includes/parser/Parser.php

   1 <?php
   2 /**
   3  * @defgroup Parser Parser
   4  *
   5  * @file
   6  * @ingroup Parser
   7  * File for Parser and related classes
   8  */
   9
  10
  11 /**
  12  * PHP Parser - Processes wiki markup (which uses a more user-friendly
  13  * syntax, such as "[[link]]" for making links), and provides a one-way
  14  * transformation of that wiki markup it into XHTML output / markup
  15  * (which in turn the browser understands, and can display).
  16  *
  17  * <pre>
  18  * There are five main entry points into the Parser class:
  19  * parse()
  20  *     produces HTML output
  21  * preSaveTransform().
  22  *     produces altered wiki markup.
  23  * preprocess()
  24  *     removes HTML comments and expands templates
  25  * cleanSig() / cleanSigInSig()
  26  *     Cleans a signature before saving it to preferences
  27  * getSection()
  28  *     Return the content of a section from an article for section editing
  29  * replaceSection()
  30  *     Replaces a section by number inside an article
  31  * getPreloadText()
  32  *     Removes <noinclude> sections, and <includeonly> tags.
  33  *
  34  * Globals used:
  35  *    objects:   $wgLang, $wgContLang
  36  *
  37  * NOT $wgUser or $wgTitle. Keep them away!
  38  *
  39  * settings:
  40  *  $wgUseDynamicDates*, $wgInterwikiMagic*,
  41  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  42  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  43  *  $wgMaxArticleSize*
  44  *
  45  *  * only within ParserOptions
  46  * </pre>
  47  *
  48  * @ingroup Parser
  49  */
  50 class Parser {
  51         /**
  52          * Update this version number when the ParserOutput format
  53          * changes in an incompatible way, so the parser cache
  54          * can automatically discard old data.
  55          */
  56         const VERSION = '1.6.5';
  57
  58         /**
  59          * Update this version number when the output of serialiseHalfParsedText()
  60          * changes in an incompatible way
  61          */
  62         const HALF_PARSED_VERSION = 2;
  63
  64         # Flags for Parser::setFunctionHook
  65         # Also available as global constants from Defines.php
  66         const SFH_NO_HASH = 1;
  67         const SFH_OBJECT_ARGS = 2;
  68
  69         # Constants needed for external link processing
  70         # Everything except bracket, space, or control characters
  71         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  72         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  73                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/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
 113         # Initialised by initialiseVariables()
 114
 115         /**
 116          * @var MagicWordArray
 117          */
 118         var $mVariables;
 119
 120         /**
 121          * @var MagicWordArray
 122          */
 123         var $mSubstWords;
 124         var $mConf, $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols; # Initialised in constructor
 125
 126         # Cleared with clearState():
 127         /**
 128          * @var ParserOutput
 129          */
 130         var $mOutput;
 131         var $mAutonumber, $mDTopen;
 132
 133         /**
 134          * @var StripState
 135          */
 136         var $mStripState;
 137
 138         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 139         /**
 140          * @var LinkHolderArray
 141          */
 142         var $mLinkHolders;
 143
 144         var $mLinkID;
 145         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 146         var $mTplExpandCache; # empty-frame expansion cache
 147         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 148         var $mExpensiveFunctionCount; # number of expensive parser function calls
 149
 150         /**
 151          * @var User
 152          */
 153         var $mUser; # User object; only used when doing pre-save transform
 154
 155         # Temporary
 156         # These are variables reset at least once per parse regardless of $clearState
 157
 158         /**
 159          * @var ParserOptions
 160          */
 161         var $mOptions;
 162
 163         /**
 164          * @var Title
 165          */
 166         var $mTitle;        # Title context, used for self-link rendering and similar things
 167         var $mOutputType;   # Output type, one of the OT_xxx constants
 168         var $ot;            # Shortcut alias, see setOutputType()
 169         var $mRevisionObject; # The revision object of the specified revision ID
 170         var $mRevisionId;   # ID to display in {{REVISIONID}} tags
 171         var $mRevisionTimestamp; # The timestamp of the specified revision ID
 172         var $mRevisionUser; # User to display in {{REVISIONUSER}} tag
 173         var $mRevIdForTs;   # The revision ID which was used to fetch the timestamp
 174
 175         /**
 176          * @var string
 177          */
 178         var $mUniqPrefix;
 179
 180         /**
 181          * Constructor
 182          */
 183         public function __construct( $conf = array() ) {
 184                 $this->mConf = $conf;
 185                 $this->mUrlProtocols = wfUrlProtocols();
 186                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 187                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/S';
 188                 if ( isset( $conf['preprocessorClass'] ) ) {
 189                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 190                 } elseif ( defined( 'MW_COMPILED' ) ) {
 191                         # Preprocessor_Hash is much faster than Preprocessor_DOM in compiled mode
 192                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 193                 } elseif ( extension_loaded( 'domxml' ) ) {
 194                         # PECL extension that conflicts with the core DOM extension (bug 13770)
 195                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 196                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 197                 } elseif ( extension_loaded( 'dom' ) ) {
 198                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 199                 } else {
 200                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 201                 }
 202                 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
 203         }
 204
 205         /**
 206          * Reduce memory usage to reduce the impact of circular references
 207          */
 208         function __destruct() {
 209                 if ( isset( $this->mLinkHolders ) ) {
 210                         unset( $this->mLinkHolders );
 211                 }
 212                 foreach ( $this as $name => $value ) {
 213                         unset( $this->$name );
 214                 }
 215         }
 216
 217         /**
 218          * Do various kinds of initialisation on the first call of the parser
 219          */
 220         function firstCallInit() {
 221                 if ( !$this->mFirstCall ) {
 222                         return;
 223                 }
 224                 $this->mFirstCall = false;
 225
 226                 wfProfileIn( __METHOD__ );
 227
 228                 CoreParserFunctions::register( $this );
 229                 CoreTagHooks::register( $this );
 230                 $this->initialiseVariables();
 231
 232                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 233                 wfProfileOut( __METHOD__ );
 234         }
 235
 236         /**
 237          * Clear Parser state
 238          *
 239          * @private
 240          */
 241         function clearState() {
 242                 wfProfileIn( __METHOD__ );
 243                 if ( $this->mFirstCall ) {
 244                         $this->firstCallInit();
 245                 }
 246                 $this->mOutput = new ParserOutput;
 247                 $this->mOptions->registerWatcher( array( $this->mOutput, 'recordOption' ) );
 248                 $this->mAutonumber = 0;
 249                 $this->mLastSection = '';
 250                 $this->mDTopen = false;
 251                 $this->mIncludeCount = array();
 252                 $this->mArgStack = false;
 253                 $this->mInPre = false;
 254                 $this->mLinkHolders = new LinkHolderArray( $this );
 255                 $this->mLinkID = 0;
 256                 $this->mRevisionObject = $this->mRevisionTimestamp =
 257                         $this->mRevisionId = $this->mRevisionUser = null;
 258                 $this->mVarCache = array();
 259                 $this->mUser = null;
 260
 261                 /**
 262                  * Prefix for temporary replacement strings for the multipass parser.
 263                  * \x07 should never appear in input as it's disallowed in XML.
 264                  * Using it at the front also gives us a little extra robustness
 265                  * since it shouldn't match when butted up against identifier-like
 266                  * string constructs.
 267                  *
 268                  * Must not consist of all title characters, or else it will change
 269                  * the behaviour of <nowiki> in a link.
 270                  */
 271                 # $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 272                 # Changed to \x7f to allow XML double-parsing -- TS
 273                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 274                 $this->mStripState = new StripState( $this->mUniqPrefix );
 275
 276
 277                 # Clear these on every parse, bug 4549
 278                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 279
 280                 $this->mShowToc = true;
 281                 $this->mForceTocPosition = false;
 282                 $this->mIncludeSizes = array(
 283                         'post-expand' => 0,
 284                         'arg' => 0,
 285                 );
 286                 $this->mPPNodeCount = 0;
 287                 $this->mDefaultSort = false;
 288                 $this->mHeadings = array();
 289                 $this->mDoubleUnderscores = array();
 290                 $this->mExpensiveFunctionCount = 0;
 291
 292                 # Fix cloning
 293                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 294                         $this->mPreprocessor = null;
 295                 }
 296
 297                 wfRunHooks( 'ParserClearState', array( &$this ) );
 298                 wfProfileOut( __METHOD__ );
 299         }
 300
 301         /**
 302          * Convert wikitext to HTML
 303          * Do not call this function recursively.
 304          *
 305          * @param $text String: text we want to parse
 306          * @param $title Title object
 307          * @param $options ParserOptions
 308          * @param $linestart boolean
 309          * @param $clearState boolean
 310          * @param $revid Int: number to pass in {{REVISIONID}}
 311          * @return ParserOutput a ParserOutput
 312          */
 313         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 314                 /**
 315                  * First pass--just handle <nowiki> sections, pass the rest off
 316                  * to internalParse() which does all the real work.
 317                  */
 318
 319                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang, $wgDisableLangConversion, $wgDisableTitleConversion;
 320                 $fname = __METHOD__.'-' . wfGetCaller();
 321                 wfProfileIn( __METHOD__ );
 322                 wfProfileIn( $fname );
 323
 324                 $this->startParse( $title, $options, self::OT_HTML, $clearState );
 325
 326                 $oldRevisionId = $this->mRevisionId;
 327                 $oldRevisionObject = $this->mRevisionObject;
 328                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 329                 $oldRevisionUser = $this->mRevisionUser;
 330                 if ( $revid !== null ) {
 331                         $this->mRevisionId = $revid;
 332                         $this->mRevisionObject = null;
 333                         $this->mRevisionTimestamp = null;
 334                         $this->mRevisionUser = null;
 335                 }
 336
 337                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 338                 # No more strip!
 339                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 340                 $text = $this->internalParse( $text );
 341
 342                 $text = $this->mStripState->unstripGeneral( $text );
 343
 344                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 345                 $fixtags = array(
 346                         # french spaces, last one Guillemet-left
 347                         # only if there is something before the space
 348                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;\\2',
 349                         # french spaces, Guillemet-right
 350                         '/(\\302\\253) /' => '\\1&#160;',
 351                         '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
 352                 );
 353                 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
 354
 355                 $text = $this->doBlockLevels( $text, $linestart );
 356
 357                 $this->replaceLinkHolders( $text );
 358
 359                 /**
 360                  * The page doesn't get language converted if
 361                  * a) It's disabled
 362                  * b) Content isn't converted
 363                  * c) It's a conversion table
 364                  */
 365                 if ( !( $wgDisableLangConversion
 366                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 367                                 || $this->mTitle->isConversionTable() ) ) {
 368
 369                         # The position of the convert() call should not be changed. it
 370                         # assumes that the links are all replaced and the only thing left
 371                         # is the <nowiki> mark.
 372
 373                         $text = $wgContLang->convert( $text );
 374                 }
 375
 376                 /**
 377                  * A converted title will be provided in the output object if title and
 378                  * content conversion are enabled, the article text does not contain
 379                  * a conversion-suppressing double-underscore tag, and no
 380                  * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
 381                  * automatic link conversion.
 382                  */
 383                 if ( !( $wgDisableLangConversion
 384                                 || $wgDisableTitleConversion
 385                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 386                                 || isset( $this->mDoubleUnderscores['notitleconvert'] )
 387                                 || $this->mOutput->getDisplayTitle() !== false ) )
 388                 {
 389                         $convruletitle = $wgContLang->getConvRuleTitle();
 390                         if ( $convruletitle ) {
 391                                 $this->mOutput->setTitleText( $convruletitle );
 392                         } else {
 393                                 $titleText = $wgContLang->convertTitle( $title );
 394                                 $this->mOutput->setTitleText( $titleText );
 395                         }
 396                 }
 397
 398                 $text = $this->mStripState->unstripNoWiki( $text );
 399
 400                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 401
 402                 $text = $this->replaceTransparentTags( $text );
 403                 $text = $this->mStripState->unstripGeneral( $text );
 404
 405                 $text = Sanitizer::normalizeCharReferences( $text );
 406
 407                 if ( ( $wgUseTidy && $this->mOptions->getTidy() ) || $wgAlwaysUseTidy ) {
 408                         $text = MWTidy::tidy( $text );
 409                 } else {
 410                         # attempt to sanitize at least some nesting problems
 411                         # (bug #2702 and quite a few others)
 412                         $tidyregs = array(
 413                                 # ''Something [http://www.cool.com cool''] -->
 414                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 415                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 416                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 417                                 # fix up an anchor inside another anchor, only
 418                                 # at least for a single single nested link (bug 3695)
 419                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 420                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 421                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 422                                 # contains a div, so fix it up here; replace
 423                                 # div with escaped text
 424                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 425                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 426                                 # remove empty italic or bold tag pairs, some
 427                                 # introduced by rules above
 428                                 '/<([bi])><\/\\1>/' => '',
 429                         );
 430
 431                         $text = preg_replace(
 432                                 array_keys( $tidyregs ),
 433                                 array_values( $tidyregs ),
 434                                 $text );
 435                 }
 436                 global $wgExpensiveParserFunctionLimit;
 437                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 438                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 439                 }
 440
 441                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 442
 443                 # Information on include size limits, for the benefit of users who try to skirt them
 444                 if ( $this->mOptions->getEnableLimitReport() ) {
 445                         $max = $this->mOptions->getMaxIncludeSize();
 446                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 447                         $limitReport =
 448                                 "NewPP limit report\n" .
 449                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->getMaxPPNodeCount()}\n" .
 450                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 451                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 452                                 $PFreport;
 453                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 454                         $text .= "\n<!-- \n$limitReport-->\n";
 455                 }
 456                 $this->mOutput->setText( $text );
 457
 458                 $this->mRevisionId = $oldRevisionId;
 459                 $this->mRevisionObject = $oldRevisionObject;
 460                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 461                 $this->mRevisionUser = $oldRevisionUser;
 462                 wfProfileOut( $fname );
 463                 wfProfileOut( __METHOD__ );
 464
 465                 return $this->mOutput;
 466         }
 467
 468         /**
 469          * Recursive parser entry point that can be called from an extension tag
 470          * hook.
 471          *
 472          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 473          *
 474          * @param $text String: text extension wants to have parsed
 475          * @param $frame PPFrame: The frame to use for expanding any template variables
 476          *
 477          * @return string
 478          */
 479         function recursiveTagParse( $text, $frame=false ) {
 480                 wfProfileIn( __METHOD__ );
 481                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 482                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 483                 $text = $this->internalParse( $text, false, $frame );
 484                 wfProfileOut( __METHOD__ );
 485                 return $text;
 486         }
 487
 488         /**
 489          * Expand templates and variables in the text, producing valid, static wikitext.
 490          * Also removes comments.
 491          */
 492         function preprocess( $text, Title $title, ParserOptions $options, $revid = null ) {
 493                 wfProfileIn( __METHOD__ );
 494                 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
 495                 if ( $revid !== null ) {
 496                         $this->mRevisionId = $revid;
 497                 }
 498                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 499                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 500                 $text = $this->replaceVariables( $text );
 501                 $text = $this->mStripState->unstripBoth( $text );
 502                 wfProfileOut( __METHOD__ );
 503                 return $text;
 504         }
 505
 506         /**
 507          * Process the wikitext for the ?preload= feature. (bug 5210)
 508          *
 509          * <noinclude>, <includeonly> etc. are parsed as for template transclusion,
 510          * comments, templates, arguments, tags hooks and parser functions are untouched.
 511          */
 512         public function getPreloadText( $text, Title $title, ParserOptions $options ) {
 513                 # Parser (re)initialisation
 514                 $this->startParse( $title, $options, self::OT_PLAIN, true );
 515
 516                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 517                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
 518                 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
 519                 $text = $this->mStripState->unstripBoth( $text );
 520                 return $text;
 521         }
 522
 523         /**
 524          * Get a random string
 525          *
 526          * @return string
 527          */
 528         static public function getRandomString() {
 529                 return dechex( mt_rand( 0, 0x7fffffff ) ) . dechex( mt_rand( 0, 0x7fffffff ) );
 530         }
 531
 532         /**
 533          * Set the current user.
 534          * Should only be used when doing pre-save transform.
 535          *
 536          * @param $user Mixed: User object or null (to reset)
 537          */
 538         function setUser( $user ) {
 539                 $this->mUser = $user;
 540         }
 541
 542         /**
 543          * Accessor for mUniqPrefix.
 544          *
 545          * @return String
 546          */
 547         public function uniqPrefix() {
 548                 if ( !isset( $this->mUniqPrefix ) ) {
 549                         # @todo FIXME: This is probably *horribly wrong*
 550                         # LanguageConverter seems to want $wgParser's uniqPrefix, however
 551                         # if this is called for a parser cache hit, the parser may not
 552                         # have ever been initialized in the first place.
 553                         # Not really sure what the heck is supposed to be going on here.
 554                         return '';
 555                         # throw new MWException( "Accessing uninitialized mUniqPrefix" );
 556                 }
 557                 return $this->mUniqPrefix;
 558         }
 559
 560         /**
 561          * Set the context title
 562          *
 563          * @param $t Title
 564          */
 565         function setTitle( $t ) {
 566                 if ( !$t || $t instanceof FakeTitle ) {
 567                         $t = Title::newFromText( 'NO TITLE' );
 568                 }
 569
 570                 if ( strval( $t->getFragment() ) !== '' ) {
 571                         # Strip the fragment to avoid various odd effects
 572                         $this->mTitle = clone $t;
 573                         $this->mTitle->setFragment( '' );
 574                 } else {
 575                         $this->mTitle = $t;
 576                 }
 577         }
 578
 579         /**
 580          * Accessor for the Title object
 581          *
 582          * @return Title object
 583          */
 584         function getTitle() {
 585                 return $this->mTitle;
 586         }
 587
 588         /**
 589          * Accessor/mutator for the Title object
 590          *
 591          * @param $x New Title object or null to just get the current one
 592          * @return Title object
 593          */
 594         function Title( $x = null ) {
 595                 return wfSetVar( $this->mTitle, $x );
 596         }
 597
 598         /**
 599          * Set the output type
 600          *
 601          * @param $ot Integer: new value
 602          */
 603         function setOutputType( $ot ) {
 604                 $this->mOutputType = $ot;
 605                 # Shortcut alias
 606                 $this->ot = array(
 607                         'html' => $ot == self::OT_HTML,
 608                         'wiki' => $ot == self::OT_WIKI,
 609                         'pre' => $ot == self::OT_PREPROCESS,
 610                         'plain' => $ot == self::OT_PLAIN,
 611                 );
 612         }
 613
 614         /**
 615          * Accessor/mutator for the output type
 616          *
 617          * @param $x New value or null to just get the current one
 618          * @return Integer
 619          */
 620         function OutputType( $x = null ) {
 621                 return wfSetVar( $this->mOutputType, $x );
 622         }
 623
 624         /**
 625          * Get the ParserOutput object
 626          *
 627          * @return ParserOutput object
 628          */
 629         function getOutput() {
 630                 return $this->mOutput;
 631         }
 632
 633         /**
 634          * Get the ParserOptions object
 635          *
 636          * @return ParserOptions object
 637          */
 638         function getOptions() {
 639                 return $this->mOptions;
 640         }
 641
 642         /**
 643          * Accessor/mutator for the ParserOptions object
 644          *
 645          * @param $x New value or null to just get the current one
 646          * @return Current ParserOptions object
 647          */
 648         function Options( $x = null ) {
 649                 return wfSetVar( $this->mOptions, $x );
 650         }
 651
 652         /**
 653          * @return int
 654          */
 655         function nextLinkID() {
 656                 return $this->mLinkID++;
 657         }
 658
 659         /**
 660          * @param $id int
 661          */
 662         function setLinkID( $id ) {
 663                 $this->mLinkID = $id;
 664         }
 665
 666         /**
 667          * @return Language
 668          */
 669         function getFunctionLang() {
 670                 global $wgLang, $wgContLang;
 671
 672                 $target = $this->mOptions->getTargetLanguage();
 673                 if ( $target !== null ) {
 674                         return $target;
 675                 } else {
 676                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $this->mTitle->getPageLanguage();
 677                 }
 678         }
 679
 680         /**
 681          * Get a User object either from $this->mUser, if set, or from the
 682          * ParserOptions object otherwise
 683          *
 684          * @return User object
 685          */
 686         function getUser() {
 687                 if ( !is_null( $this->mUser ) ) {
 688                         return $this->mUser;
 689                 }
 690                 return $this->mOptions->getUser();
 691         }
 692
 693         /**
 694          * Get a preprocessor object
 695          *
 696          * @return Preprocessor instance
 697          */
 698         function getPreprocessor() {
 699                 if ( !isset( $this->mPreprocessor ) ) {
 700                         $class = $this->mPreprocessorClass;
 701                         $this->mPreprocessor = new $class( $this );
 702                 }
 703                 return $this->mPreprocessor;
 704         }
 705
 706         /**
 707          * Replaces all occurrences of HTML-style comments and the given tags
 708          * in the text with a random marker and returns the next text. The output
 709          * parameter $matches will be an associative array filled with data in
 710          * the form:
 711          *   'UNIQ-xxxxx' => array(
 712          *     'element',
 713          *     'tag content',
 714          *     array( 'param' => 'x' ),
 715          *     '<element param="x">tag content</element>' ) )
 716          *
 717          * @param $elements array list of element names. Comments are always extracted.
 718          * @param $text string Source text string.
 719          * @param $matches array Out parameter, Array: extracted tags
 720          * @param $uniq_prefix string
 721          * @return String: stripped text
 722          */
 723         public static function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
 724                 static $n = 1;
 725                 $stripped = '';
 726                 $matches = array();
 727
 728                 $taglist = implode( '|', $elements );
 729                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 730
 731                 while ( $text != '' ) {
 732                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 733                         $stripped .= $p[0];
 734                         if ( count( $p ) < 5 ) {
 735                                 break;
 736                         }
 737                         if ( count( $p ) > 5 ) {
 738                                 # comment
 739                                 $element    = $p[4];
 740                                 $attributes = '';
 741                                 $close      = '';
 742                                 $inside     = $p[5];
 743                         } else {
 744                                 # tag
 745                                 $element    = $p[1];
 746                                 $attributes = $p[2];
 747                                 $close      = $p[3];
 748                                 $inside     = $p[4];
 749                         }
 750
 751                         $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
 752                         $stripped .= $marker;
 753
 754                         if ( $close === '/>' ) {
 755                                 # Empty element tag, <tag />
 756                                 $content = null;
 757                                 $text = $inside;
 758                                 $tail = null;
 759                         } else {
 760                                 if ( $element === '!--' ) {
 761                                         $end = '/(-->)/';
 762                                 } else {
 763                                         $end = "/(<\\/$element\\s*>)/i";
 764                                 }
 765                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 766                                 $content = $q[0];
 767                                 if ( count( $q ) < 3 ) {
 768                                         # No end tag -- let it run out to the end of the text.
 769                                         $tail = '';
 770                                         $text = '';
 771                                 } else {
 772                                         $tail = $q[1];
 773                                         $text = $q[2];
 774                                 }
 775                         }
 776
 777                         $matches[$marker] = array( $element,
 778                                 $content,
 779                                 Sanitizer::decodeTagAttributes( $attributes ),
 780                                 "<$element$attributes$close$content$tail" );
 781                 }
 782                 return $stripped;
 783         }
 784
 785         /**
 786          * Get a list of strippable XML-like elements
 787          *
 788          * @return array
 789          */
 790         function getStripList() {
 791                 return $this->mStripList;
 792         }
 793
 794         /**
 795          * Add an item to the strip state
 796          * Returns the unique tag which must be inserted into the stripped text
 797          * The tag will be replaced with the original text in unstrip()
 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          * parse the wiki syntax used to render tables
 808          *
 809          * @private
 810          */
 811         function doTableStuff( $text ) {
 812                 wfProfileIn( __METHOD__ );
 813
 814                 $lines = StringUtils::explode( "\n", $text );
 815                 $out = '';
 816                 $output =& $out;
 817
 818                 foreach ( $lines as $outLine ) {
 819                         $line = trim( $outLine );
 820
 821                         # empty line, go to next line,
 822                         # but only append \n if outside of table
 823                         if ( $line === '') {
 824                                 $output .= $outLine . "\n";
 825                                 continue;
 826                         }
 827                         $firstChars = $line[0];
 828                         if ( strlen( $line ) > 1 ) {
 829                                 $firstChars .= in_array( $line[1], array( '}', '+', '-' ) ) ? $line[1] : '';
 830                         }
 831                         $matches = array();
 832
 833                         if ( preg_match( '/^(:*)\s*\{\|(.*)$/', $line , $matches ) ) {
 834                                 $tables[] = array();
 835                                 $table =& $this->last( $tables );
 836                                 $table[0] = array(); // first row
 837                                 $currentRow =& $table[0];
 838                                 $table['indent'] = strlen( $matches[1] );
 839
 840                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 841                                 $attributes = Sanitizer::fixTagAttributes( $attributes , 'table' );
 842
 843                                 if ( $attributes !== '' ) {
 844                                         $table['attributes'] = $attributes;
 845                                 }
 846                         } elseif ( !isset( $tables[0] ) ) {
 847                                 // we're outside the table
 848
 849                                 $out .= $outLine . "\n";
 850                         } elseif ( $firstChars === '|}' ) {
 851                                 // trim the |} code from the line
 852                                 $line = substr ( $line , 2 );
 853
 854                                 // Shorthand for last row
 855                                 $lastRow =& $this->last( $table );
 856
 857                                 // a thead at the end becomes a tfoot, unless there is only one row
 858                                 // Do this before deleting empty last lines to allow headers at the bottom of tables
 859                                 if ( isset( $lastRow['type'] ) && $lastRow['type'] == 'thead' && isset( $table[1] ) ) {
 860                                         $lastRow['type'] = 'tfoot';
 861                                         for ( $i = 0; isset( $lastRow[$i] ); $i++ ) {
 862                                                 $lastRow[$i]['type'] = 'th';
 863                                         }
 864                                 }
 865
 866                                 // Delete empty last lines
 867                                 if ( empty( $lastRow ) ) {
 868                                         $lastRow = NULL;
 869                                 }
 870                                 $o = '';
 871                                 $curtable = array_pop( $tables );
 872
 873                                 #Add a line-ending before the table, but only if there isn't one already
 874                                 if ( substr( $out, -1 ) !== "\n" ) {
 875                                         $o .= "\n";
 876                                 }
 877                                 $o .= $this->generateTableHTML( $curtable ) . $line . "\n";
 878
 879                                 if ( count( $tables ) > 0 ) {
 880                                         $table =& $this->last( $tables );
 881                                         $currentRow =& $this->last( $table );
 882                                         $currentElement =& $this->last( $currentRow );
 883
 884                                         $output =& $currentElement['content'];
 885                                 } else {
 886                                         $output =& $out;
 887                                 }
 888
 889                                 $output .= $o;
 890
 891                         } elseif ( $firstChars === '|-' ) {
 892                                 // start a new row element
 893                                 // but only when we haven't started one already
 894                                 if ( count( $currentRow ) != 0 ) {
 895                                         $table[] = array();
 896                                         $currentRow =& $this->last( $table );
 897                                 }
 898                                 // Get the attributes, there's nothing else useful in $line now
 899                                 $line = substr ( $line , 2 );
 900                                 $attributes = $this->mStripState->unstripBoth( $line );
 901                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 902                                 if ( $attributes !== '' ) {
 903                                         $currentRow['attributes'] = $attributes;
 904                                 }
 905
 906                         } elseif ( $firstChars  === '|+' ) {
 907                                 // a table caption, but only proceed if there isn't one already
 908                                 if ( !isset ( $table['caption'] ) ) {
 909                                         $line = substr ( $line , 2 );
 910
 911                                         $c = $this->getCellAttr( $line , 'caption' );
 912                                         $table['caption'] = array();
 913                                         $table['caption']['content'] = $c[0];
 914                                         if ( isset( $c[1] ) ) $table['caption']['attributes'] = $c[1];
 915                                         unset( $c );
 916                                         $output =& $table['caption']['content'];
 917                                 }
 918                         } elseif ( $firstChars === '|' || $firstChars === '!' || $firstChars === '!+' ) {
 919                                 // Which kind of cells are we dealing with
 920                                 $currentTag = 'td';
 921                                 $line = substr ( $line , 1 );
 922
 923                                 if ( $firstChars === '!'  || $firstChars === '!+' ) {
 924                                         $line = str_replace ( '!!' , '||' , $line );
 925                                         $currentTag = 'th';
 926                                 }
 927
 928                                 // Split up multiple cells on the same line.
 929                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 930                                 $line = ''; // save memory
 931
 932                                 // decide whether thead to tbody
 933                                 if ( !array_key_exists( 'type', $currentRow ) ) {
 934                                         $currentRow['type'] = ( $firstChars === '!' ) ? 'thead' : 'tbody' ;
 935                                 } elseif ( $firstChars === '|' ) {
 936                                         $currentRow['type'] = 'tbody';
 937                                 }
 938
 939                                 // Loop through each table cell
 940                                 foreach ( $cells as $cell ) {
 941                                         // a new cell
 942                                         $currentRow[] = array();
 943                                         $currentElement =& $this->last( $currentRow );
 944
 945                                         $currentElement['type'] = $currentTag;
 946
 947                                         $c = $this->getCellAttr( $cell , $currentTag );
 948                                         $currentElement['content'] = $c[0];
 949                                         if ( isset( $c[1] ) ) $currentElement['attributes'] = $c[1];
 950                                         unset( $c );
 951                                 }
 952                                 $output =& $currentElement['content'];
 953
 954                         } else {
 955                                 $output .= "\n$outLine";
 956                         }
 957                 }
 958
 959                 # Remove trailing line-ending (b/c)
 960                 if ( substr( $out, -1 ) === "\n" ) {
 961                         $out = substr( $out, 0, -1 );
 962                 }
 963
 964                 # Close any unclosed tables
 965                 if ( isset( $tables ) && count( $tables ) > 0 ) {
 966                         for ( $i = 0; $i < count( $tables ); $i++ ) {
 967                                 $curtable = array_pop( $tables );
 968                                 $curtable = $this->generateTableHTML( $curtable );
 969                                 #Add a line-ending before the table, but only if there isn't one already
 970                                 if ( substr( $out, -1 ) !== "\n"  && $curtable !== "" ) {
 971                                         $out .= "\n";
 972                                 }
 973                                 $out .= $curtable;
 974                         }
 975                 }
 976
 977                 wfProfileOut( __METHOD__ );
 978
 979                 return $out;
 980         }
 981
 982         /**
 983          * Helper function for doTableStuff() separating the contents of cells from
 984          * attributes. Particularly useful as there's a possible bug and this action
 985          * is repeated twice.
 986          *
 987          * @private
 988          * @param $cell
 989          * @param $tagName
 990          * @return array
 991          */
 992         function getCellAttr ( $cell, $tagName ) {
 993                 $attributes = null;
 994
 995                 $cell = trim ( $cell );
 996
 997                 // A cell could contain both parameters and data
 998                 $cellData = explode ( '|' , $cell , 2 );
 999
1000                 // Bug 553: Note that a '|' inside an invalid link should not
1001                 // be mistaken as delimiting cell parameters
1002                 if ( strpos( $cellData[0], '[[' ) !== false ) {
1003                         $content = trim ( $cell );
1004                 }
1005                 elseif ( count ( $cellData ) == 1 ) {
1006                         $content = trim ( $cellData[0] );
1007                 } else {
1008                         $attributes = $this->mStripState->unstripBoth( $cellData[0] );
1009                         $attributes = Sanitizer::fixTagAttributes( $attributes , $tagName );
1010
1011                         $content = trim ( $cellData[1] );
1012                 }
1013                 return array( $content, $attributes );
1014         }
1015
1016
1017         /**
1018          * Helper function for doTableStuff(). This converts the structured array into html.
1019          *
1020          * @private
1021          */
1022         function generateTableHTML( &$table ) {
1023                 $return = "";
1024                 $return .= str_repeat( '<dl><dd>' , $table['indent'] );
1025                 $return .= '<table';
1026                 $return .= isset( $table['attributes'] ) ? $table['attributes'] : '';
1027                 $return .= '>';
1028                 unset( $table['attributes'] );
1029
1030                 if ( isset( $table['caption'] ) ) {
1031                         $return .= "\n<caption";
1032                         $return .= isset( $table['caption']['attributes'] ) ? $table['caption']['attributes'] : '';
1033                         $return .= '>';
1034                         $return .= $table['caption']['content'];
1035                         $return .= "\n</caption>";
1036                 }
1037                 $lastSection = '';
1038                 $empty = true;
1039                 $simple = true;
1040
1041                 // If we only have tbodies, mark table as simple
1042                 for ( $i = 0; isset( $table[$i] ); $i++ ) {
1043                         if ( !count( $table[$i] ) ) continue;
1044                         if ( !isset( $table[$i]['type'] ) ) {
1045                                 $table[$i]['type'] = 'tbody';
1046                         }
1047                         if ( !$lastSection ) {
1048                                 $lastSection = $table[$i]['type'];
1049                         } elseif ( $lastSection != $table[$i]['type'] ) {
1050                                 $simple = false;
1051                         }
1052                 }
1053                 $lastSection = '';
1054                 for ( $i = 0; isset( $table[$i] ); $i++ ) {
1055                         if ( !count( $table[$i] ) ) continue;
1056                         $empty = false; // check for empty tables
1057
1058                         if ( $table[$i]['type'] != $lastSection && !$simple ) {
1059                                 $return .= "\n<" . $table[$i]['type'] . '>';
1060                         }
1061
1062                         $return .= "\n<tr";
1063                         $return .= isset( $table[$i]['attributes'] ) ? $table[$i]['attributes'] : '';
1064                         $return .= '>';
1065                         for ( $j = 0; isset( $table[$i][$j] ); $j++ ) {
1066                                 if ( !isset( $table[$i][$j]['type'] ) ) $table[$i][$j]['type'] = 'td';
1067                                 $return .= "\n<" . $table[$i][$j]['type'];
1068                                 $return .= isset( $table[$i][$j]['attributes'] ) ? $table[$i][$j]['attributes'] : '';
1069                                 $return .= '>';
1070
1071                                 $return .= $table[$i][$j]['content'];
1072                                 if ( $table[$i][$j]['content'] != '' )
1073                                         $return .= "\n";
1074
1075                                 $return .= '</' . $table[$i][$j]['type'] . '>';
1076                                 unset( $table[$i][$j] );
1077                         }
1078                         $return .= "\n</tr>";
1079
1080                         if ( ( !isset( $table[$i + 1] ) && !$simple ) || ( isset( $table[$i + 1] ) && isset( $table[$i + 1]['type'] ) && $table[$i]['type'] != $table[$i + 1]['type'] ) ) {
1081                                 $return .= '</' . $table[$i]['type'] . '>';
1082                         }
1083                         $lastSection = $table[$i]['type'];
1084                         unset( $table[$i] );
1085                 }
1086                 if ( $empty ) {
1087                         if ( isset( $table['caption'] ) ) {
1088                                 $return .= "\n<tr><td></td></tr>";
1089                         } else {
1090                                 return '';
1091                         }
1092                 }
1093                 $return .= "\n</table>";
1094                 $return .= str_repeat( '</dd></dl>' , $table['indent'] );
1095
1096                 return $return;
1097         }
1098
1099         /**
1100          * like end() but only works on the numeric array index and php's internal pointers
1101          * returns a reference to the last element of an array much like "\$arr[-1]" in perl
1102          * ignores associative elements and will create a 0 key will a NULL value if there were
1103          * no numric elements and an array itself if not previously defined.
1104          *
1105          * @private
1106          */
1107         function &last ( &$arr ) {
1108                 for ( $i = count( $arr ); ( !isset( $arr[$i] ) && $i > 0 ); $i-- ) {  }
1109                 return $arr[$i];
1110         }
1111
1112         /**
1113          * Helper function for parse() that transforms wiki markup into
1114          * HTML. Only called for $mOutputType == self::OT_HTML.
1115          *
1116          * @private
1117          */
1118         function internalParse( $text, $isMain = true, $frame=false ) {
1119                 wfProfileIn( __METHOD__ );
1120
1121                 $origText = $text;
1122
1123                 # Hook to suspend the parser in this state
1124                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1125                         wfProfileOut( __METHOD__ );
1126                         return $text ;
1127                 }
1128
1129                 # if $frame is provided, then use $frame for replacing any variables
1130                 if ( $frame ) {
1131                         # use frame depth to infer how include/noinclude tags should be handled
1132                         # depth=0 means this is the top-level document; otherwise it's an included document
1133                         if ( !$frame->depth ) {
1134                                 $flag = 0;
1135                         } else {
1136                                 $flag = Parser::PTD_FOR_INCLUSION;
1137                         }
1138                         $dom = $this->preprocessToDom( $text, $flag );
1139                         $text = $frame->expand( $dom );
1140                 } else {
1141                         # if $frame is not provided, then use old-style replaceVariables
1142                         $text = $this->replaceVariables( $text );
1143                 }
1144
1145                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
1146                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1147
1148                 # Tables need to come after variable replacement for things to work
1149                 # properly; putting them before other transformations should keep
1150                 # exciting things like link expansions from showing up in surprising
1151                 # places.
1152                 $text = $this->doTableStuff( $text );
1153
1154                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1155
1156                 $text = $this->doDoubleUnderscore( $text );
1157
1158                 $text = $this->doHeadings( $text );
1159                 if ( $this->mOptions->getUseDynamicDates() ) {
1160                         $df = DateFormatter::getInstance();
1161                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1162                 }
1163                 $text = $this->replaceInternalLinks( $text );
1164                 $text = $this->doAllQuotes( $text );
1165                 $text = $this->replaceExternalLinks( $text );
1166
1167                 # replaceInternalLinks may sometimes leave behind
1168                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1169                 $text = str_replace( $this->mUniqPrefix.'NOPARSE', '', $text );
1170
1171                 $text = $this->doMagicLinks( $text );
1172                 $text = $this->formatHeadings( $text, $origText, $isMain );
1173
1174                 wfProfileOut( __METHOD__ );
1175                 return $text;
1176         }
1177
1178         /**
1179          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1180          * magic external links.
1181          *
1182          * DML
1183          * @private
1184          */
1185         function doMagicLinks( $text ) {
1186                 wfProfileIn( __METHOD__ );
1187                 $prots = $this->mUrlProtocols;
1188                 $urlChar = self::EXT_LINK_URL_CLASS;
1189                 $text = preg_replace_callback(
1190                         '!(?:                           # Start cases
1191                                 (<a[ \t\r\n>].*?</a>) |     # m[1]: Skip link text
1192                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1193                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
1194                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1195                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1196                                         (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1197                                         (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1198                                         [0-9Xx]                 # check digit
1199                                         \b)
1200                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1201                 wfProfileOut( __METHOD__ );
1202                 return $text;
1203         }
1204
1205         /**
1206          * @throws MWException
1207          * @param $m array
1208          * @return HTML|string
1209          */
1210         function magicLinkCallback( $m ) {
1211                 if ( isset( $m[1] ) && $m[1] !== '' ) {
1212                         # Skip anchor
1213                         return $m[0];
1214                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1215                         # Skip HTML element
1216                         return $m[0];
1217                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1218                         # Free external link
1219                         return $this->makeFreeExternalLink( $m[0] );
1220                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1221                         # RFC or PMID
1222                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1223                                 $keyword = 'RFC';
1224                                 $urlmsg = 'rfcurl';
1225                                 $CssClass = 'mw-magiclink-rfc';
1226                                 $id = $m[4];
1227                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1228                                 $keyword = 'PMID';
1229                                 $urlmsg = 'pubmedurl';
1230                                 $CssClass = 'mw-magiclink-pmid';
1231                                 $id = $m[4];
1232                         } else {
1233                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1234                                         substr( $m[0], 0, 20 ) . '"' );
1235                         }
1236                         $url = wfMsgForContent( $urlmsg, $id );
1237                         return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $CssClass );
1238                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1239                         # ISBN
1240                         $isbn = $m[5];
1241                         $num = strtr( $isbn, array(
1242                                 '-' => '',
1243                                 ' ' => '',
1244                                 'x' => 'X',
1245                         ));
1246                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1247                         return'<a href="' .
1248                                 $titleObj->escapeLocalUrl() .
1249                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1250                 } else {
1251                         return $m[0];
1252                 }
1253         }
1254
1255         /**
1256          * Make a free external link, given a user-supplied URL
1257          * @return HTML
1258          * @private
1259          */
1260         function makeFreeExternalLink( $url ) {
1261                 global $wgContLang;
1262                 wfProfileIn( __METHOD__ );
1263
1264                 $trail = '';
1265
1266                 # The characters '<' and '>' (which were escaped by
1267                 # removeHTMLtags()) should not be included in
1268                 # URLs, per RFC 2396.
1269                 $m2 = array();
1270                 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1271                         $trail = substr( $url, $m2[0][1] ) . $trail;
1272                         $url = substr( $url, 0, $m2[0][1] );
1273                 }
1274
1275                 # Move trailing punctuation to $trail
1276                 $sep = ',;\.:!?';
1277                 # If there is no left bracket, then consider right brackets fair game too
1278                 if ( strpos( $url, '(' ) === false ) {
1279                         $sep .= ')';
1280                 }
1281
1282                 $numSepChars = strspn( strrev( $url ), $sep );
1283                 if ( $numSepChars ) {
1284                         $trail = substr( $url, -$numSepChars ) . $trail;
1285                         $url = substr( $url, 0, -$numSepChars );
1286                 }
1287
1288                 $url = Sanitizer::cleanUrl( $url );
1289
1290                 # Is this an external image?
1291                 $text = $this->maybeMakeExternalImage( $url );
1292                 if ( $text === false ) {
1293                         # Not an image, make a link
1294                         $text = Linker::makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1295                                 $this->getExternalLinkAttribs( $url ) );
1296                         # Register it in the output object...
1297                         # Replace unnecessary URL escape codes with their equivalent characters
1298                         $pasteurized = self::replaceUnusualEscapes( $url );
1299                         $this->mOutput->addExternalLink( $pasteurized );
1300                 }
1301                 wfProfileOut( __METHOD__ );
1302                 return $text . $trail;
1303         }
1304
1305
1306         /**
1307          * Parse headers and return html
1308          *
1309          * @private
1310          */
1311         function doHeadings( $text ) {
1312                 wfProfileIn( __METHOD__ );
1313                 for ( $i = 6; $i >= 1; --$i ) {
1314                         $h = str_repeat( '=', $i );
1315                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1316                           "<h$i>\\1</h$i>", $text );
1317                 }
1318                 wfProfileOut( __METHOD__ );
1319                 return $text;
1320         }
1321
1322         /**
1323          * Replace single quotes with HTML markup
1324          * @private
1325          * @return string the altered text
1326          */
1327         function doAllQuotes( $text ) {
1328                 wfProfileIn( __METHOD__ );
1329                 $outtext = '';
1330                 $lines = StringUtils::explode( "\n", $text );
1331                 foreach ( $lines as $line ) {
1332                         $outtext .= $this->doQuotes( $line ) . "\n";
1333                 }
1334                 $outtext = substr( $outtext, 0,-1 );
1335                 wfProfileOut( __METHOD__ );
1336                 return $outtext;
1337         }
1338
1339         /**
1340          * Helper function for doAllQuotes()
1341          */
1342         public function doQuotes( $text ) {
1343                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1344                 if ( count( $arr ) == 1 ) {
1345                         return $text;
1346                 } else {
1347                         # First, do some preliminary work. This may shift some apostrophes from
1348                         # being mark-up to being text. It also counts the number of occurrences
1349                         # of bold and italics mark-ups.
1350                         $numbold = 0;
1351                         $numitalics = 0;
1352                         for ( $i = 0; $i < count( $arr ); $i++ ) {
1353                                 if ( ( $i % 2 ) == 1 ) {
1354                                         # If there are ever four apostrophes, assume the first is supposed to
1355                                         # be text, and the remaining three constitute mark-up for bold text.
1356                                         if ( strlen( $arr[$i] ) == 4 ) {
1357                                                 $arr[$i-1] .= "'";
1358                                                 $arr[$i] = "'''";
1359                                         } elseif ( strlen( $arr[$i] ) > 5 ) {
1360                                                 # If there are more than 5 apostrophes in a row, assume they're all
1361                                                 # text except for the last 5.
1362                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1363                                                 $arr[$i] = "'''''";
1364                                         }
1365                                         # Count the number of occurrences of bold and italics mark-ups.
1366                                         # We are not counting sequences of five apostrophes.
1367                                         if ( strlen( $arr[$i] ) == 2 ) {
1368                                                 $numitalics++;
1369                                         } elseif ( strlen( $arr[$i] ) == 3 ) {
1370                                                 $numbold++;
1371                                         } elseif ( strlen( $arr[$i] ) == 5 ) {
1372                                                 $numitalics++;
1373                                                 $numbold++;
1374                                         }
1375                                 }
1376                         }
1377
1378                         # If there is an odd number of both bold and italics, it is likely
1379                         # that one of the bold ones was meant to be an apostrophe followed
1380                         # by italics. Which one we cannot know for certain, but it is more
1381                         # likely to be one that has a single-letter word before it.
1382                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1383                                 $i = 0;
1384                                 $firstsingleletterword = -1;
1385                                 $firstmultiletterword = -1;
1386                                 $firstspace = -1;
1387                                 foreach ( $arr as $r ) {
1388                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) ) {
1389                                                 $x1 = substr( $arr[$i-1], -1 );
1390                                                 $x2 = substr( $arr[$i-1], -2, 1 );
1391                                                 if ( $x1 === ' ' ) {
1392                                                         if ( $firstspace == -1 ) {
1393                                                                 $firstspace = $i;
1394                                                         }
1395                                                 } elseif ( $x2 === ' ') {
1396                                                         if ( $firstsingleletterword == -1 ) {
1397                                                                 $firstsingleletterword = $i;
1398                                                         }
1399                                                 } else {
1400                                                         if ( $firstmultiletterword == -1 ) {
1401                                                                 $firstmultiletterword = $i;
1402                                                         }
1403                                                 }
1404                                         }
1405                                         $i++;
1406                                 }
1407
1408                                 # If there is a single-letter word, use it!
1409                                 if ( $firstsingleletterword > -1 ) {
1410                                         $arr[$firstsingleletterword] = "''";
1411                                         $arr[$firstsingleletterword-1] .= "'";
1412                                 } elseif ( $firstmultiletterword > -1 ) {
1413                                         # If not, but there's a multi-letter word, use that one.
1414                                         $arr[$firstmultiletterword] = "''";
1415                                         $arr[$firstmultiletterword-1] .= "'";
1416                                 } elseif ( $firstspace > -1 ) {
1417                                         # ... otherwise use the first one that has neither.
1418                                         # (notice that it is possible for all three to be -1 if, for example,
1419                                         # there is only one pentuple-apostrophe in the line)
1420                                         $arr[$firstspace] = "''";
1421                                         $arr[$firstspace-1] .= "'";
1422                                 }
1423                         }
1424
1425                         # Now let's actually convert our apostrophic mush to HTML!
1426                         $output = '';
1427                         $buffer = '';
1428                         $state = '';
1429                         $i = 0;
1430                         foreach ( $arr as $r ) {
1431                                 if ( ( $i % 2 ) == 0 ) {
1432                                         if ( $state === 'both' ) {
1433                                                 $buffer .= $r;
1434                                         } else {
1435                                                 $output .= $r;
1436                                         }
1437                                 } else {
1438                                         if ( strlen( $r ) == 2 ) {
1439                                                 if ( $state === 'i' ) {
1440                                                         $output .= '</i>'; $state = '';
1441                                                 } elseif ( $state === 'bi' ) {
1442                                                         $output .= '</i>'; $state = 'b';
1443                                                 } elseif ( $state === 'ib' ) {
1444                                                         $output .= '</b></i><b>'; $state = 'b';
1445                                                 } elseif ( $state === 'both' ) {
1446                                                         $output .= '<b><i>'.$buffer.'</i>'; $state = 'b';
1447                                                 } else { # $state can be 'b' or ''
1448                                                         $output .= '<i>'; $state .= 'i';
1449                                                 }
1450                                         } elseif ( strlen( $r ) == 3 ) {
1451                                                 if ( $state === 'b' ) {
1452                                                         $output .= '</b>'; $state = '';
1453                                                 } elseif ( $state === 'bi' ) {
1454                                                         $output .= '</i></b><i>'; $state = 'i';
1455                                                 } elseif ( $state === 'ib' ) {
1456                                                         $output .= '</b>'; $state = 'i';
1457                                                 } elseif ( $state === 'both' ) {
1458                                                         $output .= '<i><b>'.$buffer.'</b>'; $state = 'i';
1459                                                 } else { # $state can be 'i' or ''
1460                                                         $output .= '<b>'; $state .= 'b';
1461                                                 }
1462                                         } elseif ( strlen( $r ) == 5 ) {
1463                                                 if ( $state === 'b' ) {
1464                                                         $output .= '</b><i>'; $state = 'i';
1465                                                 } elseif ( $state === 'i' ) {
1466                                                         $output .= '</i><b>'; $state = 'b';
1467                                                 } elseif ( $state === 'bi' ) {
1468                                                         $output .= '</i></b>'; $state = '';
1469                                                 } elseif ( $state === 'ib' ) {
1470                                                         $output .= '</b></i>'; $state = '';
1471                                                 } elseif ( $state === 'both' ) {
1472                                                         $output .= '<i><b>'.$buffer.'</b></i>'; $state = '';
1473                                                 } else { # ($state == '')
1474                                                         $buffer = ''; $state = 'both';
1475                                                 }
1476                                         }
1477                                 }
1478                                 $i++;
1479                         }
1480                         # Now close all remaining tags.  Notice that the order is important.
1481                         if ( $state === 'b' || $state === 'ib' ) {
1482                                 $output .= '</b>';
1483                         }
1484                         if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1485                                 $output .= '</i>';
1486                         }
1487                         if ( $state === 'bi' ) {
1488                                 $output .= '</b>';
1489                         }
1490                         # There might be lonely ''''', so make sure we have a buffer
1491                         if ( $state === 'both' && $buffer ) {
1492                                 $output .= '<b><i>'.$buffer.'</i></b>';
1493                         }
1494                         return $output;
1495                 }
1496         }
1497
1498         /**
1499          * Replace external links (REL)
1500          *
1501          * Note: this is all very hackish and the order of execution matters a lot.
1502          * Make sure to run maintenance/parserTests.php if you change this code.
1503          *
1504          * @private
1505          */
1506         function replaceExternalLinks( $text ) {
1507                 global $wgContLang;
1508                 wfProfileIn( __METHOD__ );
1509
1510                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1511                 $s = array_shift( $bits );
1512
1513                 $i = 0;
1514                 while ( $i<count( $bits ) ) {
1515                         $url = $bits[$i++];
1516                         $protocol = $bits[$i++];
1517                         $text = $bits[$i++];
1518                         $trail = $bits[$i++];
1519
1520                         # The characters '<' and '>' (which were escaped by
1521                         # removeHTMLtags()) should not be included in
1522                         # URLs, per RFC 2396.
1523                         $m2 = array();
1524                         if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1525                                 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1526                                 $url = substr( $url, 0, $m2[0][1] );
1527                         }
1528
1529                         # If the link text is an image URL, replace it with an <img> tag
1530                         # This happened by accident in the original parser, but some people used it extensively
1531                         $img = $this->maybeMakeExternalImage( $text );
1532                         if ( $img !== false ) {
1533                                 $text = $img;
1534                         }
1535
1536                         $dtrail = '';
1537
1538                         # Set linktype for CSS - if URL==text, link is essentially free
1539                         $linktype = ( $text === $url ) ? 'free' : 'text';
1540
1541                         # No link text, e.g. [http://domain.tld/some.link]
1542                         if ( $text == '' ) {
1543                                 # Autonumber if allowed. See bug #5918
1544                                 if ( strpos( wfUrlProtocols(), substr( $protocol, 0, strpos( $protocol, ':' ) ) ) !== false ) {
1545                                         $langObj = $this->getFunctionLang();
1546                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1547                                         $linktype = 'autonumber';
1548                                 } else {
1549                                         # Otherwise just use the URL
1550                                         $text = htmlspecialchars( $url );
1551                                         $linktype = 'free';
1552                                 }
1553                         } else {
1554                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1555                                 # Check for trail
1556                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1557                         }
1558
1559                         $text = $wgContLang->markNoConversion( $text );
1560
1561                         $url = Sanitizer::cleanUrl( $url );
1562
1563                         # Use the encoded URL
1564                         # This means that users can paste URLs directly into the text
1565                         # Funny characters like ö aren't valid in URLs anyway
1566                         # This was changed in August 2004
1567                         $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1568                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1569
1570                         # Register link in the output object.
1571                         # Replace unnecessary URL escape codes with the referenced character
1572                         # This prevents spammers from hiding links from the filters
1573                         $pasteurized = self::replaceUnusualEscapes( $url );
1574                         $this->mOutput->addExternalLink( $pasteurized );
1575                 }
1576
1577                 wfProfileOut( __METHOD__ );
1578                 return $s;
1579         }
1580
1581         /**
1582          * Get an associative array of additional HTML attributes appropriate for a
1583          * particular external link.  This currently may include rel => nofollow
1584          * (depending on configuration, namespace, and the URL's domain) and/or a
1585          * target attribute (depending on configuration).
1586          *
1587          * @param $url String: optional URL, to extract the domain from for rel =>
1588          *   nofollow if appropriate
1589          * @return Array: associative array of HTML attributes
1590          */
1591         function getExternalLinkAttribs( $url = false ) {
1592                 $attribs = array();
1593                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1594                 $ns = $this->mTitle->getNamespace();
1595                 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) ) {
1596                         $attribs['rel'] = 'nofollow';
1597
1598                         global $wgNoFollowDomainExceptions;
1599                         if ( $wgNoFollowDomainExceptions ) {
1600                                 $bits = wfParseUrl( $url );
1601                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1602                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1603                                                 if ( substr( $bits['host'], -strlen( $domain ) ) == $domain ) {
1604                                                         unset( $attribs['rel'] );
1605                                                         break;
1606                                                 }
1607                                         }
1608                                 }
1609                         }
1610                 }
1611                 if ( $this->mOptions->getExternalLinkTarget() ) {
1612                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1613                 }
1614                 return $attribs;
1615         }
1616
1617         /**
1618          * Replace unusual URL escape codes with their equivalent characters
1619          *
1620          * @param $url String
1621          * @return String
1622          *
1623          * @todo  This can merge genuinely required bits in the path or query string,
1624          *        breaking legit URLs. A proper fix would treat the various parts of
1625          *        the URL differently; as a workaround, just use the output for
1626          *        statistical records, not for actual linking/output.
1627          */
1628         static function replaceUnusualEscapes( $url ) {
1629                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1630                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1631         }
1632
1633         /**
1634          * Callback function used in replaceUnusualEscapes().
1635          * Replaces unusual URL escape codes with their equivalent character
1636          */
1637         private static function replaceUnusualEscapesCallback( $matches ) {
1638                 $char = urldecode( $matches[0] );
1639                 $ord = ord( $char );
1640                 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1641                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1642                         # No, shouldn't be escaped
1643                         return $char;
1644                 } else {
1645                         # Yes, leave it escaped
1646                         return $matches[0];
1647                 }
1648         }
1649
1650         /**
1651          * make an image if it's allowed, either through the global
1652          * option, through the exception, or through the on-wiki whitelist
1653          * @private
1654          */
1655         function maybeMakeExternalImage( $url ) {
1656                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1657                 $imagesexception = !empty( $imagesfrom );
1658                 $text = false;
1659                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1660                 if ( $imagesexception && is_array( $imagesfrom ) ) {
1661                         $imagematch = false;
1662                         foreach ( $imagesfrom as $match ) {
1663                                 if ( strpos( $url, $match ) === 0 ) {
1664                                         $imagematch = true;
1665                                         break;
1666                                 }
1667                         }
1668                 } elseif ( $imagesexception ) {
1669                         $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1670                 } else {
1671                         $imagematch = false;
1672                 }
1673                 if ( $this->mOptions->getAllowExternalImages()
1674                          || ( $imagesexception && $imagematch ) ) {
1675                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1676                                 # Image found
1677                                 $text = Linker::makeExternalImage( $url );
1678                         }
1679                 }
1680                 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1681                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1682                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1683                         foreach ( $whitelist as $entry ) {
1684                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1685                                 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1686                                         continue;
1687                                 }
1688                                 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1689                                         # Image matches a whitelist entry
1690                                         $text = Linker::makeExternalImage( $url );
1691                                         break;
1692                                 }
1693                         }
1694                 }
1695                 return $text;
1696         }
1697
1698         /**
1699          * Process [[ ]] wikilinks
1700          * @return String: processed text
1701          *
1702          * @private
1703          */
1704         function replaceInternalLinks( $s ) {
1705                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1706                 return $s;
1707         }
1708
1709         /**
1710          * Process [[ ]] wikilinks (RIL)
1711          * @return LinkHolderArray
1712          *
1713          * @private
1714          */
1715         function replaceInternalLinks2( &$s ) {
1716                 global $wgContLang;
1717
1718                 wfProfileIn( __METHOD__ );
1719
1720                 wfProfileIn( __METHOD__.'-setup' );
1721                 static $tc = FALSE, $e1, $e1_img;
1722                 # the % is needed to support urlencoded titles as well
1723                 if ( !$tc ) {
1724                         $tc = Title::legalChars() . '#%';
1725                         # Match a link having the form [[namespace:link|alternate]]trail
1726                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1727                         # Match cases where there is no "]]", which might still be images
1728                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1729                 }
1730
1731                 $holders = new LinkHolderArray( $this );
1732
1733                 # split the entire text string on occurences of [[
1734                 $a = StringUtils::explode( '[[', ' ' . $s );
1735                 # get the first element (all text up to first [[), and remove the space we added
1736                 $s = $a->current();
1737                 $a->next();
1738                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1739                 $s = substr( $s, 1 );
1740
1741                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1742                 $e2 = null;
1743                 if ( $useLinkPrefixExtension ) {
1744                         # Match the end of a line for a word that's not followed by whitespace,
1745                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1746                         $e2 = wfMsgForContent( 'linkprefix' );
1747                 }
1748
1749                 if ( is_null( $this->mTitle ) ) {
1750                         wfProfileOut( __METHOD__.'-setup' );
1751                         wfProfileOut( __METHOD__ );
1752                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1753                 }
1754                 $nottalk = !$this->mTitle->isTalkPage();
1755
1756                 if ( $useLinkPrefixExtension ) {
1757                         $m = array();
1758                         if ( preg_match( $e2, $s, $m ) ) {
1759                                 $first_prefix = $m[2];
1760                         } else {
1761                                 $first_prefix = false;
1762                         }
1763                 } else {
1764                         $prefix = '';
1765                 }
1766
1767                 if ( $wgContLang->hasVariants() ) {
1768                         $selflink = $wgContLang->autoConvertToAllVariants( $this->mTitle->getPrefixedText() );
1769                 } else {
1770                         $selflink = array( $this->mTitle->getPrefixedText() );
1771                 }
1772                 $useSubpages = $this->areSubpagesAllowed();
1773                 wfProfileOut( __METHOD__.'-setup' );
1774
1775                 # Loop for each link
1776                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1777                         # Check for excessive memory usage
1778                         if ( $holders->isBig() ) {
1779                                 # Too big
1780                                 # Do the existence check, replace the link holders and clear the array
1781                                 $holders->replace( $s );
1782                                 $holders->clear();
1783                         }
1784
1785                         if ( $useLinkPrefixExtension ) {
1786                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1787                                 if ( preg_match( $e2, $s, $m ) ) {
1788                                         $prefix = $m[2];
1789                                         $s = $m[1];
1790                                 } else {
1791                                         $prefix='';
1792                                 }
1793                                 # first link
1794                                 if ( $first_prefix ) {
1795                                         $prefix = $first_prefix;
1796                                         $first_prefix = false;
1797                                 }
1798                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1799                         }
1800
1801                         $might_be_img = false;
1802
1803                         wfProfileIn( __METHOD__."-e1" );
1804                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1805                                 $text = $m[2];
1806                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1807                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1808                                 # the real problem is with the $e1 regex
1809                                 # See bug 1300.
1810                                 #
1811                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1812                                 # and no image is in sight. See bug 2095.
1813                                 #
1814                                 if ( $text !== '' &&
1815                                         substr( $m[3], 0, 1 ) === ']' &&
1816                                         strpos( $text, '[' ) !== false
1817                                 )
1818                                 {
1819                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1820                                         $m[3] = substr( $m[3], 1 );
1821                                 }
1822                                 # fix up urlencoded title texts
1823                                 if ( strpos( $m[1], '%' ) !== false ) {
1824                                         # Should anchors '#' also be rejected?
1825                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), rawurldecode( $m[1] ) );
1826                                 }
1827                                 $trail = $m[3];
1828                         } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1829                                 $might_be_img = true;
1830                                 $text = $m[2];
1831                                 if ( strpos( $m[1], '%' ) !== false ) {
1832                                         $m[1] = rawurldecode( $m[1] );
1833                                 }
1834                                 $trail = "";
1835                         } else { # Invalid form; output directly
1836                                 $s .= $prefix . '[[' . $line ;
1837                                 wfProfileOut( __METHOD__."-e1" );
1838                                 continue;
1839                         }
1840                         wfProfileOut( __METHOD__."-e1" );
1841                         wfProfileIn( __METHOD__."-misc" );
1842
1843                         # Don't allow internal links to pages containing
1844                         # PROTO: where PROTO is a valid URL protocol; these
1845                         # should be external links.
1846                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1847                                 $s .= $prefix . '[[' . $line ;
1848                                 wfProfileOut( __METHOD__."-misc" );
1849                                 continue;
1850                         }
1851
1852                         # Make subpage if necessary
1853                         if ( $useSubpages ) {
1854                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1855                         } else {
1856                                 $link = $m[1];
1857                         }
1858
1859                         $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1860                         if ( !$noforce ) {
1861                                 # Strip off leading ':'
1862                                 $link = substr( $link, 1 );
1863                         }
1864
1865                         wfProfileOut( __METHOD__."-misc" );
1866                         wfProfileIn( __METHOD__."-title" );
1867                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1868                         if ( $nt === null ) {
1869                                 $s .= $prefix . '[[' . $line;
1870                                 wfProfileOut( __METHOD__."-title" );
1871                                 continue;
1872                         }
1873
1874                         $ns = $nt->getNamespace();
1875                         $iw = $nt->getInterWiki();
1876                         wfProfileOut( __METHOD__."-title" );
1877
1878                         if ( $might_be_img ) { # if this is actually an invalid link
1879                                 wfProfileIn( __METHOD__."-might_be_img" );
1880                                 if ( $ns == NS_FILE && $noforce ) { # but might be an image
1881                                         $found = false;
1882                                         while ( true ) {
1883                                                 # look at the next 'line' to see if we can close it there
1884                                                 $a->next();
1885                                                 $next_line = $a->current();
1886                                                 if ( $next_line === false || $next_line === null ) {
1887                                                         break;
1888                                                 }
1889                                                 $m = explode( ']]', $next_line, 3 );
1890                                                 if ( count( $m ) == 3 ) {
1891                                                         # the first ]] closes the inner link, the second the image
1892                                                         $found = true;
1893                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1894                                                         $trail = $m[2];
1895                                                         break;
1896                                                 } elseif ( count( $m ) == 2 ) {
1897                                                         # if there's exactly one ]] that's fine, we'll keep looking
1898                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1899                                                 } else {
1900                                                         # if $next_line is invalid too, we need look no further
1901                                                         $text .= '[[' . $next_line;
1902                                                         break;
1903                                                 }
1904                                         }
1905                                         if ( !$found ) {
1906                                                 # we couldn't find the end of this imageLink, so output it raw
1907                                                 # but don't ignore what might be perfectly normal links in the text we've examined
1908                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1909                                                 $s .= "{$prefix}[[$link|$text";
1910                                                 # note: no $trail, because without an end, there *is* no trail
1911                                                 wfProfileOut( __METHOD__."-might_be_img" );
1912                                                 continue;
1913                                         }
1914                                 } else { # it's not an image, so output it raw
1915                                         $s .= "{$prefix}[[$link|$text";
1916                                         # note: no $trail, because without an end, there *is* no trail
1917                                         wfProfileOut( __METHOD__."-might_be_img" );
1918                                         continue;
1919                                 }
1920                                 wfProfileOut( __METHOD__."-might_be_img" );
1921                         }
1922
1923                         $wasblank = ( $text  == '' );
1924                         if ( $wasblank ) {
1925                                 $text = $link;
1926                         } else {
1927                                 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
1928                                 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
1929                                 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
1930                                 #    -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
1931                                 $text = $this->doQuotes( $text );
1932                         }
1933
1934                         # Link not escaped by : , create the various objects
1935                         if ( $noforce ) {
1936
1937                                 # Interwikis
1938                                 wfProfileIn( __METHOD__."-interwiki" );
1939                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1940                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1941                                         $s = rtrim( $s . $prefix );
1942                                         $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
1943                                         wfProfileOut( __METHOD__."-interwiki" );
1944                                         continue;
1945                                 }
1946                                 wfProfileOut( __METHOD__."-interwiki" );
1947
1948                                 if ( $ns == NS_FILE ) {
1949                                         wfProfileIn( __METHOD__."-image" );
1950                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1951                                                 if ( $wasblank ) {
1952                                                         # if no parameters were passed, $text
1953                                                         # becomes something like "File:Foo.png",
1954                                                         # which we don't want to pass on to the
1955                                                         # image generator
1956                                                         $text = '';
1957                                                 } else {
1958                                                         # recursively parse links inside the image caption
1959                                                         # actually, this will parse them in any other parameters, too,
1960                                                         # but it might be hard to fix that, and it doesn't matter ATM
1961                                                         $text = $this->replaceExternalLinks( $text );
1962                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1963                                                 }
1964                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1965                                                 $s .= $prefix . $this->armorLinks(
1966                                                         $this->makeImage( $nt, $text, $holders ) ) . $trail;
1967                                         } else {
1968                                                 $s .= $prefix . $trail;
1969                                         }
1970                                         wfProfileOut( __METHOD__."-image" );
1971                                         continue;
1972                                 }
1973
1974                                 if ( $ns == NS_CATEGORY ) {
1975                                         wfProfileIn( __METHOD__."-category" );
1976                                         $s = rtrim( $s . "\n" ); # bug 87
1977
1978                                         if ( $wasblank ) {
1979                                                 $sortkey = $this->getDefaultSort();
1980                                         } else {
1981                                                 $sortkey = $text;
1982                                         }
1983                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1984                                         $sortkey = str_replace( "\n", '', $sortkey );
1985                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1986                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1987
1988                                         /**
1989                                          * Strip the whitespace Category links produce, see bug 87
1990                                          * @todo We might want to use trim($tmp, "\n") here.
1991                                          */
1992                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
1993
1994                                         wfProfileOut( __METHOD__."-category" );
1995                                         continue;
1996                                 }
1997                         }
1998
1999                         # Self-link checking
2000                         if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
2001                                 if ( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
2002                                         $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2003                                         continue;
2004                                 }
2005                         }
2006
2007                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
2008                         # @todo FIXME: Should do batch file existence checks, see comment below
2009                         if ( $ns == NS_MEDIA ) {
2010                                 wfProfileIn( __METHOD__."-media" );
2011                                 # Give extensions a chance to select the file revision for us
2012                                 $time = $sha1 = $descQuery = false;
2013                                 wfRunHooks( 'BeforeParserFetchFileAndTitle',
2014                                         array( $this, $nt, &$time, &$sha1, &$descQuery ) );
2015                                 # Fetch and register the file (file title may be different via hooks)
2016                                 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $time, $sha1 );
2017                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2018                                 $s .= $prefix . $this->armorLinks(
2019                                         Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2020                                 wfProfileOut( __METHOD__."-media" );
2021                                 continue;
2022                         }
2023
2024                         wfProfileIn( __METHOD__."-always_known" );
2025                         # Some titles, such as valid special pages or files in foreign repos, should
2026                         # be shown as bluelinks even though they're not included in the page table
2027                         #
2028                         # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2029                         # batch file existence checks for NS_FILE and NS_MEDIA
2030                         if ( $iw == '' && $nt->isAlwaysKnown() ) {
2031                                 $this->mOutput->addLink( $nt );
2032                                 $s .= $this->makeKnownLinkHolder( $nt, $text, array(), $trail, $prefix );
2033                         } else {
2034                                 # Links will be added to the output link list after checking
2035                                 $s .= $holders->makeHolder( $nt, $text, array(), $trail, $prefix );
2036                         }
2037                         wfProfileOut( __METHOD__."-always_known" );
2038                 }
2039                 wfProfileOut( __METHOD__ );
2040                 return $holders;
2041         }
2042
2043         /**
2044          * Render a forced-blue link inline; protect against double expansion of
2045          * URLs if we're in a mode that prepends full URL prefixes to internal links.
2046          * Since this little disaster has to split off the trail text to avoid
2047          * breaking URLs in the following text without breaking trails on the
2048          * wiki links, it's been made into a horrible function.
2049          *
2050          * @param $nt Title
2051          * @param $text String
2052          * @param $query Array or String
2053          * @param $trail String
2054          * @param $prefix String
2055          * @return String: HTML-wikitext mix oh yuck
2056          */
2057         function makeKnownLinkHolder( $nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
2058                 list( $inside, $trail ) = Linker::splitTrail( $trail );
2059
2060                 if ( is_string( $query ) ) {
2061                         $query = wfCgiToArray( $query );
2062                 }
2063                 if ( $text == '' ) {
2064                         $text = htmlspecialchars( $nt->getPrefixedText() );
2065                 }
2066
2067                 $link = Linker::linkKnown( $nt, "$prefix$text$inside", array(), $query );
2068
2069                 return $this->armorLinks( $link ) . $trail;
2070         }
2071
2072         /**
2073          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2074          * going to go through further parsing steps before inline URL expansion.
2075          *
2076          * Not needed quite as much as it used to be since free links are a bit
2077          * more sensible these days. But bracketed links are still an issue.
2078          *
2079          * @param $text String: more-or-less HTML
2080          * @return String: less-or-more HTML with NOPARSE bits
2081          */
2082         function armorLinks( $text ) {
2083                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
2084                         "{$this->mUniqPrefix}NOPARSE$1", $text );
2085         }
2086
2087         /**
2088          * Return true if subpage links should be expanded on this page.
2089          * @return Boolean
2090          */
2091         function areSubpagesAllowed() {
2092                 # Some namespaces don't allow subpages
2093                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2094         }
2095
2096         /**
2097          * Handle link to subpage if necessary
2098          *
2099          * @param $target String: the source of the link
2100          * @param &$text String: the link text, modified as necessary
2101          * @return string the full name of the link
2102          * @private
2103          */
2104         function maybeDoSubpageLink( $target, &$text ) {
2105                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2106         }
2107
2108         /**#@+
2109          * Used by doBlockLevels()
2110          * @private
2111          *
2112          * @return string
2113          */
2114         function closeParagraph() {
2115                 $result = '';
2116                 if ( $this->mLastSection != '' ) {
2117                         $result = '</' . $this->mLastSection  . ">\n";
2118                 }
2119                 $this->mInPre = false;
2120                 $this->mLastSection = '';
2121                 return $result;
2122         }
2123
2124         /**
2125          * getCommon() returns the length of the longest common substring
2126          * of both arguments, starting at the beginning of both.
2127          * @private
2128          */
2129         function getCommon( $st1, $st2 ) {
2130                 $fl = strlen( $st1 );
2131                 $shorter = strlen( $st2 );
2132                 if ( $fl < $shorter ) {
2133                         $shorter = $fl;
2134                 }
2135
2136                 for ( $i = 0; $i < $shorter; ++$i ) {
2137                         if ( $st1[$i] != $st2[$i] ) {
2138                                 break;
2139                         }
2140                 }
2141                 return $i;
2142         }
2143
2144         /**
2145          * These next three functions open, continue, and close the list
2146          * element appropriate to the prefix character passed into them.
2147          * @private
2148          *
2149          * @return string
2150          */
2151         function openList( $char ) {
2152                 $result = $this->closeParagraph();
2153
2154                 if ( '*' === $char ) {
2155                         $result .= '<ul><li>';
2156                 } elseif ( '#' === $char ) {
2157                         $result .= '<ol><li>';
2158                 } elseif ( ':' === $char ) {
2159                         $result .= '<dl><dd>';
2160                 } elseif ( ';' === $char ) {
2161                         $result .= '<dl><dt>';
2162                         $this->mDTopen = true;
2163                 } else {
2164                         $result = '<!-- ERR 1 -->';
2165                 }
2166
2167                 return $result;
2168         }
2169
2170         /**
2171          * TODO: document
2172          * @param $char String
2173          * @private
2174          *
2175          * @return string
2176          */
2177         function nextItem( $char ) {
2178                 if ( '*' === $char || '#' === $char ) {
2179                         return '</li><li>';
2180                 } elseif ( ':' === $char || ';' === $char ) {
2181                         $close = '</dd>';
2182                         if ( $this->mDTopen ) {
2183                                 $close = '</dt>';
2184                         }
2185                         if ( ';' === $char ) {
2186                                 $this->mDTopen = true;
2187                                 return $close . '<dt>';
2188                         } else {
2189                                 $this->mDTopen = false;
2190                                 return $close . '<dd>';
2191                         }
2192                 }
2193                 return '<!-- ERR 2 -->';
2194         }
2195
2196         /**
2197          * TODO: document
2198          * @param $char String
2199          * @private
2200          *
2201          * @return string
2202          */
2203         function closeList( $char ) {
2204                 if ( '*' === $char ) {
2205                         $text = '</li></ul>';
2206                 } elseif ( '#' === $char ) {
2207                         $text = '</li></ol>';
2208                 } elseif ( ':' === $char ) {
2209                         if ( $this->mDTopen ) {
2210                                 $this->mDTopen = false;
2211                                 $text = '</dt></dl>';
2212                         } else {
2213                                 $text = '</dd></dl>';
2214                         }
2215                 } else {
2216                         return '<!-- ERR 3 -->';
2217                 }
2218                 return $text."\n";
2219         }
2220         /**#@-*/
2221
2222         /**
2223          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2224          *
2225          * @param $text String
2226          * @param $linestart Boolean: whether or not this is at the start of a line.
2227          * @private
2228          * @return string the lists rendered as HTML
2229          */
2230         function doBlockLevels( $text, $linestart ) {
2231                 wfProfileIn( __METHOD__ );
2232
2233                 # Parsing through the text line by line.  The main thing
2234                 # happening here is handling of block-level elements p, pre,
2235                 # and making lists from lines starting with * # : etc.
2236                 #
2237                 $textLines = StringUtils::explode( "\n", $text );
2238
2239                 $lastPrefix = $output = '';
2240                 $this->mDTopen = $inBlockElem = false;
2241                 $prefixLength = 0;
2242                 $paragraphStack = false;
2243
2244                 foreach ( $textLines as $oLine ) {
2245                         # Fix up $linestart
2246                         if ( !$linestart ) {
2247                                 $output .= $oLine;
2248                                 $linestart = true;
2249                                 continue;
2250                         }
2251                         # * = ul
2252                         # # = ol
2253                         # ; = dt
2254                         # : = dd
2255
2256                         $lastPrefixLength = strlen( $lastPrefix );
2257                         $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2258                         $preOpenMatch = preg_match( '/<pre/i', $oLine );
2259                         # If not in a <pre> element, scan for and figure out what prefixes are there.
2260                         if ( !$this->mInPre ) {
2261                                 # Multiple prefixes may abut each other for nested lists.
2262                                 $prefixLength = strspn( $oLine, '*#:;' );
2263                                 $prefix = substr( $oLine, 0, $prefixLength );
2264
2265                                 # eh?
2266                                 # ; and : are both from definition-lists, so they're equivalent
2267                                 #  for the purposes of determining whether or not we need to open/close
2268                                 #  elements.
2269                                 $prefix2 = str_replace( ';', ':', $prefix );
2270                                 $t = substr( $oLine, $prefixLength );
2271                                 $this->mInPre = (bool)$preOpenMatch;
2272                         } else {
2273                                 # Don't interpret any other prefixes in preformatted text
2274                                 $prefixLength = 0;
2275                                 $prefix = $prefix2 = '';
2276                                 $t = $oLine;
2277                         }
2278
2279                         # List generation
2280                         if ( $prefixLength && $lastPrefix === $prefix2 ) {
2281                                 # Same as the last item, so no need to deal with nesting or opening stuff
2282                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2283                                 $paragraphStack = false;
2284
2285                                 if ( substr( $prefix, -1 ) === ';') {
2286                                         # The one nasty exception: definition lists work like this:
2287                                         # ; title : definition text
2288                                         # So we check for : in the remainder text to split up the
2289                                         # title and definition, without b0rking links.
2290                                         $term = $t2 = '';
2291                                         if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2292                                                 $t = $t2;
2293                                                 $output .= $term . $this->nextItem( ':' );
2294                                         }
2295                                 }
2296                         } elseif ( $prefixLength || $lastPrefixLength ) {
2297                                 # We need to open or close prefixes, or both.
2298
2299                                 # Either open or close a level...
2300                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2301                                 $paragraphStack = false;
2302
2303                                 # Close all the prefixes which aren't shared.
2304                                 while ( $commonPrefixLength < $lastPrefixLength ) {
2305                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2306                                         --$lastPrefixLength;
2307                                 }
2308
2309                                 # Continue the current prefix if appropriate.
2310                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2311                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2312                                 }
2313
2314                                 # Open prefixes where appropriate.
2315                                 while ( $prefixLength > $commonPrefixLength ) {
2316                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2317                                         $output .= $this->openList( $char );
2318
2319                                         if ( ';' === $char ) {
2320                                                 # @todo FIXME: This is dupe of code above
2321                                                 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2322                                                         $t = $t2;
2323                                                         $output .= $term . $this->nextItem( ':' );
2324                                                 }
2325                                         }
2326                                         ++$commonPrefixLength;
2327                                 }
2328                                 $lastPrefix = $prefix2;
2329                         }
2330
2331                         # If we have no prefixes, go to paragraph mode.
2332                         if ( 0 == $prefixLength ) {
2333                                 wfProfileIn( __METHOD__."-paragraph" );
2334                                 # No prefix (not in list)--go to paragraph mode
2335                                 # XXX: use a stack for nestable elements like span, table and div
2336                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2337                                 $closematch = preg_match(
2338                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2339                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2340                                 if ( $openmatch or $closematch ) {
2341                                         $paragraphStack = false;
2342                                         # TODO bug 5718: paragraph closed
2343                                         $output .= $this->closeParagraph();
2344                                         if ( $preOpenMatch and !$preCloseMatch ) {
2345                                                 $this->mInPre = true;
2346                                         }
2347                                         $inBlockElem = !$closematch;
2348                                 } elseif ( !$inBlockElem && !$this->mInPre ) {
2349                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) ) {
2350                                                 # pre
2351                                                 if ( $this->mLastSection !== 'pre' ) {
2352                                                         $paragraphStack = false;
2353                                                         $output .= $this->closeParagraph().'<pre>';
2354                                                         $this->mLastSection = 'pre';
2355                                                 }
2356                                                 $t = substr( $t, 1 );
2357                                         } else {
2358                                                 # paragraph
2359                                                 if ( trim( $t ) === '' ) {
2360                                                         if ( $paragraphStack ) {
2361                                                                 $output .= $paragraphStack.'<br />';
2362                                                                 $paragraphStack = false;
2363                                                                 $this->mLastSection = 'p';
2364                                                         } else {
2365                                                                 if ( $this->mLastSection !== 'p' ) {
2366                                                                         $output .= $this->closeParagraph();
2367                                                                         $this->mLastSection = '';
2368                                                                         $paragraphStack = '<p>';
2369                                                                 } else {
2370                                                                         $paragraphStack = '</p><p>';
2371                                                                 }
2372                                                         }
2373                                                 } else {
2374                                                         if ( $paragraphStack ) {
2375                                                                 $output .= $paragraphStack;
2376                                                                 $paragraphStack = false;
2377                                                                 $this->mLastSection = 'p';
2378                                                         } elseif ( $this->mLastSection !== 'p' ) {
2379                                                                 $output .= $this->closeParagraph().'<p>';
2380                                                                 $this->mLastSection = 'p';
2381                                                         }
2382                                                 }
2383                                         }
2384                                 }
2385                                 wfProfileOut( __METHOD__."-paragraph" );
2386                         }
2387                         # somewhere above we forget to get out of pre block (bug 785)
2388                         if ( $preCloseMatch && $this->mInPre ) {
2389                                 $this->mInPre = false;
2390                         }
2391                         if ( $paragraphStack === false ) {
2392                                 $output .= $t."\n";
2393                         }
2394                 }
2395                 while ( $prefixLength ) {
2396                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2397                         --$prefixLength;
2398                 }
2399                 if ( $this->mLastSection != '' ) {
2400                         $output .= '</' . $this->mLastSection . '>';
2401                         $this->mLastSection = '';
2402                 }
2403
2404                 wfProfileOut( __METHOD__ );
2405                 return $output;
2406         }
2407
2408         /**
2409          * Split up a string on ':', ignoring any occurences inside tags
2410          * to prevent illegal overlapping.
2411          *
2412          * @param $str String: the string to split
2413          * @param &$before String: set to everything before the ':'
2414          * @param &$after String: set to everything after the ':'
2415          * return String: the position of the ':', or false if none found
2416          */
2417         function findColonNoLinks( $str, &$before, &$after ) {
2418                 wfProfileIn( __METHOD__ );
2419
2420                 $pos = strpos( $str, ':' );
2421                 if ( $pos === false ) {
2422                         # Nothing to find!
2423                         wfProfileOut( __METHOD__ );
2424                         return false;
2425                 }
2426
2427                 $lt = strpos( $str, '<' );
2428                 if ( $lt === false || $lt > $pos ) {
2429                         # Easy; no tag nesting to worry about
2430                         $before = substr( $str, 0, $pos );
2431                         $after = substr( $str, $pos+1 );
2432                         wfProfileOut( __METHOD__ );
2433                         return $pos;
2434                 }
2435
2436                 # Ugly state machine to walk through avoiding tags.
2437                 $state = self::COLON_STATE_TEXT;
2438                 $stack = 0;
2439                 $len = strlen( $str );
2440                 for( $i = 0; $i < $len; $i++ ) {
2441                         $c = $str[$i];
2442
2443                         switch( $state ) {
2444                         # (Using the number is a performance hack for common cases)
2445                         case 0: # self::COLON_STATE_TEXT:
2446                                 switch( $c ) {
2447                                 case "<":
2448                                         # Could be either a <start> tag or an </end> tag
2449                                         $state = self::COLON_STATE_TAGSTART;
2450                                         break;
2451                                 case ":":
2452                                         if ( $stack == 0 ) {
2453                                                 # We found it!
2454                                                 $before = substr( $str, 0, $i );
2455                                                 $after = substr( $str, $i + 1 );
2456                                                 wfProfileOut( __METHOD__ );
2457                                                 return $i;
2458                                         }
2459                                         # Embedded in a tag; don't break it.
2460                                         break;
2461                                 default:
2462                                         # Skip ahead looking for something interesting
2463                                         $colon = strpos( $str, ':', $i );
2464                                         if ( $colon === false ) {
2465                                                 # Nothing else interesting
2466                                                 wfProfileOut( __METHOD__ );
2467                                                 return false;
2468                                         }
2469                                         $lt = strpos( $str, '<', $i );
2470                                         if ( $stack === 0 ) {
2471                                                 if ( $lt === false || $colon < $lt ) {
2472                                                         # We found it!
2473                                                         $before = substr( $str, 0, $colon );
2474                                                         $after = substr( $str, $colon + 1 );
2475                                                         wfProfileOut( __METHOD__ );
2476                                                         return $i;
2477                                                 }
2478                                         }
2479                                         if ( $lt === false ) {
2480                                                 # Nothing else interesting to find; abort!
2481                                                 # We're nested, but there's no close tags left. Abort!
2482                                                 break 2;
2483                                         }
2484                                         # Skip ahead to next tag start
2485                                         $i = $lt;
2486                                         $state = self::COLON_STATE_TAGSTART;
2487                                 }
2488                                 break;
2489                         case 1: # self::COLON_STATE_TAG:
2490                                 # In a <tag>
2491                                 switch( $c ) {
2492                                 case ">":
2493                                         $stack++;
2494                                         $state = self::COLON_STATE_TEXT;
2495                                         break;
2496                                 case "/":
2497                                         # Slash may be followed by >?
2498                                         $state = self::COLON_STATE_TAGSLASH;
2499                                         break;
2500                                 default:
2501                                         # ignore
2502                                 }
2503                                 break;
2504                         case 2: # self::COLON_STATE_TAGSTART:
2505                                 switch( $c ) {
2506                                 case "/":
2507                                         $state = self::COLON_STATE_CLOSETAG;
2508                                         break;
2509                                 case "!":
2510                                         $state = self::COLON_STATE_COMMENT;
2511                                         break;
2512                                 case ">":
2513                                         # Illegal early close? This shouldn't happen D:
2514                                         $state = self::COLON_STATE_TEXT;
2515                                         break;
2516                                 default:
2517                                         $state = self::COLON_STATE_TAG;
2518                                 }
2519                                 break;
2520                         case 3: # self::COLON_STATE_CLOSETAG:
2521                                 # In a </tag>
2522                                 if ( $c === ">" ) {
2523                                         $stack--;
2524                                         if ( $stack < 0 ) {
2525                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2526                                                 wfProfileOut( __METHOD__ );
2527                                                 return false;
2528                                         }
2529                                         $state = self::COLON_STATE_TEXT;
2530                                 }
2531                                 break;
2532                         case self::COLON_STATE_TAGSLASH:
2533                                 if ( $c === ">" ) {
2534                                         # Yes, a self-closed tag <blah/>
2535                                         $state = self::COLON_STATE_TEXT;
2536                                 } else {
2537                                         # Probably we're jumping the gun, and this is an attribute
2538                                         $state = self::COLON_STATE_TAG;
2539                                 }
2540                                 break;
2541                         case 5: # self::COLON_STATE_COMMENT:
2542                                 if ( $c === "-" ) {
2543                                         $state = self::COLON_STATE_COMMENTDASH;
2544                                 }
2545                                 break;
2546                         case self::COLON_STATE_COMMENTDASH:
2547                                 if ( $c === "-" ) {
2548                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2549                                 } else {
2550                                         $state = self::COLON_STATE_COMMENT;
2551                                 }
2552                                 break;
2553                         case self::COLON_STATE_COMMENTDASHDASH:
2554                                 if ( $c === ">" ) {
2555                                         $state = self::COLON_STATE_TEXT;
2556                                 } else {
2557                                         $state = self::COLON_STATE_COMMENT;
2558                                 }
2559                                 break;
2560                         default:
2561                                 throw new MWException( "State machine error in " . __METHOD__ );
2562                         }
2563                 }
2564                 if ( $stack > 0 ) {
2565                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2566                         wfProfileOut( __METHOD__ );
2567                         return false;
2568                 }
2569                 wfProfileOut( __METHOD__ );
2570                 return false;
2571         }
2572
2573         /**
2574          * Return value of a magic variable (like PAGENAME)
2575          *
2576          * @private
2577          *
2578          * @param $index integer
2579          * @param $frame PPFrame
2580          */
2581         function getVariableValue( $index, $frame=false ) {
2582                 global $wgContLang, $wgSitename, $wgServer;
2583                 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2584
2585                 /**
2586                  * Some of these require message or data lookups and can be
2587                  * expensive to check many times.
2588                  */
2589                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2590                         if ( isset( $this->mVarCache[$index] ) ) {
2591                                 return $this->mVarCache[$index];
2592                         }
2593                 }
2594
2595                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2596                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2597
2598                 # Use the time zone
2599                 global $wgLocaltimezone;
2600                 if ( isset( $wgLocaltimezone ) ) {
2601                         $oldtz = date_default_timezone_get();
2602                         date_default_timezone_set( $wgLocaltimezone );
2603                 }
2604
2605                 $localTimestamp = date( 'YmdHis', $ts );
2606                 $localMonth = date( 'm', $ts );
2607                 $localMonth1 = date( 'n', $ts );
2608                 $localMonthName = date( 'n', $ts );
2609                 $localDay = date( 'j', $ts );
2610                 $localDay2 = date( 'd', $ts );
2611                 $localDayOfWeek = date( 'w', $ts );
2612                 $localWeek = date( 'W', $ts );
2613                 $localYear = date( 'Y', $ts );
2614                 $localHour = date( 'H', $ts );
2615                 if ( isset( $wgLocaltimezone ) ) {
2616                         date_default_timezone_set( $oldtz );
2617                 }
2618
2619                 switch ( $index ) {
2620                         case 'currentmonth':
2621                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2622                                 break;
2623                         case 'currentmonth1':
2624                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2625                                 break;
2626                         case 'currentmonthname':
2627                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2628                                 break;
2629                         case 'currentmonthnamegen':
2630                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2631                                 break;
2632                         case 'currentmonthabbrev':
2633                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2634                                 break;
2635                         case 'currentday':
2636                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2637                                 break;
2638                         case 'currentday2':
2639                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2640                                 break;
2641                         case 'localmonth':
2642                                 $value = $wgContLang->formatNum( $localMonth );
2643                                 break;
2644                         case 'localmonth1':
2645                                 $value = $wgContLang->formatNum( $localMonth1 );
2646                                 break;
2647                         case 'localmonthname':
2648                                 $value = $wgContLang->getMonthName( $localMonthName );
2649                                 break;
2650                         case 'localmonthnamegen':
2651                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2652                                 break;
2653                         case 'localmonthabbrev':
2654                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2655                                 break;
2656                         case 'localday':
2657                                 $value = $wgContLang->formatNum( $localDay );
2658                                 break;
2659                         case 'localday2':
2660                                 $value = $wgContLang->formatNum( $localDay2 );
2661                                 break;
2662                         case 'pagename':
2663                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2664                                 break;
2665                         case 'pagenamee':
2666                                 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2667                                 break;
2668                         case 'fullpagename':
2669                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2670                                 break;
2671                         case 'fullpagenamee':
2672                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2673                                 break;
2674                         case 'subpagename':
2675                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2676                                 break;
2677                         case 'subpagenamee':
2678                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2679                                 break;
2680                         case 'basepagename':
2681                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2682                                 break;
2683                         case 'basepagenamee':
2684                                 $value = wfEscapeWikiText( wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) ) );
2685                                 break;
2686                         case 'talkpagename':
2687                                 if ( $this->mTitle->canTalk() ) {
2688                                         $talkPage = $this->mTitle->getTalkPage();
2689                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2690                                 } else {
2691                                         $value = '';
2692                                 }
2693                                 break;
2694                         case 'talkpagenamee':
2695                                 if ( $this->mTitle->canTalk() ) {
2696                                         $talkPage = $this->mTitle->getTalkPage();
2697                                         $value = wfEscapeWikiText( $talkPage->getPrefixedUrl() );
2698                                 } else {
2699                                         $value = '';
2700                                 }
2701                                 break;
2702                         case 'subjectpagename':
2703                                 $subjPage = $this->mTitle->getSubjectPage();
2704                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2705                                 break;
2706                         case 'subjectpagenamee':
2707                                 $subjPage = $this->mTitle->getSubjectPage();
2708                                 $value = wfEscapeWikiText( $subjPage->getPrefixedUrl() );
2709                                 break;
2710                         case 'revisionid':
2711                                 # Let the edit saving system know we should parse the page
2712                                 # *after* a revision ID has been assigned.
2713                                 $this->mOutput->setFlag( 'vary-revision' );
2714                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2715                                 $value = $this->mRevisionId;
2716                                 break;
2717                         case 'revisionday':
2718                                 # Let the edit saving system know we should parse the page
2719                                 # *after* a revision ID has been assigned. This is for null edits.
2720                                 $this->mOutput->setFlag( 'vary-revision' );
2721                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2722                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2723                                 break;
2724                         case 'revisionday2':
2725                                 # Let the edit saving system know we should parse the page
2726                                 # *after* a revision ID has been assigned. This is for null edits.
2727                                 $this->mOutput->setFlag( 'vary-revision' );
2728                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2729                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2730                                 break;
2731                         case 'revisionmonth':
2732                                 # Let the edit saving system know we should parse the page
2733                                 # *after* a revision ID has been assigned. This is for null edits.
2734                                 $this->mOutput->setFlag( 'vary-revision' );
2735                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2736                                 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2737                                 break;
2738                         case 'revisionmonth1':
2739                                 # Let the edit saving system know we should parse the page
2740                                 # *after* a revision ID has been assigned. This is for null edits.
2741                                 $this->mOutput->setFlag( 'vary-revision' );
2742                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2743                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2744                                 break;
2745                         case 'revisionyear':
2746                                 # Let the edit saving system know we should parse the page
2747                                 # *after* a revision ID has been assigned. This is for null edits.
2748                                 $this->mOutput->setFlag( 'vary-revision' );
2749                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2750                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2751                                 break;
2752                         case 'revisiontimestamp':
2753                                 # Let the edit saving system know we should parse the page
2754                                 # *after* a revision ID has been assigned. This is for null edits.
2755                                 $this->mOutput->setFlag( 'vary-revision' );
2756                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2757                                 $value = $this->getRevisionTimestamp();
2758                                 break;
2759                         case 'revisionuser':
2760                                 # Let the edit saving system know we should parse the page
2761                                 # *after* a revision ID has been assigned. This is for null edits.
2762                                 $this->mOutput->setFlag( 'vary-revision' );
2763                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2764                                 $value = $this->getRevisionUser();
2765                                 break;
2766                         case 'namespace':
2767                                 $value = str_replace( '_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2768                                 break;
2769                         case 'namespacee':
2770                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2771                                 break;
2772                         case 'talkspace':
2773                                 $value = $this->mTitle->canTalk() ? str_replace( '_',' ',$this->mTitle->getTalkNsText() ) : '';
2774                                 break;
2775                         case 'talkspacee':
2776                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2777                                 break;
2778                         case 'subjectspace':
2779                                 $value = $this->mTitle->getSubjectNsText();
2780                                 break;
2781                         case 'subjectspacee':
2782                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2783                                 break;
2784                         case 'currentdayname':
2785                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2786                                 break;
2787                         case 'currentyear':
2788                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2789                                 break;
2790                         case 'currenttime':
2791                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2792                                 break;
2793                         case 'currenthour':
2794                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2795                                 break;
2796                         case 'currentweek':
2797                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2798                                 # int to remove the padding
2799                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2800                                 break;
2801                         case 'currentdow':
2802                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2803                                 break;
2804                         case 'localdayname':
2805                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2806                                 break;
2807                         case 'localyear':
2808                                 $value = $wgContLang->formatNum( $localYear, true );
2809                                 break;
2810                         case 'localtime':
2811                                 $value = $wgContLang->time( $localTimestamp, false, false );
2812                                 break;
2813                         case 'localhour':
2814                                 $value = $wgContLang->formatNum( $localHour, true );
2815                                 break;
2816                         case 'localweek':
2817                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2818                                 # int to remove the padding
2819                                 $value = $wgContLang->formatNum( (int)$localWeek );
2820                                 break;
2821                         case 'localdow':
2822                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2823                                 break;
2824                         case 'numberofarticles':
2825                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2826                                 break;
2827                         case 'numberoffiles':
2828                                 $value = $wgContLang->formatNum( SiteStats::images() );
2829                                 break;
2830                         case 'numberofusers':
2831                                 $value = $wgContLang->formatNum( SiteStats::users() );
2832                                 break;
2833                         case 'numberofactiveusers':
2834                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2835                                 break;
2836                         case 'numberofpages':
2837                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2838                                 break;
2839                         case 'numberofadmins':
2840                                 $value = $wgContLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2841                                 break;
2842                         case 'numberofedits':
2843                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2844                                 break;
2845                         case 'numberofviews':
2846                                 $value = $wgContLang->formatNum( SiteStats::views() );
2847                                 break;
2848                         case 'currenttimestamp':
2849                                 $value = wfTimestamp( TS_MW, $ts );
2850                                 break;
2851                         case 'localtimestamp':
2852                                 $value = $localTimestamp;
2853                                 break;
2854                         case 'currentversion':
2855                                 $value = SpecialVersion::getVersion();
2856                                 break;
2857                         case 'articlepath':
2858                                 return $wgArticlePath;
2859                         case 'sitename':
2860                                 return $wgSitename;
2861                         case 'server':
2862                                 return $wgServer;
2863                         case 'servername':
2864                                 wfSuppressWarnings(); # May give an E_WARNING in PHP < 5.3.3
2865                                 $serverName = parse_url( $wgServer, PHP_URL_HOST );
2866                                 wfRestoreWarnings();
2867                                 return $serverName ? $serverName : $wgServer;
2868                         case 'scriptpath':
2869                                 return $wgScriptPath;
2870                         case 'stylepath':
2871                                 return $wgStylePath;
2872                         case 'directionmark':
2873                                 return $wgContLang->getDirMark();
2874                         case 'contentlanguage':
2875                                 global $wgLanguageCode;
2876                                 return $wgLanguageCode;
2877                         default:
2878                                 $ret = null;
2879                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
2880                                         return $ret;
2881                                 } else {
2882                                         return null;
2883                                 }
2884                 }
2885
2886                 if ( $index )
2887                         $this->mVarCache[$index] = $value;
2888
2889                 return $value;
2890         }
2891
2892         /**
2893          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2894          *
2895          * @private
2896          */
2897         function initialiseVariables() {
2898                 wfProfileIn( __METHOD__ );
2899                 $variableIDs = MagicWord::getVariableIDs();
2900                 $substIDs = MagicWord::getSubstIDs();
2901
2902                 $this->mVariables = new MagicWordArray( $variableIDs );
2903                 $this->mSubstWords = new MagicWordArray( $substIDs );
2904                 wfProfileOut( __METHOD__ );
2905         }
2906
2907         /**
2908          * Preprocess some wikitext and return the document tree.
2909          * This is the ghost of replace_variables().
2910          *
2911          * @param $text String: The text to parse
2912          * @param $flags Integer: bitwise combination of:
2913          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2914          *                                     included. Default is to assume a direct page view.
2915          *
2916          * The generated DOM tree must depend only on the input text and the flags.
2917          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2918          *
2919          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2920          * change in the DOM tree for a given text, must be passed through the section identifier
2921          * in the section edit link and thus back to extractSections().
2922          *
2923          * The output of this function is currently only cached in process memory, but a persistent
2924          * cache may be implemented at a later date which takes further advantage of these strict
2925          * dependency requirements.
2926          *
2927          * @private
2928          *
2929          * @return PPNode
2930          */
2931         function preprocessToDom( $text, $flags = 0 ) {
2932                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2933                 return $dom;
2934         }
2935
2936         /**
2937          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2938          *
2939          * @return array
2940          */
2941         public static function splitWhitespace( $s ) {
2942                 $ltrimmed = ltrim( $s );
2943                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2944                 $trimmed = rtrim( $ltrimmed );
2945                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2946                 if ( $diff > 0 ) {
2947                         $w2 = substr( $ltrimmed, -$diff );
2948                 } else {
2949                         $w2 = '';
2950                 }
2951                 return array( $w1, $trimmed, $w2 );
2952         }
2953
2954         /**
2955          * Replace magic variables, templates, and template arguments
2956          * with the appropriate text. Templates are substituted recursively,
2957          * taking care to avoid infinite loops.
2958          *
2959          * Note that the substitution depends on value of $mOutputType:
2960          *  self::OT_WIKI: only {{subst:}} templates
2961          *  self::OT_PREPROCESS: templates but not extension tags
2962          *  self::OT_HTML: all templates and extension tags
2963          *
2964          * @param $text String: the text to transform
2965          * @param $frame PPFrame Object describing the arguments passed to the template.
2966          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2967          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2968          * @param $argsOnly Boolean: only do argument (triple-brace) expansion, not double-brace expansion
2969          * @private
2970          *
2971          * @return string
2972          */
2973         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2974                 # Is there any text? Also, Prevent too big inclusions!
2975                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2976                         return $text;
2977                 }
2978                 wfProfileIn( __METHOD__ );
2979
2980                 if ( $frame === false ) {
2981                         $frame = $this->getPreprocessor()->newFrame();
2982                 } elseif ( !( $frame instanceof PPFrame ) ) {
2983                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2984                         $frame = $this->getPreprocessor()->newCustomFrame( $frame );
2985                 }
2986
2987                 $dom = $this->preprocessToDom( $text );
2988                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2989                 $text = $frame->expand( $dom, $flags );
2990
2991                 wfProfileOut( __METHOD__ );
2992                 return $text;
2993         }
2994
2995         /**
2996          * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2997          *
2998          * @return array
2999          */
3000         static function createAssocArgs( $args ) {
3001                 $assocArgs = array();
3002                 $index = 1;
3003                 foreach ( $args as $arg ) {
3004                         $eqpos = strpos( $arg, '=' );
3005                         if ( $eqpos === false ) {
3006                                 $assocArgs[$index++] = $arg;
3007                         } else {
3008                                 $name = trim( substr( $arg, 0, $eqpos ) );
3009                                 $value = trim( substr( $arg, $eqpos+1 ) );
3010                                 if ( $value === false ) {
3011                                         $value = '';
3012                                 }
3013                                 if ( $name !== false ) {
3014                                         $assocArgs[$name] = $value;
3015                                 }
3016                         }
3017                 }
3018
3019                 return $assocArgs;
3020         }
3021
3022         /**
3023          * Warn the user when a parser limitation is reached
3024          * Will warn at most once the user per limitation type
3025          *
3026          * @param $limitationType String: should be one of:
3027          *   'expensive-parserfunction' (corresponding messages:
3028          *       'expensive-parserfunction-warning',
3029          *       'expensive-parserfunction-category')
3030          *   'post-expand-template-argument' (corresponding messages:
3031          *       'post-expand-template-argument-warning',
3032          *       'post-expand-template-argument-category')
3033          *   'post-expand-template-inclusion' (corresponding messages:
3034          *       'post-expand-template-inclusion-warning',
3035          *       'post-expand-template-inclusion-category')
3036          * @param $current Current value
3037          * @param $max Maximum allowed, when an explicit limit has been
3038          *       exceeded, provide the values (optional)
3039          */
3040         function limitationWarn( $limitationType, $current=null, $max=null) {
3041                 # does no harm if $current and $max are present but are unnecessary for the message
3042                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
3043                 $this->mOutput->addWarning( $warning );
3044                 $this->addTrackingCategory( "$limitationType-category" );
3045         }
3046
3047         /**
3048          * Return the text of a template, after recursively
3049          * replacing any variables or templates within the template.
3050          *
3051          * @param $piece Array: the parts of the template
3052          *  $piece['title']: the title, i.e. the part before the |
3053          *  $piece['parts']: the parameter array
3054          *  $piece['lineStart']: whether the brace was at the start of a line
3055          * @param $frame PPFrame The current frame, contains template arguments
3056          * @return String: the text of the template
3057          * @private
3058          */
3059         function braceSubstitution( $piece, $frame ) {
3060                 global $wgContLang, $wgNonincludableNamespaces;
3061                 wfProfileIn( __METHOD__ );
3062                 wfProfileIn( __METHOD__.'-setup' );
3063
3064                 # Flags
3065                 $found = false;             # $text has been filled
3066                 $nowiki = false;            # wiki markup in $text should be escaped
3067                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
3068                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
3069                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
3070                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
3071
3072                 # Title object, where $text came from
3073                 $title = null;
3074
3075                 # $part1 is the bit before the first |, and must contain only title characters.
3076                 # Various prefixes will be stripped from it later.
3077                 $titleWithSpaces = $frame->expand( $piece['title'] );
3078                 $part1 = trim( $titleWithSpaces );
3079                 $titleText = false;
3080
3081                 # Original title text preserved for various purposes
3082                 $originalTitle = $part1;
3083
3084                 # $args is a list of argument nodes, starting from index 0, not including $part1
3085                 # @todo FIXME: If piece['parts'] is null then the call to getLength() below won't work b/c this $args isn't an object
3086                 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
3087                 wfProfileOut( __METHOD__.'-setup' );
3088                 wfProfileIn( __METHOD__."-title-$originalTitle" );
3089
3090                 # SUBST
3091                 wfProfileIn( __METHOD__.'-modifiers' );
3092                 if ( !$found ) {
3093
3094                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3095
3096                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
3097                         # Decide whether to expand template or keep wikitext as-is.
3098                         if ( $this->ot['wiki'] ) {
3099                                 if ( $substMatch === false ) {
3100                                         $literal = true;  # literal when in PST with no prefix
3101                                 } else {
3102                                         $literal = false; # expand when in PST with subst: or safesubst:
3103                                 }
3104                         } else {
3105                                 if ( $substMatch == 'subst' ) {
3106                                         $literal = true;  # literal when not in PST with plain subst:
3107                                 } else {
3108                                         $literal = false; # expand when not in PST with safesubst: or no prefix
3109                                 }
3110                         }
3111                         if ( $literal ) {
3112                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3113                                 $isLocalObj = true;
3114                                 $found = true;
3115                         }
3116                 }
3117
3118                 # Variables
3119                 if ( !$found && $args->getLength() == 0 ) {
3120                         $id = $this->mVariables->matchStartToEnd( $part1 );
3121                         if ( $id !== false ) {
3122                                 $text = $this->getVariableValue( $id, $frame );
3123                                 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3124                                         $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3125                                 }
3126                                 $found = true;
3127                         }
3128                 }
3129
3130                 # MSG, MSGNW and RAW
3131                 if ( !$found ) {
3132                         # Check for MSGNW:
3133                         $mwMsgnw = MagicWord::get( 'msgnw' );
3134                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3135                                 $nowiki = true;
3136                         } else {
3137                                 # Remove obsolete MSG:
3138                                 $mwMsg = MagicWord::get( 'msg' );
3139                                 $mwMsg->matchStartAndRemove( $part1 );
3140                         }
3141
3142                         # Check for RAW:
3143                         $mwRaw = MagicWord::get( 'raw' );
3144                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3145                                 $forceRawInterwiki = true;
3146                         }
3147                 }
3148                 wfProfileOut( __METHOD__.'-modifiers' );
3149
3150                 # Parser functions
3151                 if ( !$found ) {
3152                         wfProfileIn( __METHOD__ . '-pfunc' );
3153
3154                         $colonPos = strpos( $part1, ':' );
3155                         if ( $colonPos !== false ) {
3156                                 # Case sensitive functions
3157                                 $function = substr( $part1, 0, $colonPos );
3158                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3159                                         $function = $this->mFunctionSynonyms[1][$function];
3160                                 } else {
3161                                         # Case insensitive functions
3162                                         $function = $wgContLang->lc( $function );
3163                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3164                                                 $function = $this->mFunctionSynonyms[0][$function];
3165                                         } else {
3166                                                 $function = false;
3167                                         }
3168                                 }
3169                                 if ( $function ) {
3170                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
3171                                         $initialArgs = array( &$this );
3172                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3173                                         if ( $flags & SFH_OBJECT_ARGS ) {
3174                                                 # Add a frame parameter, and pass the arguments as an array
3175                                                 $allArgs = $initialArgs;
3176                                                 $allArgs[] = $frame;
3177                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3178                                                         $funcArgs[] = $args->item( $i );
3179                                                 }
3180                                                 $allArgs[] = $funcArgs;
3181                                         } else {
3182                                                 # Convert arguments to plain text
3183                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3184                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
3185                                                 }
3186                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
3187                                         }
3188
3189                                         # Workaround for PHP bug 35229 and similar
3190                                         if ( !is_callable( $callback ) ) {
3191                                                 wfProfileOut( __METHOD__ . '-pfunc' );
3192                                                 wfProfileOut( __METHOD__ );
3193                                                 throw new MWException( "Tag hook for $function is not callable\n" );
3194                                         }
3195                                         $result = call_user_func_array( $callback, $allArgs );
3196                                         $found = true;
3197                                         $noparse = true;
3198                                         $preprocessFlags = 0;
3199
3200                                         if ( is_array( $result ) ) {
3201                                                 if ( isset( $result[0] ) ) {
3202                                                         $text = $result[0];
3203                                                         unset( $result[0] );
3204                                                 }
3205
3206                                                 # Extract flags into the local scope
3207                                                 # This allows callers to set flags such as nowiki, found, etc.
3208                                                 extract( $result );
3209                                         } else {
3210                                                 $text = $result;
3211                                         }
3212                                         if ( !$noparse ) {
3213                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
3214                                                 $isChildObj = true;
3215                                         }
3216                                 }
3217                         }
3218                         wfProfileOut( __METHOD__ . '-pfunc' );
3219                 }
3220
3221                 # Finish mangling title and then check for loops.
3222                 # Set $title to a Title object and $titleText to the PDBK
3223                 if ( !$found ) {
3224                         $ns = NS_TEMPLATE;
3225                         # Split the title into page and subpage
3226                         $subpage = '';
3227                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3228                         if ( $subpage !== '' ) {
3229                                 $ns = $this->mTitle->getNamespace();
3230                         }
3231                         $title = Title::newFromText( $part1, $ns );
3232                         if ( $title ) {
3233                                 $titleText = $title->getPrefixedText();
3234                                 # Check for language variants if the template is not found
3235                                 if ( $wgContLang->hasVariants() && $title->getArticleID() == 0 ) {
3236                                         $wgContLang->findVariantLink( $part1, $title, true );
3237                                 }
3238                                 # Do recursion depth check
3239                                 $limit = $this->mOptions->getMaxTemplateDepth();
3240                                 if ( $frame->depth >= $limit ) {
3241                                         $found = true;
3242                                         $text = '<span class="error">'
3243                                                 . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit )
3244                                                 . '</span>';
3245                                 }
3246                         }
3247                 }
3248
3249                 # Load from database
3250                 if ( !$found && $title ) {
3251                         wfProfileIn( __METHOD__ . '-loadtpl' );
3252                         if ( !$title->isExternal() ) {
3253                                 if ( $title->getNamespace() == NS_SPECIAL
3254                                         && $this->mOptions->getAllowSpecialInclusion()
3255                                         && $this->ot['html'] )
3256                                 {
3257                                         $text = SpecialPageFactory::capturePath( $title );
3258                                         if ( is_string( $text ) ) {
3259                                                 $found = true;
3260                                                 $isHTML = true;
3261                                                 $this->disableCache();
3262                                         }
3263                                 } elseif ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3264                                         $found = false; # access denied
3265                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3266                                 } else {
3267                                         list( $text, $title ) = $this->getTemplateDom( $title );
3268                                         if ( $text !== false ) {
3269                                                 $found = true;
3270                                                 $isChildObj = true;
3271                                         }
3272                                 }
3273
3274                                 # If the title is valid but undisplayable, make a link to it
3275                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3276                                         $text = "[[:$titleText]]";
3277                                         $found = true;
3278                                 }
3279                         } elseif ( $title->isTrans() ) {
3280                                 # Interwiki transclusion
3281                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3282                                         $text = $this->interwikiTransclude( $title, 'render' );
3283                                         $isHTML = true;
3284                                 } else {
3285                                         $text = $this->interwikiTransclude( $title, 'raw' );
3286                                         # Preprocess it like a template
3287                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3288                                         $isChildObj = true;
3289                                 }
3290                                 $found = true;
3291                         }
3292
3293                         # Do infinite loop check
3294                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3295                         if ( !$frame->loopCheck( $title ) ) {
3296                                 $found = true;
3297                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3298                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3299                         }
3300                         wfProfileOut( __METHOD__ . '-loadtpl' );
3301                 }
3302
3303                 # If we haven't found text to substitute by now, we're done
3304                 # Recover the source wikitext and return it
3305                 if ( !$found ) {
3306                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3307                         wfProfileOut( __METHOD__."-title-$originalTitle" );
3308                         wfProfileOut( __METHOD__ );
3309                         return array( 'object' => $text );
3310                 }
3311
3312                 # Expand DOM-style return values in a child frame
3313                 if ( $isChildObj ) {
3314                         # Clean up argument array
3315                         $newFrame = $frame->newChild( $args, $title );
3316
3317                         if ( $nowiki ) {
3318                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3319                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3320                                 # Expansion is eligible for the empty-frame cache
3321                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3322                                         $text = $this->mTplExpandCache[$titleText];
3323                                 } else {
3324                                         $text = $newFrame->expand( $text );
3325                                         $this->mTplExpandCache[$titleText] = $text;
3326                                 }
3327                         } else {
3328                                 # Uncached expansion
3329                                 $text = $newFrame->expand( $text );
3330                         }
3331                 }
3332                 if ( $isLocalObj && $nowiki ) {
3333                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3334                         $isLocalObj = false;
3335                 }
3336
3337                 # Replace raw HTML by a placeholder
3338                 # Add a blank line preceding, to prevent it from mucking up
3339                 # immediately preceding headings
3340                 if ( $isHTML ) {
3341                         $text = "\n\n" . $this->insertStripItem( $text );
3342                 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3343                         # Escape nowiki-style return values
3344                         $text = wfEscapeWikiText( $text );
3345                 } elseif ( is_string( $text )
3346                         && !$piece['lineStart']
3347                         && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3348                 {
3349                         # Bug 529: if the template begins with a table or block-level
3350                         # element, it should be treated as beginning a new line.
3351                         # This behaviour is somewhat controversial.
3352                         $text = "\n" . $text;
3353                 }
3354
3355                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3356                         # Error, oversize inclusion
3357                         if ( $titleText !== false ) {
3358                                 # Make a working, properly escaped link if possible (bug 23588)
3359                                 $text = "[[:$titleText]]";
3360                         } else {
3361                                 # This will probably not be a working link, but at least it may
3362                                 # provide some hint of where the problem is
3363                                 preg_replace( '/^:/', '', $originalTitle );
3364                                 $text = "[[:$originalTitle]]";
3365                         }
3366                         $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3367                         $this->limitationWarn( 'post-expand-template-inclusion' );
3368                 }
3369
3370                 if ( $isLocalObj ) {
3371                         $ret = array( 'object' => $text );
3372                 } else {
3373                         $ret = array( 'text' => $text );
3374                 }
3375
3376                 wfProfileOut( __METHOD__."-title-$originalTitle" );
3377                 wfProfileOut( __METHOD__ );
3378                 return $ret;
3379         }
3380
3381         /**
3382          * Get the semi-parsed DOM representation of a template with a given title,
3383          * and its redirect destination title. Cached.
3384          *
3385          * @return array
3386          */
3387         function getTemplateDom( $title ) {
3388                 $cacheTitle = $title;
3389                 $titleText = $title->getPrefixedDBkey();
3390
3391                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3392                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3393                         $title = Title::makeTitle( $ns, $dbk );
3394                         $titleText = $title->getPrefixedDBkey();
3395                 }
3396                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3397                         return array( $this->mTplDomCache[$titleText], $title );
3398                 }
3399
3400                 # Cache miss, go to the database
3401                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3402
3403                 if ( $text === false ) {
3404                         $this->mTplDomCache[$titleText] = false;
3405                         return array( false, $title );
3406                 }
3407
3408                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3409                 $this->mTplDomCache[ $titleText ] = $dom;
3410
3411                 if ( !$title->equals( $cacheTitle ) ) {
3412                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3413                                 array( $title->getNamespace(), $cdb = $title->getDBkey() );
3414                 }
3415
3416                 return array( $dom, $title );
3417         }
3418
3419         /**
3420          * Fetch the unparsed text of a template and register a reference to it.
3421          * @param Title $title
3422          * @return Array ( string or false, Title )
3423          */
3424         function fetchTemplateAndTitle( $title ) {
3425                 $templateCb = $this->mOptions->getTemplateCallback(); # Defaults to Parser::statelessFetchTemplate()
3426                 $stuff = call_user_func( $templateCb, $title, $this );
3427                 $text = $stuff['text'];
3428                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3429                 if ( isset( $stuff['deps'] ) ) {
3430                         foreach ( $stuff['deps'] as $dep ) {
3431                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3432                         }
3433                 }
3434                 return array( $text, $finalTitle );
3435         }
3436
3437         /**
3438          * Fetch the unparsed text of a template and register a reference to it.
3439          * @param Title $title
3440          * @return mixed string or false
3441          */
3442         function fetchTemplate( $title ) {
3443                 $rv = $this->fetchTemplateAndTitle( $title );
3444                 return $rv[0];
3445         }
3446
3447         /**
3448          * Static function to get a template
3449          * Can be overridden via ParserOptions::setTemplateCallback().
3450          *
3451          * @return array
3452          */
3453         static function statelessFetchTemplate( $title, $parser = false ) {
3454                 $text = $skip = false;
3455                 $finalTitle = $title;
3456                 $deps = array();
3457
3458                 # Loop to fetch the article, with up to 1 redirect
3459                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3460                         # Give extensions a chance to select the revision instead
3461                         $id = false; # Assume current
3462                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle',
3463                                 array( $parser, $title, &$skip, &$id ) );
3464
3465                         if ( $skip ) {
3466                                 $text = false;
3467                                 $deps[] = array(
3468                                         'title'         => $title,
3469                                         'page_id'       => $title->getArticleID(),
3470                                         'rev_id'        => null
3471                                 );
3472                                 break;
3473                         }
3474                         # Get the revision
3475                         $rev = $id
3476                                 ? Revision::newFromId( $id )
3477                                 : Revision::newFromTitle( $title );
3478                         $rev_id = $rev ? $rev->getId() : 0;
3479                         # If there is no current revision, there is no page
3480                         if ( $id === false && !$rev ) {
3481                                 $linkCache = LinkCache::singleton();
3482                                 $linkCache->addBadLinkObj( $title );
3483                         }
3484
3485                         $deps[] = array(
3486                                 'title'         => $title,
3487                                 'page_id'       => $title->getArticleID(),
3488                                 'rev_id'        => $rev_id );
3489                         if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3490                                 # We fetched a rev from a different title; register it too...
3491                                 $deps[] = array(
3492                                         'title'         => $rev->getTitle(),
3493                                         'page_id'       => $rev->getPage(),
3494                                         'rev_id'        => $rev_id );
3495                         }
3496
3497                         if ( $rev ) {
3498                                 $text = $rev->getText();
3499                         } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3500                                 global $wgContLang;
3501                                 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3502                                 if ( !$message->exists() ) {
3503                                         $text = false;
3504                                         break;
3505                                 }
3506                                 $text = $message->plain();
3507                         } else {
3508                                 break;
3509                         }
3510                         if ( $text === false ) {
3511                                 break;
3512                         }
3513                         # Redirect?
3514                         $finalTitle = $title;
3515                         $title = Title::newFromRedirect( $text );
3516                 }
3517                 return array(
3518                         'text' => $text,
3519                         'finalTitle' => $finalTitle,
3520                         'deps' => $deps );
3521         }
3522
3523         /**
3524          * Fetch a file and its title and register a reference to it.
3525          * @param Title $title
3526          * @param string $time MW timestamp
3527          * @param string $sha1 base 36 SHA-1
3528          * @return mixed File or false
3529          */
3530         function fetchFile( $title, $time = false, $sha1 = false ) {
3531                 $res = $this->fetchFileAndTitle( $title, $time, $sha1 );
3532                 return $res[0];
3533         }
3534
3535         /**
3536          * Fetch a file and its title and register a reference to it.
3537          * @param Title $title
3538          * @param string $time MW timestamp
3539          * @param string $sha1 base 36 SHA-1
3540          * @return Array ( File or false, Title of file )
3541          */
3542         function fetchFileAndTitle( $title, $time = false, $sha1 = false ) {
3543                 if ( $time === '0' ) {
3544                         $file = false; // broken thumbnail forced by hook
3545                 } elseif ( $sha1 ) { // get by (sha1,timestamp)
3546                         $file = RepoGroup::singleton()->findFileFromKey( $sha1, array( 'time' => $time ) );
3547                 } else { // get by (name,timestamp)
3548                         $file = wfFindFile( $title, array( 'time' => $time ) );
3549                 }
3550                 $time = $file ? $file->getTimestamp() : false;
3551                 $sha1 = $file ? $file->getSha1() : false;
3552                 # Register the file as a dependency...
3553                 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3554                 if ( $file && !$title->equals( $file->getTitle() ) ) {
3555                         # Update fetched file title
3556                         $title = $file->getTitle();
3557                 }
3558                 return array( $file, $title );
3559         }
3560
3561         /**
3562          * Transclude an interwiki link.
3563          *
3564          * @param $title Title
3565          * @param $action
3566          *
3567          * @return string
3568          */
3569         function interwikiTransclude( $title, $action ) {
3570                 global $wgEnableScaryTranscluding;
3571
3572                 if ( !$wgEnableScaryTranscluding ) {
3573                         return wfMsgForContent('scarytranscludedisabled');
3574                 }
3575
3576                 $url = $title->getFullUrl( "action=$action" );
3577
3578                 if ( strlen( $url ) > 255 ) {
3579                         return wfMsgForContent( 'scarytranscludetoolong' );
3580                 }
3581                 return $this->fetchScaryTemplateMaybeFromCache( $url );
3582         }
3583
3584         /**
3585          * @param $url string
3586          * @return Mixed|String
3587          */
3588         function fetchScaryTemplateMaybeFromCache( $url ) {
3589                 global $wgTranscludeCacheExpiry;
3590                 $dbr = wfGetDB( DB_SLAVE );
3591                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3592                 $obj = $dbr->selectRow( 'transcache', array('tc_time', 'tc_contents' ),
3593                                 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3594                 if ( $obj ) {
3595                         return $obj->tc_contents;
3596                 }
3597
3598                 $text = Http::get( $url );
3599                 if ( !$text ) {
3600                         return wfMsgForContent( 'scarytranscludefailed', $url );
3601                 }
3602
3603                 $dbw = wfGetDB( DB_MASTER );
3604                 $dbw->replace( 'transcache', array('tc_url'), array(
3605                         'tc_url' => $url,
3606                         'tc_time' => $dbw->timestamp( time() ),
3607                         'tc_contents' => $text)
3608                 );
3609                 return $text;
3610         }
3611
3612         /**
3613          * Triple brace replacement -- used for template arguments
3614          * @private
3615          *
3616          * @param $peice array
3617          * @param $frame PPFrame
3618          *
3619          * @return array
3620          */
3621         function argSubstitution( $piece, $frame ) {
3622                 wfProfileIn( __METHOD__ );
3623
3624                 $error = false;
3625                 $parts = $piece['parts'];
3626                 $nameWithSpaces = $frame->expand( $piece['title'] );
3627                 $argName = trim( $nameWithSpaces );
3628                 $object = false;
3629                 $text = $frame->getArgument( $argName );
3630                 if (  $text === false && $parts->getLength() > 0
3631                   && (
3632                         $this->ot['html']
3633                         || $this->ot['pre']
3634                         || ( $this->ot['wiki'] && $frame->isTemplate() )
3635                   )
3636                 ) {
3637                         # No match in frame, use the supplied default
3638                         $object = $parts->item( 0 )->getChildren();
3639                 }
3640                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3641                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3642                         $this->limitationWarn( 'post-expand-template-argument' );
3643                 }
3644
3645                 if ( $text === false && $object === false ) {
3646                         # No match anywhere
3647                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3648                 }
3649                 if ( $error !== false ) {
3650                         $text .= $error;
3651                 }
3652                 if ( $object !== false ) {
3653                         $ret = array( 'object' => $object );
3654                 } else {
3655                         $ret = array( 'text' => $text );
3656                 }
3657
3658                 wfProfileOut( __METHOD__ );
3659                 return $ret;
3660         }
3661
3662         /**
3663          * Return the text to be used for a given extension tag.
3664          * This is the ghost of strip().
3665          *
3666          * @param $params Associative array of parameters:
3667          *     name       PPNode for the tag name
3668          *     attr       PPNode for unparsed text where tag attributes are thought to be
3669          *     attributes Optional associative array of parsed attributes
3670          *     inner      Contents of extension element
3671          *     noClose    Original text did not have a close tag
3672          * @param $frame PPFrame
3673          *
3674          * @return string
3675          */
3676         function extensionSubstitution( $params, $frame ) {
3677                 $name = $frame->expand( $params['name'] );
3678                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3679                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3680                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3681
3682                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3683                         ( $this->ot['html'] || $this->ot['pre'] );
3684                 if ( $isFunctionTag ) {
3685                         $markerType = 'none';
3686                 } else {
3687                         $markerType = 'general';
3688                 }
3689                 if ( $this->ot['html'] || $isFunctionTag ) {
3690                         $name = strtolower( $name );
3691                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3692                         if ( isset( $params['attributes'] ) ) {
3693                                 $attributes = $attributes + $params['attributes'];
3694                         }
3695
3696                         if ( isset( $this->mTagHooks[$name] ) ) {
3697                                 # Workaround for PHP bug 35229 and similar
3698                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3699                                         throw new MWException( "Tag hook for $name is not callable\n" );
3700                                 }
3701                                 $output = call_user_func_array( $this->mTagHooks[$name],
3702                                         array( $content, $attributes, $this, $frame ) );
3703                         } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
3704                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3705                                 if ( !is_callable( $callback ) ) {
3706                                         throw new MWException( "Tag hook for $name is not callable\n" );
3707                                 }
3708
3709                                 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
3710                         } else {
3711                                 $output = '<span class="error">Invalid tag extension name: ' .
3712                                         htmlspecialchars( $name ) . '</span>';
3713                         }
3714
3715                         if ( is_array( $output ) ) {
3716                                 # Extract flags to local scope (to override $markerType)
3717                                 $flags = $output;
3718                                 $output = $flags[0];
3719                                 unset( $flags[0] );
3720                                 extract( $flags );
3721                         }
3722                 } else {
3723                         if ( is_null( $attrText ) ) {
3724                                 $attrText = '';
3725                         }
3726                         if ( isset( $params['attributes'] ) ) {
3727                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3728                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3729                                                 htmlspecialchars( $attrValue ) . '"';
3730                                 }
3731                         }
3732                         if ( $content === null ) {
3733                                 $output = "<$name$attrText/>";
3734                         } else {
3735                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3736                                 $output = "<$name$attrText>$content$close";
3737                         }
3738                 }
3739
3740                 if ( $markerType === 'none' ) {
3741                         return $output;
3742                 } elseif ( $markerType === 'nowiki' ) {
3743                         $this->mStripState->addNoWiki( $marker, $output );
3744                 } elseif ( $markerType === 'general' ) {
3745                         $this->mStripState->addGeneral( $marker, $output );
3746                 } else {
3747                         throw new MWException( __METHOD__.': invalid marker type' );
3748                 }
3749                 return $marker;
3750         }
3751
3752         /**
3753          * Increment an include size counter
3754          *
3755          * @param $type String: the type of expansion
3756          * @param $size Integer: the size of the text
3757          * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
3758          */
3759         function incrementIncludeSize( $type, $size ) {
3760                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3761                         return false;
3762                 } else {
3763                         $this->mIncludeSizes[$type] += $size;
3764                         return true;
3765                 }
3766         }
3767
3768         /**
3769          * Increment the expensive function count
3770          *
3771          * @return Boolean: false if the limit has been exceeded
3772          */
3773         function incrementExpensiveFunctionCount() {
3774                 global $wgExpensiveParserFunctionLimit;
3775                 $this->mExpensiveFunctionCount++;
3776                 if ( $this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit ) {
3777                         return true;
3778                 }
3779                 return false;
3780         }
3781
3782         /**
3783          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3784          * Fills $this->mDoubleUnderscores, returns the modified text
3785          */
3786         function doDoubleUnderscore( $text ) {
3787                 wfProfileIn( __METHOD__ );
3788
3789                 # The position of __TOC__ needs to be recorded
3790                 $mw = MagicWord::get( 'toc' );
3791                 if ( $mw->match( $text ) ) {
3792                         $this->mShowToc = true;
3793                         $this->mForceTocPosition = true;
3794
3795                         # Set a placeholder. At the end we'll fill it in with the TOC.
3796                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3797
3798                         # Only keep the first one.
3799                         $text = $mw->replace( '', $text );
3800                 }
3801
3802                 # Now match and remove the rest of them
3803                 $mwa = MagicWord::getDoubleUnderscoreArray();
3804                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3805
3806                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3807                         $this->mOutput->mNoGallery = true;
3808                 }
3809                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3810                         $this->mShowToc = false;
3811                 }
3812                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3813                         $this->addTrackingCategory( 'hidden-category-category' );
3814                 }
3815                 # (bug 8068) Allow control over whether robots index a page.
3816                 #
3817                 # @todo FIXME: Bug 14899: __INDEX__ always overrides __NOINDEX__ here!  This
3818                 # is not desirable, the last one on the page should win.
3819                 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3820                         $this->mOutput->setIndexPolicy( 'noindex' );
3821                         $this->addTrackingCategory( 'noindex-category' );
3822                 }
3823                 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
3824                         $this->mOutput->setIndexPolicy( 'index' );
3825                         $this->addTrackingCategory( 'index-category' );
3826                 }
3827
3828                 # Cache all double underscores in the database
3829                 foreach ( $this->mDoubleUnderscores as $key => $val ) {
3830                         $this->mOutput->setProperty( $key, '' );
3831                 }
3832
3833                 wfProfileOut( __METHOD__ );
3834                 return $text;
3835         }
3836
3837         /**
3838          * Add a tracking category, getting the title from a system message,
3839          * or print a debug message if the title is invalid.
3840          *
3841          * @param $msg String: message key
3842          * @return Boolean: whether the addition was successful
3843          */
3844         protected function addTrackingCategory( $msg ) {
3845                 $cat = wfMsgForContent( $msg );
3846
3847                 # Allow tracking categories to be disabled by setting them to "-"
3848                 if ( $cat === '-' ) {
3849                         return false;
3850                 }
3851
3852                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3853                 if ( $containerCategory ) {
3854                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3855                         return true;
3856                 } else {
3857                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3858                         return false;
3859                 }
3860         }
3861
3862         /**
3863          * This function accomplishes several tasks:
3864          * 1) Auto-number headings if that option is enabled
3865          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3866          * 3) Add a Table of contents on the top for users who have enabled the option
3867          * 4) Auto-anchor headings
3868          *
3869          * It loops through all headlines, collects the necessary data, then splits up the
3870          * string and re-inserts the newly formatted headlines.
3871          *
3872          * @param $text String
3873          * @param $origText String: original, untouched wikitext
3874          * @param $isMain Boolean
3875          * @private
3876          */
3877         function formatHeadings( $text, $origText, $isMain=true ) {
3878                 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3879
3880                 # Inhibit editsection links if requested in the page
3881                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3882                         $showEditLink = 0;
3883                 } else {
3884                         $showEditLink = $this->mOptions->getEditSection();
3885                 }
3886                 if ( $showEditLink ) {
3887                         $this->mOutput->setEditSectionTokens( true );
3888                 }
3889
3890                 # Get all headlines for numbering them and adding funky stuff like [edit]
3891                 # links - this is for later, but we need the number of headlines right now
3892                 $matches = array();
3893                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3894
3895                 # if there are fewer than 4 headlines in the article, do not show TOC
3896                 # unless it's been explicitly enabled.
3897                 $enoughToc = $this->mShowToc &&
3898                         ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
3899
3900                 # Allow user to stipulate that a page should have a "new section"
3901                 # link added via __NEWSECTIONLINK__
3902                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3903                         $this->mOutput->setNewSection( true );
3904                 }
3905
3906                 # Allow user to remove the "new section"
3907                 # link via __NONEWSECTIONLINK__
3908                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3909                         $this->mOutput->hideNewSection( true );
3910                 }
3911
3912                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3913                 # override above conditions and always show TOC above first header
3914                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3915                         $this->mShowToc = true;
3916                         $enoughToc = true;
3917                 }
3918
3919                 # headline counter
3920                 $headlineCount = 0;
3921                 $numVisible = 0;
3922
3923                 # Ugh .. the TOC should have neat indentation levels which can be
3924                 # passed to the skin functions. These are determined here
3925                 $toc = '';
3926                 $full = '';
3927                 $head = array();
3928                 $sublevelCount = array();
3929                 $levelCount = array();
3930                 $level = 0;
3931                 $prevlevel = 0;
3932                 $toclevel = 0;
3933                 $prevtoclevel = 0;
3934                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3935                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3936                 $oldType = $this->mOutputType;
3937                 $this->setOutputType( self::OT_WIKI );
3938                 $frame = $this->getPreprocessor()->newFrame();
3939                 $root = $this->preprocessToDom( $origText );
3940                 $node = $root->getFirstChild();
3941                 $byteOffset = 0;
3942                 $tocraw = array();
3943                 $refers = array();
3944
3945                 foreach ( $matches[3] as $headline ) {
3946                         $isTemplate = false;
3947                         $titleText = false;
3948                         $sectionIndex = false;
3949                         $numbering = '';
3950                         $markerMatches = array();
3951                         if ( preg_match("/^$markerRegex/", $headline, $markerMatches ) ) {
3952                                 $serial = $markerMatches[1];
3953                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3954                                 $isTemplate = ( $titleText != $baseTitleText );
3955                                 $headline = preg_replace( "/^$markerRegex/", "", $headline );
3956                         }
3957
3958                         if ( $toclevel ) {
3959                                 $prevlevel = $level;
3960                         }
3961                         $level = $matches[1][$headlineCount];
3962
3963                         if ( $level > $prevlevel ) {
3964                                 # Increase TOC level
3965                                 $toclevel++;
3966                                 $sublevelCount[$toclevel] = 0;
3967                                 if ( $toclevel<$wgMaxTocLevel ) {
3968                                         $prevtoclevel = $toclevel;
3969                                         $toc .= Linker::tocIndent();
3970                                         $numVisible++;
3971                                 }
3972                         } elseif ( $level < $prevlevel && $toclevel > 1 ) {
3973                                 # Decrease TOC level, find level to jump to
3974
3975                                 for ( $i = $toclevel; $i > 0; $i-- ) {
3976                                         if ( $levelCount[$i] == $level ) {
3977                                                 # Found last matching level
3978                                                 $toclevel = $i;
3979                                                 break;
3980                                         } elseif ( $levelCount[$i] < $level ) {
3981                                                 # Found first matching level below current level
3982                                                 $toclevel = $i + 1;
3983                                                 break;
3984                                         }
3985                                 }
3986                                 if ( $i == 0 ) {
3987                                         $toclevel = 1;
3988                                 }
3989                                 if ( $toclevel<$wgMaxTocLevel ) {
3990                                         if ( $prevtoclevel < $wgMaxTocLevel ) {
3991                                                 # Unindent only if the previous toc level was shown :p
3992                                                 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
3993                                                 $prevtoclevel = $toclevel;
3994                                         } else {
3995                                                 $toc .= Linker::tocLineEnd();
3996                                         }
3997                                 }
3998                         } else {
3999                                 # No change in level, end TOC line
4000                                 if ( $toclevel<$wgMaxTocLevel ) {
4001                                         $toc .= Linker::tocLineEnd();
4002                                 }
4003                         }
4004
4005                         $levelCount[$toclevel] = $level;
4006
4007                         # count number of headlines for each level
4008                         @$sublevelCount[$toclevel]++;
4009                         $dot = 0;
4010                         for( $i = 1; $i <= $toclevel; $i++ ) {
4011                                 if ( !empty( $sublevelCount[$i] ) ) {
4012                                         if ( $dot ) {
4013                                                 $numbering .= '.';
4014                                         }
4015                                         $numbering .= $this->getFunctionLang()->formatNum( $sublevelCount[$i] );
4016                                         $dot = 1;
4017                                 }
4018                         }
4019
4020                         # The safe header is a version of the header text safe to use for links
4021                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4022                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
4023
4024                         # Remove link placeholders by the link text.
4025                         #     <!--LINK number-->
4026                         # turns into
4027                         #     link text with suffix
4028                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
4029
4030                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
4031                         $tocline = preg_replace(
4032                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
4033                                 array( '',                          '<$1>' ),
4034                                 $safeHeadline
4035                         );
4036                         $tocline = trim( $tocline );
4037
4038                         # For the anchor, strip out HTML-y stuff period
4039                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
4040                         $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4041
4042                         # Save headline for section edit hint before it's escaped
4043                         $headlineHint = $safeHeadline;
4044
4045                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
4046                                 # For reverse compatibility, provide an id that's
4047                                 # HTML4-compatible, like we used to.
4048                                 #
4049                                 # It may be worth noting, academically, that it's possible for
4050                                 # the legacy anchor to conflict with a non-legacy headline
4051                                 # anchor on the page.  In this case likely the "correct" thing
4052                                 # would be to either drop the legacy anchors or make sure
4053                                 # they're numbered first.  However, this would require people
4054                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4055                                 # manually, so let's not bother worrying about it.
4056                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
4057                                         array( 'noninitial', 'legacy' ) );
4058                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4059
4060                                 if ( $legacyHeadline == $safeHeadline ) {
4061                                         # No reason to have both (in fact, we can't)
4062                                         $legacyHeadline = false;
4063                                 }
4064                         } else {
4065                                 $legacyHeadline = false;
4066                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
4067                                         'noninitial' );
4068                         }
4069
4070                         # HTML names must be case-insensitively unique (bug 10721).
4071                         # This does not apply to Unicode characters per
4072                         # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
4073                         # @todo FIXME: We may be changing them depending on the current locale.
4074                         $arrayKey = strtolower( $safeHeadline );
4075                         if ( $legacyHeadline === false ) {
4076                                 $legacyArrayKey = false;
4077                         } else {
4078                                 $legacyArrayKey = strtolower( $legacyHeadline );
4079                         }
4080
4081                         # count how many in assoc. array so we can track dupes in anchors
4082                         if ( isset( $refers[$arrayKey] ) ) {
4083                                 $refers[$arrayKey]++;
4084                         } else {
4085                                 $refers[$arrayKey] = 1;
4086                         }
4087                         if ( isset( $refers[$legacyArrayKey] ) ) {
4088                                 $refers[$legacyArrayKey]++;
4089                         } else {
4090                                 $refers[$legacyArrayKey] = 1;
4091                         }
4092
4093                         # Don't number the heading if it is the only one (looks silly)
4094                         if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4095                                 # the two are different if the line contains a link
4096                                 $headline = $numbering . ' ' . $headline;
4097                         }
4098
4099                         # Create the anchor for linking from the TOC to the section
4100                         $anchor = $safeHeadline;
4101                         $legacyAnchor = $legacyHeadline;
4102                         if ( $refers[$arrayKey] > 1 ) {
4103                                 $anchor .= '_' . $refers[$arrayKey];
4104                         }
4105                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
4106                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
4107                         }
4108                         if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4109                                 $toc .= Linker::tocLine( $anchor, $tocline,
4110                                         $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4111                         }
4112
4113                         # Add the section to the section tree
4114                         # Find the DOM node for this header
4115                         while ( $node && !$isTemplate ) {
4116                                 if ( $node->getName() === 'h' ) {
4117                                         $bits = $node->splitHeading();
4118                                         if ( $bits['i'] == $sectionIndex ) {
4119                                                 break;
4120                                         }
4121                                 }
4122                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4123                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4124                                 $node = $node->getNextSibling();
4125                         }
4126                         $tocraw[] = array(
4127                                 'toclevel' => $toclevel,
4128                                 'level' => $level,
4129                                 'line' => $tocline,
4130                                 'number' => $numbering,
4131                                 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4132                                 'fromtitle' => $titleText,
4133                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
4134                                 'anchor' => $anchor,
4135                         );
4136
4137                         # give headline the correct <h#> tag
4138                         if ( $showEditLink && $sectionIndex !== false ) {
4139                                 // Output edit section links as markers with styles that can be customized by skins
4140                                 if ( $isTemplate ) {
4141                                         # Put a T flag in the section identifier, to indicate to extractSections()
4142                                         # that sections inside <includeonly> should be counted.
4143                                         $editlinkArgs = array( $titleText, "T-$sectionIndex"/*, null */ );
4144                                 } else {
4145                                         $editlinkArgs = array( $this->mTitle->getPrefixedText(), $sectionIndex, $headlineHint );
4146                                 }
4147                                 // We use a bit of pesudo-xml for editsection markers. The language converter is run later on
4148                                 // Using a UNIQ style marker leads to the converter screwing up the tokens when it converts stuff
4149                                 // And trying to insert strip tags fails too. At this point all real inputted tags have already been escaped
4150                                 // so we don't have to worry about a user trying to input one of these markers directly.
4151                                 // We use a page and section attribute to stop the language converter from converting these important bits
4152                                 // of data, but put the headline hint inside a content block because the language converter is supposed to
4153                                 // be able to convert that piece of data.
4154                                 $editlink = '<mw:editsection page="' . htmlspecialchars($editlinkArgs[0]);
4155                                 $editlink .= '" section="' . htmlspecialchars($editlinkArgs[1]) .'"';
4156                                 if ( isset($editlinkArgs[2]) ) {
4157                                         $editlink .= '>' . $editlinkArgs[2] . '</mw:editsection>';
4158                                 } else {
4159                                         $editlink .= '/>';
4160                                 }
4161                         } else {
4162                                 $editlink = '';
4163                         }
4164                         $head[$headlineCount] = Linker::makeHeadline( $level,
4165                                 $matches['attrib'][$headlineCount], $anchor, $headline,
4166                                 $editlink, $legacyAnchor );
4167
4168                         $headlineCount++;
4169                 }
4170
4171                 $this->setOutputType( $oldType );
4172
4173                 # Never ever show TOC if no headers
4174                 if ( $numVisible < 1 ) {
4175                         $enoughToc = false;
4176                 }
4177
4178                 if ( $enoughToc ) {
4179                         if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4180                                 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4181                         }
4182                         $toc = Linker::tocList( $toc, $this->mOptions->getUserLang() );
4183                         $this->mOutput->setTOCHTML( $toc );
4184                 }
4185
4186                 if ( $isMain ) {
4187                         $this->mOutput->setSections( $tocraw );
4188                 }
4189
4190                 # split up and insert constructed headlines
4191
4192                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
4193                 $i = 0;
4194
4195                 foreach ( $blocks as $block ) {
4196                         if ( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
4197                                 # This is the [edit] link that appears for the top block of text when
4198                                 # section editing is enabled
4199
4200                                 # Disabled because it broke block formatting
4201                                 # For example, a bullet point in the top line
4202                                 # $full .= $sk->editSectionLink(0);
4203                         }
4204                         $full .= $block;
4205                         if ( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
4206                                 # Top anchor now in skin
4207                                 $full = $full.$toc;
4208                         }
4209
4210                         if ( !empty( $head[$i] ) ) {
4211                                 $full .= $head[$i];
4212                         }
4213                         $i++;
4214                 }
4215                 if ( $this->mForceTocPosition ) {
4216                         return str_replace( '<!--MWTOC-->', $toc, $full );
4217                 } else {
4218                         return $full;
4219                 }
4220         }
4221
4222         /**
4223          * Transform wiki markup when saving a page by doing \r\n -> \n
4224          * conversion, substitting signatures, {{subst:}} templates, etc.
4225          *
4226          * @param $text String: the text to transform
4227          * @param $title Title: the Title object for the current article
4228          * @param $user User: the User object describing the current user
4229          * @param $options ParserOptions: parsing options
4230          * @param $clearState Boolean: whether to clear the parser state first
4231          * @return String: the altered wiki markup
4232          */
4233         public function preSaveTransform( $text, Title $title, User $user, ParserOptions $options, $clearState = true ) {
4234                 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4235                 $this->setUser( $user );
4236
4237                 $pairs = array(
4238                         "\r\n" => "\n",
4239                 );
4240                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4241                 if( $options->getPreSaveTransform() ) {
4242                         $text = $this->pstPass2( $text, $user );
4243                 }
4244                 $text = $this->mStripState->unstripBoth( $text );
4245
4246                 $this->setUser( null ); #Reset
4247
4248                 return $text;
4249         }
4250
4251         /**
4252          * Pre-save transform helper function
4253          * @private
4254          */
4255         function pstPass2( $text, $user ) {
4256                 global $wgContLang, $wgLocaltimezone;
4257
4258                 # Note: This is the timestamp saved as hardcoded wikitext to
4259                 # the database, we use $wgContLang here in order to give
4260                 # everyone the same signature and use the default one rather
4261                 # than the one selected in each user's preferences.
4262                 # (see also bug 12815)
4263                 $ts = $this->mOptions->getTimestamp();
4264                 if ( isset( $wgLocaltimezone ) ) {
4265                         $tz = $wgLocaltimezone;
4266                 } else {
4267                         $tz = date_default_timezone_get();
4268                 }
4269
4270                 $unixts = wfTimestamp( TS_UNIX, $ts );
4271                 $oldtz = date_default_timezone_get();
4272                 date_default_timezone_set( $tz );
4273                 $ts = date( 'YmdHis', $unixts );
4274                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4275
4276                 # Allow translation of timezones through wiki. date() can return
4277                 # whatever crap the system uses, localised or not, so we cannot
4278                 # ship premade translations.
4279                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4280                 $msg = wfMessage( $key )->inContentLanguage();
4281                 if ( $msg->exists() ) {
4282                         $tzMsg = $msg->text();
4283                 }
4284
4285                 date_default_timezone_set( $oldtz );
4286
4287                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4288
4289                 # Variable replacement
4290                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4291                 $text = $this->replaceVariables( $text );
4292
4293                 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4294                 # which may corrupt this parser instance via its wfMsgExt( parsemag ) call-
4295
4296                 # Signatures
4297                 $sigText = $this->getUserSig( $user );
4298                 $text = strtr( $text, array(
4299                         '~~~~~' => $d,
4300                         '~~~~' => "$sigText $d",
4301                         '~~~' => $sigText
4302                 ) );
4303
4304                 # Context links: [[|name]] and [[name (context)|]]
4305                 global $wgLegalTitleChars;
4306                 $tc = "[$wgLegalTitleChars]";
4307                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4308
4309                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
4310                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
4311                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4312                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4313
4314                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4315                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4316                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4317                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4318
4319                 $t = $this->mTitle->getText();
4320                 $m = array();
4321                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4322                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4323                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4324                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4325                 } else {
4326                         # if there's no context, don't bother duplicating the title
4327                         $text = preg_replace( $p2, '[[\\1]]', $text );
4328                 }
4329
4330                 # Trim trailing whitespace
4331                 $text = rtrim( $text );
4332
4333                 return $text;
4334         }
4335
4336         /**
4337          * Fetch the user's signature text, if any, and normalize to
4338          * validated, ready-to-insert wikitext.
4339          * If you have pre-fetched the nickname or the fancySig option, you can
4340          * specify them here to save a database query.
4341          * Do not reuse this parser instance after calling getUserSig(),
4342          * as it may have changed if it's the $wgParser.
4343          *
4344          * @param $user User
4345          * @param $nickname String: nickname to use or false to use user's default nickname
4346          * @param $fancySig Boolean: whether the nicknname is the complete signature
4347          *                  or null to use default value
4348          * @return string
4349          */
4350         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4351                 global $wgMaxSigChars;
4352
4353                 $username = $user->getName();
4354
4355                 # If not given, retrieve from the user object.
4356                 if ( $nickname === false )
4357                         $nickname = $user->getOption( 'nickname' );
4358
4359                 if ( is_null( $fancySig ) ) {
4360                         $fancySig = $user->getBoolOption( 'fancysig' );
4361                 }
4362
4363                 $nickname = $nickname == null ? $username : $nickname;
4364
4365                 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4366                         $nickname = $username;
4367                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4368                 } elseif ( $fancySig !== false ) {
4369                         # Sig. might contain markup; validate this
4370                         if ( $this->validateSig( $nickname ) !== false ) {
4371                                 # Validated; clean up (if needed) and return it
4372                                 return $this->cleanSig( $nickname, true );
4373                         } else {
4374                                 # Failed to validate; fall back to the default
4375                                 $nickname = $username;
4376                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4377                         }
4378                 }
4379
4380                 # Make sure nickname doesnt get a sig in a sig
4381                 $nickname = $this->cleanSigInSig( $nickname );
4382
4383                 # If we're still here, make it a link to the user page
4384                 $userText = wfEscapeWikiText( $username );
4385                 $nickText = wfEscapeWikiText( $nickname );
4386                 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4387
4388                 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()->title( $this->getTitle() )->text();
4389         }
4390
4391         /**
4392          * Check that the user's signature contains no bad XML
4393          *
4394          * @param $text String
4395          * @return mixed An expanded string, or false if invalid.
4396          */
4397         function validateSig( $text ) {
4398                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4399         }
4400
4401         /**
4402          * Clean up signature text
4403          *
4404          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4405          * 2) Substitute all transclusions
4406          *
4407          * @param $text String
4408          * @param $parsing Whether we're cleaning (preferences save) or parsing
4409          * @return String: signature text
4410          */
4411         function cleanSig( $text, $parsing = false ) {
4412                 if ( !$parsing ) {
4413                         global $wgTitle;
4414                         $this->mOptions = new ParserOptions;
4415                         $this->clearState();
4416                         $this->setTitle( $wgTitle );
4417                         $this->setOutputType = self::OT_PREPROCESS;
4418                 }
4419
4420                 # Option to disable this feature
4421                 if ( !$this->mOptions->getCleanSignatures() ) {
4422                         return $text;
4423                 }
4424
4425                 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4426                 #  => Move this logic to braceSubstitution()
4427                 $substWord = MagicWord::get( 'subst' );
4428                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4429                 $substText = '{{' . $substWord->getSynonym( 0 );
4430
4431                 $text = preg_replace( $substRegex, $substText, $text );
4432                 $text = $this->cleanSigInSig( $text );
4433                 $dom = $this->preprocessToDom( $text );
4434                 $frame = $this->getPreprocessor()->newFrame();
4435                 $text = $frame->expand( $dom );
4436
4437                 if ( !$parsing ) {
4438                         $text = $this->mStripState->unstripBoth( $text );
4439                 }
4440
4441                 return $text;
4442         }
4443
4444         /**
4445          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4446          *
4447          * @param $text String
4448          * @return String: signature text with /~{3,5}/ removed
4449          */
4450         function cleanSigInSig( $text ) {
4451                 $text = preg_replace( '/~{3,5}/', '', $text );
4452                 return $text;
4453         }
4454
4455         /**
4456          * Set up some variables which are usually set up in parse()
4457          * so that an external function can call some class members with confidence
4458          */
4459         public function startExternalParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4460                 $this->startParse( $title, $options, $outputType, $clearState );
4461         }
4462
4463         private function startParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4464                 $this->setTitle( $title );
4465                 $this->mOptions = $options;
4466                 $this->setOutputType( $outputType );
4467                 if ( $clearState ) {
4468                         $this->clearState();
4469                 }
4470         }
4471
4472         /**
4473          * Wrapper for preprocess()
4474          *
4475          * @param $text String: the text to preprocess
4476          * @param $options ParserOptions: options
4477          * @param $title Title object or null to use $wgTitle
4478          * @return String
4479          */
4480         public function transformMsg( $text, $options, $title = null ) {
4481                 static $executing = false;
4482
4483                 # Guard against infinite recursion
4484                 if ( $executing ) {
4485                         return $text;
4486                 }
4487                 $executing = true;
4488
4489                 wfProfileIn( __METHOD__ );
4490                 if ( !$title ) {
4491                         global $wgTitle;
4492                         $title = $wgTitle;
4493                 }
4494                 if ( !$title ) {
4495                         # It's not uncommon having a null $wgTitle in scripts. See r80898
4496                         # Create a ghost title in such case
4497                         $title = Title::newFromText( 'Dwimmerlaik' );
4498                 }
4499                 $text = $this->preprocess( $text, $title, $options );
4500
4501                 $executing = false;
4502                 wfProfileOut( __METHOD__ );
4503                 return $text;
4504         }
4505
4506         /**
4507          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4508          * The callback should have the following form:
4509          *    function myParserHook( $text, $params, $parser, $frame ) { ... }
4510          *
4511          * Transform and return $text. Use $parser for any required context, e.g. use
4512          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4513          *
4514          * Hooks may return extended information by returning an array, of which the
4515          * first numbered element (index 0) must be the return string, and all other
4516          * entries are extracted into local variables within an internal function
4517          * in the Parser class.
4518          *
4519          * This interface (introduced r61913) appears to be undocumented, but
4520          * 'markerName' is used by some core tag hooks to override which strip
4521          * array their results are placed in. **Use great caution if attempting
4522          * this interface, as it is not documented and injudicious use could smash
4523          * private variables.**
4524          *
4525          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4526          * @param $callback Mixed: the callback function (and object) to use for the tag
4527          * @return The old value of the mTagHooks array associated with the hook
4528          */
4529         public function setHook( $tag, $callback ) {
4530                 $tag = strtolower( $tag );
4531                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4532                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4533                 $this->mTagHooks[$tag] = $callback;
4534                 if ( !in_array( $tag, $this->mStripList ) ) {
4535                         $this->mStripList[] = $tag;
4536                 }
4537
4538                 return $oldVal;
4539         }
4540
4541         /**
4542          * As setHook(), but letting the contents be parsed.
4543          *
4544          * Transparent tag hooks are like regular XML-style tag hooks, except they
4545          * operate late in the transformation sequence, on HTML instead of wikitext.
4546          *
4547          * This is probably obsoleted by things dealing with parser frames?
4548          * The only extension currently using it is geoserver.
4549          *
4550          * @since 1.10
4551          * @todo better document or deprecate this
4552          *
4553          * @param $tag Mixed: the tag to use, e.g. 'hook' for <hook>
4554          * @param $callback Mixed: the callback function (and object) to use for the tag
4555          * @return The old value of the mTagHooks array associated with the hook
4556          */
4557         function setTransparentTagHook( $tag, $callback ) {
4558                 $tag = strtolower( $tag );
4559                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4560                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4561                 $this->mTransparentTagHooks[$tag] = $callback;
4562
4563                 return $oldVal;
4564         }
4565
4566         /**
4567          * Remove all tag hooks
4568          */
4569         function clearTagHooks() {
4570                 $this->mTagHooks = array();
4571                 $this->mStripList = $this->mDefaultStripList;
4572         }
4573
4574         /**
4575          * Remove a specific tag hook. Should not be called on $wgParser.
4576          * Does not change the strip list.
4577          *
4578          * @param string $tag
4579          * @return void
4580          */
4581         function clearTagHook( $tag ) {
4582                 if ( isset( $this->mTagHooks[$tag] ) ) {
4583                         unset( $this->mTagHooks[$tag] );
4584                 }
4585         }
4586
4587         /**
4588          * Create a function, e.g. {{sum:1|2|3}}
4589          * The callback function should have the form:
4590          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4591          *
4592          * Or with SFH_OBJECT_ARGS:
4593          *    function myParserFunction( $parser, $frame, $args ) { ... }
4594          *
4595          * The callback may either return the text result of the function, or an array with the text
4596          * in element 0, and a number of flags in the other elements. The names of the flags are
4597          * specified in the keys. Valid flags are:
4598          *   found                     The text returned is valid, stop processing the template. This
4599          *                             is on by default.
4600          *   nowiki                    Wiki markup in the return value should be escaped
4601          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4602          *
4603          * @param $id String: The magic word ID
4604          * @param $callback Mixed: the callback function (and object) to use
4605          * @param $flags Integer: a combination of the following flags:
4606          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4607          *
4608          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4609          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4610          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4611          *     the arguments, and to control the way they are expanded.
4612          *
4613          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4614          *     arguments, for instance:
4615          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4616          *
4617          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4618          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4619          *     working if/when this is changed.
4620          *
4621          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4622          *     expansion.
4623          *
4624          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4625          *     about the methods available in PPFrame and PPNode.
4626          *
4627          * @return The old callback function for this name, if any
4628          */
4629         public function setFunctionHook( $id, $callback, $flags = 0 ) {
4630                 global $wgContLang;
4631
4632                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4633                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4634
4635                 # Add to function cache
4636                 $mw = MagicWord::get( $id );
4637                 if ( !$mw )
4638                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4639
4640                 $synonyms = $mw->getSynonyms();
4641                 $sensitive = intval( $mw->isCaseSensitive() );
4642
4643                 foreach ( $synonyms as $syn ) {
4644                         # Case
4645                         if ( !$sensitive ) {
4646                                 $syn = $wgContLang->lc( $syn );
4647                         }
4648                         # Add leading hash
4649                         if ( !( $flags & SFH_NO_HASH ) ) {
4650                                 $syn = '#' . $syn;
4651                         }
4652                         # Remove trailing colon
4653                         if ( substr( $syn, -1, 1 ) === ':' ) {
4654                                 $syn = substr( $syn, 0, -1 );
4655                         }
4656                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4657                 }
4658                 return $oldVal;
4659         }
4660
4661         /**
4662          * Get all registered function hook identifiers
4663          *
4664          * @return Array
4665          */
4666         function getFunctionHooks() {
4667                 return array_keys( $this->mFunctionHooks );
4668         }
4669
4670         /**
4671          * Create a tag function, e.g. <test>some stuff</test>.
4672          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4673          * Unlike parser functions, their content is not preprocessed.
4674          */
4675         function setFunctionTagHook( $tag, $callback, $flags ) {
4676                 $tag = strtolower( $tag );
4677                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
4678                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4679                         $this->mFunctionTagHooks[$tag] : null;
4680                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4681
4682                 if ( !in_array( $tag, $this->mStripList ) ) {
4683                         $this->mStripList[] = $tag;
4684                 }
4685
4686                 return $old;
4687         }
4688
4689         /**
4690          * @todo FIXME: Update documentation. makeLinkObj() is deprecated.
4691          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4692          * Placeholders created in Skin::makeLinkObj()
4693          * Returns an array of link CSS classes, indexed by PDBK.
4694          */
4695         function replaceLinkHolders( &$text, $options = 0 ) {
4696                 return $this->mLinkHolders->replace( $text );
4697         }
4698
4699         /**
4700          * Replace <!--LINK--> link placeholders with plain text of links
4701          * (not HTML-formatted).
4702          *
4703          * @param $text String
4704          * @return String
4705          */
4706         function replaceLinkHoldersText( $text ) {
4707                 return $this->mLinkHolders->replaceText( $text );
4708         }
4709
4710         /**
4711          * Renders an image gallery from a text with one line per image.
4712          * text labels may be given by using |-style alternative text. E.g.
4713          *   Image:one.jpg|The number "1"
4714          *   Image:tree.jpg|A tree
4715          * given as text will return the HTML of a gallery with two images,
4716          * labeled 'The number "1"' and
4717          * 'A tree'.
4718          *
4719          * @param string $text
4720          * @param array $param
4721          * @return string HTML
4722          */
4723         function renderImageGallery( $text, $params ) {
4724                 $ig = new ImageGallery();
4725                 $ig->setContextTitle( $this->mTitle );
4726                 $ig->setShowBytes( false );
4727                 $ig->setShowFilename( false );
4728                 $ig->setParser( $this );
4729                 $ig->setHideBadImages();
4730                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4731
4732                 if ( isset( $params['showfilename'] ) ) {
4733                         $ig->setShowFilename( true );
4734                 } else {
4735                         $ig->setShowFilename( false );
4736                 }
4737                 if ( isset( $params['caption'] ) ) {
4738                         $caption = $params['caption'];
4739                         $caption = htmlspecialchars( $caption );
4740                         $caption = $this->replaceInternalLinks( $caption );
4741                         $ig->setCaptionHtml( $caption );
4742                 }
4743                 if ( isset( $params['perrow'] ) ) {
4744                         $ig->setPerRow( $params['perrow'] );
4745                 }
4746                 if ( isset( $params['widths'] ) ) {
4747                         $ig->setWidths( $params['widths'] );
4748                 }
4749                 if ( isset( $params['heights'] ) ) {
4750                         $ig->setHeights( $params['heights'] );
4751                 }
4752
4753                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4754
4755                 $lines = StringUtils::explode( "\n", $text );
4756                 foreach ( $lines as $line ) {
4757                         # match lines like these:
4758                         # Image:someimage.jpg|This is some image
4759                         $matches = array();
4760                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4761                         # Skip empty lines
4762                         if ( count( $matches ) == 0 ) {
4763                                 continue;
4764                         }
4765
4766                         if ( strpos( $matches[0], '%' ) !== false ) {
4767                                 $matches[1] = rawurldecode( $matches[1] );
4768                         }
4769                         $title = Title::newFromText( $matches[1], NS_FILE );
4770                         if ( is_null( $title ) ) {
4771                                 # Bogus title. Ignore these so we don't bomb out later.
4772                                 continue;
4773                         }
4774
4775                         $label = '';
4776                         $alt = '';
4777                         if ( isset( $matches[3] ) ) {
4778                                 // look for an |alt= definition while trying not to break existing
4779                                 // captions with multiple pipes (|) in it, until a more sensible grammar
4780                                 // is defined for images in galleries
4781
4782                                 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
4783                                 $altmatches = StringUtils::explode('|', $matches[3]);
4784                                 $magicWordAlt = MagicWord::get( 'img_alt' );
4785
4786                                 foreach ( $altmatches as $altmatch ) {
4787                                         $match = $magicWordAlt->matchVariableStartToEnd( $altmatch );
4788                                         if ( $match ) {
4789                                                 $alt = $this->stripAltText( $match, false );
4790                                         }
4791                                         else {
4792                                                 // concatenate all other pipes
4793                                                 $label .= '|' . $altmatch;
4794                                         }
4795                                 }
4796                                 // remove the first pipe
4797                                 $label = substr( $label, 1 );
4798                         }
4799
4800                         $ig->add( $title, $label, $alt );
4801                 }
4802                 return $ig->toHTML();
4803         }
4804
4805         function getImageParams( $handler ) {
4806                 if ( $handler ) {
4807                         $handlerClass = get_class( $handler );
4808                 } else {
4809                         $handlerClass = '';
4810                 }
4811                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4812                         # Initialise static lists
4813                         static $internalParamNames = array(
4814                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4815                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4816                                         'bottom', 'text-bottom' ),
4817                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4818                                         'upright', 'border', 'link', 'alt' ),
4819                         );
4820                         static $internalParamMap;
4821                         if ( !$internalParamMap ) {
4822                                 $internalParamMap = array();
4823                                 foreach ( $internalParamNames as $type => $names ) {
4824                                         foreach ( $names as $name ) {
4825                                                 $magicName = str_replace( '-', '_', "img_$name" );
4826                                                 $internalParamMap[$magicName] = array( $type, $name );
4827                                         }
4828                                 }
4829                         }
4830
4831                         # Add handler params
4832                         $paramMap = $internalParamMap;
4833                         if ( $handler ) {
4834                                 $handlerParamMap = $handler->getParamMap();
4835                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4836                                         $paramMap[$magic] = array( 'handler', $paramName );
4837                                 }
4838                         }
4839                         $this->mImageParams[$handlerClass] = $paramMap;
4840                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4841                 }
4842                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4843         }
4844
4845         /**
4846          * Parse image options text and use it to make an image
4847          *
4848          * @param $title Title
4849          * @param $options String
4850          * @param $holders LinkHolderArray
4851          * @return string HTML
4852          */
4853         function makeImage( $title, $options, $holders = false ) {
4854                 # Check if the options text is of the form "options|alt text"
4855                 # Options are:
4856                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4857                 #  * left       no resizing, just left align. label is used for alt= only
4858                 #  * right      same, but right aligned
4859                 #  * none       same, but not aligned
4860                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4861                 #  * center     center the image
4862                 #  * frame      Keep original image size, no magnify-button.
4863                 #  * framed     Same as "frame"
4864                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4865                 #  * upright    reduce width for upright images, rounded to full __0 px
4866                 #  * border     draw a 1px border around the image
4867                 #  * alt        Text for HTML alt attribute (defaults to empty)
4868                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4869                 # vertical-align values (no % or length right now):
4870                 #  * baseline
4871                 #  * sub
4872                 #  * super
4873                 #  * top
4874                 #  * text-top
4875                 #  * middle
4876                 #  * bottom
4877                 #  * text-bottom
4878
4879                 $parts = StringUtils::explode( "|", $options );
4880
4881                 # Give extensions a chance to select the file revision for us
4882                 $time = $sha1 = $descQuery = false;
4883                 wfRunHooks( 'BeforeParserFetchFileAndTitle',
4884                         array( $this, $title, &$time, &$sha1, &$descQuery ) );
4885                 # Fetch and register the file (file title may be different via hooks)
4886                 list( $file, $title ) = $this->fetchFileAndTitle( $title, $time, $sha1 );
4887
4888                 # Get parameter map
4889                 $handler = $file ? $file->getHandler() : false;
4890
4891                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4892
4893                 if ( !$file ) {
4894                         $this->addTrackingCategory( 'broken-file-category' );
4895                 }
4896
4897                 # Process the input parameters
4898                 $caption = '';
4899                 $params = array( 'frame' => array(), 'handler' => array(),
4900                         'horizAlign' => array(), 'vertAlign' => array() );
4901                 foreach ( $parts as $part ) {
4902                         $part = trim( $part );
4903                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4904                         $validated = false;
4905                         if ( isset( $paramMap[$magicName] ) ) {
4906                                 list( $type, $paramName ) = $paramMap[$magicName];
4907
4908                                 # Special case; width and height come in one variable together
4909                                 if ( $type === 'handler' && $paramName === 'width' ) {
4910                                         $m = array();
4911                                         # (bug 13500) In both cases (width/height and width only),
4912                                         # permit trailing "px" for backward compatibility.
4913                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4914                                                 $width = intval( $m[1] );
4915                                                 $height = intval( $m[2] );
4916                                                 if ( $handler->validateParam( 'width', $width ) ) {
4917                                                         $params[$type]['width'] = $width;
4918                                                         $validated = true;
4919                                                 }
4920                                                 if ( $handler->validateParam( 'height', $height ) ) {
4921                                                         $params[$type]['height'] = $height;
4922                                                         $validated = true;
4923                                                 }
4924                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4925                                                 $width = intval( $value );
4926                                                 if ( $handler->validateParam( 'width', $width ) ) {
4927                                                         $params[$type]['width'] = $width;
4928                                                         $validated = true;
4929                                                 }
4930                                         } # else no validation -- bug 13436
4931                                 } else {
4932                                         if ( $type === 'handler' ) {
4933                                                 # Validate handler parameter
4934                                                 $validated = $handler->validateParam( $paramName, $value );
4935                                         } else {
4936                                                 # Validate internal parameters
4937                                                 switch( $paramName ) {
4938                                                 case 'manualthumb':
4939                                                 case 'alt':
4940                                                         # @todo FIXME: Possibly check validity here for
4941                                                         # manualthumb? downstream behavior seems odd with
4942                                                         # missing manual thumbs.
4943                                                         $validated = true;
4944                                                         $value = $this->stripAltText( $value, $holders );
4945                                                         break;
4946                                                 case 'link':
4947                                                         $chars = self::EXT_LINK_URL_CLASS;
4948                                                         $prots = $this->mUrlProtocols;
4949                                                         if ( $value === '' ) {
4950                                                                 $paramName = 'no-link';
4951                                                                 $value = true;
4952                                                                 $validated = true;
4953                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4954                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4955                                                                         $paramName = 'link-url';
4956                                                                         $this->mOutput->addExternalLink( $value );
4957                                                                         if ( $this->mOptions->getExternalLinkTarget() ) {
4958                                                                                 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
4959                                                                         }
4960                                                                         $validated = true;
4961                                                                 }
4962                                                         } else {
4963                                                                 $linkTitle = Title::newFromText( $value );
4964                                                                 if ( $linkTitle ) {
4965                                                                         $paramName = 'link-title';
4966                                                                         $value = $linkTitle;
4967                                                                         $this->mOutput->addLink( $linkTitle );
4968                                                                         $validated = true;
4969                                                                 }
4970                                                         }
4971                                                         break;
4972                                                 default:
4973                                                         # Most other things appear to be empty or numeric...
4974                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4975                                                 }
4976                                         }
4977
4978                                         if ( $validated ) {
4979                                                 $params[$type][$paramName] = $value;
4980                                         }
4981                                 }
4982                         }
4983                         if ( !$validated ) {
4984                                 $caption = $part;
4985                         }
4986                 }
4987
4988                 # Process alignment parameters
4989                 if ( $params['horizAlign'] ) {
4990                         $params['frame']['align'] = key( $params['horizAlign'] );
4991                 }
4992                 if ( $params['vertAlign'] ) {
4993                         $params['frame']['valign'] = key( $params['vertAlign'] );
4994                 }
4995
4996                 $params['frame']['caption'] = $caption;
4997
4998                 # Will the image be presented in a frame, with the caption below?
4999                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
5000                                                  isset( $params['frame']['framed'] ) ||
5001                                                  isset( $params['frame']['thumbnail'] ) ||
5002                                                  isset( $params['frame']['manualthumb'] );
5003
5004                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
5005                 # came to also set the caption, ordinary text after the image -- which
5006                 # makes no sense, because that just repeats the text multiple times in
5007                 # screen readers.  It *also* came to set the title attribute.
5008                 #
5009                 # Now that we have an alt attribute, we should not set the alt text to
5010                 # equal the caption: that's worse than useless, it just repeats the
5011                 # text.  This is the framed/thumbnail case.  If there's no caption, we
5012                 # use the unnamed parameter for alt text as well, just for the time be-
5013                 # ing, if the unnamed param is set and the alt param is not.
5014                 #
5015                 # For the future, we need to figure out if we want to tweak this more,
5016                 # e.g., introducing a title= parameter for the title; ignoring the un-
5017                 # named parameter entirely for images without a caption; adding an ex-
5018                 # plicit caption= parameter and preserving the old magic unnamed para-
5019                 # meter for BC; ...
5020                 if ( $imageIsFramed ) { # Framed image
5021                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5022                                 # No caption or alt text, add the filename as the alt text so
5023                                 # that screen readers at least get some description of the image
5024                                 $params['frame']['alt'] = $title->getText();
5025                         }
5026                         # Do not set $params['frame']['title'] because tooltips don't make sense
5027                         # for framed images
5028                 } else { # Inline image
5029                         if ( !isset( $params['frame']['alt'] ) ) {
5030                                 # No alt text, use the "caption" for the alt text
5031                                 if ( $caption !== '') {
5032                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5033                                 } else {
5034                                         # No caption, fall back to using the filename for the
5035                                         # alt text
5036                                         $params['frame']['alt'] = $title->getText();
5037                                 }
5038                         }
5039                         # Use the "caption" for the tooltip text
5040                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5041                 }
5042
5043                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
5044
5045                 # Linker does the rest
5046                 $ret = Linker::makeImageLink2( $title, $file, $params['frame'], $params['handler'],
5047                         $time, $descQuery, $this->mOptions->getThumbSize() );
5048
5049                 # Give the handler a chance to modify the parser object
5050                 if ( $handler ) {
5051                         $handler->parserTransformHook( $this, $file );
5052                 }
5053
5054                 return $ret;
5055         }
5056
5057         /**
5058          * @param $caption
5059          * @param $holders LinkHolderArray
5060          * @return mixed|String
5061          */
5062         protected function stripAltText( $caption, $holders ) {
5063                 # Strip bad stuff out of the title (tooltip).  We can't just use
5064                 # replaceLinkHoldersText() here, because if this function is called
5065                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5066                 if ( $holders ) {
5067                         $tooltip = $holders->replaceText( $caption );
5068                 } else {
5069                         $tooltip = $this->replaceLinkHoldersText( $caption );
5070                 }
5071
5072                 # make sure there are no placeholders in thumbnail attributes
5073                 # that are later expanded to html- so expand them now and
5074                 # remove the tags
5075                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5076                 $tooltip = Sanitizer::stripAllTags( $tooltip );
5077
5078                 return $tooltip;
5079         }
5080
5081         /**
5082          * Set a flag in the output object indicating that the content is dynamic and
5083          * shouldn't be cached.
5084          */
5085         function disableCache() {
5086                 wfDebug( "Parser output marked as uncacheable.\n" );
5087                 $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
5088                 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5089         }
5090
5091         /**
5092          * Callback from the Sanitizer for expanding items found in HTML attribute
5093          * values, so they can be safely tested and escaped.
5094          *
5095          * @param $text String
5096          * @param $frame PPFrame
5097          * @return String
5098          */
5099         function attributeStripCallback( &$text, $frame = false ) {
5100                 $text = $this->replaceVariables( $text, $frame );
5101                 $text = $this->mStripState->unstripBoth( $text );
5102                 return $text;
5103         }
5104
5105         /**
5106          * Accessor
5107          *
5108          * @return array
5109          */
5110         function getTags() {
5111                 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ) );
5112         }
5113
5114         /**
5115          * Replace transparent tags in $text with the values given by the callbacks.
5116          *
5117          * Transparent tag hooks are like regular XML-style tag hooks, except they
5118          * operate late in the transformation sequence, on HTML instead of wikitext.
5119          */
5120         function replaceTransparentTags( $text ) {
5121                 $matches = array();
5122                 $elements = array_keys( $this->mTransparentTagHooks );
5123                 $text = self::extractTagsAndParams( $elements, $text, $matches, $this->mUniqPrefix );
5124                 $replacements = array();
5125
5126                 foreach ( $matches as $marker => $data ) {
5127                         list( $element, $content, $params, $tag ) = $data;
5128                         $tagName = strtolower( $element );
5129                         if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5130                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
5131                         } else {
5132                                 $output = $tag;
5133                         }
5134                         $replacements[$marker] = $output;
5135                 }
5136                 return strtr( $text, $replacements );
5137         }
5138
5139         /**
5140          * Break wikitext input into sections, and either pull or replace
5141          * some particular section's text.
5142          *
5143          * External callers should use the getSection and replaceSection methods.
5144          *
5145          * @param $text String: Page wikitext
5146          * @param $section String: a section identifier string of the form:
5147          *   <flag1> - <flag2> - ... - <section number>
5148          *
5149          * Currently the only recognised flag is "T", which means the target section number
5150          * was derived during a template inclusion parse, in other words this is a template
5151          * section edit link. If no flags are given, it was an ordinary section edit link.
5152          * This flag is required to avoid a section numbering mismatch when a section is
5153          * enclosed by <includeonly> (bug 6563).
5154          *
5155          * The section number 0 pulls the text before the first heading; other numbers will
5156          * pull the given section along with its lower-level subsections. If the section is
5157          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5158          *
5159          * @param $mode String: one of "get" or "replace"
5160          * @param $newText String: replacement text for section data.
5161          * @return String: for "get", the extracted section text.
5162          *                 for "replace", the whole page with the section replaced.
5163          *                 If the page is empty and section 0 is requested, $text (as '')
5164          *                  is returned
5165          */
5166         private function extractSections( $text, $section, $mode, $newText='' ) {
5167                 global $wgTitle; # not generally used but removes an ugly failure mode
5168                 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5169                 $outText = '';
5170                 $frame = $this->getPreprocessor()->newFrame();
5171
5172                 # Process section extraction flags
5173                 $flags = 0;
5174                 $sectionParts = explode( '-', $section );
5175                 $sectionIndex = array_pop( $sectionParts );
5176                 foreach ( $sectionParts as $part ) {
5177                         if ( $part === 'T' ) {
5178                                 $flags |= self::PTD_FOR_INCLUSION;
5179                         }
5180                 }
5181                 # Preprocess the text
5182                 $root = $this->preprocessToDom( $text, $flags );
5183
5184                 # <h> nodes indicate section breaks
5185                 # They can only occur at the top level, so we can find them by iterating the root's children
5186                 $node = $root->getFirstChild();
5187
5188                 # Find the target section
5189                 if ( $sectionIndex == 0 ) {
5190                         # Section zero doesn't nest, level=big
5191                         $targetLevel = 1000;
5192                         if ( !$node ) {
5193                                 # The page definitely exists - we checked that earlier - so it must be blank: see bug #14005
5194                                 return $text;
5195                         }
5196                 } else {
5197                         while ( $node ) {
5198                                 if ( $node->getName() === 'h' ) {
5199                                         $bits = $node->splitHeading();
5200                                         if ( $bits['i'] == $sectionIndex ) {
5201                                                 $targetLevel = $bits['level'];
5202                                                 break;
5203                                         }
5204                                 }
5205                                 if ( $mode === 'replace' ) {
5206                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5207                                 }
5208                                 $node = $node->getNextSibling();
5209                         }
5210                 }
5211
5212                 if ( !$node ) {
5213                         # Not found
5214                         if ( $mode === 'get' ) {
5215                                 return $newText;
5216                         } else {
5217                                 return $text;
5218                         }
5219                 }
5220
5221                 # Find the end of the section, including nested sections
5222                 do {
5223                         if ( $node->getName() === 'h' ) {
5224                                 $bits = $node->splitHeading();
5225                                 $curLevel = $bits['level'];
5226                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5227                                         break;
5228                                 }
5229                         }
5230                         if ( $mode === 'get' ) {
5231                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5232                         }
5233                         $node = $node->getNextSibling();
5234                 } while ( $node );
5235
5236                 # Write out the remainder (in replace mode only)
5237                 if ( $mode === 'replace' ) {
5238                         # Output the replacement text
5239                         # Add two newlines on -- trailing whitespace in $newText is conventionally
5240                         # stripped by the editor, so we need both newlines to restore the paragraph gap
5241                         # Only add trailing whitespace if there is newText
5242                         if ( $newText != "" ) {
5243                                 $outText .= $newText . "\n\n";
5244                         }
5245
5246                         while ( $node ) {
5247                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5248                                 $node = $node->getNextSibling();
5249                         }
5250                 }
5251
5252                 if ( is_string( $outText ) ) {
5253                         # Re-insert stripped tags
5254                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5255                 }
5256
5257                 return $outText;
5258         }
5259
5260         /**
5261          * This function returns the text of a section, specified by a number ($section).
5262          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5263          * the first section before any such heading (section 0).
5264          *
5265          * If a section contains subsections, these are also returned.
5266          *
5267          * @param $text String: text to look in
5268          * @param $section String: section identifier
5269          * @param $deftext String: default to return if section is not found
5270          * @return string text of the requested section
5271          */
5272         public function getSection( $text, $section, $deftext='' ) {
5273                 return $this->extractSections( $text, $section, "get", $deftext );
5274         }
5275
5276         /**
5277          * This function returns $oldtext after the content of the section
5278          * specified by $section has been replaced with $text.
5279          *
5280          * @param $oldtext String: former text of the article
5281          * @param $section Numeric: section identifier
5282          * @param $text String: replacing text
5283          * @return String: modified text
5284          */
5285         public function replaceSection( $oldtext, $section, $text ) {
5286                 return $this->extractSections( $oldtext, $section, "replace", $text );
5287         }
5288
5289         /**
5290          * Get the ID of the revision we are parsing
5291          *
5292          * @return Mixed: integer or null
5293          */
5294         function getRevisionId() {
5295                 return $this->mRevisionId;
5296         }
5297
5298         /**
5299          * Get the revision object for $this->mRevisionId
5300          *
5301          * @return Revision|null either a Revision object or null
5302          */
5303         protected function getRevisionObject() {
5304                 if ( !is_null( $this->mRevisionObject ) ) {
5305                         return $this->mRevisionObject;
5306                 }
5307                 if ( is_null( $this->mRevisionId ) ) {
5308                         return null;
5309                 }
5310
5311                 $this->mRevisionObject = Revision::newFromId( $this->mRevisionId );
5312                 return $this->mRevisionObject;
5313         }
5314
5315         /**
5316          * Get the timestamp associated with the current revision, adjusted for
5317          * the default server-local timestamp
5318          */
5319         function getRevisionTimestamp() {
5320                 if ( is_null( $this->mRevisionTimestamp ) ) {
5321                         wfProfileIn( __METHOD__ );
5322
5323                         $revObject = $this->getRevisionObject();
5324                         $timestamp = $revObject ? $revObject->getTimestamp() : false;
5325
5326                         if( $timestamp !== false ) {
5327                                 global $wgContLang;
5328
5329                                 # The cryptic '' timezone parameter tells to use the site-default
5330                                 # timezone offset instead of the user settings.
5331                                 #
5332                                 # Since this value will be saved into the parser cache, served
5333                                 # to other users, and potentially even used inside links and such,
5334                                 # it needs to be consistent for all visitors.
5335                                 $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5336                         }
5337
5338                         wfProfileOut( __METHOD__ );
5339                 }
5340                 return $this->mRevisionTimestamp;
5341         }
5342
5343         /**
5344          * Get the name of the user that edited the last revision
5345          *
5346          * @return String: user name
5347          */
5348         function getRevisionUser() {
5349                 if( is_null( $this->mRevisionUser ) ) {
5350                         $revObject = $this->getRevisionObject();
5351
5352                         # if this template is subst: the revision id will be blank,
5353                         # so just use the current user's name
5354                         if( $revObject ) {
5355                                 $this->mRevisionUser = $revObject->getUserText();
5356                         } elseif( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5357                                 $this->mRevisionUser = $this->getUser()->getName();
5358                         }
5359                 }
5360                 return $this->mRevisionUser;
5361         }
5362
5363         /**
5364          * Mutator for $mDefaultSort
5365          *
5366          * @param $sort New value
5367          */
5368         public function setDefaultSort( $sort ) {
5369                 $this->mDefaultSort = $sort;
5370                 $this->mOutput->setProperty( 'defaultsort', $sort );
5371         }
5372
5373         /**
5374          * Accessor for $mDefaultSort
5375          * Will use the empty string if none is set.
5376          *
5377          * This value is treated as a prefix, so the
5378          * empty string is equivalent to sorting by
5379          * page name.
5380          *
5381          * @return string
5382          */
5383         public function getDefaultSort() {
5384                 if ( $this->mDefaultSort !== false ) {
5385                         return $this->mDefaultSort;
5386                 } else {
5387                         return '';
5388                 }
5389         }
5390
5391         /**
5392          * Accessor for $mDefaultSort
5393          * Unlike getDefaultSort(), will return false if none is set
5394          *
5395          * @return string or false
5396          */
5397         public function getCustomDefaultSort() {
5398                 return $this->mDefaultSort;
5399         }
5400
5401         /**
5402          * Try to guess the section anchor name based on a wikitext fragment
5403          * presumably extracted from a heading, for example "Header" from
5404          * "== Header ==".
5405          */
5406         public function guessSectionNameFromWikiText( $text ) {
5407                 # Strip out wikitext links(they break the anchor)
5408                 $text = $this->stripSectionName( $text );
5409                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5410                 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5411         }
5412
5413         /**
5414          * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5415          * instead.  For use in redirects, since IE6 interprets Redirect: headers
5416          * as something other than UTF-8 (apparently?), resulting in breakage.
5417          *
5418          * @param $text String: The section name
5419          * @return string An anchor
5420          */
5421         public function guessLegacySectionNameFromWikiText( $text ) {
5422                 # Strip out wikitext links(they break the anchor)
5423                 $text = $this->stripSectionName( $text );
5424                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5425                 return '#' . Sanitizer::escapeId( $text, array( 'noninitial', 'legacy' ) );
5426         }
5427
5428         /**
5429          * Strips a text string of wikitext for use in a section anchor
5430          *
5431          * Accepts a text string and then removes all wikitext from the
5432          * string and leaves only the resultant text (i.e. the result of
5433          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5434          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5435          * to create valid section anchors by mimicing the output of the
5436          * parser when headings are parsed.
5437          *
5438          * @param $text String: text string to be stripped of wikitext
5439          * for use in a Section anchor
5440          * @return Filtered text string
5441          */
5442         public function stripSectionName( $text ) {
5443                 # Strip internal link markup
5444                 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5445                 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5446
5447                 # Strip external link markup
5448                 # @todo FIXME: Not tolerant to blank link text
5449                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5450                 # on how many empty links there are on the page - need to figure that out.
5451                 $text = preg_replace( '/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5452
5453                 # Parse wikitext quotes (italics & bold)
5454                 $text = $this->doQuotes( $text );
5455
5456                 # Strip HTML tags
5457                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5458                 return $text;
5459         }
5460
5461         /**
5462          * strip/replaceVariables/unstrip for preprocessor regression testing
5463          *
5464          * @return string
5465          */
5466         function testSrvus( $text, Title $title, ParserOptions $options, $outputType = self::OT_HTML ) {
5467                 $this->startParse( $title, $options, $outputType, true );
5468
5469                 $text = $this->replaceVariables( $text );
5470                 $text = $this->mStripState->unstripBoth( $text );
5471                 $text = Sanitizer::removeHTMLtags( $text );
5472                 return $text;
5473         }
5474
5475         function testPst( $text, Title $title, ParserOptions $options ) {
5476                 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
5477         }
5478
5479         function testPreprocess( $text, Title $title, ParserOptions $options ) {
5480                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5481         }
5482
5483         /**
5484          * Call a callback function on all regions of the given text that are not
5485          * inside strip markers, and replace those regions with the return value
5486          * of the callback. For example, with input:
5487          *
5488          *  aaa<MARKER>bbb
5489          *
5490          * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5491          * two strings will be replaced with the value returned by the callback in
5492          * each case.
5493          *
5494          * @return string
5495          */
5496         function markerSkipCallback( $s, $callback ) {
5497                 $i = 0;
5498                 $out = '';
5499                 while ( $i < strlen( $s ) ) {
5500                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5501                         if ( $markerStart === false ) {
5502                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5503                                 break;
5504                         } else {
5505                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5506                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5507                                 if ( $markerEnd === false ) {
5508                                         $out .= substr( $s, $markerStart );
5509                                         break;
5510                                 } else {
5511                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5512                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5513                                         $i = $markerEnd;
5514                                 }
5515                         }
5516                 }
5517                 return $out;
5518         }
5519
5520         /**
5521          * Save the parser state required to convert the given half-parsed text to
5522          * HTML. "Half-parsed" in this context means the output of
5523          * recursiveTagParse() or internalParse(). This output has strip markers
5524          * from replaceVariables (extensionSubstitution() etc.), and link
5525          * placeholders from replaceLinkHolders().
5526          *
5527          * Returns an array which can be serialized and stored persistently. This
5528          * array can later be loaded into another parser instance with
5529          * unserializeHalfParsedText(). The text can then be safely incorporated into
5530          * the return value of a parser hook.
5531          *
5532          * @return array
5533          */
5534         function serializeHalfParsedText( $text ) {
5535                 wfProfileIn( __METHOD__ );
5536                 $data = array(
5537                         'text' => $text,
5538                         'version' => self::HALF_PARSED_VERSION,
5539                         'stripState' => $this->mStripState->getSubState( $text ),
5540                         'linkHolders' => $this->mLinkHolders->getSubArray( $text )
5541                 );
5542                 wfProfileOut( __METHOD__ );
5543                 return $data;
5544         }
5545
5546         /**
5547          * Load the parser state given in the $data array, which is assumed to
5548          * have been generated by serializeHalfParsedText(). The text contents is
5549          * extracted from the array, and its markers are transformed into markers
5550          * appropriate for the current Parser instance. This transformed text is
5551          * returned, and can be safely included in the return value of a parser
5552          * hook.
5553          *
5554          * If the $data array has been stored persistently, the caller should first
5555          * check whether it is still valid, by calling isValidHalfParsedText().
5556          *
5557          * @param $data Serialized data
5558          * @return String
5559          */
5560         function unserializeHalfParsedText( $data ) {
5561                 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
5562                         throw new MWException( __METHOD__.': invalid version' );
5563                 }
5564
5565                 # First, extract the strip state.
5566                 $texts = array( $data['text'] );
5567                 $texts = $this->mStripState->merge( $data['stripState'], $texts );
5568
5569                 # Now renumber links
5570                 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
5571
5572                 # Should be good to go.
5573                 return $texts[0];
5574         }
5575
5576         /**
5577          * Returns true if the given array, presumed to be generated by
5578          * serializeHalfParsedText(), is compatible with the current version of the
5579          * parser.
5580          *
5581          * @param $data Array
5582          *
5583          * @return bool
5584          */
5585         function isValidHalfParsedText( $data ) {
5586                 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
5587         }
5588 }