':' );
2311                                         }
2312                                 }
2313                         } elseif ( $prefixLength || $lastPrefixLength ) {
2314                                 # We need to open or close prefixes, or both.
2315
2316                                 # Either open or close a level...
2317                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2318                                 $paragraphStack = false;
2319
2320                                 # Close all the prefixes which aren't shared.
2321                                 while ( $commonPrefixLength < $lastPrefixLength ) {
2322                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2323                                         --$lastPrefixLength;
2324                                 }
2325
2326                                 # Continue the current prefix if appropriate.
2327                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2328                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2329                                 }
2330
2331                                 # Open prefixes where appropriate.
2332                                 while ( $prefixLength > $commonPrefixLength ) {
2333                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2334                                         $output .= $this->openList( $char );
2335
2336                                         if ( ';' === $char ) {
2337                                                 # FIXME: This is dupe of code above
2338                                                 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2339                                                         $t = $t2;
2340                                                         $output .= $term . $this->nextItem( ':' );
2341                                                 }
2342                                         }
2343                                         ++$commonPrefixLength;
2344                                 }
2345                                 $lastPrefix = $prefix2;
2346                         }
2347
2348                         # If we have no prefixes, go to paragraph mode.
2349                         if ( 0 == $prefixLength ) {
2350                                 wfProfileIn( __METHOD__."-paragraph" );
2351                                 # No prefix (not in list)--go to paragraph mode
2352                                 # XXX: use a stack for nestable elements like span, table and div
2353                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2354                                 $closematch = preg_match(
2355                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2356                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2357                                 if ( $openmatch or $closematch ) {
2358                                         $paragraphStack = false;
2359                                         # TODO bug 5718: paragraph closed
2360                                         $output .= $this->closeParagraph();
2361                                         if ( $preOpenMatch and !$preCloseMatch ) {
2362                                                 $this->mInPre = true;
2363                                         }
2364                                         $inBlockElem = !$closematch;
2365                                 } elseif ( !$inBlockElem && !$this->mInPre ) {
2366                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) ) {
2367                                                 # pre
2368                                                 if ( $this->mLastSection !== 'pre' ) {
2369                                                         $paragraphStack = false;
2370                                                         $output .= $this->closeParagraph().'<pre>';
2371                                                         $this->mLastSection = 'pre';
2372                                                 }
2373                                                 $t = substr( $t, 1 );
2374                                         } else {
2375                                                 # paragraph
2376                                                 if ( trim( $t ) === '' ) {
2377                                                         if ( $paragraphStack ) {
2378                                                                 $output .= $paragraphStack.'<br />';
2379                                                                 $paragraphStack = false;
2380                                                                 $this->mLastSection = 'p';
2381                                                         } else {
2382                                                                 if ( $this->mLastSection !== 'p' ) {
2383                                                                         $output .= $this->closeParagraph();
2384                                                                         $this->mLastSection = '';
2385                                                                         $paragraphStack = '<p>';
2386                                                                 } else {
2387                                                                         $paragraphStack = '</p><p>';
2388                                                                 }
2389                                                         }
2390                                                 } else {
2391                                                         if ( $paragraphStack ) {
2392                                                                 $output .= $paragraphStack;
2393                                                                 $paragraphStack = false;
2394                                                                 $this->mLastSection = 'p';
2395                                                         } elseif ( $this->mLastSection !== 'p' ) {
2396                                                                 $output .= $this->closeParagraph().'<p>';
2397                                                                 $this->mLastSection = 'p';
2398                                                         }
2399                                                 }
2400                                         }
2401                                 }
2402                                 wfProfileOut( __METHOD__."-paragraph" );
2403                         }
2404                         # somewhere above we forget to get out of pre block (bug 785)
2405                         if ( $preCloseMatch && $this->mInPre ) {
2406                                 $this->mInPre = false;
2407                         }
2408                         if ( $paragraphStack === false ) {
2409                                 $output .= $t."\n";
2410                         }
2411                 }
2412                 while ( $prefixLength ) {
2413                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2414                         --$prefixLength;
2415                 }
2416                 if ( $this->mLastSection != '' ) {
2417                         $output .= '</' . $this->mLastSection . '>';
2418                         $this->mLastSection = '';
2419                 }
2420
2421                 wfProfileOut( __METHOD__ );
2422                 return $output;
2423         }
2424
2425         /**
2426          * Split up a string on ':', ignoring any occurences inside tags
2427          * to prevent illegal overlapping.
2428          *
2429          * @param $str String: the string to split
2430          * @param &$before String: set to everything before the ':'
2431          * @param &$after String: set to everything after the ':'
2432          * return String: the position of the ':', or false if none found
2433          */
2434         function findColonNoLinks( $str, &$before, &$after ) {
2435                 wfProfileIn( __METHOD__ );
2436
2437                 $pos = strpos( $str, ':' );
2438                 if ( $pos === false ) {
2439                         # Nothing to find!
2440                         wfProfileOut( __METHOD__ );
2441                         return false;
2442                 }
2443
2444                 $lt = strpos( $str, '<' );
2445                 if ( $lt === false || $lt > $pos ) {
2446                         # Easy; no tag nesting to worry about
2447                         $before = substr( $str, 0, $pos );
2448                         $after = substr( $str, $pos+1 );
2449                         wfProfileOut( __METHOD__ );
2450                         return $pos;
2451                 }
2452
2453                 # Ugly state machine to walk through avoiding tags.
2454                 $state = self::COLON_STATE_TEXT;
2455                 $stack = 0;
2456                 $len = strlen( $str );
2457                 for( $i = 0; $i < $len; $i++ ) {
2458                         $c = $str[$i];
2459
2460                         switch( $state ) {
2461                         # (Using the number is a performance hack for common cases)
2462                         case 0: # self::COLON_STATE_TEXT:
2463                                 switch( $c ) {
2464                                 case "<":
2465                                         # Could be either a <start> tag or an </end> tag
2466                                         $state = self::COLON_STATE_TAGSTART;
2467                                         break;
2468                                 case ":":
2469                                         if ( $stack == 0 ) {
2470                                                 # We found it!
2471                                                 $before = substr( $str, 0, $i );
2472                                                 $after = substr( $str, $i + 1 );
2473                                                 wfProfileOut( __METHOD__ );
2474                                                 return $i;
2475                                         }
2476                                         # Embedded in a tag; don't break it.
2477                                         break;
2478                                 default:
2479                                         # Skip ahead looking for something interesting
2480                                         $colon = strpos( $str, ':', $i );
2481                                         if ( $colon === false ) {
2482                                                 # Nothing else interesting
2483                                                 wfProfileOut( __METHOD__ );
2484                                                 return false;
2485                                         }
2486                                         $lt = strpos( $str, '<', $i );
2487                                         if ( $stack === 0 ) {
2488                                                 if ( $lt === false || $colon < $lt ) {
2489                                                         # We found it!
2490                                                         $before = substr( $str, 0, $colon );
2491                                                         $after = substr( $str, $colon + 1 );
2492                                                         wfProfileOut( __METHOD__ );
2493                                                         return $i;
2494                                                 }
2495                                         }
2496                                         if ( $lt === false ) {
2497                                                 # Nothing else interesting to find; abort!
2498                                                 # We're nested, but there's no close tags left. Abort!
2499                                                 break 2;
2500                                         }
2501                                         # Skip ahead to next tag start
2502                                         $i = $lt;
2503                                         $state = self::COLON_STATE_TAGSTART;
2504                                 }
2505                                 break;
2506                         case 1: # self::COLON_STATE_TAG:
2507                                 # In a <tag>
2508                                 switch( $c ) {
2509                                 case ">":
2510                                         $stack++;
2511                                         $state = self::COLON_STATE_TEXT;
2512                                         break;
2513                                 case "/":
2514                                         # Slash may be followed by >?
2515                                         $state = self::COLON_STATE_TAGSLASH;
2516                                         break;
2517                                 default:
2518                                         # ignore
2519                                 }
2520                                 break;
2521                         case 2: # self::COLON_STATE_TAGSTART:
2522                                 switch( $c ) {
2523                                 case "/":
2524                                         $state = self::COLON_STATE_CLOSETAG;
2525                                         break;
2526                                 case "!":
2527                                         $state = self::COLON_STATE_COMMENT;
2528                                         break;
2529                                 case ">":
2530                                         # Illegal early close? This shouldn't happen D:
2531                                         $state = self::COLON_STATE_TEXT;
2532                                         break;
2533                                 default:
2534                                         $state = self::COLON_STATE_TAG;
2535                                 }
2536                                 break;
2537                         case 3: # self::COLON_STATE_CLOSETAG:
2538                                 # In a </tag>
2539                                 if ( $c === ">" ) {
2540                                         $stack--;
2541                                         if ( $stack < 0 ) {
2542                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2543                                                 wfProfileOut( __METHOD__ );
2544                                                 return false;
2545                                         }
2546                                         $state = self::COLON_STATE_TEXT;
2547                                 }
2548                                 break;
2549                         case self::COLON_STATE_TAGSLASH:
2550                                 if ( $c === ">" ) {
2551                                         # Yes, a self-closed tag <blah/>
2552                                         $state = self::COLON_STATE_TEXT;
2553                                 } else {
2554                                         # Probably we're jumping the gun, and this is an attribute
2555                                         $state = self::COLON_STATE_TAG;
2556                                 }
2557                                 break;
2558                         case 5: # self::COLON_STATE_COMMENT:
2559                                 if ( $c === "-" ) {
2560                                         $state = self::COLON_STATE_COMMENTDASH;
2561                                 }
2562                                 break;
2563                         case self::COLON_STATE_COMMENTDASH:
2564                                 if ( $c === "-" ) {
2565                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2566                                 } else {
2567                                         $state = self::COLON_STATE_COMMENT;
2568                                 }
2569                                 break;
2570                         case self::COLON_STATE_COMMENTDASHDASH:
2571                                 if ( $c === ">" ) {
2572                                         $state = self::COLON_STATE_TEXT;
2573                                 } else {
2574                                         $state = self::COLON_STATE_COMMENT;
2575                                 }
2576                                 break;
2577                         default:
2578                                 throw new MWException( "State machine error in " . __METHOD__ );
2579                         }
2580                 }
2581                 if ( $stack > 0 ) {
2582                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2583                         wfProfileOut( __METHOD__ );
2584                         return false;
2585                 }
2586                 wfProfileOut( __METHOD__ );
2587                 return false;
2588         }
2589
2590         /**
2591          * Return value of a magic variable (like PAGENAME)
2592          *
2593          * @private
2594          *
2595          * @param $index integer
2596          * @param $frame PPFrame
2597          */
2598         function getVariableValue( $index, $frame=false ) {
2599                 global $wgContLang, $wgSitename, $wgServer;
2600                 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2601
2602                 /**
2603                  * Some of these require message or data lookups and can be
2604                  * expensive to check many times.
2605                  */
2606                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2607                         if ( isset( $this->mVarCache[$index] ) ) {
2608                                 return $this->mVarCache[$index];
2609                         }
2610                 }
2611
2612                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2613                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2614
2615                 # Use the time zone
2616                 global $wgLocaltimezone;
2617                 if ( isset( $wgLocaltimezone ) ) {
2618                         $oldtz = date_default_timezone_get();
2619                         date_default_timezone_set( $wgLocaltimezone );
2620                 }
2621
2622                 $localTimestamp = date( 'YmdHis', $ts );
2623                 $localMonth = date( 'm', $ts );
2624                 $localMonth1 = date( 'n', $ts );
2625                 $localMonthName = date( 'n', $ts );
2626                 $localDay = date( 'j', $ts );
2627                 $localDay2 = date( 'd', $ts );
2628                 $localDayOfWeek = date( 'w', $ts );
2629                 $localWeek = date( 'W', $ts );
2630                 $localYear = date( 'Y', $ts );
2631                 $localHour = date( 'H', $ts );
2632                 if ( isset( $wgLocaltimezone ) ) {
2633                         date_default_timezone_set( $oldtz );
2634                 }
2635
2636                 switch ( $index ) {
2637                         case 'currentmonth':
2638                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2639                                 break;
2640                         case 'currentmonth1':
2641                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2642                                 break;
2643                         case 'currentmonthname':
2644                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2645                                 break;
2646                         case 'currentmonthnamegen':
2647                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2648                                 break;
2649                         case 'currentmonthabbrev':
2650                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2651                                 break;
2652                         case 'currentday':
2653                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2654                                 break;
2655                         case 'currentday2':
2656                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2657                                 break;
2658                         case 'localmonth':
2659                                 $value = $wgContLang->formatNum( $localMonth );
2660                                 break;
2661                         case 'localmonth1':
2662                                 $value = $wgContLang->formatNum( $localMonth1 );
2663                                 break;
2664                         case 'localmonthname':
2665                                 $value = $wgContLang->getMonthName( $localMonthName );
2666                                 break;
2667                         case 'localmonthnamegen':
2668                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2669                                 break;
2670                         case 'localmonthabbrev':
2671                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2672                                 break;
2673                         case 'localday':
2674                                 $value = $wgContLang->formatNum( $localDay );
2675                                 break;
2676                         case 'localday2':
2677                                 $value = $wgContLang->formatNum( $localDay2 );
2678                                 break;
2679                         case 'pagename':
2680                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2681                                 break;
2682                         case 'pagenamee':
2683                                 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2684                                 break;
2685                         case 'fullpagename':
2686                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2687                                 break;
2688                         case 'fullpagenamee':
2689                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2690                                 break;
2691                         case 'subpagename':
2692                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2693                                 break;
2694                         case 'subpagenamee':
2695                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2696                                 break;
2697                         case 'basepagename':
2698                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2699                                 break;
2700                         case 'basepagenamee':
2701                                 $value = wfEscapeWikiText( wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) ) );
2702                                 break;
2703                         case 'talkpagename':
2704                                 if ( $this->mTitle->canTalk() ) {
2705                                         $talkPage = $this->mTitle->getTalkPage();
2706                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2707                                 } else {
2708                                         $value = '';
2709                                 }
2710                                 break;
2711                         case 'talkpagenamee':
2712                                 if ( $this->mTitle->canTalk() ) {
2713                                         $talkPage = $this->mTitle->getTalkPage();
2714                                         $value = wfEscapeWikiText( $talkPage->getPrefixedUrl() );
2715                                 } else {
2716                                         $value = '';
2717                                 }
2718                                 break;
2719                         case 'subjectpagename':
2720                                 $subjPage = $this->mTitle->getSubjectPage();
2721                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2722                                 break;
2723                         case 'subjectpagenamee':
2724                                 $subjPage = $this->mTitle->getSubjectPage();
2725                                 $value = wfEscapeWikiText( $subjPage->getPrefixedUrl() );
2726                                 break;
2727                         case 'revisionid':
2728                                 # Let the edit saving system know we should parse the page
2729                                 # *after* a revision ID has been assigned.
2730                                 $this->mOutput->setFlag( 'vary-revision' );
2731                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2732                                 $value = $this->mRevisionId;
2733                                 break;
2734                         case 'revisionday':
2735                                 # Let the edit saving system know we should parse the page
2736                                 # *after* a revision ID has been assigned. This is for null edits.
2737                                 $this->mOutput->setFlag( 'vary-revision' );
2738                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2739                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2740                                 break;
2741                         case 'revisionday2':
2742                                 # Let the edit saving system know we should parse the page
2743                                 # *after* a revision ID has been assigned. This is for null edits.
2744                                 $this->mOutput->setFlag( 'vary-revision' );
2745                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2746                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2747                                 break;
2748                         case 'revisionmonth':
2749                                 # Let the edit saving system know we should parse the page
2750                                 # *after* a revision ID has been assigned. This is for null edits.
2751                                 $this->mOutput->setFlag( 'vary-revision' );
2752                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2753                                 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2754                                 break;
2755                         case 'revisionmonth1':
2756                                 # Let the edit saving system know we should parse the page
2757                                 # *after* a revision ID has been assigned. This is for null edits.
2758                                 $this->mOutput->setFlag( 'vary-revision' );
2759                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2760                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2761                                 break;
2762                         case 'revisionyear':
2763                                 # Let the edit saving system know we should parse the page
2764                                 # *after* a revision ID has been assigned. This is for null edits.
2765                                 $this->mOutput->setFlag( 'vary-revision' );
2766                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2767                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2768                                 break;
2769                         case 'revisiontimestamp':
2770                                 # Let the edit saving system know we should parse the page
2771                                 # *after* a revision ID has been assigned. This is for null edits.
2772                                 $this->mOutput->setFlag( 'vary-revision' );
2773                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2774                                 $value = $this->getRevisionTimestamp();
2775                                 break;
2776                         case 'revisionuser':
2777                                 # Let the edit saving system know we should parse the page
2778                                 # *after* a revision ID has been assigned. This is for null edits.
2779                                 $this->mOutput->setFlag( 'vary-revision' );
2780                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2781                                 $value = $this->getRevisionUser();
2782                                 break;
2783                         case 'namespace':
2784                                 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2785                                 break;
2786                         case 'namespacee':
2787                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2788                                 break;
2789                         case 'talkspace':
2790                                 $value = $this->mTitle->canTalk() ? str_replace( '_',' ',$this->mTitle->getTalkNsText() ) : '';
2791                                 break;
2792                         case 'talkspacee':
2793                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2794                                 break;
2795                         case 'subjectspace':
2796                                 $value = $this->mTitle->getSubjectNsText();
2797                                 break;
2798                         case 'subjectspacee':
2799                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2800                                 break;
2801                         case 'currentdayname':
2802                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2803                                 break;
2804                         case 'currentyear':
2805                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2806                                 break;
2807                         case 'currenttime':
2808                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2809                                 break;
2810                         case 'currenthour':
2811                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2812                                 break;
2813                         case 'currentweek':
2814                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2815                                 # int to remove the padding
2816                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2817                                 break;
2818                         case 'currentdow':
2819                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2820                                 break;
2821                         case 'localdayname':
2822                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2823                                 break;
2824                         case 'localyear':
2825                                 $value = $wgContLang->formatNum( $localYear, true );
2826                                 break;
2827                         case 'localtime':
2828                                 $value = $wgContLang->time( $localTimestamp, false, false );
2829                                 break;
2830                         case 'localhour':
2831                                 $value = $wgContLang->formatNum( $localHour, true );
2832                                 break;
2833                         case 'localweek':
2834                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2835                                 # int to remove the padding
2836                                 $value = $wgContLang->formatNum( (int)$localWeek );
2837                                 break;
2838                         case 'localdow':
2839                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2840                                 break;
2841                         case 'numberofarticles':
2842                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2843                                 break;
2844                         case 'numberoffiles':
2845                                 $value = $wgContLang->formatNum( SiteStats::images() );
2846                                 break;
2847                         case 'numberofusers':
2848                                 $value = $wgContLang->formatNum( SiteStats::users() );
2849                                 break;
2850                         case 'numberofactiveusers':
2851                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2852                                 break;
2853                         case 'numberofpages':
2854                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2855                                 break;
2856                         case 'numberofadmins':
2857                                 $value = $wgContLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2858                                 break;
2859                         case 'numberofedits':
2860                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2861                                 break;
2862                         case 'numberofviews':
2863                                 $value = $wgContLang->formatNum( SiteStats::views() );
2864                                 break;
2865                         case 'currenttimestamp':
2866                                 $value = wfTimestamp( TS_MW, $ts );
2867                                 break;
2868                         case 'localtimestamp':
2869                                 $value = $localTimestamp;
2870                                 break;
2871                         case 'currentversion':
2872                                 $value = SpecialVersion::getVersion();
2873                                 break;
2874                         case 'articlepath':
2875                                 return $wgArticlePath;
2876                         case 'sitename':
2877                                 return $wgSitename;
2878                         case 'server':
2879                                 return $wgServer;
2880                         case 'servername':
2881                                 wfSuppressWarnings(); # May give an E_WARNING in PHP < 5.3.3
2882                                 $serverName = parse_url( $wgServer, PHP_URL_HOST );
2883                                 wfRestoreWarnings();
2884                                 return $serverName ? $serverName : $wgServer;
2885                         case 'scriptpath':
2886                                 return $wgScriptPath;
2887                         case 'stylepath':
2888                                 return $wgStylePath;
2889                         case 'directionmark':
2890                                 return $wgContLang->getDirMark();
2891                         case 'contentlanguage':
2892                                 global $wgLanguageCode;
2893                                 return $wgLanguageCode;
2894                         default:
2895                                 $ret = null;
2896                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
2897                                         return $ret;
2898                                 } else {
2899                                         return null;
2900                                 }
2901                 }
2902
2903                 if ( $index )
2904                         $this->mVarCache[$index] = $value;
2905
2906                 return $value;
2907         }
2908
2909         /**
2910          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2911          *
2912          * @private
2913          */
2914         function initialiseVariables() {
2915                 wfProfileIn( __METHOD__ );
2916                 $variableIDs = MagicWord::getVariableIDs();
2917                 $substIDs = MagicWord::getSubstIDs();
2918
2919                 $this->mVariables = new MagicWordArray( $variableIDs );
2920                 $this->mSubstWords = new MagicWordArray( $substIDs );
2921                 wfProfileOut( __METHOD__ );
2922         }
2923
2924         /**
2925          * Preprocess some wikitext and return the document tree.
2926          * This is the ghost of replace_variables().
2927          *
2928          * @param $text String: The text to parse
2929          * @param $flags Integer: bitwise combination of:
2930          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2931          *                                     included. Default is to assume a direct page view.
2932          *
2933          * The generated DOM tree must depend only on the input text and the flags.
2934          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2935          *
2936          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2937          * change in the DOM tree for a given text, must be passed through the section identifier
2938          * in the section edit link and thus back to extractSections().
2939          *
2940          * The output of this function is currently only cached in process memory, but a persistent
2941          * cache may be implemented at a later date which takes further advantage of these strict
2942          * dependency requirements.
2943          *
2944          * @private
2945          *
2946          * @return PPNode
2947          */
2948         function preprocessToDom( $text, $flags = 0 ) {
2949                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2950                 return $dom;
2951         }
2952
2953         /**
2954          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2955          *
2956          * @return array
2957          */
2958         public static function splitWhitespace( $s ) {
2959                 $ltrimmed = ltrim( $s );
2960                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2961                 $trimmed = rtrim( $ltrimmed );
2962                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2963                 if ( $diff > 0 ) {
2964                         $w2 = substr( $ltrimmed, -$diff );
2965                 } else {
2966                         $w2 = '';
2967                 }
2968                 return array( $w1, $trimmed, $w2 );
2969         }
2970
2971         /**
2972          * Replace magic variables, templates, and template arguments
2973          * with the appropriate text. Templates are substituted recursively,
2974          * taking care to avoid infinite loops.
2975          *
2976          * Note that the substitution depends on value of $mOutputType:
2977          *  self::OT_WIKI: only {{subst:}} templates
2978          *  self::OT_PREPROCESS: templates but not extension tags
2979          *  self::OT_HTML: all templates and extension tags
2980          *
2981          * @param $text String: the text to transform
2982          * @param $frame PPFrame Object describing the arguments passed to the template.
2983          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2984          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2985          * @param $argsOnly Boolean: only do argument (triple-brace) expansion, not double-brace expansion
2986          * @private
2987          *
2988          * @return string
2989          */
2990         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2991                 # Is there any text? Also, Prevent too big inclusions!
2992                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2993                         return $text;
2994                 }
2995                 wfProfileIn( __METHOD__ );
2996
2997                 if ( $frame === false ) {
2998                         $frame = $this->getPreprocessor()->newFrame();
2999                 } elseif ( !( $frame instanceof PPFrame ) ) {
3000                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
3001                         $frame = $this->getPreprocessor()->newCustomFrame( $frame );
3002                 }
3003
3004                 $dom = $this->preprocessToDom( $text );
3005                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
3006                 $text = $frame->expand( $dom, $flags );
3007
3008                 wfProfileOut( __METHOD__ );
3009                 return $text;
3010         }
3011
3012         /**
3013          * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3014          *
3015          * @return array
3016          */
3017         static function createAssocArgs( $args ) {
3018                 $assocArgs = array();
3019                 $index = 1;
3020                 foreach ( $args as $arg ) {
3021                         $eqpos = strpos( $arg, '=' );
3022                         if ( $eqpos === false ) {
3023                                 $assocArgs[$index++] = $arg;
3024                         } else {
3025                                 $name = trim( substr( $arg, 0, $eqpos ) );
3026                                 $value = trim( substr( $arg, $eqpos+1 ) );
3027                                 if ( $value === false ) {
3028                                         $value = '';
3029                                 }
3030                                 if ( $name !== false ) {
3031                                         $assocArgs[$name] = $value;
3032                                 }
3033                         }
3034                 }
3035
3036                 return $assocArgs;
3037         }
3038
3039         /**
3040          * Warn the user when a parser limitation is reached
3041          * Will warn at most once the user per limitation type
3042          *
3043          * @param $limitationType String: should be one of:
3044          *   'expensive-parserfunction' (corresponding messages:
3045          *       'expensive-parserfunction-warning',
3046          *       'expensive-parserfunction-category')
3047          *   'post-expand-template-argument' (corresponding messages:
3048          *       'post-expand-template-argument-warning',
3049          *       'post-expand-template-argument-category')
3050          *   'post-expand-template-inclusion' (corresponding messages:
3051          *       'post-expand-template-inclusion-warning',
3052          *       'post-expand-template-inclusion-category')
3053          * @param $current Current value
3054          * @param $max Maximum allowed, when an explicit limit has been
3055          *       exceeded, provide the values (optional)
3056          */
3057         function limitationWarn( $limitationType, $current=null, $max=null) {
3058                 # does no harm if $current and $max are present but are unnecessary for the message
3059                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
3060                 $this->mOutput->addWarning( $warning );
3061                 $this->addTrackingCategory( "$limitationType-category" );
3062         }
3063
3064         /**
3065          * Return the text of a template, after recursively
3066          * replacing any variables or templates within the template.
3067          *
3068          * @param $piece Array: the parts of the template
3069          *  $piece['title']: the title, i.e. the part before the |
3070          *  $piece['parts']: the parameter array
3071          *  $piece['lineStart']: whether the brace was at the start of a line
3072          * @param $frame PPFrame The current frame, contains template arguments
3073          * @return String: the text of the template
3074          * @private
3075          */
3076         function braceSubstitution( $piece, $frame ) {
3077                 global $wgContLang, $wgNonincludableNamespaces;
3078                 wfProfileIn( __METHOD__ );
3079                 wfProfileIn( __METHOD__.'-setup' );
3080
3081                 # Flags
3082                 $found = false;             # $text has been filled
3083                 $nowiki = false;            # wiki markup in $text should be escaped
3084                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
3085                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
3086                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
3087                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
3088
3089                 # Title object, where $text came from
3090                 $title = null;
3091
3092                 # $part1 is the bit before the first |, and must contain only title characters.
3093                 # Various prefixes will be stripped from it later.
3094                 $titleWithSpaces = $frame->expand( $piece['title'] );
3095                 $part1 = trim( $titleWithSpaces );
3096                 $titleText = false;
3097
3098                 # Original title text preserved for various purposes
3099                 $originalTitle = $part1;
3100
3101                 # $args is a list of argument nodes, starting from index 0, not including $part1
3102                 # *** FIXME if piece['parts'] is null then the call to getLength() below won't work b/c this $args isn't an object
3103                 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
3104                 wfProfileOut( __METHOD__.'-setup' );
3105
3106                 # SUBST
3107                 wfProfileIn( __METHOD__.'-modifiers' );
3108                 if ( !$found ) {
3109
3110                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3111
3112                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
3113                         # Decide whether to expand template or keep wikitext as-is.
3114                         if ( $this->ot['wiki'] ) {
3115                                 if ( $substMatch === false ) {
3116                                         $literal = true;  # literal when in PST with no prefix
3117                                 } else {
3118                                         $literal = false; # expand when in PST with subst: or safesubst:
3119                                 }
3120                         } else {
3121                                 if ( $substMatch == 'subst' ) {
3122                                         $literal = true;  # literal when not in PST with plain subst:
3123                                 } else {
3124                                         $literal = false; # expand when not in PST with safesubst: or no prefix
3125                                 }
3126                         }
3127                         if ( $literal ) {
3128                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3129                                 $isLocalObj = true;
3130                                 $found = true;
3131                         }
3132                 }
3133
3134                 # Variables
3135                 if ( !$found && $args->getLength() == 0 ) {
3136                         $id = $this->mVariables->matchStartToEnd( $part1 );
3137                         if ( $id !== false ) {
3138                                 $text = $this->getVariableValue( $id, $frame );
3139                                 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3140                                         $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3141                                 }
3142                                 $found = true;
3143                         }
3144                 }
3145
3146                 # MSG, MSGNW and RAW
3147                 if ( !$found ) {
3148                         # Check for MSGNW:
3149                         $mwMsgnw = MagicWord::get( 'msgnw' );
3150                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3151                                 $nowiki = true;
3152                         } else {
3153                                 # Remove obsolete MSG:
3154                                 $mwMsg = MagicWord::get( 'msg' );
3155                                 $mwMsg->matchStartAndRemove( $part1 );
3156                         }
3157
3158                         # Check for RAW:
3159                         $mwRaw = MagicWord::get( 'raw' );
3160                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3161                                 $forceRawInterwiki = true;
3162                         }
3163                 }
3164                 wfProfileOut( __METHOD__.'-modifiers' );
3165
3166                 # Parser functions
3167                 if ( !$found ) {
3168                         wfProfileIn( __METHOD__ . '-pfunc' );
3169
3170                         $colonPos = strpos( $part1, ':' );
3171                         if ( $colonPos !== false ) {
3172                                 # Case sensitive functions
3173                                 $function = substr( $part1, 0, $colonPos );
3174                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3175                                         $function = $this->mFunctionSynonyms[1][$function];
3176                                 } else {
3177                                         # Case insensitive functions
3178                                         $function = $wgContLang->lc( $function );
3179                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3180                                                 $function = $this->mFunctionSynonyms[0][$function];
3181                                         } else {
3182                                                 $function = false;
3183                                         }
3184                                 }
3185                                 if ( $function ) {
3186                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3187                                         $initialArgs = array( &$this );
3188                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3189                                         if ( $flags & SFH_OBJECT_ARGS ) {
3190                                                 # Add a frame parameter, and pass the arguments as an array
3191                                                 $allArgs = $initialArgs;
3192                                                 $allArgs[] = $frame;
3193                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3194                                                         $funcArgs[] = $args->item( $i );
3195                                                 }
3196                                                 $allArgs[] = $funcArgs;
3197                                         } else {
3198                                                 # Convert arguments to plain text
3199                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3200                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
3201                                                 }
3202                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3203                                         }
3204
3205                                         # Workaround for PHP bug 35229 and similar
3206                                         if ( !is_callable( $callback ) ) {
3207                                                 wfProfileOut( __METHOD__ . '-pfunc' );
3208                                                 wfProfileOut( __METHOD__ );
3209                                                 throw new MWException( "Tag hook for $function is not callable\n" );
3210                                         }
3211                                         $result = call_user_func_array( $callback, $allArgs );
3212                                         $found = true;
3213                                         $noparse = true;
3214                                         $preprocessFlags = 0;
3215
3216                                         if ( is_array( $result ) ) {
3217                                                 if ( isset( $result[0] ) ) {
3218                                                         $text = $result[0];
3219                                                         unset( $result[0] );
3220                                                 }
3221
3222                                                 # Extract flags into the local scope
3223                                                 # This allows callers to set flags such as nowiki, found, etc.
3224                                                 extract( $result );
3225                                         } else {
3226                                                 $text = $result;
3227                                         }
3228                                         if ( !$noparse ) {
3229                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
3230                                                 $isChildObj = true;
3231                                         }
3232                                 }
3233                         }
3234                         wfProfileOut( __METHOD__ . '-pfunc' );
3235                 }
3236
3237                 # Finish mangling title and then check for loops.
3238                 # Set $title to a Title object and $titleText to the PDBK
3239                 if ( !$found ) {
3240                         $ns = NS_TEMPLATE;
3241                         # Split the title into page and subpage
3242                         $subpage = '';
3243                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3244                         if ( $subpage !== '' ) {
3245                                 $ns = $this->mTitle->getNamespace();
3246                         }
3247                         $title = Title::newFromText( $part1, $ns );
3248                         if ( $title ) {
3249                                 $titleText = $title->getPrefixedText();
3250                                 # Check for language variants if the template is not found
3251                                 if ( $wgContLang->hasVariants() && $title->getArticleID() == 0 ) {
3252                                         $wgContLang->findVariantLink( $part1, $title, true );
3253                                 }
3254                                 # Do recursion depth check
3255                                 $limit = $this->mOptions->getMaxTemplateDepth();
3256                                 if ( $frame->depth >= $limit ) {
3257                                         $found = true;
3258                                         $text = '<span class="error">'
3259                                                 . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit )
3260                                                 . '</span>';
3261                                 }
3262                         }
3263                 }
3264
3265                 # Load from database
3266                 if ( !$found && $title ) {
3267                         wfProfileIn( __METHOD__ . '-loadtpl' );
3268                         if ( !$title->isExternal() ) {
3269                                 if ( $title->getNamespace() == NS_SPECIAL
3270                                         && $this->mOptions->getAllowSpecialInclusion()
3271                                         && $this->ot['html'] )
3272                                 {
3273                                         $text = SpecialPageFactory::capturePath( $title );
3274                                         if ( is_string( $text ) ) {
3275                                                 $found = true;
3276                                                 $isHTML = true;
3277                                                 $this->disableCache();
3278                                         }
3279                                 } elseif ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3280                                         $found = false; # access denied
3281                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3282                                 } else {
3283                                         list( $text, $title ) = $this->getTemplateDom( $title );
3284                                         if ( $text !== false ) {
3285                                                 $found = true;
3286                                                 $isChildObj = true;
3287                                         }
3288                                 }
3289
3290                                 # If the title is valid but undisplayable, make a link to it
3291                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3292                                         $text = "[[:$titleText]]";
3293                                         $found = true;
3294                                 }
3295                         } elseif ( $title->isTrans() ) {
3296                                 # Interwiki transclusion
3297                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3298                                         $text = $this->interwikiTransclude( $title, 'render' );
3299                                         $isHTML = true;
3300                                 } else {
3301                                         $text = $this->interwikiTransclude( $title, 'raw' );
3302                                         # Preprocess it like a template
3303                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3304                                         $isChildObj = true;
3305                                 }
3306                                 $found = true;
3307                         }
3308
3309                         # Do infinite loop check
3310                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3311                         if ( !$frame->loopCheck( $title ) ) {
3312                                 $found = true;
3313                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3314                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3315                         }
3316                         wfProfileOut( __METHOD__ . '-loadtpl' );
3317                 }
3318
3319                 # If we haven't found text to substitute by now, we're done
3320                 # Recover the source wikitext and return it
3321                 if ( !$found ) {
3322                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3323                         wfProfileOut( __METHOD__ );
3324                         return array( 'object' => $text );
3325                 }
3326
3327                 # Expand DOM-style return values in a child frame
3328                 if ( $isChildObj ) {
3329                         # Clean up argument array
3330                         $newFrame = $frame->newChild( $args, $title );
3331
3332                         if ( $nowiki ) {
3333                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3334                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3335                                 # Expansion is eligible for the empty-frame cache
3336                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3337                                         $text = $this->mTplExpandCache[$titleText];
3338                                 } else {
3339                                         $text = $newFrame->expand( $text );
3340                                         $this->mTplExpandCache[$titleText] = $text;
3341                                 }
3342                         } else {
3343                                 # Uncached expansion
3344                                 $text = $newFrame->expand( $text );
3345                         }
3346                 }
3347                 if ( $isLocalObj && $nowiki ) {
3348                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3349                         $isLocalObj = false;
3350                 }
3351
3352                 # Replace raw HTML by a placeholder
3353                 # Add a blank line preceding, to prevent it from mucking up
3354                 # immediately preceding headings
3355                 if ( $isHTML ) {
3356                         $text = "\n\n" . $this->insertStripItem( $text );
3357                 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3358                         # Escape nowiki-style return values
3359                         $text = wfEscapeWikiText( $text );
3360                 } elseif ( is_string( $text )
3361                         && !$piece['lineStart']
3362                         && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3363                 {
3364                         # Bug 529: if the template begins with a table or block-level
3365                         # element, it should be treated as beginning a new line.
3366                         # This behaviour is somewhat controversial.
3367                         $text = "\n" . $text;
3368                 }
3369
3370                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3371                         # Error, oversize inclusion
3372                         if ( $titleText !== false ) {
3373                                 # Make a working, properly escaped link if possible (bug 23588)
3374                                 $text = "[[:$titleText]]";
3375                         } else {
3376                                 # This will probably not be a working link, but at least it may
3377                                 # provide some hint of where the problem is
3378                                 preg_replace( '/^:/', '', $originalTitle );
3379                                 $text = "[[:$originalTitle]]";
3380                         }
3381                         $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3382                         $this->limitationWarn( 'post-expand-template-inclusion' );
3383                 }
3384
3385                 if ( $isLocalObj ) {
3386                         $ret = array( 'object' => $text );
3387                 } else {
3388                         $ret = array( 'text' => $text );
3389                 }
3390
3391                 wfProfileOut( __METHOD__ );
3392                 return $ret;
3393         }
3394
3395         /**
3396          * Get the semi-parsed DOM representation of a template with a given title,
3397          * and its redirect destination title. Cached.
3398          *
3399          * @return array
3400          */
3401         function getTemplateDom( $title ) {
3402                 $cacheTitle = $title;
3403                 $titleText = $title->getPrefixedDBkey();
3404
3405                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3406                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3407                         $title = Title::makeTitle( $ns, $dbk );
3408                         $titleText = $title->getPrefixedDBkey();
3409                 }
3410                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3411                         return array( $this->mTplDomCache[$titleText], $title );
3412                 }
3413
3414                 # Cache miss, go to the database
3415                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3416
3417                 if ( $text === false ) {
3418                         $this->mTplDomCache[$titleText] = false;
3419                         return array( false, $title );
3420                 }
3421
3422                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3423                 $this->mTplDomCache[ $titleText ] = $dom;
3424
3425                 if ( !$title->equals( $cacheTitle ) ) {
3426                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3427                                 array( $title->getNamespace(), $cdb = $title->getDBkey() );
3428                 }
3429
3430                 return array( $dom, $title );
3431         }
3432
3433         /**
3434          * Fetch the unparsed text of a template and register a reference to it.
3435          * @param Title $title
3436          * @return Array ( string or false, Title )
3437          */
3438         function fetchTemplateAndTitle( $title ) {
3439                 $templateCb = $this->mOptions->getTemplateCallback(); # Defaults to Parser::statelessFetchTemplate()
3440                 $stuff = call_user_func( $templateCb, $title, $this );
3441                 $text = $stuff['text'];
3442                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3443                 if ( isset( $stuff['deps'] ) ) {
3444                         foreach ( $stuff['deps'] as $dep ) {
3445                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3446                         }
3447                 }
3448                 return array( $text, $finalTitle );
3449         }
3450
3451         /**
3452          * Fetch the unparsed text of a template and register a reference to it.
3453          * @param Title $title
3454          * @return mixed string or false
3455          */
3456         function fetchTemplate( $title ) {
3457                 $rv = $this->fetchTemplateAndTitle( $title );
3458                 return $rv[0];
3459         }
3460
3461         /**
3462          * Static function to get a template
3463          * Can be overridden via ParserOptions::setTemplateCallback().
3464          *
3465          * @return array
3466          */
3467         static function statelessFetchTemplate( $title, $parser = false ) {
3468                 $text = $skip = false;
3469                 $finalTitle = $title;
3470                 $deps = array();
3471
3472                 # Loop to fetch the article, with up to 1 redirect
3473                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3474                         # Give extensions a chance to select the revision instead
3475                         $id = false; # Assume current
3476                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle',
3477                                 array( $parser, $title, &$skip, &$id ) );
3478
3479                         if ( $skip ) {
3480                                 $text = false;
3481                                 $deps[] = array(
3482                                         'title'         => $title,
3483                                         'page_id'       => $title->getArticleID(),
3484                                         'rev_id'        => null
3485                                 );
3486                                 break;
3487                         }
3488                         # Get the revision
3489                         $rev = $id
3490                                 ? Revision::newFromId( $id )
3491                                 : Revision::newFromTitle( $title );
3492                         $rev_id = $rev ? $rev->getId() : 0;
3493                         # If there is no current revision, there is no page
3494                         if ( $id === false && !$rev ) {
3495                                 $linkCache = LinkCache::singleton();
3496                                 $linkCache->addBadLinkObj( $title );
3497                         }
3498
3499                         $deps[] = array(
3500                                 'title'         => $title,
3501                                 'page_id'       => $title->getArticleID(),
3502                                 'rev_id'        => $rev_id );
3503                         if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3504                                 # We fetched a rev from a different title; register it too...
3505                                 $deps[] = array(
3506                                         'title'         => $rev->getTitle(),
3507                                         'page_id'       => $rev->getPage(),
3508                                         'rev_id'        => $rev_id );
3509                         }
3510
3511                         if ( $rev ) {
3512                                 $text = $rev->getText();
3513                         } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3514                                 global $wgContLang;
3515                                 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3516                                 if ( !$message->exists() ) {
3517                                         $text = false;
3518                                         break;
3519                                 }
3520                                 $text = $message->plain();
3521                         } else {
3522                                 break;
3523                         }
3524                         if ( $text === false ) {
3525                                 break;
3526                         }
3527                         # Redirect?
3528                         $finalTitle = $title;
3529                         $title = Title::newFromRedirect( $text );
3530                 }
3531                 return array(
3532                         'text' => $text,
3533                         'finalTitle' => $finalTitle,
3534                         'deps' => $deps );
3535         }
3536
3537         /**
3538          * Fetch a file and its title and register a reference to it.
3539          * @param Title $title
3540          * @param string $time MW timestamp
3541          * @param string $sha1 base 36 SHA-1
3542          * @return mixed File or false
3543          */
3544         function fetchFile( $title, $time = false, $sha1 = false ) {
3545                 $res = $this->fetchFileAndTitle( $title, $time, $sha1 );
3546                 return $res[0];
3547         }
3548
3549         /**
3550          * Fetch a file and its title and register a reference to it.
3551          * @param Title $title
3552          * @param string $time MW timestamp
3553          * @param string $sha1 base 36 SHA-1
3554          * @return Array ( File or false, Title of file )
3555          */
3556         function fetchFileAndTitle( $title, $time = false, $sha1 = false ) {
3557                 if ( $time === '0' ) {
3558                         $file = false; // broken thumbnail forced by hook
3559                 } elseif ( $sha1 ) { // get by (sha1,timestamp)
3560                         $file = RepoGroup::singleton()->findFileFromKey( $sha1, array( 'time' => $time ) );
3561                 } else { // get by (name,timestamp)
3562                         $file = wfFindFile( $title, array( 'time' => $time ) );
3563                 }
3564                 $time = $file ? $file->getTimestamp() : false;
3565                 $sha1 = $file ? $file->getSha1() : false;
3566                 # Register the file as a dependency...
3567                 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3568                 if ( $file && !$title->equals( $file->getTitle() ) ) {
3569                         # We fetched a rev from a different title; register it too...
3570                         $this->mOutput->addImage( $file->getTitle()->getDBkey(), $time, $sha1 );
3571                         # Update fetched file title
3572                         $title = $file->getTitle();
3573                 }
3574                 return array( $file, $title );
3575         }
3576
3577         /**
3578          * Transclude an interwiki link.
3579          *
3580          * @return string
3581          */
3582         function interwikiTransclude( $title, $action ) {
3583                 global $wgEnableScaryTranscluding;
3584
3585                 if ( !$wgEnableScaryTranscluding ) {
3586                         return wfMsgForContent('scarytranscludedisabled');
3587                 }
3588
3589                 $url = $title->getFullUrl( "action=$action" );
3590
3591                 if ( strlen( $url ) > 255 ) {
3592                         return wfMsgForContent( 'scarytranscludetoolong' );
3593                 }
3594                 return $this->fetchScaryTemplateMaybeFromCache( $url );
3595         }
3596
3597         /**
3598          * @param $url string
3599          * @return Mixed|String
3600          */
3601         function fetchScaryTemplateMaybeFromCache( $url ) {
3602                 global $wgTranscludeCacheExpiry;
3603                 $dbr = wfGetDB( DB_SLAVE );
3604                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3605                 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3606                                 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3607                 if ( $obj ) {
3608                         return $obj->tc_contents;
3609                 }
3610
3611                 $text = Http::get( $url );
3612                 if ( !$text ) {
3613                         return wfMsgForContent( 'scarytranscludefailed', $url );
3614                 }
3615
3616                 $dbw = wfGetDB( DB_MASTER );
3617                 $dbw->replace( 'transcache', array('tc_url'), array(
3618                         'tc_url' => $url,
3619                         'tc_time' => $dbw->timestamp( time() ),
3620                         'tc_contents' => $text)
3621                 );
3622                 return $text;
3623         }
3624
3625         /**
3626          * Triple brace replacement -- used for template arguments
3627          * @private
3628          *
3629          * @param $peice array
3630          * @param $frame PPFrame
3631          *
3632          * @return array
3633          */
3634         function argSubstitution( $piece, $frame ) {
3635                 wfProfileIn( __METHOD__ );
3636
3637                 $error = false;
3638                 $parts = $piece['parts'];
3639                 $nameWithSpaces = $frame->expand( $piece['title'] );
3640                 $argName = trim( $nameWithSpaces );
3641                 $object = false;
3642                 $text = $frame->getArgument( $argName );
3643                 if (  $text === false && $parts->getLength() > 0
3644                   && (
3645                         $this->ot['html']
3646                         || $this->ot['pre']
3647                         || ( $this->ot['wiki'] && $frame->isTemplate() )
3648                   )
3649                 ) {
3650                         # No match in frame, use the supplied default
3651                         $object = $parts->item( 0 )->getChildren();
3652                 }
3653                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3654                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3655                         $this->limitationWarn( 'post-expand-template-argument' );
3656                 }
3657
3658                 if ( $text === false && $object === false ) {
3659                         # No match anywhere
3660                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3661                 }
3662                 if ( $error !== false ) {
3663                         $text .= $error;
3664                 }
3665                 if ( $object !== false ) {
3666                         $ret = array( 'object' => $object );
3667                 } else {
3668                         $ret = array( 'text' => $text );
3669                 }
3670
3671                 wfProfileOut( __METHOD__ );
3672                 return $ret;
3673         }
3674
3675         /**
3676          * Return the text to be used for a given extension tag.
3677          * This is the ghost of strip().
3678          *
3679          * @param $params Associative array of parameters:
3680          *     name       PPNode for the tag name
3681          *     attr       PPNode for unparsed text where tag attributes are thought to be
3682          *     attributes Optional associative array of parsed attributes
3683          *     inner      Contents of extension element
3684          *     noClose    Original text did not have a close tag
3685          * @param $frame PPFrame
3686          *
3687          * @return string
3688          */
3689         function extensionSubstitution( $params, $frame ) {
3690                 $name = $frame->expand( $params['name'] );
3691                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3692                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3693                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3694
3695                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3696                         ( $this->ot['html'] || $this->ot['pre'] );
3697                 if ( $isFunctionTag ) {
3698                         $markerType = 'none';
3699                 } else {
3700                         $markerType = 'general';
3701                 }
3702                 if ( $this->ot['html'] || $isFunctionTag ) {
3703                         $name = strtolower( $name );
3704                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3705                         if ( isset( $params['attributes'] ) ) {
3706                                 $attributes = $attributes + $params['attributes'];
3707                         }
3708
3709                         if ( isset( $this->mTagHooks[$name] ) ) {
3710                                 # Workaround for PHP bug 35229 and similar
3711                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3712                                         throw new MWException( "Tag hook for $name is not callable\n" );
3713                                 }
3714                                 $output = call_user_func_array( $this->mTagHooks[$name],
3715                                         array( $content, $attributes, $this, $frame ) );
3716                         } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3717                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3718                                 if ( !is_callable( $callback ) ) {
3719                                         throw new MWException( "Tag hook for $name is not callable\n" );
3720                                 }
3721
3722                                 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3723                         } else {
3724                                 $output = '<span class="error">Invalid tag extension name: ' .
3725                                         htmlspecialchars( $name ) . '</span>';
3726                         }
3727
3728                         if ( is_array( $output ) ) {
3729                                 # Extract flags to local scope (to override $markerType)
3730                                 $flags = $output;
3731                                 $output = $flags[0];
3732                                 unset( $flags[0] );
3733                                 extract( $flags );
3734                         }
3735                 } else {
3736                         if ( is_null( $attrText ) ) {
3737                                 $attrText = '';
3738                         }
3739                         if ( isset( $params['attributes'] ) ) {
3740                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3741                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3742                                                 htmlspecialchars( $attrValue ) . '"';
3743                                 }
3744                         }
3745                         if ( $content === null ) {
3746                                 $output = "<$name$attrText/>";
3747                         } else {
3748                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3749                                 $output = "<$name$attrText>$content$close";
3750                         }
3751                 }
3752
3753                 if ( $markerType === 'none' ) {
3754                         return $output;
3755                 } elseif ( $markerType === 'nowiki' ) {
3756                         $this->mStripState->addNoWiki( $marker, $output );
3757                 } elseif ( $markerType === 'general' ) {
3758                         $this->mStripState->addGeneral( $marker, $output );
3759                 } else {
3760                         throw new MWException( __METHOD__.': invalid marker type' );
3761                 }
3762                 return $marker;
3763         }
3764
3765         /**
3766          * Increment an include size counter
3767          *
3768          * @param $type String: the type of expansion
3769          * @param $size Integer: the size of the text
3770          * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
3771          */
3772         function incrementIncludeSize( $type, $size ) {
3773                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3774                         return false;
3775                 } else {
3776                         $this->mIncludeSizes[$type] += $size;
3777                         return true;
3778                 }
3779         }
3780
3781         /**
3782          * Increment the expensive function count
3783          *
3784          * @return Boolean: false if the limit has been exceeded
3785          */
3786         function incrementExpensiveFunctionCount() {
3787                 global $wgExpensiveParserFunctionLimit;
3788                 $this->mExpensiveFunctionCount++;
3789                 if ( $this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit ) {
3790                         return true;
3791                 }
3792                 return false;
3793         }
3794
3795         /**
3796          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3797          * Fills $this->mDoubleUnderscores, returns the modified text
3798          */
3799         function doDoubleUnderscore( $text ) {
3800                 wfProfileIn( __METHOD__ );
3801
3802                 # The position of __TOC__ needs to be recorded
3803                 $mw = MagicWord::get( 'toc' );
3804                 if ( $mw->match( $text ) ) {
3805                         $this->mShowToc = true;
3806                         $this->mForceTocPosition = true;
3807
3808                         # Set a placeholder. At the end we'll fill it in with the TOC.
3809                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3810
3811                         # Only keep the first one.
3812                         $text = $mw->replace( '', $text );
3813                 }
3814
3815                 # Now match and remove the rest of them
3816                 $mwa = MagicWord::getDoubleUnderscoreArray();
3817                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3818
3819                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3820                         $this->mOutput->mNoGallery = true;
3821                 }
3822                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3823                         $this->mShowToc = false;
3824                 }
3825                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3826                         $this->addTrackingCategory( 'hidden-category-category' );
3827                 }
3828                 # (bug 8068) Allow control over whether robots index a page.
3829                 #
3830                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3831                 # is not desirable, the last one on the page should win.
3832                 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3833                         $this->mOutput->setIndexPolicy( 'noindex' );
3834                         $this->addTrackingCategory( 'noindex-category' );
3835                 }
3836                 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3837                         $this->mOutput->setIndexPolicy( 'index' );
3838                         $this->addTrackingCategory( 'index-category' );
3839                 }
3840
3841                 # Cache all double underscores in the database
3842                 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3843                         $this->mOutput->setProperty( $key, '' );
3844                 }
3845
3846                 wfProfileOut( __METHOD__ );
3847                 return $text;
3848         }
3849
3850         /**
3851          * Add a tracking category, getting the title from a system message,
3852          * or print a debug message if the title is invalid.
3853          *
3854          * @param $msg String: message key
3855          * @return Boolean: whether the addition was successful
3856          */
3857         protected function addTrackingCategory( $msg ) {
3858                 $cat = wfMsgForContent( $msg );
3859
3860                 # Allow tracking categories to be disabled by setting them to "-"
3861                 if ( $cat === '-' ) {
3862                         return false;
3863                 }
3864
3865                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3866                 if ( $containerCategory ) {
3867                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3868                         return true;
3869                 } else {
3870                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3871                         return false;
3872                 }
3873         }
3874
3875         /**
3876          * This function accomplishes several tasks:
3877          * 1) Auto-number headings if that option is enabled
3878          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3879          * 3) Add a Table of contents on the top for users who have enabled the option
3880          * 4) Auto-anchor headings
3881          *
3882          * It loops through all headlines, collects the necessary data, then splits up the
3883          * string and re-inserts the newly formatted headlines.
3884          *
3885          * @param $text String
3886          * @param $origText String: original, untouched wikitext
3887          * @param $isMain Boolean
3888          * @private
3889          */
3890         function formatHeadings( $text, $origText, $isMain=true ) {
3891                 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3892
3893                 # Inhibit editsection links if requested in the page
3894                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3895                         $showEditLink = 0;
3896                 } else {
3897                         $showEditLink = $this->mOptions->getEditSection();
3898                 }
3899                 if ( $showEditLink ) {
3900                         $this->mOutput->setEditSectionTokens( true );
3901                 }
3902
3903                 # Get all headlines for numbering them and adding funky stuff like [edit]
3904                 # links - this is for later, but we need the number of headlines right now
3905                 $matches = array();
3906                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3907
3908                 # if there are fewer than 4 headlines in the article, do not show TOC
3909                 # unless it's been explicitly enabled.
3910                 $enoughToc = $this->mShowToc &&
3911                         ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3912
3913                 # Allow user to stipulate that a page should have a "new section"
3914                 # link added via __NEWSECTIONLINK__
3915                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3916                         $this->mOutput->setNewSection( true );
3917                 }
3918
3919                 # Allow user to remove the "new section"
3920                 # link via __NONEWSECTIONLINK__
3921                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3922                         $this->mOutput->hideNewSection( true );
3923                 }
3924
3925                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3926                 # override above conditions and always show TOC above first header
3927                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3928                         $this->mShowToc = true;
3929                         $enoughToc = true;
3930                 }
3931
3932                 # headline counter
3933                 $headlineCount = 0;
3934                 $numVisible = 0;
3935
3936                 # Ugh .. the TOC should have neat indentation levels which can be
3937                 # passed to the skin functions. These are determined here
3938                 $toc = '';
3939                 $full = '';
3940                 $head = array();
3941                 $sublevelCount = array();
3942                 $levelCount = array();
3943                 $level = 0;
3944                 $prevlevel = 0;
3945                 $toclevel = 0;
3946                 $prevtoclevel = 0;
3947                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3948                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3949                 $oldType = $this->mOutputType;
3950                 $this->setOutputType( self::OT_WIKI );
3951                 $frame = $this->getPreprocessor()->newFrame();
3952                 $root = $this->preprocessToDom( $origText );
3953                 $node = $root->getFirstChild();
3954                 $byteOffset = 0;
3955                 $tocraw = array();
3956                 $refers = array();
3957
3958                 foreach ( $matches[3] as $headline ) {
3959                         $isTemplate = false;
3960                         $titleText = false;
3961                         $sectionIndex = false;
3962                         $numbering = '';
3963                         $markerMatches = array();
3964                         if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
3965                                 $serial = $markerMatches[1];
3966                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3967                                 $isTemplate = ( $titleText != $baseTitleText );
3968                                 $headline = preg_replace( "/^$markerRegex/", "", $headline );
3969                         }
3970
3971                         if ( $toclevel ) {
3972                                 $prevlevel = $level;
3973                         }
3974                         $level = $matches[1][$headlineCount];
3975
3976                         if ( $level > $prevlevel ) {
3977                                 # Increase TOC level
3978                                 $toclevel++;
3979                                 $sublevelCount[$toclevel] = 0;
3980                                 if ( $toclevel<$wgMaxTocLevel ) {
3981                                         $prevtoclevel = $toclevel;
3982                                         $toc .= Linker::tocIndent();
3983                                         $numVisible++;
3984                                 }
3985                         } elseif ( $level < $prevlevel && $toclevel > 1 ) {
3986                                 # Decrease TOC level, find level to jump to
3987
3988                                 for ( $i = $toclevel; $i > 0; $i-- ) {
3989                                         if ( $levelCount[$i] == $level ) {
3990                                                 # Found last matching level
3991                                                 $toclevel = $i;
3992                                                 break;
3993                                         } elseif ( $levelCount[$i] < $level ) {
3994                                                 # Found first matching level below current level
3995                                                 $toclevel = $i + 1;
3996                                                 break;
3997                                         }
3998                                 }
3999                                 if ( $i == 0 ) {
4000                                         $toclevel = 1;
4001                                 }
4002                                 if ( $toclevel<$wgMaxTocLevel ) {
4003                                         if ( $prevtoclevel < $wgMaxTocLevel ) {
4004                                                 # Unindent only if the previous toc level was shown :p
4005                                                 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4006                                                 $prevtoclevel = $toclevel;
4007                                         } else {
4008                                                 $toc .= Linker::tocLineEnd();
4009                                         }
4010                                 }
4011                         } else {
4012                                 # No change in level, end TOC line
4013                                 if ( $toclevel<$wgMaxTocLevel ) {
4014                                         $toc .= Linker::tocLineEnd();
4015                                 }
4016                         }
4017
4018                         $levelCount[$toclevel] = $level;
4019
4020                         # count number of headlines for each level
4021                         @$sublevelCount[$toclevel]++;
4022                         $dot = 0;
4023                         for( $i = 1; $i <= $toclevel; $i++ ) {
4024                                 if ( !empty( $sublevelCount[$i] ) ) {
4025                                         if ( $dot ) {
4026                                                 $numbering .= '.';
4027                                         }
4028                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
4029                                         $dot = 1;
4030                                 }
4031                         }
4032
4033                         # The safe header is a version of the header text safe to use for links
4034                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4035                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
4036
4037                         # Remove link placeholders by the link text.
4038                         #     <!--LINK number-->
4039                         # turns into
4040                         #     link text with suffix
4041                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
4042
4043                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
4044                         $tocline = preg_replace(
4045                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
4046                                 array( '',                          '<$1>' ),
4047                                 $safeHeadline
4048                         );
4049                         $tocline = trim( $tocline );
4050
4051                         # For the anchor, strip out HTML-y stuff period
4052                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
4053                         $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4054
4055                         # Save headline for section edit hint before it's escaped
4056                         $headlineHint = $safeHeadline;
4057
4058                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
4059                                 # For reverse compatibility, provide an id that's
4060                                 # HTML4-compatible, like we used to.
4061                                 #
4062                                 # It may be worth noting, academically, that it's possible for
4063                                 # the legacy anchor to conflict with a non-legacy headline
4064                                 # anchor on the page.  In this case likely the "correct" thing
4065                                 # would be to either drop the legacy anchors or make sure
4066                                 # they're numbered first.  However, this would require people
4067                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4068                                 # manually, so let's not bother worrying about it.
4069                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
4070                                         array( 'noninitial', 'legacy' ) );
4071                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4072
4073                                 if ( $legacyHeadline == $safeHeadline ) {
4074                                         # No reason to have both (in fact, we can't)
4075                                         $legacyHeadline = false;
4076                                 }
4077                         } else {
4078                                 $legacyHeadline = false;
4079                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
4080                                         'noninitial' );
4081                         }
4082
4083                         # HTML names must be case-insensitively unique (bug 10721).
4084                         # This does not apply to Unicode characters per
4085                         # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
4086                         # FIXME: We may be changing them depending on the current locale.
4087                         $arrayKey = strtolower( $safeHeadline );
4088                         if ( $legacyHeadline === false ) {
4089                                 $legacyArrayKey = false;
4090                         } else {
4091                                 $legacyArrayKey = strtolower( $legacyHeadline );
4092                         }
4093
4094                         # count how many in assoc. array so we can track dupes in anchors
4095                         if ( isset( $refers[$arrayKey] ) ) {
4096                                 $refers[$arrayKey]++;
4097                         } else {
4098                                 $refers[$arrayKey] = 1;
4099                         }
4100                         if ( isset( $refers[$legacyArrayKey] ) ) {
4101                                 $refers[$legacyArrayKey]++;
4102                         } else {
4103                                 $refers[$legacyArrayKey] = 1;
4104                         }
4105
4106                         # Don't number the heading if it is the only one (looks silly)
4107                         if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4108                                 # the two are different if the line contains a link
4109                                 $headline = $numbering . ' ' . $headline;
4110                         }
4111
4112                         # Create the anchor for linking from the TOC to the section
4113                         $anchor = $safeHeadline;
4114                         $legacyAnchor = $legacyHeadline;
4115                         if ( $refers[$arrayKey] > 1 ) {
4116                                 $anchor .= '_' . $refers[$arrayKey];
4117                         }
4118                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
4119                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
4120                         }
4121                         if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4122                                 $toc .= Linker::tocLine( $anchor, $tocline,
4123                                         $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4124                         }
4125
4126                         # Add the section to the section tree
4127                         # Find the DOM node for this header
4128                         while ( $node && !$isTemplate ) {
4129                                 if ( $node->getName() === 'h' ) {
4130                                         $bits = $node->splitHeading();
4131                                         if ( $bits['i'] == $sectionIndex ) {
4132                                                 break;
4133                                         }
4134                                 }
4135                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4136                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4137                                 $node = $node->getNextSibling();
4138                         }
4139                         $tocraw[] = array(
4140                                 'toclevel' => $toclevel,
4141                                 'level' => $level,
4142                                 'line' => $tocline,
4143                                 'number' => $numbering,
4144                                 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4145                                 'fromtitle' => $titleText,
4146                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
4147                                 'anchor' => $anchor,
4148                         );
4149
4150                         # give headline the correct <h#> tag
4151                         if ( $showEditLink && $sectionIndex !== false ) {
4152                                 // Output edit section links as markers with styles that can be customized by skins
4153                                 if ( $isTemplate ) {
4154                                         # Put a T flag in the section identifier, to indicate to extractSections()
4155                                         # that sections inside <includeonly> should be counted.
4156                                         $editlinkArgs = array( $titleText, "T-$sectionIndex"/*, null */ );
4157                                 } else {
4158                                         $editlinkArgs = array( $this->mTitle->getPrefixedText(), $sectionIndex, $headlineHint );
4159                                 }
4160                                 // We use a bit of pesudo-xml for editsection markers. The language converter is run later on
4161                                 // Using a UNIQ style marker leads to the converter screwing up the tokens when it converts stuff
4162                                 // And trying to insert strip tags fails too. At this point all real inputted tags have already been escaped
4163                                 // so we don't have to worry about a user trying to input one of these markers directly.
4164                                 // We use a page and section attribute to stop the language converter from converting these important bits
4165                                 // of data, but put the headline hint inside a content block because the language converter is supposed to
4166                                 // be able to convert that piece of data.
4167                                 $editlink = '<mw:editsection page="' . htmlspecialchars($editlinkArgs[0]);
4168                                 $editlink .= '" section="' . htmlspecialchars($editlinkArgs[1]) .'"';
4169                                 if ( isset($editlinkArgs[2]) ) {
4170                                         $editlink .= '>' . $editlinkArgs[2] . '</mw:editsection>';
4171                                 } else {
4172                                         $editlink .= '/>';
4173                                 }
4174                         } else {
4175                                 $editlink = '';
4176                         }
4177                         $head[$headlineCount] = Linker::makeHeadline( $level,
4178                                 $matches['attrib'][$headlineCount], $anchor, $headline,
4179                                 $editlink, $legacyAnchor );
4180
4181                         $headlineCount++;
4182                 }
4183
4184                 $this->setOutputType( $oldType );
4185
4186                 # Never ever show TOC if no headers
4187                 if ( $numVisible < 1 ) {
4188                         $enoughToc = false;
4189                 }
4190
4191                 if ( $enoughToc ) {
4192                         if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4193                                 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4194                         }
4195                         $toc = Linker::tocList( $toc, $this->mOptions->getUserLang() );
4196                         $this->mOutput->setTOCHTML( $toc );
4197                 }
4198
4199                 if ( $isMain ) {
4200                         $this->mOutput->setSections( $tocraw );
4201                 }
4202
4203                 # split up and insert constructed headlines
4204
4205                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
4206                 $i = 0;
4207
4208                 foreach ( $blocks as $block ) {
4209                         if ( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
4210                                 # This is the [edit] link that appears for the top block of text when
4211                                 # section editing is enabled
4212
4213                                 # Disabled because it broke block formatting
4214                                 # For example, a bullet point in the top line
4215                                 # $full .= $sk->editSectionLink(0);
4216                         }
4217                         $full .= $block;
4218                         if ( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
4219                                 # Top anchor now in skin
4220                                 $full = $full.$toc;
4221                         }
4222
4223                         if ( !empty( $head[$i] ) ) {
4224                                 $full .= $head[$i];
4225                         }
4226                         $i++;
4227                 }
4228                 if ( $this->mForceTocPosition ) {
4229                         return str_replace( '<!--MWTOC-->', $toc, $full );
4230                 } else {
4231                         return $full;
4232                 }
4233         }
4234
4235         /**
4236          * Transform wiki markup when saving a page by doing \r\n -> \n
4237          * conversion, substitting signatures, {{subst:}} templates, etc.
4238          *
4239          * @param $text String: the text to transform
4240          * @param $title Title: the Title object for the current article
4241          * @param $user User: the User object describing the current user
4242          * @param $options ParserOptions: parsing options
4243          * @param $clearState Boolean: whether to clear the parser state first
4244          * @return String: the altered wiki markup
4245          */
4246         public function preSaveTransform( $text, Title $title, User $user, ParserOptions $options, $clearState = true ) {
4247                 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4248                 $this->setUser( $user );
4249
4250                 $pairs = array(
4251                         "\r\n" => "\n",
4252                 );
4253                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4254                 if( $options->getPreSaveTransform() ) {
4255                         $text = $this->pstPass2( $text, $user );
4256                 }
4257                 $text = $this->mStripState->unstripBoth( $text );
4258
4259                 $this->setUser( null ); #Reset
4260
4261                 return $text;
4262         }
4263
4264         /**
4265          * Pre-save transform helper function
4266          * @private
4267          */
4268         function pstPass2( $text, $user ) {
4269                 global $wgContLang, $wgLocaltimezone;
4270
4271                 # Note: This is the timestamp saved as hardcoded wikitext to
4272                 # the database, we use $wgContLang here in order to give
4273                 # everyone the same signature and use the default one rather
4274                 # than the one selected in each user's preferences.
4275                 # (see also bug 12815)
4276                 $ts = $this->mOptions->getTimestamp();
4277                 if ( isset( $wgLocaltimezone ) ) {
4278                         $tz = $wgLocaltimezone;
4279                 } else {
4280                         $tz = date_default_timezone_get();
4281                 }
4282
4283                 $unixts = wfTimestamp( TS_UNIX, $ts );
4284                 $oldtz = date_default_timezone_get();
4285                 date_default_timezone_set( $tz );
4286                 $ts = date( 'YmdHis', $unixts );
4287                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4288
4289                 # Allow translation of timezones through wiki. date() can return
4290                 # whatever crap the system uses, localised or not, so we cannot
4291                 # ship premade translations.
4292                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4293                 $msg = wfMessage( $key )->inContentLanguage();
4294                 if ( $msg->exists() ) {
4295                         $tzMsg = $msg->text();
4296                 }
4297
4298                 date_default_timezone_set( $oldtz );
4299
4300                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4301
4302                 # Variable replacement
4303                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4304                 $text = $this->replaceVariables( $text );
4305
4306                 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4307                 # which may corrupt this parser instance via its wfMsgExt( parsemag ) call-
4308
4309                 # Signatures
4310                 $sigText = $this->getUserSig( $user );
4311                 $text = strtr( $text, array(
4312                         '~~~~~' => $d,
4313                         '~~~~' => "$sigText $d",
4314                         '~~~' => $sigText
4315                 ) );
4316
4317                 # Context links: [[|name]] and [[name (context)|]]
4318                 global $wgLegalTitleChars;
4319                 $tc = "[$wgLegalTitleChars]";
4320                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4321
4322                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4323                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
4324                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4325                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4326
4327                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4328                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4329                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4330                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4331
4332                 $t = $this->mTitle->getText();
4333                 $m = array();
4334                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4335                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4336                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4337                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4338                 } else {
4339                         # if there's no context, don't bother duplicating the title
4340                         $text = preg_replace( $p2, '[[\\1]]', $text );
4341                 }
4342
4343                 # Trim trailing whitespace
4344                 $text = rtrim( $text );
4345
4346                 return $text;
4347         }
4348
4349         /**
4350          * Fetch the user's signature text, if any, and normalize to
4351          * validated, ready-to-insert wikitext.
4352          * If you have pre-fetched the nickname or the fancySig option, you can
4353          * specify them here to save a database query.
4354          * Do not reuse this parser instance after calling getUserSig(),
4355          * as it may have changed if it's the $wgParser.
4356          *
4357          * @param $user User
4358          * @param $nickname String: nickname to use or false to use user's default nickname
4359          * @param $fancySig Boolean: whether the nicknname is the complete signature
4360          *                  or null to use default value
4361          * @return string
4362          */
4363         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4364                 global $wgMaxSigChars;
4365
4366                 $username = $user->getName();
4367
4368                 # If not given, retrieve from the user object.
4369                 if ( $nickname === false )
4370                         $nickname = $user->getOption( 'nickname' );
4371
4372                 if ( is_null( $fancySig ) ) {
4373                         $fancySig = $user->getBoolOption( 'fancysig' );
4374                 }
4375
4376                 $nickname = $nickname == null ? $username : $nickname;
4377
4378                 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4379                         $nickname = $username;
4380                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4381                 } elseif ( $fancySig !== false ) {
4382                         # Sig. might contain markup; validate this
4383                         if ( $this->validateSig( $nickname ) !== false ) {
4384                                 # Validated; clean up (if needed) and return it
4385                                 return $this->cleanSig( $nickname, true );
4386                         } else {
4387                                 # Failed to validate; fall back to the default
4388                                 $nickname = $username;
4389                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4390                         }
4391                 }
4392
4393                 # Make sure nickname doesnt get a sig in a sig
4394                 $nickname = $this->cleanSigInSig( $nickname );
4395
4396                 # If we're still here, make it a link to the user page
4397                 $userText = wfEscapeWikiText( $username );
4398                 $nickText = wfEscapeWikiText( $nickname );
4399                 if ( $user->isAnon() )  {
4400                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4401                 } else {
4402                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4403                 }
4404         }
4405
4406         /**
4407          * Check that the user's signature contains no bad XML
4408          *
4409          * @param $text String
4410          * @return mixed An expanded string, or false if invalid.
4411          */
4412         function validateSig( $text ) {
4413                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4414         }
4415
4416         /**
4417          * Clean up signature text
4418          *
4419          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4420          * 2) Substitute all transclusions
4421          *
4422          * @param $text String
4423          * @param $parsing Whether we're cleaning (preferences save) or parsing
4424          * @return String: signature text
4425          */
4426         function cleanSig( $text, $parsing = false ) {
4427                 if ( !$parsing ) {
4428                         global $wgTitle;
4429                         $this->mOptions = new ParserOptions;
4430                         $this->clearState();
4431                         $this->setTitle( $wgTitle );
4432                         $this->setOutputType = self::OT_PREPROCESS;
4433                 }
4434
4435                 # Option to disable this feature
4436                 if ( !$this->mOptions->getCleanSignatures() ) {
4437                         return $text;
4438                 }
4439
4440                 # FIXME: regex doesn't respect extension tags or nowiki
4441                 #  => Move this logic to braceSubstitution()
4442                 $substWord = MagicWord::get( 'subst' );
4443                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4444                 $substText = '{{' . $substWord->getSynonym( 0 );
4445
4446                 $text = preg_replace( $substRegex, $substText, $text );
4447                 $text = $this->cleanSigInSig( $text );
4448                 $dom = $this->preprocessToDom( $text );
4449                 $frame = $this->getPreprocessor()->newFrame();
4450                 $text = $frame->expand( $dom );
4451
4452                 if ( !$parsing ) {
4453                         $text = $this->mStripState->unstripBoth( $text );
4454                 }
4455
4456                 return $text;
4457         }
4458
4459         /**
4460          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4461          *
4462          * @param $text String
4463          * @return String: signature text with /~{3,5}/ removed
4464          */
4465         function cleanSigInSig( $text ) {
4466                 $text = preg_replace( '/~{3,5}/', '', $text );
4467                 return $text;
4468         }
4469
4470         /**
4471          * Set up some variables which are usually set up in parse()
4472          * so that an external function can call some class members with confidence
4473          */
4474         public function startExternalParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4475                 $this->startParse( $title, $options, $outputType, $clearState );
4476         }
4477
4478         private function startParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4479                 $this->setTitle( $title );
4480                 $this->mOptions = $options;
4481                 $this->setOutputType( $outputType );
4482                 if ( $clearState ) {
4483                         $this->clearState();
4484                 }
4485         }
4486
4487         /**
4488          * Wrapper for preprocess()
4489          *
4490          * @param $text String: the text to preprocess
4491          * @param $options ParserOptions: options
4492          * @param $title Title object or null to use $wgTitle
4493          * @return String
4494          */
4495         public function transformMsg( $text, $options, $title = null ) {
4496                 static $executing = false;
4497
4498                 # Guard against infinite recursion
4499                 if ( $executing ) {
4500                         return $text;
4501                 }
4502                 $executing = true;
4503
4504                 wfProfileIn( __METHOD__ );
4505                 if ( !$title ) {
4506                         global $wgTitle;
4507                         $title = $wgTitle;
4508                 }
4509                 if ( !$title ) {
4510                         # It's not uncommon having a null $wgTitle in scripts. See r80898
4511                         # Create a ghost title in such case
4512                         $title = Title::newFromText( 'Dwimmerlaik' );
4513                 }
4514                 $text = $this->preprocess( $text, $title, $options );
4515
4516                 $executing = false;
4517                 wfProfileOut( __METHOD__ );
4518                 return $text;
4519         }
4520
4521         /**
4522          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4523          * The callback should have the following form:
4524          *    function myParserHook( $text, $params, $parser, $frame ) { ... }
4525          *
4526          * Transform and return $text. Use $parser for any required context, e.g. use
4527          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4528          *
4529          * Hooks may return extended information by returning an array, of which the
4530          * first numbered element (index 0) must be the return string, and all other
4531          * entries are extracted into local variables within an internal function
4532          * in the Parser class.
4533          *
4534          * This interface (introduced r61913) appears to be undocumented, but
4535          * 'markerName' is used by some core tag hooks to override which strip
4536          * array their results are placed in. **Use great caution if attempting
4537          * this interface, as it is not documented and injudicious use could smash
4538          * private variables.**
4539          *
4540          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4541          * @param $callback Mixed: the callback function (and object) to use for the tag
4542          * @return The old value of the mTagHooks array associated with the hook
4543          */
4544         public function setHook( $tag, $callback ) {
4545                 $tag = strtolower( $tag );
4546                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4547                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4548                 $this->mTagHooks[$tag] = $callback;
4549                 if ( !in_array( $tag, $this->mStripList ) ) {
4550                         $this->mStripList[] = $tag;
4551                 }
4552
4553                 return $oldVal;
4554         }
4555
4556         /**
4557          * As setHook(), but letting the contents be parsed.
4558          *
4559          * Transparent tag hooks are like regular XML-style tag hooks, except they
4560          * operate late in the transformation sequence, on HTML instead of wikitext.
4561          *
4562          * This is probably obsoleted by things dealing with parser frames?
4563          * The only extension currently using it is geoserver.
4564          *
4565          * @since 1.10
4566          * @todo better document or deprecate this
4567          *
4568          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4569          * @param $callback Mixed: the callback function (and object) to use for the tag
4570          * @return The old value of the mTagHooks array associated with the hook
4571          */
4572         function setTransparentTagHook( $tag, $callback ) {
4573                 $tag = strtolower( $tag );
4574                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4575                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4576                 $this->mTransparentTagHooks[$tag] = $callback;
4577
4578                 return $oldVal;
4579         }
4580
4581         /**
4582          * Remove all tag hooks
4583          */
4584         function clearTagHooks() {
4585                 $this->mTagHooks = array();
4586                 $this->mStripList = $this->mDefaultStripList;
4587         }
4588
4589         /**
4590          * Create a function, e.g. {{sum:1|2|3}}
4591          * The callback function should have the form:
4592          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4593          *
4594          * Or with SFH_OBJECT_ARGS:
4595          *    function myParserFunction( $parser, $frame, $args ) { ... }
4596          *
4597          * The callback may either return the text result of the function, or an array with the text
4598          * in element 0, and a number of flags in the other elements. The names of the flags are
4599          * specified in the keys. Valid flags are:
4600          *   found                     The text returned is valid, stop processing the template. This
4601          *                             is on by default.
4602          *   nowiki                    Wiki markup in the return value should be escaped
4603          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4604          *
4605          * @param $id String: The magic word ID
4606          * @param $callback Mixed: the callback function (and object) to use
4607          * @param $flags Integer: a combination of the following flags:
4608          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4609          *
4610          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4611          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4612          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4613          *     the arguments, and to control the way they are expanded.
4614          *
4615          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4616          *     arguments, for instance:
4617          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4618          *
4619          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4620          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4621          *     working if/when this is changed.
4622          *
4623          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4624          *     expansion.
4625          *
4626          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4627          *     about the methods available in PPFrame and PPNode.
4628          *
4629          * @return The old callback function for this name, if any
4630          */
4631         public function setFunctionHook( $id, $callback, $flags = 0 ) {
4632                 global $wgContLang;
4633
4634                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4635                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4636
4637                 # Add to function cache
4638                 $mw = MagicWord::get( $id );
4639                 if ( !$mw )
4640                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4641
4642                 $synonyms = $mw->getSynonyms();
4643                 $sensitive = intval( $mw->isCaseSensitive() );
4644
4645                 foreach ( $synonyms as $syn ) {
4646                         # Case
4647                         if ( !$sensitive ) {
4648                                 $syn = $wgContLang->lc( $syn );
4649                         }
4650                         # Add leading hash
4651                         if ( !( $flags & SFH_NO_HASH ) ) {
4652                                 $syn = '#' . $syn;
4653                         }
4654                         # Remove trailing colon
4655                         if ( substr( $syn, -1, 1 ) === ':' ) {
4656                                 $syn = substr( $syn, 0, -1 );
4657                         }
4658                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4659                 }
4660                 return $oldVal;
4661         }
4662
4663         /**
4664          * Get all registered function hook identifiers
4665          *
4666          * @return Array
4667          */
4668         function getFunctionHooks() {
4669                 return array_keys( $this->mFunctionHooks );
4670         }
4671
4672         /**
4673          * Create a tag function, e.g. <test>some stuff</test>.
4674          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4675          * Unlike parser functions, their content is not preprocessed.
4676          */
4677         function setFunctionTagHook( $tag, $callback, $flags ) {
4678                 $tag = strtolower( $tag );
4679                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4680                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4681                         $this->mFunctionTagHooks[$tag] : null;
4682                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4683
4684                 if ( !in_array( $tag, $this->mStripList ) ) {
4685                         $this->mStripList[] = $tag;
4686                 }
4687
4688                 return $old;
4689         }
4690
4691         /**
4692          * FIXME: update documentation. makeLinkObj() is deprecated.
4693          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4694          * Placeholders created in Skin::makeLinkObj()
4695          * Returns an array of link CSS classes, indexed by PDBK.
4696          */
4697         function replaceLinkHolders( &$text, $options = 0 ) {
4698                 return $this->mLinkHolders->replace( $text );
4699         }
4700
4701         /**
4702          * Replace <!--LINK--> link placeholders with plain text of links
4703          * (not HTML-formatted).
4704          *
4705          * @param $text String
4706          * @return String
4707          */
4708         function replaceLinkHoldersText( $text ) {
4709                 return $this->mLinkHolders->replaceText( $text );
4710         }
4711
4712         /**
4713          * Renders an image gallery from a text with one line per image.
4714          * text labels may be given by using |-style alternative text. E.g.
4715          *   Image:one.jpg|The number "1"
4716          *   Image:tree.jpg|A tree
4717          * given as text will return the HTML of a gallery with two images,
4718          * labeled 'The number "1"' and
4719          * 'A tree'.
4720          *
4721          * @param string $text
4722          * @param array $param
4723          * @return string HTML
4724          */
4725         function renderImageGallery( $text, $params ) {
4726                 $ig = new ImageGallery();
4727                 $ig->setContextTitle( $this->mTitle );
4728                 $ig->setShowBytes( false );
4729                 $ig->setShowFilename( false );
4730                 $ig->setParser( $this );
4731                 $ig->setHideBadImages();
4732                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4733
4734                 if ( isset( $params['showfilename'] ) ) {
4735                         $ig->setShowFilename( true );
4736                 } else {
4737                         $ig->setShowFilename( false );
4738                 }
4739                 if ( isset( $params['caption'] ) ) {
4740                         $caption = $params['caption'];
4741                         $caption = htmlspecialchars( $caption );
4742                         $caption = $this->replaceInternalLinks( $caption );
4743                         $ig->setCaptionHtml( $caption );
4744                 }
4745                 if ( isset( $params['perrow'] ) ) {
4746                         $ig->setPerRow( $params['perrow'] );
4747                 }
4748                 if ( isset( $params['widths'] ) ) {
4749                         $ig->setWidths( $params['widths'] );
4750                 }
4751                 if ( isset( $params['heights'] ) ) {
4752                         $ig->setHeights( $params['heights'] );
4753                 }
4754
4755                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4756
4757                 $lines = StringUtils::explode( "\n", $text );
4758                 foreach ( $lines as $line ) {
4759                         # match lines like these:
4760                         # Image:someimage.jpg|This is some image
4761                         $matches = array();
4762                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4763                         # Skip empty lines
4764                         if ( count( $matches ) == 0 ) {
4765                                 continue;
4766                         }
4767
4768                         if ( strpos( $matches[0], '%' ) !== false ) {
4769                                 $matches[1] = rawurldecode( $matches[1] );
4770                         }
4771                         $title = Title::newFromText( $matches[1], NS_FILE );
4772                         if ( is_null( $title ) ) {
4773                                 # Bogus title. Ignore these so we don't bomb out later.
4774                                 continue;
4775                         }
4776
4777                         $label = '';
4778                         $alt = '';
4779                         if ( isset( $matches[3] ) ) {
4780                                 // look for an |alt= definition while trying not to break existing
4781                                 // captions with multiple pipes (|) in it, until a more sensible grammar
4782                                 // is defined for images in galleries
4783
4784                                 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
4785                                 $altmatches = StringUtils::explode('|', $matches[3]);
4786                                 $magicWordAlt = MagicWord::get( 'img_alt' );
4787
4788                                 foreach ( $altmatches as $altmatch ) {
4789                                         $match = $magicWordAlt->matchVariableStartToEnd( $altmatch );
4790                                         if ( $match ) {
4791                                                 $alt = $this->stripAltText( $match, false );
4792                                         }
4793                                         else {
4794                                                 // concatenate all other pipes
4795                                                 $label .= '|' . $altmatch;
4796                                         }
4797                                 }
4798                                 // remove the first pipe
4799                                 $label = substr( $label, 1 );
4800                         }
4801
4802                         $ig->add( $title, $label, $alt );
4803                 }
4804                 return $ig->toHTML();
4805         }
4806
4807         function getImageParams( $handler ) {
4808                 if ( $handler ) {
4809                         $handlerClass = get_class( $handler );
4810                 } else {
4811                         $handlerClass = '';
4812                 }
4813                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4814                         # Initialise static lists
4815                         static $internalParamNames = array(
4816                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4817                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4818                                         'bottom', 'text-bottom' ),
4819                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4820                                         'upright', 'border', 'link', 'alt' ),
4821                         );
4822                         static $internalParamMap;
4823                         if ( !$internalParamMap ) {
4824                                 $internalParamMap = array();
4825                                 foreach ( $internalParamNames as $type => $names ) {
4826                                         foreach ( $names as $name ) {
4827                                                 $magicName = str_replace( '-', '_', "img_$name" );
4828                                                 $internalParamMap[$magicName] = array( $type, $name );
4829                                         }
4830                                 }
4831                         }
4832
4833                         # Add handler params
4834                         $paramMap = $internalParamMap;
4835                         if ( $handler ) {
4836                                 $handlerParamMap = $handler->getParamMap();
4837                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4838                                         $paramMap[$magic] = array( 'handler', $paramName );
4839                                 }
4840                         }
4841                         $this->mImageParams[$handlerClass] = $paramMap;
4842                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4843                 }
4844                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4845         }
4846
4847         /**
4848          * Parse image options text and use it to make an image
4849          *
4850          * @param $title Title
4851          * @param $options String
4852          * @param $holders LinkHolderArray
4853          * @return string HTML
4854          */
4855         function makeImage( $title, $options, $holders = false ) {
4856                 # Check if the options text is of the form "options|alt text"
4857                 # Options are:
4858                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4859                 #  * left       no resizing, just left align. label is used for alt= only
4860                 #  * right      same, but right aligned
4861                 #  * none       same, but not aligned
4862                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4863                 #  * center     center the image
4864                 #  * frame      Keep original image size, no magnify-button.
4865                 #  * framed     Same as "frame"
4866                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4867                 #  * upright    reduce width for upright images, rounded to full __0 px
4868                 #  * border     draw a 1px border around the image
4869                 #  * alt        Text for HTML alt attribute (defaults to empty)
4870                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4871                 # vertical-align values (no % or length right now):
4872                 #  * baseline
4873                 #  * sub
4874                 #  * super
4875                 #  * top
4876                 #  * text-top
4877                 #  * middle
4878                 #  * bottom
4879                 #  * text-bottom
4880
4881                 $parts = StringUtils::explode( "|", $options );
4882
4883                 # Give extensions a chance to select the file revision for us
4884                 $time = $sha1 = $descQuery = false;
4885                 wfRunHooks( 'BeforeParserFetchFileAndTitle',
4886                         array( $this, $title, &$time, &$sha1, &$descQuery ) );
4887                 # Fetch and register the file (file title may be different via hooks)
4888                 list( $file, $title ) = $this->fetchFileAndTitle( $title, $time, $sha1 );
4889
4890                 # Get parameter map
4891                 $handler = $file ? $file->getHandler() : false;
4892
4893                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4894
4895                 if ( !$file ) {
4896                         $this->addTrackingCategory( 'broken-file-category' );
4897                 }
4898
4899                 # Process the input parameters
4900                 $caption = '';
4901                 $params = array( 'frame' => array(), 'handler' => array(),
4902                         'horizAlign' => array(), 'vertAlign' => array() );
4903                 foreach ( $parts as $part ) {
4904                         $part = trim( $part );
4905                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4906                         $validated = false;
4907                         if ( isset( $paramMap[$magicName] ) ) {
4908                                 list( $type, $paramName ) = $paramMap[$magicName];
4909
4910                                 # Special case; width and height come in one variable together
4911                                 if ( $type === 'handler' && $paramName === 'width' ) {
4912                                         $m = array();
4913                                         # (bug 13500) In both cases (width/height and width only),
4914                                         # permit trailing "px" for backward compatibility.
4915                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4916                                                 $width = intval( $m[1] );
4917                                                 $height = intval( $m[2] );
4918                                                 if ( $handler->validateParam( 'width', $width ) ) {
4919                                                         $params[$type]['width'] = $width;
4920                                                         $validated = true;
4921                                                 }
4922                                                 if ( $handler->validateParam( 'height', $height ) ) {
4923                                                         $params[$type]['height'] = $height;
4924                                                         $validated = true;
4925                                                 }
4926                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4927                                                 $width = intval( $value );
4928                                                 if ( $handler->validateParam( 'width', $width ) ) {
4929                                                         $params[$type]['width'] = $width;
4930                                                         $validated = true;
4931                                                 }
4932                                         } # else no validation -- bug 13436
4933                                 } else {
4934                                         if ( $type === 'handler' ) {
4935                                                 # Validate handler parameter
4936                                                 $validated = $handler->validateParam( $paramName, $value );
4937                                         } else {
4938                                                 # Validate internal parameters
4939                                                 switch( $paramName ) {
4940                                                 case 'manualthumb':
4941                                                 case 'alt':
4942                                                         # @todo Fixme: possibly check validity here for
4943                                                         # manualthumb? downstream behavior seems odd with
4944                                                         # missing manual thumbs.
4945                                                         $validated = true;
4946                                                         $value = $this->stripAltText( $value, $holders );
4947                                                         break;
4948                                                 case 'link':
4949                                                         $chars = self::EXT_LINK_URL_CLASS;
4950                                                         $prots = $this->mUrlProtocols;
4951                                                         if ( $value === '' ) {
4952                                                                 $paramName = 'no-link';
4953                                                                 $value = true;
4954                                                                 $validated = true;
4955                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4956                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4957                                                                         $paramName = 'link-url';
4958                                                                         $this->mOutput->addExternalLink( $value );
4959                                                                         if ( $this->mOptions->getExternalLinkTarget() ) {
4960                                                                                 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
4961                                                                         }
4962                                                                         $validated = true;
4963                                                                 }
4964                                                         } else {
4965                                                                 $linkTitle = Title::newFromText( $value );
4966                                                                 if ( $linkTitle ) {
4967                                                                         $paramName = 'link-title';
4968                                                                         $value = $linkTitle;
4969                                                                         $this->mOutput->addLink( $linkTitle );
4970                                                                         $validated = true;
4971                                                                 }
4972                                                         }
4973                                                         break;
4974                                                 default:
4975                                                         # Most other things appear to be empty or numeric...
4976                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4977                                                 }
4978                                         }
4979
4980                                         if ( $validated ) {
4981                                                 $params[$type][$paramName] = $value;
4982                                         }
4983                                 }
4984                         }
4985                         if ( !$validated ) {
4986                                 $caption = $part;
4987                         }
4988                 }
4989
4990                 # Process alignment parameters
4991                 if ( $params['horizAlign'] ) {
4992                         $params['frame']['align'] = key( $params['horizAlign'] );
4993                 }
4994                 if ( $params['vertAlign'] ) {
4995                         $params['frame']['valign'] = key( $params['vertAlign'] );
4996                 }
4997
4998                 $params['frame']['caption'] = $caption;
4999
5000                 # Will the image be presented in a frame, with the caption below?
5001                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
5002                                                  isset( $params['frame']['framed'] ) ||
5003                                                  isset( $params['frame']['thumbnail'] ) ||
5004                                                  isset( $params['frame']['manualthumb'] );
5005
5006                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
5007                 # came to also set the caption, ordinary text after the image -- which
5008                 # makes no sense, because that just repeats the text multiple times in
5009                 # screen readers.  It *also* came to set the title attribute.
5010                 #
5011                 # Now that we have an alt attribute, we should not set the alt text to
5012                 # equal the caption: that's worse than useless, it just repeats the
5013                 # text.  This is the framed/thumbnail case.  If there's no caption, we
5014                 # use the unnamed parameter for alt text as well, just for the time be-
5015                 # ing, if the unnamed param is set and the alt param is not.
5016                 #
5017                 # For the future, we need to figure out if we want to tweak this more,
5018                 # e.g., introducing a title= parameter for the title; ignoring the un-
5019                 # named parameter entirely for images without a caption; adding an ex-
5020                 # plicit caption= parameter and preserving the old magic unnamed para-
5021                 # meter for BC; ...
5022                 if ( $imageIsFramed ) { # Framed image
5023                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5024                                 # No caption or alt text, add the filename as the alt text so
5025                                 # that screen readers at least get some description of the image
5026                                 $params['frame']['alt'] = $title->getText();
5027                         }
5028                         # Do not set $params['frame']['title'] because tooltips don't make sense
5029                         # for framed images
5030                 } else { # Inline image
5031                         if ( !isset( $params['frame']['alt'] ) ) {
5032                                 # No alt text, use the "caption" for the alt text
5033                                 if ( $caption !== '') {
5034                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5035                                 } else {
5036                                         # No caption, fall back to using the filename for the
5037                                         # alt text
5038                                         $params['frame']['alt'] = $title->getText();
5039                                 }
5040                         }
5041                         # Use the "caption" for the tooltip text
5042                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5043                 }
5044
5045                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
5046
5047                 # Linker does the rest
5048                 $ret = Linker::makeImageLink2( $title, $file, $params['frame'], $params['handler'],
5049                         $time, $descQuery, $this->mOptions->getThumbSize() );
5050
5051                 # Give the handler a chance to modify the parser object
5052                 if ( $handler ) {
5053                         $handler->parserTransformHook( $this, $file );
5054                 }
5055
5056                 return $ret;
5057         }
5058
5059         protected function stripAltText( $caption, $holders ) {
5060                 # Strip bad stuff out of the title (tooltip).  We can't just use
5061                 # replaceLinkHoldersText() here, because if this function is called
5062                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5063                 if ( $holders ) {
5064                         $tooltip = $holders->replaceText( $caption );
5065                 } else {
5066                         $tooltip = $this->replaceLinkHoldersText( $caption );
5067                 }
5068
5069                 # make sure there are no placeholders in thumbnail attributes
5070                 # that are later expanded to html- so expand them now and
5071                 # remove the tags
5072                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5073                 $tooltip = Sanitizer::stripAllTags( $tooltip );
5074
5075                 return $tooltip;
5076         }
5077
5078         /**
5079          * Set a flag in the output object indicating that the content is dynamic and
5080          * shouldn't be cached.
5081          */
5082         function disableCache() {
5083                 wfDebug( "Parser output marked as uncacheable.\n" );
5084                 $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
5085                 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5086         }
5087
5088         /**
5089          * Callback from the Sanitizer for expanding items found in HTML attribute
5090          * values, so they can be safely tested and escaped.
5091          *
5092          * @param $text String
5093          * @param $frame PPFrame
5094          * @return String
5095          */
5096         function attributeStripCallback( &$text, $frame = false ) {
5097                 $text = $this->replaceVariables( $text, $frame );
5098                 $text = $this->mStripState->unstripBoth( $text );
5099                 return $text;
5100         }
5101
5102         /**
5103          * Accessor
5104          *
5105          * @return array
5106          */
5107         function getTags() {
5108                 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ) );
5109         }
5110
5111         /**
5112          * Replace transparent tags in $text with the values given by the callbacks.
5113          *
5114          * Transparent tag hooks are like regular XML-style tag hooks, except they
5115          * operate late in the transformation sequence, on HTML instead of wikitext.
5116          */
5117         function replaceTransparentTags( $text ) {
5118                 $matches = array();
5119                 $elements = array_keys( $this->mTransparentTagHooks );
5120                 $text = $this->extractTagsAndParams( $elements, $text, $matches, $this->mUniqPrefix );
5121
5122                 foreach ( $matches as $marker => $data ) {
5123                         list( $element, $content, $params, $tag ) = $data;
5124                         $tagName = strtolower( $element );
5125                         if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5126                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
5127                         } else {
5128                                 $output = $tag;
5129                         }
5130                         $this->mStripState->addGeneral( $marker, $output );
5131                 }
5132                 return $text;
5133         }
5134
5135         /**
5136          * Break wikitext input into sections, and either pull or replace
5137          * some particular section's text.
5138          *
5139          * External callers should use the getSection and replaceSection methods.
5140          *
5141          * @param $text String: Page wikitext
5142          * @param $section String: a section identifier string of the form:
5143          *   <flag1> - <flag2> - ... - <section number>
5144          *
5145          * Currently the only recognised flag is "T", which means the target section number
5146          * was derived during a template inclusion parse, in other words this is a template
5147          * section edit link. If no flags are given, it was an ordinary section edit link.
5148          * This flag is required to avoid a section numbering mismatch when a section is
5149          * enclosed by <includeonly> (bug 6563).
5150          *
5151          * The section number 0 pulls the text before the first heading; other numbers will
5152          * pull the given section along with its lower-level subsections. If the section is
5153          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5154          *
5155          * @param $mode String: one of "get" or "replace"
5156          * @param $newText String: replacement text for section data.
5157          * @return String: for "get", the extracted section text.
5158          *                 for "replace", the whole page with the section replaced.
5159          */
5160         private function extractSections( $text, $section, $mode, $newText='' ) {
5161                 global $wgTitle; # not generally used but removes an ugly failure mode
5162                 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5163                 $outText = '';
5164                 $frame = $this->getPreprocessor()->newFrame();
5165
5166                 # Process section extraction flags
5167                 $flags = 0;
5168                 $sectionParts = explode( '-', $section );
5169                 $sectionIndex = array_pop( $sectionParts );
5170                 foreach ( $sectionParts as $part ) {
5171                         if ( $part === 'T' ) {
5172                                 $flags |= self::PTD_FOR_INCLUSION;
5173                         }
5174                 }
5175                 # Preprocess the text
5176                 $root = $this->preprocessToDom( $text, $flags );
5177
5178                 # <h> nodes indicate section breaks
5179                 # They can only occur at the top level, so we can find them by iterating the root's children
5180                 $node = $root->getFirstChild();
5181
5182                 # Find the target section
5183                 if ( $sectionIndex == 0 ) {
5184                         # Section zero doesn't nest, level=big
5185                         $targetLevel = 1000;
5186                         if ( !$node ) {
5187                                 # The page definitely exists - we checked that earlier - so it must be blank: see bug #14005
5188                                 return $text;
5189                         }
5190                 } else {
5191                         while ( $node ) {
5192                                 if ( $node->getName() === 'h' ) {
5193                                         $bits = $node->splitHeading();
5194                                         if ( $bits['i'] == $sectionIndex ) {
5195                                                 $targetLevel = $bits['level'];
5196                                                 break;
5197                                         }
5198                                 }
5199                                 if ( $mode === 'replace' ) {
5200                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5201                                 }
5202                                 $node = $node->getNextSibling();
5203                         }
5204                 }
5205
5206                 if ( !$node ) {
5207                         # Not found
5208                         if ( $mode === 'get' ) {
5209                                 return $newText;
5210                         } else {
5211                                 return $text;
5212                         }
5213                 }
5214
5215                 # Find the end of the section, including nested sections
5216                 do {
5217                         if ( $node->getName() === 'h' ) {
5218                                 $bits = $node->splitHeading();
5219                                 $curLevel = $bits['level'];
5220                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5221                                         break;
5222                                 }
5223                         }
5224                         if ( $mode === 'get' ) {
5225                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5226                         }
5227                         $node = $node->getNextSibling();
5228                 } while ( $node );
5229
5230                 # Write out the remainder (in replace mode only)
5231                 if ( $mode === 'replace' ) {
5232                         # Output the replacement text
5233                         # Add two newlines on -- trailing whitespace in $newText is conventionally
5234                         # stripped by the editor, so we need both newlines to restore the paragraph gap
5235                         # Only add trailing whitespace if there is newText
5236                         if ( $newText != "" ) {
5237                                 $outText .= $newText . "\n\n";
5238                         }
5239
5240                         while ( $node ) {
5241                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5242                                 $node = $node->getNextSibling();
5243                         }
5244                 }
5245
5246                 if ( is_string( $outText ) ) {
5247                         # Re-insert stripped tags
5248                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5249                 }
5250
5251                 return $outText;
5252         }
5253
5254         /**
5255          * This function returns the text of a section, specified by a number ($section).
5256          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5257          * the first section before any such heading (section 0).
5258          *
5259          * If a section contains subsections, these are also returned.
5260          *
5261          * @param $text String: text to look in
5262          * @param $section String: section identifier
5263          * @param $deftext String: default to return if section is not found
5264          * @return string text of the requested section
5265          */
5266         public function getSection( $text, $section, $deftext='' ) {
5267                 return $this->extractSections( $text, $section, "get", $deftext );
5268         }
5269
5270         /**
5271          * This function returns $oldtext after the content of the section
5272          * specified by $section has been replaced with $text.
5273          *
5274          * @param $oldtext String: former text of the article
5275          * @param $section Numeric: section identifier
5276          * @param $text String: replacing text
5277          * @return String: modified text
5278          */
5279         public function replaceSection( $oldtext, $section, $text ) {
5280                 return $this->extractSections( $oldtext, $section, "replace", $text );
5281         }
5282
5283         /**
5284          * Get the ID of the revision we are parsing
5285          *
5286          * @return Mixed: integer or null
5287          */
5288         function getRevisionId() {
5289                 return $this->mRevisionId;
5290         }
5291
5292         /**
5293          * Get the revision object for $this->mRevisionId
5294          *
5295          * @return Revision|null either a Revision object or null
5296          */
5297         protected function getRevisionObject() {
5298                 if ( !is_null( $this->mRevisionObject ) ) {
5299                         return $this->mRevisionObject;
5300                 }
5301                 if ( is_null( $this->mRevisionId ) ) {
5302                         return null;
5303                 }
5304
5305                 $this->mRevisionObject = Revision::newFromId( $this->mRevisionId );
5306                 return $this->mRevisionObject;
5307         }
5308
5309         /**
5310          * Get the timestamp associated with the current revision, adjusted for
5311          * the default server-local timestamp
5312          */
5313         function getRevisionTimestamp() {
5314                 if ( is_null( $this->mRevisionTimestamp ) ) {
5315                         wfProfileIn( __METHOD__ );
5316
5317                         $revObject = $this->getRevisionObject();
5318                         $timestamp = $revObject ? $revObject->getTimestamp() : false;
5319
5320                         if( $timestamp !== false ) {
5321                                 global $wgContLang;
5322
5323                                 # The cryptic '' timezone parameter tells to use the site-default
5324                                 # timezone offset instead of the user settings.
5325                                 #
5326                                 # Since this value will be saved into the parser cache, served
5327                                 # to other users, and potentially even used inside links and such,
5328                                 # it needs to be consistent for all visitors.
5329                                 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5330                         }
5331
5332                         wfProfileOut( __METHOD__ );
5333                 }
5334                 return $this->mRevisionTimestamp;
5335         }
5336
5337         /**
5338          * Get the name of the user that edited the last revision
5339          *
5340          * @return String: user name
5341          */
5342         function getRevisionUser() {
5343                 if( is_null( $this->mRevisionUser ) ) {
5344                         $revObject = $this->getRevisionObject();
5345
5346                         # if this template is subst: the revision id will be blank,
5347                         # so just use the current user's name
5348                         if( $revObject ) {
5349                                 $this->mRevisionUser = $revObject->getUserText();
5350                         } elseif( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5351                                 $this->mRevisionUser = $this->getUser()->getName();
5352                         }
5353                 }
5354                 return $this->mRevisionUser;
5355         }
5356
5357         /**
5358          * Mutator for $mDefaultSort
5359          *
5360          * @param $sort New value
5361          */
5362         public function setDefaultSort( $sort ) {
5363                 $this->mDefaultSort = $sort;
5364                 $this->mOutput->setProperty( 'defaultsort', $sort );
5365         }
5366
5367         /**
5368          * Accessor for $mDefaultSort
5369          * Will use the empty string if none is set.
5370          *
5371          * This value is treated as a prefix, so the
5372          * empty string is equivalent to sorting by
5373          * page name.
5374          *
5375          * @return string
5376          */
5377         public function getDefaultSort() {
5378                 if ( $this->mDefaultSort !== false ) {
5379                         return $this->mDefaultSort;
5380                 } else {
5381                         return '';
5382                 }
5383         }
5384
5385         /**
5386          * Accessor for $mDefaultSort
5387          * Unlike getDefaultSort(), will return false if none is set
5388          *
5389          * @return string or false
5390          */
5391         public function getCustomDefaultSort() {
5392                 return $this->mDefaultSort;
5393         }
5394
5395         /**
5396          * Try to guess the section anchor name based on a wikitext fragment
5397          * presumably extracted from a heading, for example "Header" from
5398          * "== Header ==".
5399          */
5400         public function guessSectionNameFromWikiText( $text ) {
5401                 # Strip out wikitext links(they break the anchor)
5402                 $text = $this->stripSectionName( $text );
5403                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5404                 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5405         }
5406
5407         /**
5408          * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5409          * instead.  For use in redirects, since IE6 interprets Redirect: headers
5410          * as something other than UTF-8 (apparently?), resulting in breakage.
5411          *
5412          * @param $text String: The section name
5413          * @return string An anchor
5414          */
5415         public function guessLegacySectionNameFromWikiText( $text ) {
5416                 # Strip out wikitext links(they break the anchor)
5417                 $text = $this->stripSectionName( $text );
5418                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5419                 return '#' . Sanitizer::escapeId( $text, array( 'noninitial', 'legacy' ) );
5420         }
5421
5422         /**
5423          * Strips a text string of wikitext for use in a section anchor
5424          *
5425          * Accepts a text string and then removes all wikitext from the
5426          * string and leaves only the resultant text (i.e. the result of
5427          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5428          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5429          * to create valid section anchors by mimicing the output of the
5430          * parser when headings are parsed.
5431          *
5432          * @param $text String: text string to be stripped of wikitext
5433          * for use in a Section anchor
5434          * @return Filtered text string
5435          */
5436         public function stripSectionName( $text ) {
5437                 # Strip internal link markup
5438                 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5439                 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5440
5441                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5442                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5443                 # on how many empty links there are on the page - need to figure that out.
5444                 $text = preg_replace( '/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5445
5446                 # Parse wikitext quotes (italics & bold)
5447                 $text = $this->doQuotes( $text );
5448
5449                 # Strip HTML tags
5450                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5451                 return $text;
5452         }
5453
5454         /**
5455          * strip/replaceVariables/unstrip for preprocessor regression testing
5456          *
5457          * @return string
5458          */
5459         function testSrvus( $text, $title, ParserOptions $options, $outputType = self::OT_HTML ) {
5460                 if ( !$title instanceof Title ) {
5461                         $title = Title::newFromText( $title );
5462                 }
5463                 $this->startParse( $title, $options, $outputType, true );
5464
5465                 $text = $this->replaceVariables( $text );
5466                 $text = $this->mStripState->unstripBoth( $text );
5467                 $text = Sanitizer::removeHTMLtags( $text );
5468                 return $text;
5469         }
5470
5471         function testPst( $text, $title, $options ) {
5472                 global $wgUser;
5473                 if ( !$title instanceof Title ) {
5474                         $title = Title::newFromText( $title );
5475                 }
5476                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5477         }
5478
5479         function testPreprocess( $text, $title, $options ) {
5480                 if ( !$title instanceof Title ) {
5481                         $title = Title::newFromText( $title );
5482                 }
5483                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5484         }
5485
5486         /**
5487          * Call a callback function on all regions of the given text that are not
5488          * inside strip markers, and replace those regions with the return value
5489          * of the callback. For example, with input:
5490          *
5491          *  aaa<MARKER>bbb
5492          *
5493          * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5494          * two strings will be replaced with the value returned by the callback in
5495          * each case.
5496          *
5497          * @return string
5498          */
5499         function markerSkipCallback( $s, $callback ) {
5500                 $i = 0;
5501                 $out = '';
5502                 while ( $i < strlen( $s ) ) {
5503                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5504                         if ( $markerStart === false ) {
5505                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5506                                 break;
5507                         } else {
5508                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5509                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5510                                 if ( $markerEnd === false ) {
5511                                         $out .= substr( $s, $markerStart );
5512                                         break;
5513                                 } else {
5514                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5515                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5516                                         $i = $markerEnd;
5517                                 }
5518                         }
5519                 }
5520                 return $out;
5521         }
5522
5523         /**
5524          * Save the parser state required to convert the given half-parsed text to
5525          * HTML. "Half-parsed" in this context means the output of
5526          * recursiveTagParse() or internalParse(). This output has strip markers
5527          * from replaceVariables (extensionSubstitution() etc.), and link
5528          * placeholders from replaceLinkHolders().
5529          *
5530          * Returns an array which can be serialized and stored persistently. This
5531          * array can later be loaded into another parser instance with
5532          * unserializeHalfParsedText(). The text can then be safely incorporated into
5533          * the return value of a parser hook.
5534          *
5535          * @return array
5536          */
5537         function serializeHalfParsedText( $text ) {
5538                 wfProfileIn( __METHOD__ );
5539                 $data = array(
5540                         'text' => $text,
5541                         'version' => self::HALF_PARSED_VERSION,
5542                         'stripState' => $this->mStripState->getSubState( $text ),
5543                         'linkHolders' => $this->mLinkHolders->getSubArray( $text )
5544                 );
5545                 wfProfileOut( __METHOD__ );
5546                 return $data;
5547         }
5548
5549         /**
5550          * Load the parser state given in the $data array, which is assumed to
5551          * have been generated by serializeHalfParsedText(). The text contents is
5552          * extracted from the array, and its markers are transformed into markers
5553          * appropriate for the current Parser instance. This transformed text is
5554          * returned, and can be safely included in the return value of a parser
5555          * hook.
5556          *
5557          * If the $data array has been stored persistently, the caller should first
5558          * check whether it is still valid, by calling isValidHalfParsedText().
5559          *
5560          * @param $data Serialized data
5561          * @return String
5562          */
5563         function unserializeHalfParsedText( $data ) {
5564                 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
5565                         throw new MWException( __METHOD__.': invalid version' );
5566                 }
5567
5568                 # First, extract the strip state.
5569                 $texts = array( $data['text'] );
5570                 $texts = $this->mStripState->merge( $data['stripState'], $texts );
5571
5572                 # Now renumber links
5573                 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
5574
5575                 # Should be good to go.
5576                 return $texts[0];
5577         }
5578
5579         /**
5580          * Returns true if the given array, presumed to be generated by
5581          * serializeHalfParsedText(), is compatible with the current version of the
5582          * parser.
5583          *
5584          * @param $data Array
5585          *
5586          * @return bool
5587          */
5588         function isValidHalfParsedText( $data ) {
5589                 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
5590         }
5591 }