includes/parser/Parser.php

   1 <?php
   2 /**
   3  * PHP parser that converts wiki markup to HTML.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup Parser
  22  */
  23
  24 /**
  25  * @defgroup Parser Parser
  26  */
  27
  28 /**
  29  * PHP Parser - Processes wiki markup (which uses a more user-friendly
  30  * syntax, such as "[[link]]" for making links), and provides a one-way
  31  * transformation of that wiki markup it into (X)HTML output / markup
  32  * (which in turn the browser understands, and can display).
  33  *
  34  * There are seven main entry points into the Parser class:
  35  *
  36  * - Parser::parse()
  37  *     produces HTML output
  38  * - Parser::preSaveTransform().
  39  *     produces altered wiki markup.
  40  * - Parser::preprocess()
  41  *     removes HTML comments and expands templates
  42  * - Parser::cleanSig() and Parser::cleanSigInSig()
  43  *     Cleans a signature before saving it to preferences
  44  * - Parser::getSection()
  45  *     Return the content of a section from an article for section editing
  46  * - Parser::replaceSection()
  47  *     Replaces a section by number inside an article
  48  * - Parser::getPreloadText()
  49  *     Removes <noinclude> sections, and <includeonly> tags.
  50  *
  51  * Globals used:
  52  *    object: $wgContLang
  53  *
  54  * @warning $wgUser or $wgTitle or $wgRequest or $wgLang. Keep them away!
  55  *
  56  * @par Settings:
  57  * $wgNamespacesWithSubpages
  58  *
  59  * @par Settings only within ParserOptions:
  60  * $wgAllowExternalImages
  61  * $wgAllowSpecialInclusion
  62  * $wgInterwikiMagic
  63  * $wgMaxArticleSize
  64  *
  65  * @ingroup Parser
  66  */
  67 class Parser {
  68         /**
  69          * Update this version number when the ParserOutput format
  70          * changes in an incompatible way, so the parser cache
  71          * can automatically discard old data.
  72          */
  73         const VERSION = '1.6.4';
  74
  75         /**
  76          * Update this version number when the output of serialiseHalfParsedText()
  77          * changes in an incompatible way
  78          */
  79         const HALF_PARSED_VERSION = 2;
  80
  81         # Flags for Parser::setFunctionHook
  82         # Also available as global constants from Defines.php
  83         const SFH_NO_HASH = 1;
  84         const SFH_OBJECT_ARGS = 2;
  85
  86         # Constants needed for external link processing
  87         # Everything except bracket, space, or control characters
  88         # \p{Zs} is unicode 'separator, space' category. It covers the space 0x20
  89         # as well as U+3000 is IDEOGRAPHIC SPACE for bug 19052
  90         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F\p{Zs}]';
  91         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F\p{Zs}]+)
  92                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sxu';
  93
  94         # State constants for the definition list colon extraction
  95         const COLON_STATE_TEXT = 0;
  96         const COLON_STATE_TAG = 1;
  97         const COLON_STATE_TAGSTART = 2;
  98         const COLON_STATE_CLOSETAG = 3;
  99         const COLON_STATE_TAGSLASH = 4;
 100         const COLON_STATE_COMMENT = 5;
 101         const COLON_STATE_COMMENTDASH = 6;
 102         const COLON_STATE_COMMENTDASHDASH = 7;
 103
 104         # Flags for preprocessToDom
 105         const PTD_FOR_INCLUSION = 1;
 106
 107         # Allowed values for $this->mOutputType
 108         # Parameter to startExternalParse().
 109         const OT_HTML = 1; # like parse()
 110         const OT_WIKI = 2; # like preSaveTransform()
 111         const OT_PREPROCESS = 3; # like preprocess()
 112         const OT_MSG = 3;
 113         const OT_PLAIN = 4; # like extractSections() - portions of the original are returned unchanged.
 114
 115         # Marker Suffix needs to be accessible staticly.
 116         const MARKER_SUFFIX = "-QINU\x7f";
 117
 118         # Persistent:
 119         var $mTagHooks = array();
 120         var $mTransparentTagHooks = array();
 121         var $mFunctionHooks = array();
 122         var $mFunctionSynonyms = array( 0 => array(), 1 => array() );
 123         var $mFunctionTagHooks = array();
 124         var $mStripList = array();
 125         var $mDefaultStripList = array();
 126         var $mVarCache = array();
 127         var $mImageParams = array();
 128         var $mImageParamsMagicArray = array();
 129         var $mMarkerIndex = 0;
 130         var $mFirstCall = true;
 131
 132         # Initialised by initialiseVariables()
 133
 134         /**
 135          * @var MagicWordArray
 136          */
 137         var $mVariables;
 138
 139         /**
 140          * @var MagicWordArray
 141          */
 142         var $mSubstWords;
 143         var $mConf, $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols; # Initialised in constructor
 144
 145         # Cleared with clearState():
 146         /**
 147          * @var ParserOutput
 148          */
 149         var $mOutput;
 150         var $mAutonumber, $mDTopen;
 151
 152         /**
 153          * @var StripState
 154          */
 155         var $mStripState;
 156
 157         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 158         /**
 159          * @var LinkHolderArray
 160          */
 161         var $mLinkHolders;
 162
 163         var $mLinkID;
 164         var $mIncludeSizes, $mPPNodeCount, $mGeneratedPPNodeCount, $mHighestExpansionDepth;
 165         var $mDefaultSort;
 166         var $mTplExpandCache; # empty-frame expansion cache
 167         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 168         var $mExpensiveFunctionCount; # number of expensive parser function calls
 169         var $mShowToc, $mForceTocPosition;
 170
 171         /**
 172          * @var User
 173          */
 174         var $mUser; # User object; only used when doing pre-save transform
 175
 176         # Temporary
 177         # These are variables reset at least once per parse regardless of $clearState
 178
 179         /**
 180          * @var ParserOptions
 181          */
 182         var $mOptions;
 183
 184         /**
 185          * @var Title
 186          */
 187         var $mTitle;        # Title context, used for self-link rendering and similar things
 188         var $mOutputType;   # Output type, one of the OT_xxx constants
 189         var $ot;            # Shortcut alias, see setOutputType()
 190         var $mRevisionObject; # The revision object of the specified revision ID
 191         var $mRevisionId;   # ID to display in {{REVISIONID}} tags
 192         var $mRevisionTimestamp; # The timestamp of the specified revision ID
 193         var $mRevisionUser; # User to display in {{REVISIONUSER}} tag
 194         var $mRevIdForTs;   # The revision ID which was used to fetch the timestamp
 195         var $mInputSize = false; # For {{PAGESIZE}} on current page.
 196
 197         /**
 198          * @var string
 199          */
 200         var $mUniqPrefix;
 201
 202         /**
 203          * @var Array with the language name of each language link (i.e. the
 204          * interwiki prefix) in the key, value arbitrary. Used to avoid sending
 205          * duplicate language links to the ParserOutput.
 206          */
 207         var $mLangLinkLanguages;
 208
 209         /**
 210          * Constructor
 211          *
 212          * @param $conf array
 213          */
 214         public function __construct( $conf = array() ) {
 215                 $this->mConf = $conf;
 216                 $this->mUrlProtocols = wfUrlProtocols();
 217                 $this->mExtLinkBracketedRegex = '/\[(((?i)' . $this->mUrlProtocols . ')' .
 218                         self::EXT_LINK_URL_CLASS . '+)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/Su';
 219                 if ( isset( $conf['preprocessorClass'] ) ) {
 220                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 221                 } elseif ( defined( 'HPHP_VERSION' ) ) {
 222                         # Preprocessor_Hash is much faster than Preprocessor_DOM under HipHop
 223                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 224                 } elseif ( extension_loaded( 'domxml' ) ) {
 225                         # PECL extension that conflicts with the core DOM extension (bug 13770)
 226                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 227                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 228                 } elseif ( extension_loaded( 'dom' ) ) {
 229                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 230                 } else {
 231                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 232                 }
 233                 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
 234         }
 235
 236         /**
 237          * Reduce memory usage to reduce the impact of circular references
 238          */
 239         function __destruct() {
 240                 if ( isset( $this->mLinkHolders ) ) {
 241                         unset( $this->mLinkHolders );
 242                 }
 243                 foreach ( $this as $name => $value ) {
 244                         unset( $this->$name );
 245                 }
 246         }
 247
 248         /**
 249          * Allow extensions to clean up when the parser is cloned
 250          */
 251         function __clone() {
 252                 wfRunHooks( 'ParserCloned', array( $this ) );
 253         }
 254
 255         /**
 256          * Do various kinds of initialisation on the first call of the parser
 257          */
 258         function firstCallInit() {
 259                 if ( !$this->mFirstCall ) {
 260                         return;
 261                 }
 262                 $this->mFirstCall = false;
 263
 264                 wfProfileIn( __METHOD__ );
 265
 266                 CoreParserFunctions::register( $this );
 267                 CoreTagHooks::register( $this );
 268                 $this->initialiseVariables();
 269
 270                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 271                 wfProfileOut( __METHOD__ );
 272         }
 273
 274         /**
 275          * Clear Parser state
 276          *
 277          * @private
 278          */
 279         function clearState() {
 280                 wfProfileIn( __METHOD__ );
 281                 if ( $this->mFirstCall ) {
 282                         $this->firstCallInit();
 283                 }
 284                 $this->mOutput = new ParserOutput;
 285                 $this->mOptions->registerWatcher( array( $this->mOutput, 'recordOption' ) );
 286                 $this->mAutonumber = 0;
 287                 $this->mLastSection = '';
 288                 $this->mDTopen = false;
 289                 $this->mIncludeCount = array();
 290                 $this->mArgStack = false;
 291                 $this->mInPre = false;
 292                 $this->mLinkHolders = new LinkHolderArray( $this );
 293                 $this->mLinkID = 0;
 294                 $this->mRevisionObject = $this->mRevisionTimestamp =
 295                         $this->mRevisionId = $this->mRevisionUser = null;
 296                 $this->mVarCache = array();
 297                 $this->mUser = null;
 298                 $this->mLangLinkLanguages = array();
 299
 300                 /**
 301                  * Prefix for temporary replacement strings for the multipass parser.
 302                  * \x07 should never appear in input as it's disallowed in XML.
 303                  * Using it at the front also gives us a little extra robustness
 304                  * since it shouldn't match when butted up against identifier-like
 305                  * string constructs.
 306                  *
 307                  * Must not consist of all title characters, or else it will change
 308                  * the behavior of <nowiki> in a link.
 309                  */
 310                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 311                 $this->mStripState = new StripState( $this->mUniqPrefix );
 312
 313                 # Clear these on every parse, bug 4549
 314                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 315
 316                 $this->mShowToc = true;
 317                 $this->mForceTocPosition = false;
 318                 $this->mIncludeSizes = array(
 319                         'post-expand' => 0,
 320                         'arg' => 0,
 321                 );
 322                 $this->mPPNodeCount = 0;
 323                 $this->mGeneratedPPNodeCount = 0;
 324                 $this->mHighestExpansionDepth = 0;
 325                 $this->mDefaultSort = false;
 326                 $this->mHeadings = array();
 327                 $this->mDoubleUnderscores = array();
 328                 $this->mExpensiveFunctionCount = 0;
 329
 330                 # Fix cloning
 331                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 332                         $this->mPreprocessor = null;
 333                 }
 334
 335                 wfRunHooks( 'ParserClearState', array( &$this ) );
 336                 wfProfileOut( __METHOD__ );
 337         }
 338
 339         /**
 340          * Convert wikitext to HTML
 341          * Do not call this function recursively.
 342          *
 343          * @param string $text text we want to parse
 344          * @param $title Title object
 345          * @param $options ParserOptions
 346          * @param $linestart boolean
 347          * @param $clearState boolean
 348          * @param int $revid number to pass in {{REVISIONID}}
 349          * @return ParserOutput a ParserOutput
 350          */
 351         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 352                 /**
 353                  * First pass--just handle <nowiki> sections, pass the rest off
 354                  * to internalParse() which does all the real work.
 355                  */
 356
 357                 global $wgUseTidy, $wgAlwaysUseTidy;
 358                 $fname = __METHOD__ . '-' . wfGetCaller();
 359                 wfProfileIn( __METHOD__ );
 360                 wfProfileIn( $fname );
 361
 362                 $this->startParse( $title, $options, self::OT_HTML, $clearState );
 363
 364                 $this->mInputSize = strlen( $text );
 365                 if ( $this->mOptions->getEnableLimitReport() ) {
 366                         $this->mOutput->resetParseStartTime();
 367                 }
 368
 369                 # Remove the strip marker tag prefix from the input, if present.
 370                 if ( $clearState ) {
 371                         $text = str_replace( $this->mUniqPrefix, '', $text );
 372                 }
 373
 374                 $oldRevisionId = $this->mRevisionId;
 375                 $oldRevisionObject = $this->mRevisionObject;
 376                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 377                 $oldRevisionUser = $this->mRevisionUser;
 378                 if ( $revid !== null ) {
 379                         $this->mRevisionId = $revid;
 380                         $this->mRevisionObject = null;
 381                         $this->mRevisionTimestamp = null;
 382                         $this->mRevisionUser = null;
 383                 }
 384
 385                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 386                 # No more strip!
 387                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 388                 $text = $this->internalParse( $text );
 389                 wfRunHooks( 'ParserAfterParse', array( &$this, &$text, &$this->mStripState ) );
 390
 391                 $text = $this->mStripState->unstripGeneral( $text );
 392
 393                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 394                 $fixtags = array(
 395                         # french spaces, last one Guillemet-left
 396                         # only if there is something before the space
 397                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;',
 398                         # french spaces, Guillemet-right
 399                         '/(\\302\\253) /' => '\\1&#160;',
 400                         '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
 401                 );
 402                 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
 403
 404                 $text = $this->doBlockLevels( $text, $linestart );
 405
 406                 $this->replaceLinkHolders( $text );
 407
 408                 /**
 409                  * The input doesn't get language converted if
 410                  * a) It's disabled
 411                  * b) Content isn't converted
 412                  * c) It's a conversion table
 413                  * d) it is an interface message (which is in the user language)
 414                  */
 415                 if ( !( $options->getDisableContentConversion()
 416                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] ) ) )
 417                 {
 418                         if ( !$this->mOptions->getInterfaceMessage() ) {
 419                                 # The position of the convert() call should not be changed. it
 420                                 # assumes that the links are all replaced and the only thing left
 421                                 # is the <nowiki> mark.
 422                                 $text = $this->getConverterLanguage()->convert( $text );
 423                         }
 424                 }
 425
 426                 /**
 427                  * A converted title will be provided in the output object if title and
 428                  * content conversion are enabled, the article text does not contain
 429                  * a conversion-suppressing double-underscore tag, and no
 430                  * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
 431                  * automatic link conversion.
 432                  */
 433                 if ( !( $options->getDisableTitleConversion()
 434                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 435                                 || isset( $this->mDoubleUnderscores['notitleconvert'] )
 436                                 || $this->mOutput->getDisplayTitle() !== false ) )
 437                 {
 438                         $convruletitle = $this->getConverterLanguage()->getConvRuleTitle();
 439                         if ( $convruletitle ) {
 440                                 $this->mOutput->setTitleText( $convruletitle );
 441                         } else {
 442                                 $titleText = $this->getConverterLanguage()->convertTitle( $title );
 443                                 $this->mOutput->setTitleText( $titleText );
 444                         }
 445                 }
 446
 447                 $text = $this->mStripState->unstripNoWiki( $text );
 448
 449                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 450
 451                 $text = $this->replaceTransparentTags( $text );
 452                 $text = $this->mStripState->unstripGeneral( $text );
 453
 454                 $text = Sanitizer::normalizeCharReferences( $text );
 455
 456                 if ( ( $wgUseTidy && $this->mOptions->getTidy() ) || $wgAlwaysUseTidy ) {
 457                         $text = MWTidy::tidy( $text );
 458                 } else {
 459                         # attempt to sanitize at least some nesting problems
 460                         # (bug #2702 and quite a few others)
 461                         $tidyregs = array(
 462                                 # ''Something [http://www.cool.com cool''] -->
 463                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 464                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 465                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 466                                 # fix up an anchor inside another anchor, only
 467                                 # at least for a single single nested link (bug 3695)
 468                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 469                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 470                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 471                                 # contains a div, so fix it up here; replace
 472                                 # div with escaped text
 473                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 474                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 475                                 # remove empty italic or bold tag pairs, some
 476                                 # introduced by rules above
 477                                 '/<([bi])><\/\\1>/' => '',
 478                         );
 479
 480                         $text = preg_replace(
 481                                 array_keys( $tidyregs ),
 482                                 array_values( $tidyregs ),
 483                                 $text );
 484                 }
 485
 486                 if ( $this->mExpensiveFunctionCount > $this->mOptions->getExpensiveParserFunctionLimit() ) {
 487                         $this->limitationWarn( 'expensive-parserfunction',
 488                                 $this->mExpensiveFunctionCount,
 489                                 $this->mOptions->getExpensiveParserFunctionLimit()
 490                         );
 491                 }
 492
 493                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 494
 495                 # Information on include size limits, for the benefit of users who try to skirt them
 496                 if ( $this->mOptions->getEnableLimitReport() ) {
 497                         $max = $this->mOptions->getMaxIncludeSize();
 498
 499                         $cpuTime = $this->mOutput->getTimeSinceStart( 'cpu' );
 500                         if ( $cpuTime !== null ) {
 501                                 $this->mOutput->setLimitReportData( 'limitreport-cputime',
 502                                         sprintf( "%.3f", $cpuTime )
 503                                 );
 504                         }
 505
 506                         $wallTime = $this->mOutput->getTimeSinceStart( 'wall' );
 507                         $this->mOutput->setLimitReportData( 'limitreport-walltime',
 508                                 sprintf( "%.3f", $wallTime )
 509                         );
 510
 511                         $this->mOutput->setLimitReportData( 'limitreport-ppvisitednodes',
 512                                 array( $this->mPPNodeCount, $this->mOptions->getMaxPPNodeCount() )
 513                         );
 514                         $this->mOutput->setLimitReportData( 'limitreport-ppgeneratednodes',
 515                                 array( $this->mGeneratedPPNodeCount, $this->mOptions->getMaxGeneratedPPNodeCount() )
 516                         );
 517                         $this->mOutput->setLimitReportData( 'limitreport-postexpandincludesize',
 518                                 array( $this->mIncludeSizes['post-expand'], $max )
 519                         );
 520                         $this->mOutput->setLimitReportData( 'limitreport-templateargumentsize',
 521                                 array( $this->mIncludeSizes['arg'], $max )
 522                         );
 523                         $this->mOutput->setLimitReportData( 'limitreport-expansiondepth',
 524                                 array( $this->mHighestExpansionDepth, $this->mOptions->getMaxPPExpandDepth() )
 525                         );
 526                         $this->mOutput->setLimitReportData( 'limitreport-expensivefunctioncount',
 527                                 array( $this->mExpensiveFunctionCount, $this->mOptions->getExpensiveParserFunctionLimit() )
 528                         );
 529                         wfRunHooks( 'ParserLimitReportPrepare', array( $this, $this->mOutput ) );
 530
 531                         $limitReport = "NewPP limit report\n";
 532                         foreach ( $this->mOutput->getLimitReportData() as $key => $value ) {
 533                                 if ( wfRunHooks( 'ParserLimitReportFormat',
 534                                         array( $key, $value, &$limitReport, false, false )
 535                                 ) ) {
 536                                         $keyMsg = wfMessage( $key )->inLanguage( 'en' )->useDatabase( false );
 537                                         $valueMsg = wfMessage( array( "$key-value-text", "$key-value" ) )
 538                                                 ->inLanguage( 'en' )->useDatabase( false );
 539                                         if ( !$valueMsg->exists() ) {
 540                                                 $valueMsg = new RawMessage( '$1' );
 541                                         }
 542                                         if ( !$keyMsg->isDisabled() && !$valueMsg->isDisabled() ) {
 543                                                 $valueMsg->params( $value );
 544                                                 $limitReport .= "{$keyMsg->text()}: {$valueMsg->text()}\n";
 545                                         }
 546                                 }
 547                         }
 548                         // Since we're not really outputting HTML, decode the entities and
 549                         // then re-encode the things that need hiding inside HTML comments.
 550                         $limitReport = htmlspecialchars_decode( $limitReport );
 551                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 552
 553                         // Sanitize for comment. Note '‐' in the replacement is U+2010,
 554                         // which looks much like the problematic '-'.
 555                         $limitReport = str_replace( array( '-', '&' ), array( '‐', '&amp;' ), $limitReport );
 556                         $text .= "\n<!-- \n$limitReport-->\n";
 557
 558                         if ( $this->mGeneratedPPNodeCount > $this->mOptions->getMaxGeneratedPPNodeCount() / 10 ) {
 559                                 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount . ' ' .
 560                                         $this->mTitle->getPrefixedDBkey() );
 561                         }
 562                 }
 563                 $this->mOutput->setText( $text );
 564
 565                 $this->mRevisionId = $oldRevisionId;
 566                 $this->mRevisionObject = $oldRevisionObject;
 567                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 568                 $this->mRevisionUser = $oldRevisionUser;
 569                 $this->mInputSize = false;
 570                 wfProfileOut( $fname );
 571                 wfProfileOut( __METHOD__ );
 572
 573                 return $this->mOutput;
 574         }
 575
 576         /**
 577          * Recursive parser entry point that can be called from an extension tag
 578          * hook.
 579          *
 580          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 581          *
 582          * @param string $text text extension wants to have parsed
 583          * @param $frame PPFrame: The frame to use for expanding any template variables
 584          *
 585          * @return string
 586          */
 587         function recursiveTagParse( $text, $frame = false ) {
 588                 wfProfileIn( __METHOD__ );
 589                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 590                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 591                 $text = $this->internalParse( $text, false, $frame );
 592                 wfProfileOut( __METHOD__ );
 593                 return $text;
 594         }
 595
 596         /**
 597          * Expand templates and variables in the text, producing valid, static wikitext.
 598          * Also removes comments.
 599          * @return mixed|string
 600          */
 601         function preprocess( $text, Title $title = null, ParserOptions $options, $revid = null ) {
 602                 wfProfileIn( __METHOD__ );
 603                 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
 604                 if ( $revid !== null ) {
 605                         $this->mRevisionId = $revid;
 606                 }
 607                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 608                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 609                 $text = $this->replaceVariables( $text );
 610                 $text = $this->mStripState->unstripBoth( $text );
 611                 wfProfileOut( __METHOD__ );
 612                 return $text;
 613         }
 614
 615         /**
 616          * Recursive parser entry point that can be called from an extension tag
 617          * hook.
 618          *
 619          * @param string $text text to be expanded
 620          * @param $frame PPFrame: The frame to use for expanding any template variables
 621          * @return String
 622          * @since 1.19
 623          */
 624         public function recursivePreprocess( $text, $frame = false ) {
 625                 wfProfileIn( __METHOD__ );
 626                 $text = $this->replaceVariables( $text, $frame );
 627                 $text = $this->mStripState->unstripBoth( $text );
 628                 wfProfileOut( __METHOD__ );
 629                 return $text;
 630         }
 631
 632         /**
 633          * Process the wikitext for the "?preload=" feature. (bug 5210)
 634          *
 635          * "<noinclude>", "<includeonly>" etc. are parsed as for template
 636          * transclusion, comments, templates, arguments, tags hooks and parser
 637          * functions are untouched.
 638          *
 639          * @param $text String
 640          * @param $title Title
 641          * @param $options ParserOptions
 642          * @return String
 643          */
 644         public function getPreloadText( $text, Title $title, ParserOptions $options ) {
 645                 # Parser (re)initialisation
 646                 $this->startParse( $title, $options, self::OT_PLAIN, true );
 647
 648                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 649                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
 650                 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
 651                 $text = $this->mStripState->unstripBoth( $text );
 652                 return $text;
 653         }
 654
 655         /**
 656          * Get a random string
 657          *
 658          * @return string
 659          */
 660         public static function getRandomString() {
 661                 return wfRandomString( 16 );
 662         }
 663
 664         /**
 665          * Set the current user.
 666          * Should only be used when doing pre-save transform.
 667          *
 668          * @param $user Mixed: User object or null (to reset)
 669          */
 670         function setUser( $user ) {
 671                 $this->mUser = $user;
 672         }
 673
 674         /**
 675          * Accessor for mUniqPrefix.
 676          *
 677          * @return String
 678          */
 679         public function uniqPrefix() {
 680                 if ( !isset( $this->mUniqPrefix ) ) {
 681                         # @todo FIXME: This is probably *horribly wrong*
 682                         # LanguageConverter seems to want $wgParser's uniqPrefix, however
 683                         # if this is called for a parser cache hit, the parser may not
 684                         # have ever been initialized in the first place.
 685                         # Not really sure what the heck is supposed to be going on here.
 686                         return '';
 687                         # throw new MWException( "Accessing uninitialized mUniqPrefix" );
 688                 }
 689                 return $this->mUniqPrefix;
 690         }
 691
 692         /**
 693          * Set the context title
 694          *
 695          * @param $t Title
 696          */
 697         function setTitle( $t ) {
 698                 if ( !$t || $t instanceof FakeTitle ) {
 699                         $t = Title::newFromText( 'NO TITLE' );
 700                 }
 701
 702                 if ( strval( $t->getFragment() ) !== '' ) {
 703                         # Strip the fragment to avoid various odd effects
 704                         $this->mTitle = clone $t;
 705                         $this->mTitle->setFragment( '' );
 706                 } else {
 707                         $this->mTitle = $t;
 708                 }
 709         }
 710
 711         /**
 712          * Accessor for the Title object
 713          *
 714          * @return Title object
 715          */
 716         function getTitle() {
 717                 return $this->mTitle;
 718         }
 719
 720         /**
 721          * Accessor/mutator for the Title object
 722          *
 723          * @param $x Title object or null to just get the current one
 724          * @return Title object
 725          */
 726         function Title( $x = null ) {
 727                 return wfSetVar( $this->mTitle, $x );
 728         }
 729
 730         /**
 731          * Set the output type
 732          *
 733          * @param $ot Integer: new value
 734          */
 735         function setOutputType( $ot ) {
 736                 $this->mOutputType = $ot;
 737                 # Shortcut alias
 738                 $this->ot = array(
 739                         'html' => $ot == self::OT_HTML,
 740                         'wiki' => $ot == self::OT_WIKI,
 741                         'pre' => $ot == self::OT_PREPROCESS,
 742                         'plain' => $ot == self::OT_PLAIN,
 743                 );
 744         }
 745
 746         /**
 747          * Accessor/mutator for the output type
 748          *
 749          * @param int|null $x New value or null to just get the current one
 750          * @return Integer
 751          */
 752         function OutputType( $x = null ) {
 753                 return wfSetVar( $this->mOutputType, $x );
 754         }
 755
 756         /**
 757          * Get the ParserOutput object
 758          *
 759          * @return ParserOutput object
 760          */
 761         function getOutput() {
 762                 return $this->mOutput;
 763         }
 764
 765         /**
 766          * Get the ParserOptions object
 767          *
 768          * @return ParserOptions object
 769          */
 770         function getOptions() {
 771                 return $this->mOptions;
 772         }
 773
 774         /**
 775          * Accessor/mutator for the ParserOptions object
 776          *
 777          * @param $x ParserOptions New value or null to just get the current one
 778          * @return ParserOptions Current ParserOptions object
 779          */
 780         function Options( $x = null ) {
 781                 return wfSetVar( $this->mOptions, $x );
 782         }
 783
 784         /**
 785          * @return int
 786          */
 787         function nextLinkID() {
 788                 return $this->mLinkID++;
 789         }
 790
 791         /**
 792          * @param $id int
 793          */
 794         function setLinkID( $id ) {
 795                 $this->mLinkID = $id;
 796         }
 797
 798         /**
 799          * Get a language object for use in parser functions such as {{FORMATNUM:}}
 800          * @return Language
 801          */
 802         function getFunctionLang() {
 803                 return $this->getTargetLanguage();
 804         }
 805
 806         /**
 807          * Get the target language for the content being parsed. This is usually the
 808          * language that the content is in.
 809          *
 810          * @since 1.19
 811          *
 812          * @throws MWException
 813          * @return Language|null
 814          */
 815         public function getTargetLanguage() {
 816                 $target = $this->mOptions->getTargetLanguage();
 817
 818                 if ( $target !== null ) {
 819                         return $target;
 820                 } elseif ( $this->mOptions->getInterfaceMessage() ) {
 821                         return $this->mOptions->getUserLangObj();
 822                 } elseif ( is_null( $this->mTitle ) ) {
 823                         throw new MWException( __METHOD__ . ': $this->mTitle is null' );
 824                 }
 825
 826                 return $this->mTitle->getPageLanguage();
 827         }
 828
 829         /**
 830          * Get the language object for language conversion
 831          */
 832         function getConverterLanguage() {
 833                 return $this->getTargetLanguage();
 834         }
 835
 836         /**
 837          * Get a User object either from $this->mUser, if set, or from the
 838          * ParserOptions object otherwise
 839          *
 840          * @return User object
 841          */
 842         function getUser() {
 843                 if ( !is_null( $this->mUser ) ) {
 844                         return $this->mUser;
 845                 }
 846                 return $this->mOptions->getUser();
 847         }
 848
 849         /**
 850          * Get a preprocessor object
 851          *
 852          * @return Preprocessor instance
 853          */
 854         function getPreprocessor() {
 855                 if ( !isset( $this->mPreprocessor ) ) {
 856                         $class = $this->mPreprocessorClass;
 857                         $this->mPreprocessor = new $class( $this );
 858                 }
 859                 return $this->mPreprocessor;
 860         }
 861
 862         /**
 863          * Replaces all occurrences of HTML-style comments and the given tags
 864          * in the text with a random marker and returns the next text. The output
 865          * parameter $matches will be an associative array filled with data in
 866          * the form:
 867          *
 868          * @code
 869          *   'UNIQ-xxxxx' => array(
 870          *     'element',
 871          *     'tag content',
 872          *     array( 'param' => 'x' ),
 873          *     '<element param="x">tag content</element>' ) )
 874          * @endcode
 875          *
 876          * @param array $elements list of element names. Comments are always extracted.
 877          * @param string $text Source text string.
 878          * @param array $matches Out parameter, Array: extracted tags
 879          * @param $uniq_prefix string
 880          * @return String: stripped text
 881          */
 882         public static function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
 883                 static $n = 1;
 884                 $stripped = '';
 885                 $matches = array();
 886
 887                 $taglist = implode( '|', $elements );
 888                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 889
 890                 while ( $text != '' ) {
 891                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 892                         $stripped .= $p[0];
 893                         if ( count( $p ) < 5 ) {
 894                                 break;
 895                         }
 896                         if ( count( $p ) > 5 ) {
 897                                 # comment
 898                                 $element = $p[4];
 899                                 $attributes = '';
 900                                 $close = '';
 901                                 $inside = $p[5];
 902                         } else {
 903                                 # tag
 904                                 $element = $p[1];
 905                                 $attributes = $p[2];
 906                                 $close = $p[3];
 907                                 $inside = $p[4];
 908                         }
 909
 910                         $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
 911                         $stripped .= $marker;
 912
 913                         if ( $close === '/>' ) {
 914                                 # Empty element tag, <tag />
 915                                 $content = null;
 916                                 $text = $inside;
 917                                 $tail = null;
 918                         } else {
 919                                 if ( $element === '!--' ) {
 920                                         $end = '/(-->)/';
 921                                 } else {
 922                                         $end = "/(<\\/$element\\s*>)/i";
 923                                 }
 924                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 925                                 $content = $q[0];
 926                                 if ( count( $q ) < 3 ) {
 927                                         # No end tag -- let it run out to the end of the text.
 928                                         $tail = '';
 929                                         $text = '';
 930                                 } else {
 931                                         $tail = $q[1];
 932                                         $text = $q[2];
 933                                 }
 934                         }
 935
 936                         $matches[$marker] = array( $element,
 937                                 $content,
 938                                 Sanitizer::decodeTagAttributes( $attributes ),
 939                                 "<$element$attributes$close$content$tail" );
 940                 }
 941                 return $stripped;
 942         }
 943
 944         /**
 945          * Get a list of strippable XML-like elements
 946          *
 947          * @return array
 948          */
 949         function getStripList() {
 950                 return $this->mStripList;
 951         }
 952
 953         /**
 954          * Add an item to the strip state
 955          * Returns the unique tag which must be inserted into the stripped text
 956          * The tag will be replaced with the original text in unstrip()
 957          *
 958          * @param $text string
 959          *
 960          * @return string
 961          */
 962         function insertStripItem( $text ) {
 963                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 964                 $this->mMarkerIndex++;
 965                 $this->mStripState->addGeneral( $rnd, $text );
 966                 return $rnd;
 967         }
 968
 969         /**
 970          * parse the wiki syntax used to render tables
 971          *
 972          * @private
 973          * @return string
 974          */
 975         function doTableStuff( $text ) {
 976                 wfProfileIn( __METHOD__ );
 977
 978                 $lines = StringUtils::explode( "\n", $text );
 979                 $out = '';
 980                 $td_history = array(); # Is currently a td tag open?
 981                 $last_tag_history = array(); # Save history of last lag activated (td, th or caption)
 982                 $tr_history = array(); # Is currently a tr tag open?
 983                 $tr_attributes = array(); # history of tr attributes
 984                 $has_opened_tr = array(); # Did this table open a <tr> element?
 985                 $indent_level = 0; # indent level of the table
 986
 987                 foreach ( $lines as $outLine ) {
 988                         $line = trim( $outLine );
 989
 990                         if ( $line === '' ) { # empty line, go to next line
 991                                 $out .= $outLine . "\n";
 992                                 continue;
 993                         }
 994
 995                         $first_character = $line[0];
 996                         $matches = array();
 997
 998                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line, $matches ) ) {
 999                                 # First check if we are starting a new table
1000                                 $indent_level = strlen( $matches[1] );
1001
1002                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
1003                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'table' );
1004
1005                                 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1006                                 array_push( $td_history, false );
1007                                 array_push( $last_tag_history, '' );
1008                                 array_push( $tr_history, false );
1009                                 array_push( $tr_attributes, '' );
1010                                 array_push( $has_opened_tr, false );
1011                         } elseif ( count( $td_history ) == 0 ) {
1012                                 # Don't do any of the following
1013                                 $out .= $outLine . "\n";
1014                                 continue;
1015                         } elseif ( substr( $line, 0, 2 ) === '|}' ) {
1016                                 # We are ending a table
1017                                 $line = '</table>' . substr( $line, 2 );
1018                                 $last_tag = array_pop( $last_tag_history );
1019
1020                                 if ( !array_pop( $has_opened_tr ) ) {
1021                                         $line = "<tr><td></td></tr>{$line}";
1022                                 }
1023
1024                                 if ( array_pop( $tr_history ) ) {
1025                                         $line = "</tr>{$line}";
1026                                 }
1027
1028                                 if ( array_pop( $td_history ) ) {
1029                                         $line = "</{$last_tag}>{$line}";
1030                                 }
1031                                 array_pop( $tr_attributes );
1032                                 $outLine = $line . str_repeat( '</dd></dl>', $indent_level );
1033                         } elseif ( substr( $line, 0, 2 ) === '|-' ) {
1034                                 # Now we have a table row
1035                                 $line = preg_replace( '#^\|-+#', '', $line );
1036
1037                                 # Whats after the tag is now only attributes
1038                                 $attributes = $this->mStripState->unstripBoth( $line );
1039                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
1040                                 array_pop( $tr_attributes );
1041                                 array_push( $tr_attributes, $attributes );
1042
1043                                 $line = '';
1044                                 $last_tag = array_pop( $last_tag_history );
1045                                 array_pop( $has_opened_tr );
1046                                 array_push( $has_opened_tr, true );
1047
1048                                 if ( array_pop( $tr_history ) ) {
1049                                         $line = '</tr>';
1050                                 }
1051
1052                                 if ( array_pop( $td_history ) ) {
1053                                         $line = "</{$last_tag}>{$line}";
1054                                 }
1055
1056                                 $outLine = $line;
1057                                 array_push( $tr_history, false );
1058                                 array_push( $td_history, false );
1059                                 array_push( $last_tag_history, '' );
1060                         } elseif ( $first_character === '|' || $first_character === '!' || substr( $line, 0, 2 ) === '|+' ) {
1061                                 # This might be cell elements, td, th or captions
1062                                 if ( substr( $line, 0, 2 ) === '|+' ) {
1063                                         $first_character = '+';
1064                                         $line = substr( $line, 1 );
1065                                 }
1066
1067                                 $line = substr( $line, 1 );
1068
1069                                 if ( $first_character === '!' ) {
1070                                         $line = str_replace( '!!', '||', $line );
1071                                 }
1072
1073                                 # Split up multiple cells on the same line.
1074                                 # FIXME : This can result in improper nesting of tags processed
1075                                 # by earlier parser steps, but should avoid splitting up eg
1076                                 # attribute values containing literal "||".
1077                                 $cells = StringUtils::explodeMarkup( '||', $line );
1078
1079                                 $outLine = '';
1080
1081                                 # Loop through each table cell
1082                                 foreach ( $cells as $cell ) {
1083                                         $previous = '';
1084                                         if ( $first_character !== '+' ) {
1085                                                 $tr_after = array_pop( $tr_attributes );
1086                                                 if ( !array_pop( $tr_history ) ) {
1087                                                         $previous = "<tr{$tr_after}>\n";
1088                                                 }
1089                                                 array_push( $tr_history, true );
1090                                                 array_push( $tr_attributes, '' );
1091                                                 array_pop( $has_opened_tr );
1092                                                 array_push( $has_opened_tr, true );
1093                                         }
1094
1095                                         $last_tag = array_pop( $last_tag_history );
1096
1097                                         if ( array_pop( $td_history ) ) {
1098                                                 $previous = "</{$last_tag}>\n{$previous}";
1099                                         }
1100
1101                                         if ( $first_character === '|' ) {
1102                                                 $last_tag = 'td';
1103                                         } elseif ( $first_character === '!' ) {
1104                                                 $last_tag = 'th';
1105                                         } elseif ( $first_character === '+' ) {
1106                                                 $last_tag = 'caption';
1107                                         } else {
1108                                                 $last_tag = '';
1109                                         }
1110
1111                                         array_push( $last_tag_history, $last_tag );
1112
1113                                         # A cell could contain both parameters and data
1114                                         $cell_data = explode( '|', $cell, 2 );
1115
1116                                         # Bug 553: Note that a '|' inside an invalid link should not
1117                                         # be mistaken as delimiting cell parameters
1118                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
1119                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
1120                                         } elseif ( count( $cell_data ) == 1 ) {
1121                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
1122                                         } else {
1123                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
1124                                                 $attributes = Sanitizer::fixTagAttributes( $attributes, $last_tag );
1125                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
1126                                         }
1127
1128                                         $outLine .= $cell;
1129                                         array_push( $td_history, true );
1130                                 }
1131                         }
1132                         $out .= $outLine . "\n";
1133                 }
1134
1135                 # Closing open td, tr && table
1136                 while ( count( $td_history ) > 0 ) {
1137                         if ( array_pop( $td_history ) ) {
1138                                 $out .= "</td>\n";
1139                         }
1140                         if ( array_pop( $tr_history ) ) {
1141                                 $out .= "</tr>\n";
1142                         }
1143                         if ( !array_pop( $has_opened_tr ) ) {
1144                                 $out .= "<tr><td></td></tr>\n";
1145                         }
1146
1147                         $out .= "</table>\n";
1148                 }
1149
1150                 # Remove trailing line-ending (b/c)
1151                 if ( substr( $out, -1 ) === "\n" ) {
1152                         $out = substr( $out, 0, -1 );
1153                 }
1154
1155                 # special case: don't return empty table
1156                 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1157                         $out = '';
1158                 }
1159
1160                 wfProfileOut( __METHOD__ );
1161
1162                 return $out;
1163         }
1164
1165         /**
1166          * Helper function for parse() that transforms wiki markup into
1167          * HTML. Only called for $mOutputType == self::OT_HTML.
1168          *
1169          * @private
1170          *
1171          * @param $text string
1172          * @param $isMain bool
1173          * @param $frame bool
1174          *
1175          * @return string
1176          */
1177         function internalParse( $text, $isMain = true, $frame = false ) {
1178                 wfProfileIn( __METHOD__ );
1179
1180                 $origText = $text;
1181
1182                 # Hook to suspend the parser in this state
1183                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1184                         wfProfileOut( __METHOD__ );
1185                         return $text;
1186                 }
1187
1188                 # if $frame is provided, then use $frame for replacing any variables
1189                 if ( $frame ) {
1190                         # use frame depth to infer how include/noinclude tags should be handled
1191                         # depth=0 means this is the top-level document; otherwise it's an included document
1192                         if ( !$frame->depth ) {
1193                                 $flag = 0;
1194                         } else {
1195                                 $flag = Parser::PTD_FOR_INCLUSION;
1196                         }
1197                         $dom = $this->preprocessToDom( $text, $flag );
1198                         $text = $frame->expand( $dom );
1199                 } else {
1200                         # if $frame is not provided, then use old-style replaceVariables
1201                         $text = $this->replaceVariables( $text );
1202                 }
1203
1204                 wfRunHooks( 'InternalParseBeforeSanitize', array( &$this, &$text, &$this->mStripState ) );
1205                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
1206                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1207
1208                 # Tables need to come after variable replacement for things to work
1209                 # properly; putting them before other transformations should keep
1210                 # exciting things like link expansions from showing up in surprising
1211                 # places.
1212                 $text = $this->doTableStuff( $text );
1213
1214                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1215
1216                 $text = $this->doDoubleUnderscore( $text );
1217
1218                 $text = $this->doHeadings( $text );
1219                 $text = $this->replaceInternalLinks( $text );
1220                 $text = $this->doAllQuotes( $text );
1221                 $text = $this->replaceExternalLinks( $text );
1222
1223                 # replaceInternalLinks may sometimes leave behind
1224                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1225                 $text = str_replace( $this->mUniqPrefix . 'NOPARSE', '', $text );
1226
1227                 $text = $this->doMagicLinks( $text );
1228                 $text = $this->formatHeadings( $text, $origText, $isMain );
1229
1230                 wfProfileOut( __METHOD__ );
1231                 return $text;
1232         }
1233
1234         /**
1235          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1236          * magic external links.
1237          *
1238          * DML
1239          * @private
1240          *
1241          * @param $text string
1242          *
1243          * @return string
1244          */
1245         function doMagicLinks( $text ) {
1246                 wfProfileIn( __METHOD__ );
1247                 $prots = wfUrlProtocolsWithoutProtRel();
1248                 $urlChar = self::EXT_LINK_URL_CLASS;
1249                 $text = preg_replace_callback(
1250                         '!(?:                           # Start cases
1251                                 (<a[ \t\r\n>].*?</a>) |     # m[1]: Skip link text
1252                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1253                                 (\\b(?i:$prots)$urlChar+) |  # m[3]: Free external links" . '
1254                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1255                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1256                                         (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1257                                         (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1258                                         [0-9Xx]                 # check digit
1259                                         \b)
1260                         )!xu', array( &$this, 'magicLinkCallback' ), $text );
1261                 wfProfileOut( __METHOD__ );
1262                 return $text;
1263         }
1264
1265         /**
1266          * @throws MWException
1267          * @param $m array
1268          * @return HTML|string
1269          */
1270         function magicLinkCallback( $m ) {
1271                 if ( isset( $m[1] ) && $m[1] !== '' ) {
1272                         # Skip anchor
1273                         return $m[0];
1274                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1275                         # Skip HTML element
1276                         return $m[0];
1277                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1278                         # Free external link
1279                         return $this->makeFreeExternalLink( $m[0] );
1280                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1281                         # RFC or PMID
1282                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1283                                 $keyword = 'RFC';
1284                                 $urlmsg = 'rfcurl';
1285                                 $CssClass = 'mw-magiclink-rfc';
1286                                 $id = $m[4];
1287                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1288                                 $keyword = 'PMID';
1289                                 $urlmsg = 'pubmedurl';
1290                                 $CssClass = 'mw-magiclink-pmid';
1291                                 $id = $m[4];
1292                         } else {
1293                                 throw new MWException( __METHOD__ . ': unrecognised match type "' .
1294                                         substr( $m[0], 0, 20 ) . '"' );
1295                         }
1296                         $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1297                         return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $CssClass );
1298                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1299                         # ISBN
1300                         $isbn = $m[5];
1301                         $num = strtr( $isbn, array(
1302                                 '-' => '',
1303                                 ' ' => '',
1304                                 'x' => 'X',
1305                         ));
1306                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1307                         return '<a href="' .
1308                                 htmlspecialchars( $titleObj->getLocalURL() ) .
1309                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1310                 } else {
1311                         return $m[0];
1312                 }
1313         }
1314
1315         /**
1316          * Make a free external link, given a user-supplied URL
1317          *
1318          * @param $url string
1319          *
1320          * @return string HTML
1321          * @private
1322          */
1323         function makeFreeExternalLink( $url ) {
1324                 wfProfileIn( __METHOD__ );
1325
1326                 $trail = '';
1327
1328                 # The characters '<' and '>' (which were escaped by
1329                 # removeHTMLtags()) should not be included in
1330                 # URLs, per RFC 2396.
1331                 $m2 = array();
1332                 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1333                         $trail = substr( $url, $m2[0][1] ) . $trail;
1334                         $url = substr( $url, 0, $m2[0][1] );
1335                 }
1336
1337                 # Move trailing punctuation to $trail
1338                 $sep = ',;\.:!?';
1339                 # If there is no left bracket, then consider right brackets fair game too
1340                 if ( strpos( $url, '(' ) === false ) {
1341                         $sep .= ')';
1342                 }
1343
1344                 $numSepChars = strspn( strrev( $url ), $sep );
1345                 if ( $numSepChars ) {
1346                         $trail = substr( $url, -$numSepChars ) . $trail;
1347                         $url = substr( $url, 0, -$numSepChars );
1348                 }
1349
1350                 $url = Sanitizer::cleanUrl( $url );
1351
1352                 # Is this an external image?
1353                 $text = $this->maybeMakeExternalImage( $url );
1354                 if ( $text === false ) {
1355                         # Not an image, make a link
1356                         $text = Linker::makeExternalLink( $url,
1357                                 $this->getConverterLanguage()->markNoConversion( $url, true ),
1358                                 true, 'free',
1359                                 $this->getExternalLinkAttribs( $url ) );
1360                         # Register it in the output object...
1361                         # Replace unnecessary URL escape codes with their equivalent characters
1362                         $pasteurized = self::replaceUnusualEscapes( $url );
1363                         $this->mOutput->addExternalLink( $pasteurized );
1364                 }
1365                 wfProfileOut( __METHOD__ );
1366                 return $text . $trail;
1367         }
1368
1369         /**
1370          * Parse headers and return html
1371          *
1372          * @private
1373          *
1374          * @param $text string
1375          *
1376          * @return string
1377          */
1378         function doHeadings( $text ) {
1379                 wfProfileIn( __METHOD__ );
1380                 for ( $i = 6; $i >= 1; --$i ) {
1381                         $h = str_repeat( '=', $i );
1382                         $text = preg_replace( "/^$h(.+)$h\\s*$/m", "<h$i>\\1</h$i>", $text );
1383                 }
1384                 wfProfileOut( __METHOD__ );
1385                 return $text;
1386         }
1387
1388         /**
1389          * Replace single quotes with HTML markup
1390          * @private
1391          *
1392          * @param $text string
1393          *
1394          * @return string the altered text
1395          */
1396         function doAllQuotes( $text ) {
1397                 wfProfileIn( __METHOD__ );
1398                 $outtext = '';
1399                 $lines = StringUtils::explode( "\n", $text );
1400                 foreach ( $lines as $line ) {
1401                         $outtext .= $this->doQuotes( $line ) . "\n";
1402                 }
1403                 $outtext = substr( $outtext, 0, -1 );
1404                 wfProfileOut( __METHOD__ );
1405                 return $outtext;
1406         }
1407
1408         /**
1409          * Helper function for doAllQuotes()
1410          *
1411          * @param $text string
1412          *
1413          * @return string
1414          */
1415         public function doQuotes( $text ) {
1416                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1417                 $countarr = count( $arr );
1418                 if ( $countarr == 1 ) {
1419                         return $text;
1420                 }
1421
1422                 // First, do some preliminary work. This may shift some apostrophes from
1423                 // being mark-up to being text. It also counts the number of occurrences
1424                 // of bold and italics mark-ups.
1425                 $numbold = 0;
1426                 $numitalics = 0;
1427                 for ( $i = 1; $i < $countarr; $i += 2 ) {
1428                         $thislen = strlen( $arr[$i] );
1429                         // If there are ever four apostrophes, assume the first is supposed to
1430                         // be text, and the remaining three constitute mark-up for bold text.
1431                         // (bug 13227: ''''foo'''' turns into ' ''' foo ' ''')
1432                         if ( $thislen == 4 ) {
1433                                 $arr[$i - 1] .= "'";
1434                                 $arr[$i] = "'''";
1435                                 $thislen = 3;
1436                         } elseif ( $thislen > 5 ) {
1437                                 // If there are more than 5 apostrophes in a row, assume they're all
1438                                 // text except for the last 5.
1439                                 // (bug 13227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1440                                 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1441                                 $arr[$i] = "'''''";
1442                                 $thislen = 5;
1443                         }
1444                         // Count the number of occurrences of bold and italics mark-ups.
1445                         if ( $thislen == 2 ) {
1446                                 $numitalics++;
1447                         } elseif ( $thislen == 3 ) {
1448                                 $numbold++;
1449                         } elseif ( $thislen == 5 ) {
1450                                 $numitalics++;
1451                                 $numbold++;
1452                         }
1453                 }
1454
1455                 // If there is an odd number of both bold and italics, it is likely
1456                 // that one of the bold ones was meant to be an apostrophe followed
1457                 // by italics. Which one we cannot know for certain, but it is more
1458                 // likely to be one that has a single-letter word before it.
1459                 if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1460                         $firstsingleletterword = -1;
1461                         $firstmultiletterword = -1;
1462                         $firstspace = -1;
1463                         for ( $i = 1; $i < $countarr; $i += 2 ) {
1464                                 if ( strlen( $arr[$i] ) == 3 ) {
1465                                         $x1 = substr( $arr[$i - 1], -1 );
1466                                         $x2 = substr( $arr[$i - 1], -2, 1 );
1467                                         if ( $x1 === ' ' ) {
1468                                                 if ( $firstspace == -1 ) {
1469                                                         $firstspace = $i;
1470                                                 }
1471                                         } elseif ( $x2 === ' ' ) {
1472                                                 if ( $firstsingleletterword == -1 ) {
1473                                                         $firstsingleletterword = $i;
1474                                                         // if $firstsingleletterword is set, we don't
1475                                                         // look at the other options, so we can bail early.
1476                                                         break;
1477                                                 }
1478                                         } else {
1479                                                 if ( $firstmultiletterword == -1 ) {
1480                                                         $firstmultiletterword = $i;
1481                                                 }
1482                                         }
1483                                 }
1484                         }
1485
1486                         // If there is a single-letter word, use it!
1487                         if ( $firstsingleletterword > -1 ) {
1488                                 $arr[$firstsingleletterword] = "''";
1489                                 $arr[$firstsingleletterword - 1] .= "'";
1490                         } elseif ( $firstmultiletterword > -1 ) {
1491                                 // If not, but there's a multi-letter word, use that one.
1492                                 $arr[$firstmultiletterword] = "''";
1493                                 $arr[$firstmultiletterword - 1] .= "'";
1494                         } elseif ( $firstspace > -1 ) {
1495                                 // ... otherwise use the first one that has neither.
1496                                 // (notice that it is possible for all three to be -1 if, for example,
1497                                 // there is only one pentuple-apostrophe in the line)
1498                                 $arr[$firstspace] = "''";
1499                                 $arr[$firstspace - 1] .= "'";
1500                         }
1501                 }
1502
1503                 // Now let's actually convert our apostrophic mush to HTML!
1504                 $output = '';
1505                 $buffer = '';
1506                 $state = '';
1507                 $i = 0;
1508                 foreach ( $arr as $r ) {
1509                         if ( ( $i % 2 ) == 0 ) {
1510                                 if ( $state === 'both' ) {
1511                                         $buffer .= $r;
1512                                 } else {
1513                                         $output .= $r;
1514                                 }
1515                         } else {
1516                                 $thislen = strlen( $r );
1517                                 if ( $thislen == 2 ) {
1518                                         if ( $state === 'i' ) {
1519                                                 $output .= '</i>';
1520                                                 $state = '';
1521                                         } elseif ( $state === 'bi' ) {
1522                                                 $output .= '</i>';
1523                                                 $state = 'b';
1524                                         } elseif ( $state === 'ib' ) {
1525                                                 $output .= '</b></i><b>';
1526                                                 $state = 'b';
1527                                         } elseif ( $state === 'both' ) {
1528                                                 $output .= '<b><i>' . $buffer . '</i>';
1529                                                 $state = 'b';
1530                                         } else { // $state can be 'b' or ''
1531                                                 $output .= '<i>';
1532                                                 $state .= 'i';
1533                                         }
1534                                 } elseif ( $thislen == 3 ) {
1535                                         if ( $state === 'b' ) {
1536                                                 $output .= '</b>';
1537                                                 $state = '';
1538                                         } elseif ( $state === 'bi' ) {
1539                                                 $output .= '</i></b><i>';
1540                                                 $state = 'i';
1541                                         } elseif ( $state === 'ib' ) {
1542                                                 $output .= '</b>';
1543                                                 $state = 'i';
1544                                         } elseif ( $state === 'both' ) {
1545                                                 $output .= '<i><b>' . $buffer . '</b>';
1546                                                 $state = 'i';
1547                                         } else { // $state can be 'i' or ''
1548                                                 $output .= '<b>';
1549                                                 $state .= 'b';
1550                                         }
1551                                 } elseif ( $thislen == 5 ) {
1552                                         if ( $state === 'b' ) {
1553                                                 $output .= '</b><i>';
1554                                                 $state = 'i';
1555                                         } elseif ( $state === 'i' ) {
1556                                                 $output .= '</i><b>';
1557                                                 $state = 'b';
1558                                         } elseif ( $state === 'bi' ) {
1559                                                 $output .= '</i></b>';
1560                                                 $state = '';
1561                                         } elseif ( $state === 'ib' ) {
1562                                                 $output .= '</b></i>';
1563                                                 $state = '';
1564                                         } elseif ( $state === 'both' ) {
1565                                                 $output .= '<i><b>' . $buffer . '</b></i>';
1566                                                 $state = '';
1567                                         } else { // ($state == '')
1568                                                 $buffer = '';
1569                                                 $state = 'both';
1570                                         }
1571                                 }
1572                         }
1573                         $i++;
1574                 }
1575                 // Now close all remaining tags.  Notice that the order is important.
1576                 if ( $state === 'b' || $state === 'ib' ) {
1577                         $output .= '</b>';
1578                 }
1579                 if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1580                         $output .= '</i>';
1581                 }
1582                 if ( $state === 'bi' ) {
1583                         $output .= '</b>';
1584                 }
1585                 // There might be lonely ''''', so make sure we have a buffer
1586                 if ( $state === 'both' && $buffer ) {
1587                         $output .= '<b><i>' . $buffer . '</i></b>';
1588                 }
1589                 return $output;
1590         }
1591
1592         /**
1593          * Replace external links (REL)
1594          *
1595          * Note: this is all very hackish and the order of execution matters a lot.
1596          * Make sure to run tests/parserTests.php if you change this code.
1597          *
1598          * @private
1599          *
1600          * @param $text string
1601          *
1602          * @throws MWException
1603          * @return string
1604          */
1605         function replaceExternalLinks( $text ) {
1606                 wfProfileIn( __METHOD__ );
1607
1608                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1609                 if ( $bits === false ) {
1610                         wfProfileOut( __METHOD__ );
1611                         throw new MWException( "PCRE needs to be compiled with --enable-unicode-properties in order for MediaWiki to function" );
1612                 }
1613                 $s = array_shift( $bits );
1614
1615                 $i = 0;
1616                 while ( $i < count( $bits ) ) {
1617                         $url = $bits[$i++];
1618                         $i++; // protocol
1619                         $text = $bits[$i++];
1620                         $trail = $bits[$i++];
1621
1622                         # The characters '<' and '>' (which were escaped by
1623                         # removeHTMLtags()) should not be included in
1624                         # URLs, per RFC 2396.
1625                         $m2 = array();
1626                         if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1627                                 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1628                                 $url = substr( $url, 0, $m2[0][1] );
1629                         }
1630
1631                         # If the link text is an image URL, replace it with an <img> tag
1632                         # This happened by accident in the original parser, but some people used it extensively
1633                         $img = $this->maybeMakeExternalImage( $text );
1634                         if ( $img !== false ) {
1635                                 $text = $img;
1636                         }
1637
1638                         $dtrail = '';
1639
1640                         # Set linktype for CSS - if URL==text, link is essentially free
1641                         $linktype = ( $text === $url ) ? 'free' : 'text';
1642
1643                         # No link text, e.g. [http://domain.tld/some.link]
1644                         if ( $text == '' ) {
1645                                 # Autonumber
1646                                 $langObj = $this->getTargetLanguage();
1647                                 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1648                                 $linktype = 'autonumber';
1649                         } else {
1650                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1651                                 # Check for trail
1652                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1653                         }
1654
1655                         $text = $this->getConverterLanguage()->markNoConversion( $text );
1656
1657                         $url = Sanitizer::cleanUrl( $url );
1658
1659                         # Use the encoded URL
1660                         # This means that users can paste URLs directly into the text
1661                         # Funny characters like ö aren't valid in URLs anyway
1662                         # This was changed in August 2004
1663                         $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1664                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1665
1666                         # Register link in the output object.
1667                         # Replace unnecessary URL escape codes with the referenced character
1668                         # This prevents spammers from hiding links from the filters
1669                         $pasteurized = self::replaceUnusualEscapes( $url );
1670                         $this->mOutput->addExternalLink( $pasteurized );
1671                 }
1672
1673                 wfProfileOut( __METHOD__ );
1674                 return $s;
1675         }
1676         /**
1677          * Get the rel attribute for a particular external link.
1678          *
1679          * @since 1.21
1680          * @param string|bool $url optional URL, to extract the domain from for rel =>
1681          *   nofollow if appropriate
1682          * @param $title Title optional Title, for wgNoFollowNsExceptions lookups
1683          * @return string|null rel attribute for $url
1684          */
1685         public static function getExternalLinkRel( $url = false, $title = null ) {
1686                 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1687                 $ns = $title ? $title->getNamespace() : false;
1688                 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) &&
1689                                 !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions ) )
1690                 {
1691                         return 'nofollow';
1692                 }
1693                 return null;
1694         }
1695         /**
1696          * Get an associative array of additional HTML attributes appropriate for a
1697          * particular external link.  This currently may include rel => nofollow
1698          * (depending on configuration, namespace, and the URL's domain) and/or a
1699          * target attribute (depending on configuration).
1700          *
1701          * @param string|bool $url optional URL, to extract the domain from for rel =>
1702          *   nofollow if appropriate
1703          * @return Array associative array of HTML attributes
1704          */
1705         function getExternalLinkAttribs( $url = false ) {
1706                 $attribs = array();
1707                 $attribs['rel'] = self::getExternalLinkRel( $url, $this->mTitle );
1708
1709                 if ( $this->mOptions->getExternalLinkTarget() ) {
1710                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1711                 }
1712                 return $attribs;
1713         }
1714
1715         /**
1716          * Replace unusual URL escape codes with their equivalent characters
1717          *
1718          * @param $url String
1719          * @return String
1720          *
1721          * @todo  This can merge genuinely required bits in the path or query string,
1722          *        breaking legit URLs. A proper fix would treat the various parts of
1723          *        the URL differently; as a workaround, just use the output for
1724          *        statistical records, not for actual linking/output.
1725          */
1726         static function replaceUnusualEscapes( $url ) {
1727                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1728                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1729         }
1730
1731         /**
1732          * Callback function used in replaceUnusualEscapes().
1733          * Replaces unusual URL escape codes with their equivalent character
1734          *
1735          * @param $matches array
1736          *
1737          * @return string
1738          */
1739         private static function replaceUnusualEscapesCallback( $matches ) {
1740                 $char = urldecode( $matches[0] );
1741                 $ord = ord( $char );
1742                 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1743                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1744                         # No, shouldn't be escaped
1745                         return $char;
1746                 } else {
1747                         # Yes, leave it escaped
1748                         return $matches[0];
1749                 }
1750         }
1751
1752         /**
1753          * make an image if it's allowed, either through the global
1754          * option, through the exception, or through the on-wiki whitelist
1755          * @private
1756          *
1757          * $param $url string
1758          *
1759          * @return string
1760          */
1761         function maybeMakeExternalImage( $url ) {
1762                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1763                 $imagesexception = !empty( $imagesfrom );
1764                 $text = false;
1765                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1766                 if ( $imagesexception && is_array( $imagesfrom ) ) {
1767                         $imagematch = false;
1768                         foreach ( $imagesfrom as $match ) {
1769                                 if ( strpos( $url, $match ) === 0 ) {
1770                                         $imagematch = true;
1771                                         break;
1772                                 }
1773                         }
1774                 } elseif ( $imagesexception ) {
1775                         $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1776                 } else {
1777                         $imagematch = false;
1778                 }
1779                 if ( $this->mOptions->getAllowExternalImages()
1780                          || ( $imagesexception && $imagematch ) ) {
1781                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1782                                 # Image found
1783                                 $text = Linker::makeExternalImage( $url );
1784                         }
1785                 }
1786                 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1787                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1788                         $whitelist = explode( "\n", wfMessage( 'external_image_whitelist' )->inContentLanguage()->text() );
1789                         foreach ( $whitelist as $entry ) {
1790                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1791                                 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1792                                         continue;
1793                                 }
1794                                 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1795                                         # Image matches a whitelist entry
1796                                         $text = Linker::makeExternalImage( $url );
1797                                         break;
1798                                 }
1799                         }
1800                 }
1801                 return $text;
1802         }
1803
1804         /**
1805          * Process [[ ]] wikilinks
1806          *
1807          * @param $s string
1808          *
1809          * @return String: processed text
1810          *
1811          * @private
1812          */
1813         function replaceInternalLinks( $s ) {
1814                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1815                 return $s;
1816         }
1817
1818         /**
1819          * Process [[ ]] wikilinks (RIL)
1820          * @param $s
1821          * @throws MWException
1822          * @return LinkHolderArray
1823          *
1824          * @private
1825          */
1826         function replaceInternalLinks2( &$s ) {
1827                 wfProfileIn( __METHOD__ );
1828
1829                 wfProfileIn( __METHOD__ . '-setup' );
1830                 static $tc = false, $e1, $e1_img;
1831                 # the % is needed to support urlencoded titles as well
1832                 if ( !$tc ) {
1833                         $tc = Title::legalChars() . '#%';
1834                         # Match a link having the form [[namespace:link|alternate]]trail
1835                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1836                         # Match cases where there is no "]]", which might still be images
1837                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1838                 }
1839
1840                 $holders = new LinkHolderArray( $this );
1841
1842                 # split the entire text string on occurrences of [[
1843                 $a = StringUtils::explode( '[[', ' ' . $s );
1844                 # get the first element (all text up to first [[), and remove the space we added
1845                 $s = $a->current();
1846                 $a->next();
1847                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1848                 $s = substr( $s, 1 );
1849
1850                 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
1851                 $e2 = null;
1852                 if ( $useLinkPrefixExtension ) {
1853                         # Match the end of a line for a word that's not followed by whitespace,
1854                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1855                         $e2 = wfMessage( 'linkprefix' )->inContentLanguage()->text();
1856                 }
1857
1858                 if ( is_null( $this->mTitle ) ) {
1859                         wfProfileOut( __METHOD__ . '-setup' );
1860                         wfProfileOut( __METHOD__ );
1861                         throw new MWException( __METHOD__ . ": \$this->mTitle is null\n" );
1862                 }
1863                 $nottalk = !$this->mTitle->isTalkPage();
1864
1865                 if ( $useLinkPrefixExtension ) {
1866                         $m = array();
1867                         if ( preg_match( $e2, $s, $m ) ) {
1868                                 $first_prefix = $m[2];
1869                         } else {
1870                                 $first_prefix = false;
1871                         }
1872                 } else {
1873                         $prefix = '';
1874                 }
1875
1876                 $useSubpages = $this->areSubpagesAllowed();
1877                 wfProfileOut( __METHOD__ . '-setup' );
1878
1879                 # Loop for each link
1880                 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
1881                         # Check for excessive memory usage
1882                         if ( $holders->isBig() ) {
1883                                 # Too big
1884                                 # Do the existence check, replace the link holders and clear the array
1885                                 $holders->replace( $s );
1886                                 $holders->clear();
1887                         }
1888
1889                         if ( $useLinkPrefixExtension ) {
1890                                 wfProfileIn( __METHOD__ . '-prefixhandling' );
1891                                 if ( preg_match( $e2, $s, $m ) ) {
1892                                         $prefix = $m[2];
1893                                         $s = $m[1];
1894                                 } else {
1895                                         $prefix = '';
1896                                 }
1897                                 # first link
1898                                 if ( $first_prefix ) {
1899                                         $prefix = $first_prefix;
1900                                         $first_prefix = false;
1901                                 }
1902                                 wfProfileOut( __METHOD__ . '-prefixhandling' );
1903                         }
1904
1905                         $might_be_img = false;
1906
1907                         wfProfileIn( __METHOD__ . "-e1" );
1908                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1909                                 $text = $m[2];
1910                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1911                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1912                                 # the real problem is with the $e1 regex
1913                                 # See bug 1300.
1914                                 #
1915                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1916                                 # and no image is in sight. See bug 2095.
1917                                 #
1918                                 if ( $text !== '' &&
1919                                         substr( $m[3], 0, 1 ) === ']' &&
1920                                         strpos( $text, '[' ) !== false
1921                                 )
1922                                 {
1923                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1924                                         $m[3] = substr( $m[3], 1 );
1925                                 }
1926                                 # fix up urlencoded title texts
1927                                 if ( strpos( $m[1], '%' ) !== false ) {
1928                                         # Should anchors '#' also be rejected?
1929                                         $m[1] = str_replace( array( '<', '>' ), array( '&lt;', '&gt;' ), rawurldecode( $m[1] ) );
1930                                 }
1931                                 $trail = $m[3];
1932                         } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1933                                 $might_be_img = true;
1934                                 $text = $m[2];
1935                                 if ( strpos( $m[1], '%' ) !== false ) {
1936                                         $m[1] = rawurldecode( $m[1] );
1937                                 }
1938                                 $trail = "";
1939                         } else { # Invalid form; output directly
1940                                 $s .= $prefix . '[[' . $line;
1941                                 wfProfileOut( __METHOD__ . "-e1" );
1942                                 continue;
1943                         }
1944                         wfProfileOut( __METHOD__ . "-e1" );
1945                         wfProfileIn( __METHOD__ . "-misc" );
1946
1947                         # Don't allow internal links to pages containing
1948                         # PROTO: where PROTO is a valid URL protocol; these
1949                         # should be external links.
1950                         if ( preg_match( '/^(?i:' . $this->mUrlProtocols . ')/', $m[1] ) ) {
1951                                 $s .= $prefix . '[[' . $line;
1952                                 wfProfileOut( __METHOD__ . "-misc" );
1953                                 continue;
1954                         }
1955
1956                         # Make subpage if necessary
1957                         if ( $useSubpages ) {
1958                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1959                         } else {
1960                                 $link = $m[1];
1961                         }
1962
1963                         $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1964                         if ( !$noforce ) {
1965                                 # Strip off leading ':'
1966                                 $link = substr( $link, 1 );
1967                         }
1968
1969                         wfProfileOut( __METHOD__ . "-misc" );
1970                         wfProfileIn( __METHOD__ . "-title" );
1971                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1972                         if ( $nt === null ) {
1973                                 $s .= $prefix . '[[' . $line;
1974                                 wfProfileOut( __METHOD__ . "-title" );
1975                                 continue;
1976                         }
1977
1978                         $ns = $nt->getNamespace();
1979                         $iw = $nt->getInterWiki();
1980                         wfProfileOut( __METHOD__ . "-title" );
1981
1982                         if ( $might_be_img ) { # if this is actually an invalid link
1983                                 wfProfileIn( __METHOD__ . "-might_be_img" );
1984                                 if ( $ns == NS_FILE && $noforce ) { # but might be an image
1985                                         $found = false;
1986                                         while ( true ) {
1987                                                 # look at the next 'line' to see if we can close it there
1988                                                 $a->next();
1989                                                 $next_line = $a->current();
1990                                                 if ( $next_line === false || $next_line === null ) {
1991                                                         break;
1992                                                 }
1993                                                 $m = explode( ']]', $next_line, 3 );
1994                                                 if ( count( $m ) == 3 ) {
1995                                                         # the first ]] closes the inner link, the second the image
1996                                                         $found = true;
1997                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1998                                                         $trail = $m[2];
1999                                                         break;
2000                                                 } elseif ( count( $m ) == 2 ) {
2001                                                         # if there's exactly one ]] that's fine, we'll keep looking
2002                                                         $text .= "[[{$m[0]}]]{$m[1]}";
2003                                                 } else {
2004                                                         # if $next_line is invalid too, we need look no further
2005                                                         $text .= '[[' . $next_line;
2006                                                         break;
2007                                                 }
2008                                         }
2009                                         if ( !$found ) {
2010                                                 # we couldn't find the end of this imageLink, so output it raw
2011                                                 # but don't ignore what might be perfectly normal links in the text we've examined
2012                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
2013                                                 $s .= "{$prefix}[[$link|$text";
2014                                                 # note: no $trail, because without an end, there *is* no trail
2015                                                 wfProfileOut( __METHOD__ . "-might_be_img" );
2016                                                 continue;
2017                                         }
2018                                 } else { # it's not an image, so output it raw
2019                                         $s .= "{$prefix}[[$link|$text";
2020                                         # note: no $trail, because without an end, there *is* no trail
2021                                         wfProfileOut( __METHOD__ . "-might_be_img" );
2022                                         continue;
2023                                 }
2024                                 wfProfileOut( __METHOD__ . "-might_be_img" );
2025                         }
2026
2027                         $wasblank = ( $text == '' );
2028                         if ( $wasblank ) {
2029                                 $text = $link;
2030                         } else {
2031                                 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
2032                                 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2033                                 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2034                                 #    -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2035                                 $text = $this->doQuotes( $text );
2036                         }
2037
2038                         # Link not escaped by : , create the various objects
2039                         if ( $noforce ) {
2040                                 # Interwikis
2041                                 wfProfileIn( __METHOD__ . "-interwiki" );
2042                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && Language::fetchLanguageName( $iw, null, 'mw' ) ) {
2043                                         // XXX: the above check prevents links to sites with identifiers that are not language codes
2044
2045                                         # Bug 24502: filter duplicates
2046                                         if ( !isset( $this->mLangLinkLanguages[$iw] ) ) {
2047                                                 $this->mLangLinkLanguages[$iw] = true;
2048                                                 $this->mOutput->addLanguageLink( $nt->getFullText() );
2049                                         }
2050
2051                                         $s = rtrim( $s . $prefix );
2052                                         $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
2053                                         wfProfileOut( __METHOD__ . "-interwiki" );
2054                                         continue;
2055                                 }
2056                                 wfProfileOut( __METHOD__ . "-interwiki" );
2057
2058                                 if ( $ns == NS_FILE ) {
2059                                         wfProfileIn( __METHOD__ . "-image" );
2060                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
2061                                                 if ( $wasblank ) {
2062                                                         # if no parameters were passed, $text
2063                                                         # becomes something like "File:Foo.png",
2064                                                         # which we don't want to pass on to the
2065                                                         # image generator
2066                                                         $text = '';
2067                                                 } else {
2068                                                         # recursively parse links inside the image caption
2069                                                         # actually, this will parse them in any other parameters, too,
2070                                                         # but it might be hard to fix that, and it doesn't matter ATM
2071                                                         $text = $this->replaceExternalLinks( $text );
2072                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
2073                                                 }
2074                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2075                                                 $s .= $prefix . $this->armorLinks(
2076                                                         $this->makeImage( $nt, $text, $holders ) ) . $trail;
2077                                         } else {
2078                                                 $s .= $prefix . $trail;
2079                                         }
2080                                         wfProfileOut( __METHOD__ . "-image" );
2081                                         continue;
2082                                 }
2083
2084                                 if ( $ns == NS_CATEGORY ) {
2085                                         wfProfileIn( __METHOD__ . "-category" );
2086                                         $s = rtrim( $s . "\n" ); # bug 87
2087
2088                                         if ( $wasblank ) {
2089                                                 $sortkey = $this->getDefaultSort();
2090                                         } else {
2091                                                 $sortkey = $text;
2092                                         }
2093                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
2094                                         $sortkey = str_replace( "\n", '', $sortkey );
2095                                         $sortkey = $this->getConverterLanguage()->convertCategoryKey( $sortkey );
2096                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
2097
2098                                         /**
2099                                          * Strip the whitespace Category links produce, see bug 87
2100                                          * @todo We might want to use trim($tmp, "\n") here.
2101                                          */
2102                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
2103
2104                                         wfProfileOut( __METHOD__ . "-category" );
2105                                         continue;
2106                                 }
2107                         }
2108
2109                         # Self-link checking
2110                         if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
2111                                 if ( $nt->equals( $this->mTitle ) || ( !$nt->isKnown() && in_array(
2112                                         $this->mTitle->getPrefixedText(),
2113                                         $this->getConverterLanguage()->autoConvertToAllVariants( $nt->getPrefixedText() ),
2114                                         true
2115                                 ) ) ) {
2116                                         $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2117                                         continue;
2118                                 }
2119                         }
2120
2121                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
2122                         # @todo FIXME: Should do batch file existence checks, see comment below
2123                         if ( $ns == NS_MEDIA ) {
2124                                 wfProfileIn( __METHOD__ . "-media" );
2125                                 # Give extensions a chance to select the file revision for us
2126                                 $options = array();
2127                                 $descQuery = false;
2128                                 wfRunHooks( 'BeforeParserFetchFileAndTitle',
2129                                         array( $this, $nt, &$options, &$descQuery ) );
2130                                 # Fetch and register the file (file title may be different via hooks)
2131                                 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2132                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2133                                 $s .= $prefix . $this->armorLinks(
2134                                         Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2135                                 wfProfileOut( __METHOD__ . "-media" );
2136                                 continue;
2137                         }
2138
2139                         wfProfileIn( __METHOD__ . "-always_known" );
2140                         # Some titles, such as valid special pages or files in foreign repos, should
2141                         # be shown as bluelinks even though they're not included in the page table
2142                         #
2143                         # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2144                         # batch file existence checks for NS_FILE and NS_MEDIA
2145                         if ( $iw == '' && $nt->isAlwaysKnown() ) {
2146                                 $this->mOutput->addLink( $nt );
2147                                 $s .= $this->makeKnownLinkHolder( $nt, $text, array(), $trail, $prefix );
2148                         } else {
2149                                 # Links will be added to the output link list after checking
2150                                 $s .= $holders->makeHolder( $nt, $text, array(), $trail, $prefix );
2151                         }
2152                         wfProfileOut( __METHOD__ . "-always_known" );
2153                 }
2154                 wfProfileOut( __METHOD__ );
2155                 return $holders;
2156         }
2157
2158         /**
2159          * Render a forced-blue link inline; protect against double expansion of
2160          * URLs if we're in a mode that prepends full URL prefixes to internal links.
2161          * Since this little disaster has to split off the trail text to avoid
2162          * breaking URLs in the following text without breaking trails on the
2163          * wiki links, it's been made into a horrible function.
2164          *
2165          * @param $nt Title
2166          * @param $text String
2167          * @param array $query or String
2168          * @param $trail String
2169          * @param $prefix String
2170          * @return String: HTML-wikitext mix oh yuck
2171          */
2172         function makeKnownLinkHolder( $nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
2173                 list( $inside, $trail ) = Linker::splitTrail( $trail );
2174
2175                 if ( is_string( $query ) ) {
2176                         $query = wfCgiToArray( $query );
2177                 }
2178                 if ( $text == '' ) {
2179                         $text = htmlspecialchars( $nt->getPrefixedText() );
2180                 }
2181
2182                 $link = Linker::linkKnown( $nt, "$prefix$text$inside", array(), $query );
2183
2184                 return $this->armorLinks( $link ) . $trail;
2185         }
2186
2187         /**
2188          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2189          * going to go through further parsing steps before inline URL expansion.
2190          *
2191          * Not needed quite as much as it used to be since free links are a bit
2192          * more sensible these days. But bracketed links are still an issue.
2193          *
2194          * @param string $text more-or-less HTML
2195          * @return String: less-or-more HTML with NOPARSE bits
2196          */
2197         function armorLinks( $text ) {
2198                 return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
2199                         "{$this->mUniqPrefix}NOPARSE$1", $text );
2200         }
2201
2202         /**
2203          * Return true if subpage links should be expanded on this page.
2204          * @return Boolean
2205          */
2206         function areSubpagesAllowed() {
2207                 # Some namespaces don't allow subpages
2208                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2209         }
2210
2211         /**
2212          * Handle link to subpage if necessary
2213          *
2214          * @param string $target the source of the link
2215          * @param &$text String: the link text, modified as necessary
2216          * @return string the full name of the link
2217          * @private
2218          */
2219         function maybeDoSubpageLink( $target, &$text ) {
2220                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2221         }
2222
2223         /**#@+
2224          * Used by doBlockLevels()
2225          * @private
2226          *
2227          * @return string
2228          */
2229         function closeParagraph() {
2230                 $result = '';
2231                 if ( $this->mLastSection != '' ) {
2232                         $result = '</' . $this->mLastSection . ">\n";
2233                 }
2234                 $this->mInPre = false;
2235                 $this->mLastSection = '';
2236                 return $result;
2237         }
2238
2239         /**
2240          * getCommon() returns the length of the longest common substring
2241          * of both arguments, starting at the beginning of both.
2242          * @private
2243          *
2244          * @param $st1 string
2245          * @param $st2 string
2246          *
2247          * @return int
2248          */
2249         function getCommon( $st1, $st2 ) {
2250                 $fl = strlen( $st1 );
2251                 $shorter = strlen( $st2 );
2252                 if ( $fl < $shorter ) {
2253                         $shorter = $fl;
2254                 }
2255
2256                 for ( $i = 0; $i < $shorter; ++$i ) {
2257                         if ( $st1[$i] != $st2[$i] ) {
2258                                 break;
2259                         }
2260                 }
2261                 return $i;
2262         }
2263
2264         /**
2265          * These next three functions open, continue, and close the list
2266          * element appropriate to the prefix character passed into them.
2267          * @private
2268          *
2269          * @param $char string
2270          *
2271          * @return string
2272          */
2273         function openList( $char ) {
2274                 $result = $this->closeParagraph();
2275
2276                 if ( '*' === $char ) {
2277                         $result .= '<ul><li>';
2278                 } elseif ( '#' === $char ) {
2279                         $result .= '<ol><li>';
2280                 } elseif ( ':' === $char ) {
2281                         $result .= '<dl><dd>';
2282                 } elseif ( ';' === $char ) {
2283                         $result .= '<dl><dt>';
2284                         $this->mDTopen = true;
2285                 } else {
2286                         $result = '<!-- ERR 1 -->';
2287                 }
2288
2289                 return $result;
2290         }
2291
2292         /**
2293          * TODO: document
2294          * @param $char String
2295          * @private
2296          *
2297          * @return string
2298          */
2299         function nextItem( $char ) {
2300                 if ( '*' === $char || '#' === $char ) {
2301                         return '</li><li>';
2302                 } elseif ( ':' === $char || ';' === $char ) {
2303                         $close = '</dd>';
2304                         if ( $this->mDTopen ) {
2305                                 $close = '</dt>';
2306                         }
2307                         if ( ';' === $char ) {
2308                                 $this->mDTopen = true;
2309                                 return $close . '<dt>';
2310                         } else {
2311                                 $this->mDTopen = false;
2312                                 return $close . '<dd>';
2313                         }
2314                 }
2315                 return '<!-- ERR 2 -->';
2316         }
2317
2318         /**
2319          * TODO: document
2320          * @param $char String
2321          * @private
2322          *
2323          * @return string
2324          */
2325         function closeList( $char ) {
2326                 if ( '*' === $char ) {
2327                         $text = '</li></ul>';
2328                 } elseif ( '#' === $char ) {
2329                         $text = '</li></ol>';
2330                 } elseif ( ':' === $char ) {
2331                         if ( $this->mDTopen ) {
2332                                 $this->mDTopen = false;
2333                                 $text = '</dt></dl>';
2334                         } else {
2335                                 $text = '</dd></dl>';
2336                         }
2337                 } else {
2338                         return '<!-- ERR 3 -->';
2339                 }
2340                 return $text . "\n";
2341         }
2342         /**#@-*/
2343
2344         /**
2345          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2346          *
2347          * @param $text String
2348          * @param $linestart Boolean: whether or not this is at the start of a line.
2349          * @private
2350          * @return string the lists rendered as HTML
2351          */
2352         function doBlockLevels( $text, $linestart ) {
2353                 wfProfileIn( __METHOD__ );
2354
2355                 # Parsing through the text line by line.  The main thing
2356                 # happening here is handling of block-level elements p, pre,
2357                 # and making lists from lines starting with * # : etc.
2358                 #
2359                 $textLines = StringUtils::explode( "\n", $text );
2360
2361                 $lastPrefix = $output = '';
2362                 $this->mDTopen = $inBlockElem = false;
2363                 $prefixLength = 0;
2364                 $paragraphStack = false;
2365                 $inBlockquote = false;
2366
2367                 foreach ( $textLines as $oLine ) {
2368                         # Fix up $linestart
2369                         if ( !$linestart ) {
2370                                 $output .= $oLine;
2371                                 $linestart = true;
2372                                 continue;
2373                         }
2374                         # * = ul
2375                         # # = ol
2376                         # ; = dt
2377                         # : = dd
2378
2379                         $lastPrefixLength = strlen( $lastPrefix );
2380                         $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2381                         $preOpenMatch = preg_match( '/<pre/i', $oLine );
2382                         # If not in a <pre> element, scan for and figure out what prefixes are there.
2383                         if ( !$this->mInPre ) {
2384                                 # Multiple prefixes may abut each other for nested lists.
2385                                 $prefixLength = strspn( $oLine, '*#:;' );
2386                                 $prefix = substr( $oLine, 0, $prefixLength );
2387
2388                                 # eh?
2389                                 # ; and : are both from definition-lists, so they're equivalent
2390                                 #  for the purposes of determining whether or not we need to open/close
2391                                 #  elements.
2392                                 $prefix2 = str_replace( ';', ':', $prefix );
2393                                 $t = substr( $oLine, $prefixLength );
2394                                 $this->mInPre = (bool)$preOpenMatch;
2395                         } else {
2396                                 # Don't interpret any other prefixes in preformatted text
2397                                 $prefixLength = 0;
2398                                 $prefix = $prefix2 = '';
2399                                 $t = $oLine;
2400                         }
2401
2402                         # List generation
2403                         if ( $prefixLength && $lastPrefix === $prefix2 ) {
2404                                 # Same as the last item, so no need to deal with nesting or opening stuff
2405                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2406                                 $paragraphStack = false;
2407
2408                                 if ( substr( $prefix, -1 ) === ';' ) {
2409                                         # The one nasty exception: definition lists work like this:
2410                                         # ; title : definition text
2411                                         # So we check for : in the remainder text to split up the
2412                                         # title and definition, without b0rking links.
2413                                         $term = $t2 = '';
2414                                         if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2415                                                 $t = $t2;
2416                                                 $output .= $term . $this->nextItem( ':' );
2417                                         }
2418                                 }
2419                         } elseif ( $prefixLength || $lastPrefixLength ) {
2420                                 # We need to open or close prefixes, or both.
2421
2422                                 # Either open or close a level...
2423                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2424                                 $paragraphStack = false;
2425
2426                                 # Close all the prefixes which aren't shared.
2427                                 while ( $commonPrefixLength < $lastPrefixLength ) {
2428                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength - 1] );
2429                                         --$lastPrefixLength;
2430                                 }
2431
2432                                 # Continue the current prefix if appropriate.
2433                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2434                                         $output .= $this->nextItem( $prefix[$commonPrefixLength - 1] );
2435                                 }
2436
2437                                 # Open prefixes where appropriate.
2438                                 while ( $prefixLength > $commonPrefixLength ) {
2439                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2440                                         $output .= $this->openList( $char );
2441
2442                                         if ( ';' === $char ) {
2443                                                 # @todo FIXME: This is dupe of code above
2444                                                 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2445                                                         $t = $t2;
2446                                                         $output .= $term . $this->nextItem( ':' );
2447                                                 }
2448                                         }
2449                                         ++$commonPrefixLength;
2450                                 }
2451                                 $lastPrefix = $prefix2;
2452                         }
2453
2454                         # If we have no prefixes, go to paragraph mode.
2455                         if ( 0 == $prefixLength ) {
2456                                 wfProfileIn( __METHOD__ . "-paragraph" );
2457                                 # No prefix (not in list)--go to paragraph mode
2458                                 # XXX: use a stack for nestable elements like span, table and div
2459                                 $openmatch = preg_match( '/(?:<table|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<dl|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2460                                 $closematch = preg_match(
2461                                         '/(?:<\\/table|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|' .
2462                                         '<td|<th|<\\/?blockquote|<\\/?div|<hr|<\\/pre|<\\/p|' . $this->mUniqPrefix . '-pre|<\\/li|<\\/ul|<\\/ol|<\\/dl|<\\/?center)/iS', $t );
2463                                 if ( $openmatch or $closematch ) {
2464                                         $paragraphStack = false;
2465                                         # TODO bug 5718: paragraph closed
2466                                         $output .= $this->closeParagraph();
2467                                         if ( $preOpenMatch and !$preCloseMatch ) {
2468                                                 $this->mInPre = true;
2469                                         }
2470                                         $bqOffset = 0;
2471                                         while ( preg_match( '/<(\\/?)blockquote[\s>]/i', $t, $bqMatch, PREG_OFFSET_CAPTURE, $bqOffset ) ) {
2472                                                 $inBlockquote = !$bqMatch[1][0]; // is this a close tag?
2473                                                 $bqOffset = $bqMatch[0][1] + strlen( $bqMatch[0][0] );
2474                                         }
2475                                         $inBlockElem = !$closematch;
2476                                 } elseif ( !$inBlockElem && !$this->mInPre ) {
2477                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) and !$inBlockquote ) {
2478                                                 # pre
2479                                                 if ( $this->mLastSection !== 'pre' ) {
2480                                                         $paragraphStack = false;
2481                                                         $output .= $this->closeParagraph() . '<pre>';
2482                                                         $this->mLastSection = 'pre';
2483                                                 }
2484                                                 $t = substr( $t, 1 );
2485                                         } else {
2486                                                 # paragraph
2487                                                 if ( trim( $t ) === '' ) {
2488                                                         if ( $paragraphStack ) {
2489                                                                 $output .= $paragraphStack . '<br />';
2490                                                                 $paragraphStack = false;
2491                                                                 $this->mLastSection = 'p';
2492                                                         } else {
2493                                                                 if ( $this->mLastSection !== 'p' ) {
2494                                                                         $output .= $this->closeParagraph();
2495                                                                         $this->mLastSection = '';
2496                                                                         $paragraphStack = '<p>';
2497                                                                 } else {
2498                                                                         $paragraphStack = '</p><p>';
2499                                                                 }
2500                                                         }
2501                                                 } else {
2502                                                         if ( $paragraphStack ) {
2503                                                                 $output .= $paragraphStack;
2504                                                                 $paragraphStack = false;
2505                                                                 $this->mLastSection = 'p';
2506                                                         } elseif ( $this->mLastSection !== 'p' ) {
2507                                                                 $output .= $this->closeParagraph() . '<p>';
2508                                                                 $this->mLastSection = 'p';
2509                                                         }
2510                                                 }
2511                                         }
2512                                 }
2513                                 wfProfileOut( __METHOD__ . "-paragraph" );
2514                         }
2515                         # somewhere above we forget to get out of pre block (bug 785)
2516                         if ( $preCloseMatch && $this->mInPre ) {
2517                                 $this->mInPre = false;
2518                         }
2519                         if ( $paragraphStack === false ) {
2520                                 $output .= $t . "\n";
2521                         }
2522                 }
2523                 while ( $prefixLength ) {
2524                         $output .= $this->closeList( $prefix2[$prefixLength - 1] );
2525                         --$prefixLength;
2526                 }
2527                 if ( $this->mLastSection != '' ) {
2528                         $output .= '</' . $this->mLastSection . '>';
2529                         $this->mLastSection = '';
2530                 }
2531
2532                 wfProfileOut( __METHOD__ );
2533                 return $output;
2534         }
2535
2536         /**
2537          * Split up a string on ':', ignoring any occurrences inside tags
2538          * to prevent illegal overlapping.
2539          *
2540          * @param string $str the string to split
2541          * @param &$before String set to everything before the ':'
2542          * @param &$after String set to everything after the ':'
2543          * @throws MWException
2544          * @return String the position of the ':', or false if none found
2545          */
2546         function findColonNoLinks( $str, &$before, &$after ) {
2547                 wfProfileIn( __METHOD__ );
2548
2549                 $pos = strpos( $str, ':' );
2550                 if ( $pos === false ) {
2551                         # Nothing to find!
2552                         wfProfileOut( __METHOD__ );
2553                         return false;
2554                 }
2555
2556                 $lt = strpos( $str, '<' );
2557                 if ( $lt === false || $lt > $pos ) {
2558                         # Easy; no tag nesting to worry about
2559                         $before = substr( $str, 0, $pos );
2560                         $after = substr( $str, $pos + 1 );
2561                         wfProfileOut( __METHOD__ );
2562                         return $pos;
2563                 }
2564
2565                 # Ugly state machine to walk through avoiding tags.
2566                 $state = self::COLON_STATE_TEXT;
2567                 $stack = 0;
2568                 $len = strlen( $str );
2569                 for ( $i = 0; $i < $len; $i++ ) {
2570                         $c = $str[$i];
2571
2572                         switch ( $state ) {
2573                         # (Using the number is a performance hack for common cases)
2574                         case 0: # self::COLON_STATE_TEXT:
2575                                 switch ( $c ) {
2576                                 case "<":
2577                                         # Could be either a <start> tag or an </end> tag
2578                                         $state = self::COLON_STATE_TAGSTART;
2579                                         break;
2580                                 case ":":
2581                                         if ( $stack == 0 ) {
2582                                                 # We found it!
2583                                                 $before = substr( $str, 0, $i );
2584                                                 $after = substr( $str, $i + 1 );
2585                                                 wfProfileOut( __METHOD__ );
2586                                                 return $i;
2587                                         }
2588                                         # Embedded in a tag; don't break it.
2589                                         break;
2590                                 default:
2591                                         # Skip ahead looking for something interesting
2592                                         $colon = strpos( $str, ':', $i );
2593                                         if ( $colon === false ) {
2594                                                 # Nothing else interesting
2595                                                 wfProfileOut( __METHOD__ );
2596                                                 return false;
2597                                         }
2598                                         $lt = strpos( $str, '<', $i );
2599                                         if ( $stack === 0 ) {
2600                                                 if ( $lt === false || $colon < $lt ) {
2601                                                         # We found it!
2602                                                         $before = substr( $str, 0, $colon );
2603                                                         $after = substr( $str, $colon + 1 );
2604                                                         wfProfileOut( __METHOD__ );
2605                                                         return $i;
2606                                                 }
2607                                         }
2608                                         if ( $lt === false ) {
2609                                                 # Nothing else interesting to find; abort!
2610                                                 # We're nested, but there's no close tags left. Abort!
2611                                                 break 2;
2612                                         }
2613                                         # Skip ahead to next tag start
2614                                         $i = $lt;
2615                                         $state = self::COLON_STATE_TAGSTART;
2616                                 }
2617                                 break;
2618                         case 1: # self::COLON_STATE_TAG:
2619                                 # In a <tag>
2620                                 switch ( $c ) {
2621                                 case ">":
2622                                         $stack++;
2623                                         $state = self::COLON_STATE_TEXT;
2624                                         break;
2625                                 case "/":
2626                                         # Slash may be followed by >?
2627                                         $state = self::COLON_STATE_TAGSLASH;
2628                                         break;
2629                                 default:
2630                                         # ignore
2631                                 }
2632                                 break;
2633                         case 2: # self::COLON_STATE_TAGSTART:
2634                                 switch ( $c ) {
2635                                 case "/":
2636                                         $state = self::COLON_STATE_CLOSETAG;
2637                                         break;
2638                                 case "!":
2639                                         $state = self::COLON_STATE_COMMENT;
2640                                         break;
2641                                 case ">":
2642                                         # Illegal early close? This shouldn't happen D:
2643                                         $state = self::COLON_STATE_TEXT;
2644                                         break;
2645                                 default:
2646                                         $state = self::COLON_STATE_TAG;
2647                                 }
2648                                 break;
2649                         case 3: # self::COLON_STATE_CLOSETAG:
2650                                 # In a </tag>
2651                                 if ( $c === ">" ) {
2652                                         $stack--;
2653                                         if ( $stack < 0 ) {
2654                                                 wfDebug( __METHOD__ . ": Invalid input; too many close tags\n" );
2655                                                 wfProfileOut( __METHOD__ );
2656                                                 return false;
2657                                         }
2658                                         $state = self::COLON_STATE_TEXT;
2659                                 }
2660                                 break;
2661                         case self::COLON_STATE_TAGSLASH:
2662                                 if ( $c === ">" ) {
2663                                         # Yes, a self-closed tag <blah/>
2664                                         $state = self::COLON_STATE_TEXT;
2665                                 } else {
2666                                         # Probably we're jumping the gun, and this is an attribute
2667                                         $state = self::COLON_STATE_TAG;
2668                                 }
2669                                 break;
2670                         case 5: # self::COLON_STATE_COMMENT:
2671                                 if ( $c === "-" ) {
2672                                         $state = self::COLON_STATE_COMMENTDASH;
2673                                 }
2674                                 break;
2675                         case self::COLON_STATE_COMMENTDASH:
2676                                 if ( $c === "-" ) {
2677                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2678                                 } else {
2679                                         $state = self::COLON_STATE_COMMENT;
2680                                 }
2681                                 break;
2682                         case self::COLON_STATE_COMMENTDASHDASH:
2683                                 if ( $c === ">" ) {
2684                                         $state = self::COLON_STATE_TEXT;
2685                                 } else {
2686                                         $state = self::COLON_STATE_COMMENT;
2687                                 }
2688                                 break;
2689                         default:
2690                                 wfProfileOut( __METHOD__ );
2691                                 throw new MWException( "State machine error in " . __METHOD__ );
2692                         }
2693                 }
2694                 if ( $stack > 0 ) {
2695                         wfDebug( __METHOD__ . ": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2696                         wfProfileOut( __METHOD__ );
2697                         return false;
2698                 }
2699                 wfProfileOut( __METHOD__ );
2700                 return false;
2701         }
2702
2703         /**
2704          * Return value of a magic variable (like PAGENAME)
2705          *
2706          * @private
2707          *
2708          * @param $index integer
2709          * @param bool|\PPFrame $frame
2710          *
2711          * @throws MWException
2712          * @return string
2713          */
2714         function getVariableValue( $index, $frame = false ) {
2715                 global $wgContLang, $wgSitename, $wgServer;
2716                 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2717
2718                 if ( is_null( $this->mTitle ) ) {
2719                         // If no title set, bad things are going to happen
2720                         // later. Title should always be set since this
2721                         // should only be called in the middle of a parse
2722                         // operation (but the unit-tests do funky stuff)
2723                         throw new MWException( __METHOD__ . ' Should only be '
2724                                 . ' called while parsing (no title set)' );
2725                 }
2726
2727                 /**
2728                  * Some of these require message or data lookups and can be
2729                  * expensive to check many times.
2730                  */
2731                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2732                         if ( isset( $this->mVarCache[$index] ) ) {
2733                                 return $this->mVarCache[$index];
2734                         }
2735                 }
2736
2737                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2738                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2739
2740                 $pageLang = $this->getFunctionLang();
2741
2742                 switch ( $index ) {
2743                         case 'currentmonth':
2744                                 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'm' ) );
2745                                 break;
2746                         case 'currentmonth1':
2747                                 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2748                                 break;
2749                         case 'currentmonthname':
2750                                 $value = $pageLang->getMonthName( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2751                                 break;
2752                         case 'currentmonthnamegen':
2753                                 $value = $pageLang->getMonthNameGen( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2754                                 break;
2755                         case 'currentmonthabbrev':
2756                                 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2757                                 break;
2758                         case 'currentday':
2759                                 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'j' ) );
2760                                 break;
2761                         case 'currentday2':
2762                                 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'd' ) );
2763                                 break;
2764                         case 'localmonth':
2765                                 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'm' ) );
2766                                 break;
2767                         case 'localmonth1':
2768                                 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2769                                 break;
2770                         case 'localmonthname':
2771                                 $value = $pageLang->getMonthName( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2772                                 break;
2773                         case 'localmonthnamegen':
2774                                 $value = $pageLang->getMonthNameGen( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2775                                 break;
2776                         case 'localmonthabbrev':
2777                                 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2778                                 break;
2779                         case 'localday':
2780                                 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'j' ) );
2781                                 break;
2782                         case 'localday2':
2783                                 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'd' ) );
2784                                 break;
2785                         case 'pagename':
2786                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2787                                 break;
2788                         case 'pagenamee':
2789                                 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2790                                 break;
2791                         case 'fullpagename':
2792                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2793                                 break;
2794                         case 'fullpagenamee':
2795                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2796                                 break;
2797                         case 'subpagename':
2798                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2799                                 break;
2800                         case 'subpagenamee':
2801                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2802                                 break;
2803                         case 'rootpagename':
2804                                 $value = wfEscapeWikiText( $this->mTitle->getRootText() );
2805                                 break;
2806                         case 'rootpagenamee':
2807                                 $value = wfEscapeWikiText( wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getRootText() ) ) );
2808                                 break;
2809                         case 'basepagename':
2810                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2811                                 break;
2812                         case 'basepagenamee':
2813                                 $value = wfEscapeWikiText( wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) ) );
2814                                 break;
2815                         case 'talkpagename':
2816                                 if ( $this->mTitle->canTalk() ) {
2817                                         $talkPage = $this->mTitle->getTalkPage();
2818                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2819                                 } else {
2820                                         $value = '';
2821                                 }
2822                                 break;
2823                         case 'talkpagenamee':
2824                                 if ( $this->mTitle->canTalk() ) {
2825                                         $talkPage = $this->mTitle->getTalkPage();
2826                                         $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2827                                 } else {
2828                                         $value = '';
2829                                 }
2830                                 break;
2831                         case 'subjectpagename':
2832                                 $subjPage = $this->mTitle->getSubjectPage();
2833                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2834                                 break;
2835                         case 'subjectpagenamee':
2836                                 $subjPage = $this->mTitle->getSubjectPage();
2837                                 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2838                                 break;
2839                         case 'pageid': // requested in bug 23427
2840                                 $pageid = $this->getTitle()->getArticleID();
2841                                 if ( $pageid == 0 ) {
2842                                         # 0 means the page doesn't exist in the database,
2843                                         # which means the user is previewing a new page.
2844                                         # The vary-revision flag must be set, because the magic word
2845                                         # will have a different value once the page is saved.
2846                                         $this->mOutput->setFlag( 'vary-revision' );
2847                                         wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2848                                 }
2849                                 $value = $pageid ? $pageid : null;
2850                                 break;
2851                         case 'revisionid':
2852                                 # Let the edit saving system know we should parse the page
2853                                 # *after* a revision ID has been assigned.
2854                                 $this->mOutput->setFlag( 'vary-revision' );
2855                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2856                                 $value = $this->mRevisionId;
2857                                 break;
2858                         case 'revisionday':
2859                                 # Let the edit saving system know we should parse the page
2860                                 # *after* a revision ID has been assigned. This is for null edits.
2861                                 $this->mOutput->setFlag( 'vary-revision' );
2862                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2863                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2864                                 break;
2865                         case 'revisionday2':
2866                                 # Let the edit saving system know we should parse the page
2867                                 # *after* a revision ID has been assigned. This is for null edits.
2868                                 $this->mOutput->setFlag( 'vary-revision' );
2869                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2870                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2871                                 break;
2872                         case 'revisionmonth':
2873                                 # Let the edit saving system know we should parse the page
2874                                 # *after* a revision ID has been assigned. This is for null edits.
2875                                 $this->mOutput->setFlag( 'vary-revision' );
2876                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2877                                 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2878                                 break;
2879                         case 'revisionmonth1':
2880                                 # Let the edit saving system know we should parse the page
2881                                 # *after* a revision ID has been assigned. This is for null edits.
2882                                 $this->mOutput->setFlag( 'vary-revision' );
2883                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2884                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2885                                 break;
2886                         case 'revisionyear':
2887                                 # Let the edit saving system know we should parse the page
2888                                 # *after* a revision ID has been assigned. This is for null edits.
2889                                 $this->mOutput->setFlag( 'vary-revision' );
2890                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2891                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2892                                 break;
2893                         case 'revisiontimestamp':
2894                                 # Let the edit saving system know we should parse the page
2895                                 # *after* a revision ID has been assigned. This is for null edits.
2896                                 $this->mOutput->setFlag( 'vary-revision' );
2897                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2898                                 $value = $this->getRevisionTimestamp();
2899                                 break;
2900                         case 'revisionuser':
2901                                 # Let the edit saving system know we should parse the page
2902                                 # *after* a revision ID has been assigned. This is for null edits.
2903                                 $this->mOutput->setFlag( 'vary-revision' );
2904                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2905                                 $value = $this->getRevisionUser();
2906                                 break;
2907                         case 'namespace':
2908                                 $value = str_replace( '_', ' ', $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2909                                 break;
2910                         case 'namespacee':
2911                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2912                                 break;
2913                         case 'namespacenumber':
2914                                 $value = $this->mTitle->getNamespace();
2915                                 break;
2916                         case 'talkspace':
2917                                 $value = $this->mTitle->canTalk() ? str_replace( '_', ' ', $this->mTitle->getTalkNsText() ) : '';
2918                                 break;
2919                         case 'talkspacee':
2920                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2921                                 break;
2922                         case 'subjectspace':
2923                                 $value = str_replace( '_', ' ', $this->mTitle->getSubjectNsText() );
2924                                 break;
2925                         case 'subjectspacee':
2926                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2927                                 break;
2928                         case 'currentdayname':
2929                                 $value = $pageLang->getWeekdayName( MWTimestamp::getInstance( $ts )->format( 'w' ) + 1 );
2930                                 break;
2931                         case 'currentyear':
2932                                 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'Y' ), true );
2933                                 break;
2934                         case 'currenttime':
2935                                 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2936                                 break;
2937                         case 'currenthour':
2938                                 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'H' ), true );
2939                                 break;
2940                         case 'currentweek':
2941                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2942                                 # int to remove the padding
2943                                 $value = $pageLang->formatNum( (int)MWTimestamp::getInstance( $ts )->format( 'W' ) );
2944                                 break;
2945                         case 'currentdow':
2946                                 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'w' ) );
2947                                 break;
2948                         case 'localdayname':
2949                                 $value = $pageLang->getWeekdayName( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) + 1 );
2950                                 break;
2951                         case 'localyear':
2952                                 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'Y' ), true );
2953                                 break;
2954                         case 'localtime':
2955                                 $value = $pageLang->time( MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' ), false, false );
2956                                 break;
2957                         case 'localhour':
2958                                 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'H' ), true );
2959                                 break;
2960                         case 'localweek':
2961                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2962                                 # int to remove the padding
2963                                 $value = $pageLang->formatNum( (int)MWTimestamp::getLocalInstance( $ts )->format( 'W' ) );
2964                                 break;
2965                         case 'localdow':
2966                                 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) );
2967                                 break;
2968                         case 'numberofarticles':
2969                                 $value = $pageLang->formatNum( SiteStats::articles() );
2970                                 break;
2971                         case 'numberoffiles':
2972                                 $value = $pageLang->formatNum( SiteStats::images() );
2973                                 break;
2974                         case 'numberofusers':
2975                                 $value = $pageLang->formatNum( SiteStats::users() );
2976                                 break;
2977                         case 'numberofactiveusers':
2978                                 $value = $pageLang->formatNum( SiteStats::activeUsers() );
2979                                 break;
2980                         case 'numberofpages':
2981                                 $value = $pageLang->formatNum( SiteStats::pages() );
2982                                 break;
2983                         case 'numberofadmins':
2984                                 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2985                                 break;
2986                         case 'numberofedits':
2987                                 $value = $pageLang->formatNum( SiteStats::edits() );
2988                                 break;
2989                         case 'numberofviews':
2990                                 global $wgDisableCounters;
2991                                 $value = !$wgDisableCounters ? $pageLang->formatNum( SiteStats::views() ) : '';
2992                                 break;
2993                         case 'currenttimestamp':
2994                                 $value = wfTimestamp( TS_MW, $ts );
2995                                 break;
2996                         case 'localtimestamp':
2997                                 $value = MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' );
2998                                 break;
2999                         case 'currentversion':
3000                                 $value = SpecialVersion::getVersion();
3001                                 break;
3002                         case 'articlepath':
3003                                 return $wgArticlePath;
3004                         case 'sitename':
3005                                 return $wgSitename;
3006                         case 'server':
3007                                 return $wgServer;
3008                         case 'servername':
3009                                 $serverParts = wfParseUrl( $wgServer );
3010                                 return $serverParts && isset( $serverParts['host'] ) ? $serverParts['host'] : $wgServer;
3011                         case 'scriptpath':
3012                                 return $wgScriptPath;
3013                         case 'stylepath':
3014                                 return $wgStylePath;
3015                         case 'directionmark':
3016                                 return $pageLang->getDirMark();
3017                         case 'contentlanguage':
3018                                 global $wgLanguageCode;
3019                                 return $wgLanguageCode;
3020                         default:
3021                                 $ret = null;
3022                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
3023                                         return $ret;
3024                                 } else {
3025                                         return null;
3026                                 }
3027                 }
3028
3029                 if ( $index ) {
3030                         $this->mVarCache[$index] = $value;
3031                 }
3032
3033                 return $value;
3034         }
3035
3036         /**
3037          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
3038          *
3039          * @private
3040          */
3041         function initialiseVariables() {
3042                 wfProfileIn( __METHOD__ );
3043                 $variableIDs = MagicWord::getVariableIDs();
3044                 $substIDs = MagicWord::getSubstIDs();
3045
3046                 $this->mVariables = new MagicWordArray( $variableIDs );
3047                 $this->mSubstWords = new MagicWordArray( $substIDs );
3048                 wfProfileOut( __METHOD__ );
3049         }
3050
3051         /**
3052          * Preprocess some wikitext and return the document tree.
3053          * This is the ghost of replace_variables().
3054          *
3055          * @param string $text The text to parse
3056          * @param $flags Integer: bitwise combination of:
3057          *          self::PTD_FOR_INCLUSION    Handle "<noinclude>" and "<includeonly>" as if the text is being
3058          *                                     included. Default is to assume a direct page view.
3059          *
3060          * The generated DOM tree must depend only on the input text and the flags.
3061          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
3062          *
3063          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
3064          * change in the DOM tree for a given text, must be passed through the section identifier
3065          * in the section edit link and thus back to extractSections().
3066          *
3067          * The output of this function is currently only cached in process memory, but a persistent
3068          * cache may be implemented at a later date which takes further advantage of these strict
3069          * dependency requirements.
3070          *
3071          * @private
3072          *
3073          * @return PPNode
3074          */
3075         function preprocessToDom( $text, $flags = 0 ) {
3076                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
3077                 return $dom;
3078         }
3079
3080         /**
3081          * Return a three-element array: leading whitespace, string contents, trailing whitespace
3082          *
3083          * @param $s string
3084          *
3085          * @return array
3086          */
3087         public static function splitWhitespace( $s ) {
3088                 $ltrimmed = ltrim( $s );
3089                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
3090                 $trimmed = rtrim( $ltrimmed );
3091                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
3092                 if ( $diff > 0 ) {
3093                         $w2 = substr( $ltrimmed, -$diff );
3094                 } else {
3095                         $w2 = '';
3096                 }
3097                 return array( $w1, $trimmed, $w2 );
3098         }
3099
3100         /**
3101          * Replace magic variables, templates, and template arguments
3102          * with the appropriate text. Templates are substituted recursively,
3103          * taking care to avoid infinite loops.
3104          *
3105          * Note that the substitution depends on value of $mOutputType:
3106          *  self::OT_WIKI: only {{subst:}} templates
3107          *  self::OT_PREPROCESS: templates but not extension tags
3108          *  self::OT_HTML: all templates and extension tags
3109          *
3110          * @param string $text the text to transform
3111          * @param $frame PPFrame Object describing the arguments passed to the template.
3112          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
3113          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
3114          * @param $argsOnly Boolean only do argument (triple-brace) expansion, not double-brace expansion
3115          * @private
3116          *
3117          * @return string
3118          */
3119         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3120                 # Is there any text? Also, Prevent too big inclusions!
3121                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
3122                         return $text;
3123                 }
3124                 wfProfileIn( __METHOD__ );
3125
3126                 if ( $frame === false ) {
3127                         $frame = $this->getPreprocessor()->newFrame();
3128                 } elseif ( !( $frame instanceof PPFrame ) ) {
3129                         wfDebug( __METHOD__ . " called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
3130                         $frame = $this->getPreprocessor()->newCustomFrame( $frame );
3131                 }
3132
3133                 $dom = $this->preprocessToDom( $text );
3134                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
3135                 $text = $frame->expand( $dom, $flags );
3136
3137                 wfProfileOut( __METHOD__ );
3138                 return $text;
3139         }
3140
3141         /**
3142          * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3143          *
3144          * @param $args array
3145          *
3146          * @return array
3147          */
3148         static function createAssocArgs( $args ) {
3149                 $assocArgs = array();
3150                 $index = 1;
3151                 foreach ( $args as $arg ) {
3152                         $eqpos = strpos( $arg, '=' );
3153                         if ( $eqpos === false ) {
3154                                 $assocArgs[$index++] = $arg;
3155                         } else {
3156                                 $name = trim( substr( $arg, 0, $eqpos ) );
3157                                 $value = trim( substr( $arg, $eqpos + 1 ) );
3158                                 if ( $value === false ) {
3159                                         $value = '';
3160                                 }
3161                                 if ( $name !== false ) {
3162                                         $assocArgs[$name] = $value;
3163                                 }
3164                         }
3165                 }
3166
3167                 return $assocArgs;
3168         }
3169
3170         /**
3171          * Warn the user when a parser limitation is reached
3172          * Will warn at most once the user per limitation type
3173          *
3174          * @param string $limitationType should be one of:
3175          *   'expensive-parserfunction' (corresponding messages:
3176          *       'expensive-parserfunction-warning',
3177          *       'expensive-parserfunction-category')
3178          *   'post-expand-template-argument' (corresponding messages:
3179          *       'post-expand-template-argument-warning',
3180          *       'post-expand-template-argument-category')
3181          *   'post-expand-template-inclusion' (corresponding messages:
3182          *       'post-expand-template-inclusion-warning',
3183          *       'post-expand-template-inclusion-category')
3184          *   'node-count-exceeded' (corresponding messages:
3185          *       'node-count-exceeded-warning',
3186          *       'node-count-exceeded-category')
3187          *   'expansion-depth-exceeded' (corresponding messages:
3188          *       'expansion-depth-exceeded-warning',
3189          *       'expansion-depth-exceeded-category')
3190          * @param int|null $current Current value
3191          * @param int|null $max Maximum allowed, when an explicit limit has been
3192          *       exceeded, provide the values (optional)
3193          */
3194         function limitationWarn( $limitationType, $current = '', $max = '' ) {
3195                 # does no harm if $current and $max are present but are unnecessary for the message
3196                 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
3197                         ->inLanguage( $this->mOptions->getUserLangObj() )->text();
3198                 $this->mOutput->addWarning( $warning );
3199                 $this->addTrackingCategory( "$limitationType-category" );
3200         }
3201
3202         /**
3203          * Return the text of a template, after recursively
3204          * replacing any variables or templates within the template.
3205          *
3206          * @param array $piece the parts of the template
3207          *  $piece['title']: the title, i.e. the part before the |
3208          *  $piece['parts']: the parameter array
3209          *  $piece['lineStart']: whether the brace was at the start of a line
3210          * @param $frame PPFrame The current frame, contains template arguments
3211          * @throws MWException
3212          * @return String: the text of the template
3213          * @private
3214          */
3215         function braceSubstitution( $piece, $frame ) {
3216                 wfProfileIn( __METHOD__ );
3217                 wfProfileIn( __METHOD__ . '-setup' );
3218
3219                 # Flags
3220                 $found = false;             # $text has been filled
3221                 $nowiki = false;            # wiki markup in $text should be escaped
3222                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
3223                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
3224                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
3225                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
3226
3227                 # Title object, where $text came from
3228                 $title = false;
3229
3230                 # $part1 is the bit before the first |, and must contain only title characters.
3231                 # Various prefixes will be stripped from it later.
3232                 $titleWithSpaces = $frame->expand( $piece['title'] );
3233                 $part1 = trim( $titleWithSpaces );
3234                 $titleText = false;
3235
3236                 # Original title text preserved for various purposes
3237                 $originalTitle = $part1;
3238
3239                 # $args is a list of argument nodes, starting from index 0, not including $part1
3240                 # @todo FIXME: If piece['parts'] is null then the call to getLength() below won't work b/c this $args isn't an object
3241                 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
3242                 wfProfileOut( __METHOD__ . '-setup' );
3243
3244                 $titleProfileIn = null; // profile templates
3245
3246                 # SUBST
3247                 wfProfileIn( __METHOD__ . '-modifiers' );
3248                 if ( !$found ) {
3249
3250                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3251
3252                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
3253                         # Decide whether to expand template or keep wikitext as-is.
3254                         if ( $this->ot['wiki'] ) {
3255                                 if ( $substMatch === false ) {
3256                                         $literal = true;  # literal when in PST with no prefix
3257                                 } else {
3258                                         $literal = false; # expand when in PST with subst: or safesubst:
3259                                 }
3260                         } else {
3261                                 if ( $substMatch == 'subst' ) {
3262                                         $literal = true;  # literal when not in PST with plain subst:
3263                                 } else {
3264                                         $literal = false; # expand when not in PST with safesubst: or no prefix
3265                                 }
3266                         }
3267                         if ( $literal ) {
3268                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3269                                 $isLocalObj = true;
3270                                 $found = true;
3271                         }
3272                 }
3273
3274                 # Variables
3275                 if ( !$found && $args->getLength() == 0 ) {
3276                         $id = $this->mVariables->matchStartToEnd( $part1 );
3277                         if ( $id !== false ) {
3278                                 $text = $this->getVariableValue( $id, $frame );
3279                                 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3280                                         $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3281                                 }
3282                                 $found = true;
3283                         }
3284                 }
3285
3286                 # MSG, MSGNW and RAW
3287                 if ( !$found ) {
3288                         # Check for MSGNW:
3289                         $mwMsgnw = MagicWord::get( 'msgnw' );
3290                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3291                                 $nowiki = true;
3292                         } else {
3293                                 # Remove obsolete MSG:
3294                                 $mwMsg = MagicWord::get( 'msg' );
3295                                 $mwMsg->matchStartAndRemove( $part1 );
3296                         }
3297
3298                         # Check for RAW:
3299                         $mwRaw = MagicWord::get( 'raw' );
3300                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3301                                 $forceRawInterwiki = true;
3302                         }
3303                 }
3304                 wfProfileOut( __METHOD__ . '-modifiers' );
3305
3306                 # Parser functions
3307                 if ( !$found ) {
3308                         wfProfileIn( __METHOD__ . '-pfunc' );
3309
3310                         $colonPos = strpos( $part1, ':' );
3311                         if ( $colonPos !== false ) {
3312                                 $func = substr( $part1, 0, $colonPos );
3313                                 $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3314                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3315                                         $funcArgs[] = $args->item( $i );
3316                                 }
3317                                 try {
3318                                         $result = $this->callParserFunction( $frame, $func, $funcArgs );
3319                                 } catch ( Exception $ex ) {
3320                                         wfProfileOut( __METHOD__ . '-pfunc' );
3321                                         wfProfileOut( __METHOD__ );
3322                                         throw $ex;
3323                                 }
3324
3325                                 # The interface for parser functions allows for extracting
3326                                 # flags into the local scope. Extract any forwarded flags
3327                                 # here.
3328                                 extract( $result );
3329                         }
3330                         wfProfileOut( __METHOD__ . '-pfunc' );
3331                 }
3332
3333                 # Finish mangling title and then check for loops.
3334                 # Set $title to a Title object and $titleText to the PDBK
3335                 if ( !$found ) {
3336                         $ns = NS_TEMPLATE;
3337                         # Split the title into page and subpage
3338                         $subpage = '';
3339                         $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3340                         if ( $part1 !== $relative ) {
3341                                 $part1 = $relative;
3342                                 $ns = $this->mTitle->getNamespace();
3343                         }
3344                         $title = Title::newFromText( $part1, $ns );
3345                         if ( $title ) {
3346                                 $titleText = $title->getPrefixedText();
3347                                 # Check for language variants if the template is not found
3348                                 if ( $this->getConverterLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3349                                         $this->getConverterLanguage()->findVariantLink( $part1, $title, true );
3350                                 }
3351                                 # Do recursion depth check
3352                                 $limit = $this->mOptions->getMaxTemplateDepth();
3353                                 if ( $frame->depth >= $limit ) {
3354                                         $found = true;
3355                                         $text = '<span class="error">'
3356                                                 . wfMessage( 'parser-template-recursion-depth-warning' )
3357                                                         ->numParams( $limit )->inContentLanguage()->text()
3358                                                 . '</span>';
3359                                 }
3360                         }
3361                 }
3362
3363                 # Load from database
3364                 if ( !$found && $title ) {
3365                         if ( !Profiler::instance()->isPersistent() ) {
3366                                 # Too many unique items can kill profiling DBs/collectors
3367                                 $titleProfileIn = __METHOD__ . "-title-" . $title->getPrefixedDBkey();
3368                                 wfProfileIn( $titleProfileIn ); // template in
3369                         }
3370                         wfProfileIn( __METHOD__ . '-loadtpl' );
3371                         if ( !$title->isExternal() ) {
3372                                 if ( $title->isSpecialPage()
3373                                         && $this->mOptions->getAllowSpecialInclusion()
3374                                         && $this->ot['html'] )
3375                                 {
3376                                         // Pass the template arguments as URL parameters.
3377                                         // "uselang" will have no effect since the Language object
3378                                         // is forced to the one defined in ParserOptions.
3379                                         $pageArgs = array();
3380                                         for ( $i = 0; $i < $args->getLength(); $i++ ) {
3381                                                 $bits = $args->item( $i )->splitArg();
3382                                                 if ( strval( $bits['index'] ) === '' ) {
3383                                                         $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3384                                                         $value = trim( $frame->expand( $bits['value'] ) );
3385                                                         $pageArgs[$name] = $value;
3386                                                 }
3387                                         }
3388
3389                                         // Create a new context to execute the special page
3390                                         $context = new RequestContext;
3391                                         $context->setTitle( $title );
3392                                         $context->setRequest( new FauxRequest( $pageArgs ) );
3393                                         $context->setUser( $this->getUser() );
3394                                         $context->setLanguage( $this->mOptions->getUserLangObj() );
3395                                         $ret = SpecialPageFactory::capturePath( $title, $context );
3396                                         if ( $ret ) {
3397                                                 $text = $context->getOutput()->getHTML();
3398                                                 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3399                                                 $found = true;
3400                                                 $isHTML = true;
3401                                                 $this->disableCache();
3402                                         }
3403                                 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3404                                         $found = false; # access denied
3405                                         wfDebug( __METHOD__ . ": template inclusion denied for " . $title->getPrefixedDBkey() );
3406                                 } else {
3407                                         list( $text, $title ) = $this->getTemplateDom( $title );
3408                                         if ( $text !== false ) {
3409                                                 $found = true;
3410                                                 $isChildObj = true;
3411                                         }
3412                                 }
3413
3414                                 # If the title is valid but undisplayable, make a link to it
3415                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3416                                         $text = "[[:$titleText]]";
3417                                         $found = true;
3418                                 }
3419                         } elseif ( $title->isTrans() ) {
3420                                 # Interwiki transclusion
3421                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3422                                         $text = $this->interwikiTransclude( $title, 'render' );
3423                                         $isHTML = true;
3424                                 } else {
3425                                         $text = $this->interwikiTransclude( $title, 'raw' );
3426                                         # Preprocess it like a template
3427                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3428                                         $isChildObj = true;
3429                                 }
3430                                 $found = true;
3431                         }
3432
3433                         # Do infinite loop check
3434                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3435                         if ( !$frame->loopCheck( $title ) ) {
3436                                 $found = true;
3437                                 $text = '<span class="error">'
3438                                         . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3439                                         . '</span>';
3440                                 wfDebug( __METHOD__ . ": template loop broken at '$titleText'\n" );
3441                         }
3442                         wfProfileOut( __METHOD__ . '-loadtpl' );
3443                 }
3444
3445                 # If we haven't found text to substitute by now, we're done
3446                 # Recover the source wikitext and return it
3447                 if ( !$found ) {
3448                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3449                         if ( $titleProfileIn ) {
3450                                 wfProfileOut( $titleProfileIn ); // template out
3451                         }
3452                         wfProfileOut( __METHOD__ );
3453                         return array( 'object' => $text );
3454                 }
3455
3456                 # Expand DOM-style return values in a child frame
3457                 if ( $isChildObj ) {
3458                         # Clean up argument array
3459                         $newFrame = $frame->newChild( $args, $title );
3460
3461                         if ( $nowiki ) {
3462                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3463                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3464                                 # Expansion is eligible for the empty-frame cache
3465                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3466                                         $text = $this->mTplExpandCache[$titleText];
3467                                 } else {
3468                                         $text = $newFrame->expand( $text );
3469                                         $this->mTplExpandCache[$titleText] = $text;
3470                                 }
3471                         } else {
3472                                 # Uncached expansion
3473                                 $text = $newFrame->expand( $text );
3474                         }
3475                 }
3476                 if ( $isLocalObj && $nowiki ) {
3477                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3478                         $isLocalObj = false;
3479                 }
3480
3481                 if ( $titleProfileIn ) {
3482                         wfProfileOut( $titleProfileIn ); // template out
3483                 }
3484
3485                 # Replace raw HTML by a placeholder
3486                 if ( $isHTML ) {
3487                         $text = $this->insertStripItem( $text );
3488                 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3489                         # Escape nowiki-style return values
3490                         $text = wfEscapeWikiText( $text );
3491                 } elseif ( is_string( $text )
3492                         && !$piece['lineStart']
3493                         && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3494                 {
3495                         # Bug 529: if the template begins with a table or block-level
3496                         # element, it should be treated as beginning a new line.
3497                         # This behavior is somewhat controversial.
3498                         $text = "\n" . $text;
3499                 }
3500
3501                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3502                         # Error, oversize inclusion
3503                         if ( $titleText !== false ) {
3504                                 # Make a working, properly escaped link if possible (bug 23588)
3505                                 $text = "[[:$titleText]]";
3506                         } else {
3507                                 # This will probably not be a working link, but at least it may
3508                                 # provide some hint of where the problem is
3509                                 preg_replace( '/^:/', '', $originalTitle );
3510                                 $text = "[[:$originalTitle]]";
3511                         }
3512                         $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3513                         $this->limitationWarn( 'post-expand-template-inclusion' );
3514                 }
3515
3516                 if ( $isLocalObj ) {
3517                         $ret = array( 'object' => $text );
3518                 } else {
3519                         $ret = array( 'text' => $text );
3520                 }
3521
3522                 wfProfileOut( __METHOD__ );
3523                 return $ret;
3524         }
3525
3526         /**
3527          * Call a parser function and return an array with text and flags.
3528          *
3529          * The returned array will always contain a boolean 'found', indicating
3530          * whether the parser function was found or not. It may also contain the
3531          * following:
3532          *  text: string|object, resulting wikitext or PP DOM object
3533          *  isHTML: bool, $text is HTML, armour it against wikitext transformation
3534          *  isChildObj: bool, $text is a DOM node needing expansion in a child frame
3535          *  isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3536          *  nowiki: bool, wiki markup in $text should be escaped
3537          *
3538          * @since 1.21
3539          * @param $frame PPFrame The current frame, contains template arguments
3540          * @param $function string Function name
3541          * @param $args array Arguments to the function
3542          * @return array
3543          */
3544         public function callParserFunction( $frame, $function, array $args = array() ) {
3545                 global $wgContLang;
3546
3547                 wfProfileIn( __METHOD__ );
3548
3549                 # Case sensitive functions
3550                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3551                         $function = $this->mFunctionSynonyms[1][$function];
3552                 } else {
3553                         # Case insensitive functions
3554                         $function = $wgContLang->lc( $function );
3555                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3556                                 $function = $this->mFunctionSynonyms[0][$function];
3557                         } else {
3558                                 wfProfileOut( __METHOD__ );
3559                                 return array( 'found' => false );
3560                         }
3561                 }
3562
3563                 wfProfileIn( __METHOD__ . '-pfunc-' . $function );
3564                 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3565
3566                 # Workaround for PHP bug 35229 and similar
3567                 if ( !is_callable( $callback ) ) {
3568                         wfProfileOut( __METHOD__ . '-pfunc-' . $function );
3569                         wfProfileOut( __METHOD__ );
3570                         throw new MWException( "Tag hook for $function is not callable\n" );
3571                 }
3572
3573                 $allArgs = array( &$this );
3574                 if ( $flags & SFH_OBJECT_ARGS ) {
3575                         # Convert arguments to PPNodes and collect for appending to $allArgs
3576                         $funcArgs = array();
3577                         foreach ( $args as $k => $v ) {
3578                                 if ( $v instanceof PPNode || $k === 0 ) {
3579                                         $funcArgs[] = $v;
3580                                 } else {
3581                                         $funcArgs[] = $this->mPreprocessor->newPartNodeArray( array( $k => $v ) )->item( 0 );
3582                                 }
3583                         }
3584
3585                         # Add a frame parameter, and pass the arguments as an array
3586                         $allArgs[] = $frame;
3587                         $allArgs[] = $funcArgs;
3588                 } else {
3589                         # Convert arguments to plain text and append to $allArgs
3590                         foreach ( $args as $k => $v ) {
3591                                 if ( $v instanceof PPNode ) {
3592                                         $allArgs[] = trim( $frame->expand( $v ) );
3593                                 } elseif ( is_int( $k ) && $k >= 0 ) {
3594                                         $allArgs[] = trim( $v );
3595                                 } else {
3596                                         $allArgs[] = trim( "$k=$v" );
3597                                 }
3598                         }
3599                 }
3600
3601                 $result = call_user_func_array( $callback, $allArgs );
3602
3603                 # The interface for function hooks allows them to return a wikitext
3604                 # string or an array containing the string and any flags. This mungs
3605                 # things around to match what this method should return.
3606                 if ( !is_array( $result ) ) {
3607                         $result = array(
3608                                 'found' => true,
3609                                 'text' => $result,
3610                         );
3611                 } else {
3612                         if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3613                                 $result['text'] = $result[0];
3614                         }
3615                         unset( $result[0] );
3616                         $result += array(
3617                                 'found' => true,
3618                         );
3619                 }
3620
3621                 $noparse = true;
3622                 $preprocessFlags = 0;
3623                 if ( isset( $result['noparse'] ) ) {
3624                         $noparse = $result['noparse'];
3625                 }
3626                 if ( isset( $result['preprocessFlags'] ) ) {
3627                         $preprocessFlags = $result['preprocessFlags'];
3628                 }
3629
3630                 if ( !$noparse ) {
3631                         $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3632                         $result['isChildObj'] = true;
3633                 }
3634                 wfProfileOut( __METHOD__ . '-pfunc-' . $function );
3635                 wfProfileOut( __METHOD__ );
3636
3637                 return $result;
3638         }
3639
3640         /**
3641          * Get the semi-parsed DOM representation of a template with a given title,
3642          * and its redirect destination title. Cached.
3643          *
3644          * @param $title Title
3645          *
3646          * @return array
3647          */
3648         function getTemplateDom( $title ) {
3649                 $cacheTitle = $title;
3650                 $titleText = $title->getPrefixedDBkey();
3651
3652                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3653                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3654                         $title = Title::makeTitle( $ns, $dbk );
3655                         $titleText = $title->getPrefixedDBkey();
3656                 }
3657                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3658                         return array( $this->mTplDomCache[$titleText], $title );
3659                 }
3660
3661                 # Cache miss, go to the database
3662                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3663
3664                 if ( $text === false ) {
3665                         $this->mTplDomCache[$titleText] = false;
3666                         return array( false, $title );
3667                 }
3668
3669                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3670                 $this->mTplDomCache[$titleText] = $dom;
3671
3672                 if ( !$title->equals( $cacheTitle ) ) {
3673                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3674                                 array( $title->getNamespace(), $cdb = $title->getDBkey() );
3675                 }
3676
3677                 return array( $dom, $title );
3678         }
3679
3680         /**
3681          * Fetch the unparsed text of a template and register a reference to it.
3682          * @param Title $title
3683          * @return Array ( string or false, Title )
3684          */
3685         function fetchTemplateAndTitle( $title ) {
3686                 $templateCb = $this->mOptions->getTemplateCallback(); # Defaults to Parser::statelessFetchTemplate()
3687                 $stuff = call_user_func( $templateCb, $title, $this );
3688                 $text = $stuff['text'];
3689                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3690                 if ( isset( $stuff['deps'] ) ) {
3691                         foreach ( $stuff['deps'] as $dep ) {
3692                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3693                                 if ( $dep['title']->equals( $this->getTitle() ) ) {
3694                                         // If we transclude ourselves, the final result
3695                                         // will change based on the new version of the page
3696                                         $this->mOutput->setFlag( 'vary-revision' );
3697                                 }
3698                         }
3699                 }
3700                 return array( $text, $finalTitle );
3701         }
3702
3703         /**
3704          * Fetch the unparsed text of a template and register a reference to it.
3705          * @param Title $title
3706          * @return mixed string or false
3707          */
3708         function fetchTemplate( $title ) {
3709                 $rv = $this->fetchTemplateAndTitle( $title );
3710                 return $rv[0];
3711         }
3712
3713         /**
3714          * Static function to get a template
3715          * Can be overridden via ParserOptions::setTemplateCallback().
3716          *
3717          * @param $title  Title
3718          * @param $parser Parser
3719          *
3720          * @return array
3721          */
3722         static function statelessFetchTemplate( $title, $parser = false ) {
3723                 $text = $skip = false;
3724                 $finalTitle = $title;
3725                 $deps = array();
3726
3727                 # Loop to fetch the article, with up to 1 redirect
3728                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3729                         # Give extensions a chance to select the revision instead
3730                         $id = false; # Assume current
3731                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle',
3732                                 array( $parser, $title, &$skip, &$id ) );
3733
3734                         if ( $skip ) {
3735                                 $text = false;
3736                                 $deps[] = array(
3737                                         'title' => $title,
3738                                         'page_id' => $title->getArticleID(),
3739                                         'rev_id' => null
3740                                 );
3741                                 break;
3742                         }
3743                         # Get the revision
3744                         $rev = $id
3745                                 ? Revision::newFromId( $id )
3746                                 : Revision::newFromTitle( $title, false, Revision::READ_NORMAL );
3747                         $rev_id = $rev ? $rev->getId() : 0;
3748                         # If there is no current revision, there is no page
3749                         if ( $id === false && !$rev ) {
3750                                 $linkCache = LinkCache::singleton();
3751                                 $linkCache->addBadLinkObj( $title );
3752                         }
3753
3754                         $deps[] = array(
3755                                 'title' => $title,
3756                                 'page_id' => $title->getArticleID(),
3757                                 'rev_id' => $rev_id );
3758                         if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3759                                 # We fetched a rev from a different title; register it too...
3760                                 $deps[] = array(
3761                                         'title' => $rev->getTitle(),
3762                                         'page_id' => $rev->getPage(),
3763                                         'rev_id' => $rev_id );
3764                         }
3765
3766                         if ( $rev ) {
3767                                 $content = $rev->getContent();
3768                                 $text = $content ? $content->getWikitextForTransclusion() : null;
3769
3770                                 if ( $text === false || $text === null ) {
3771                                         $text = false;
3772                                         break;
3773                                 }
3774                         } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3775                                 global $wgContLang;
3776                                 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3777                                 if ( !$message->exists() ) {
3778                                         $text = false;
3779                                         break;
3780                                 }
3781                                 $content = $message->content();
3782                                 $text = $message->plain();
3783                         } else {
3784                                 break;
3785                         }
3786                         if ( !$content ) {
3787                                 break;
3788                         }
3789                         # Redirect?
3790                         $finalTitle = $title;
3791                         $title = $content->getRedirectTarget();
3792                 }
3793                 return array(
3794                         'text' => $text,
3795                         'finalTitle' => $finalTitle,
3796                         'deps' => $deps );
3797         }
3798
3799         /**
3800          * Fetch a file and its title and register a reference to it.
3801          * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3802          * @param Title $title
3803          * @param array $options Array of options to RepoGroup::findFile
3804          * @return File|bool
3805          */
3806         function fetchFile( $title, $options = array() ) {
3807                 $res = $this->fetchFileAndTitle( $title, $options );
3808                 return $res[0];
3809         }
3810
3811         /**
3812          * Fetch a file and its title and register a reference to it.
3813          * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3814          * @param Title $title
3815          * @param array $options Array of options to RepoGroup::findFile
3816          * @return Array ( File or false, Title of file )
3817          */
3818         function fetchFileAndTitle( $title, $options = array() ) {
3819                 $file = $this->fetchFileNoRegister( $title, $options );
3820
3821                 $time = $file ? $file->getTimestamp() : false;
3822                 $sha1 = $file ? $file->getSha1() : false;
3823                 # Register the file as a dependency...
3824                 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3825                 if ( $file && !$title->equals( $file->getTitle() ) ) {
3826                         # Update fetched file title
3827                         $title = $file->getTitle();
3828                         if ( is_null( $file->getRedirectedTitle() ) ) {
3829                                 # This file was not a redirect, but the title does not match.
3830                                 # Register under the new name because otherwise the link will
3831                                 # get lost.
3832                                 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3833                         }
3834                 }
3835                 return array( $file, $title );
3836         }
3837
3838         /**
3839          * Helper function for fetchFileAndTitle.
3840          *
3841          * Also useful if you need to fetch a file but not use it yet,
3842          * for example to get the file's handler.
3843          *
3844          * @param Title $title
3845          * @param array $options Array of options to RepoGroup::findFile
3846          * @return File or false
3847          */
3848         protected function fetchFileNoRegister( $title, $options = array() ) {
3849                 if ( isset( $options['broken'] ) ) {
3850                         $file = false; // broken thumbnail forced by hook
3851                 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3852                         $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3853                 } else { // get by (name,timestamp)
3854                         $file = wfFindFile( $title, $options );
3855                 }
3856                 return $file;
3857         }
3858
3859         /**
3860          * Transclude an interwiki link.
3861          *
3862          * @param $title Title
3863          * @param $action
3864          *
3865          * @return string
3866          */
3867         function interwikiTransclude( $title, $action ) {
3868                 global $wgEnableScaryTranscluding;
3869
3870                 if ( !$wgEnableScaryTranscluding ) {
3871                         return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3872                 }
3873
3874                 $url = $title->getFullURL( array( 'action' => $action ) );
3875
3876                 if ( strlen( $url ) > 255 ) {
3877                         return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3878                 }
3879                 return $this->fetchScaryTemplateMaybeFromCache( $url );
3880         }
3881
3882         /**
3883          * @param $url string
3884          * @return Mixed|String
3885          */
3886         function fetchScaryTemplateMaybeFromCache( $url ) {
3887                 global $wgTranscludeCacheExpiry;
3888                 $dbr = wfGetDB( DB_SLAVE );
3889                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3890                 $obj = $dbr->selectRow( 'transcache', array( 'tc_time', 'tc_contents' ),
3891                                 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3892                 if ( $obj ) {
3893                         return $obj->tc_contents;
3894                 }
3895
3896                 $req = MWHttpRequest::factory( $url );
3897                 $status = $req->execute(); // Status object
3898                 if ( $status->isOK() ) {
3899                         $text = $req->getContent();
3900                 } elseif ( $req->getStatus() != 200 ) { // Though we failed to fetch the content, this status is useless.
3901                         return wfMessage( 'scarytranscludefailed-httpstatus', $url, $req->getStatus() /* HTTP status */ )->inContentLanguage()->text();
3902                 } else {
3903                         return wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3904                 }
3905
3906                 $dbw = wfGetDB( DB_MASTER );
3907                 $dbw->replace( 'transcache', array( 'tc_url' ), array(
3908                         'tc_url' => $url,
3909                         'tc_time' => $dbw->timestamp( time() ),
3910                         'tc_contents' => $text
3911                 ) );
3912                 return $text;
3913         }
3914
3915         /**
3916          * Triple brace replacement -- used for template arguments
3917          * @private
3918          *
3919          * @param $piece array
3920          * @param $frame PPFrame
3921          *
3922          * @return array
3923          */
3924         function argSubstitution( $piece, $frame ) {
3925                 wfProfileIn( __METHOD__ );
3926
3927                 $error = false;
3928                 $parts = $piece['parts'];
3929                 $nameWithSpaces = $frame->expand( $piece['title'] );
3930                 $argName = trim( $nameWithSpaces );
3931                 $object = false;
3932                 $text = $frame->getArgument( $argName );
3933                 if ( $text === false && $parts->getLength() > 0
3934                   && (
3935                         $this->ot['html']
3936                         || $this->ot['pre']
3937                         || ( $this->ot['wiki'] && $frame->isTemplate() )
3938                   )
3939                 ) {
3940                         # No match in frame, use the supplied default
3941                         $object = $parts->item( 0 )->getChildren();
3942                 }
3943                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3944                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3945                         $this->limitationWarn( 'post-expand-template-argument' );
3946                 }
3947
3948                 if ( $text === false && $object === false ) {
3949                         # No match anywhere
3950                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3951                 }
3952                 if ( $error !== false ) {
3953                         $text .= $error;
3954                 }
3955                 if ( $object !== false ) {
3956                         $ret = array( 'object' => $object );
3957                 } else {
3958                         $ret = array( 'text' => $text );
3959                 }
3960
3961                 wfProfileOut( __METHOD__ );
3962                 return $ret;
3963         }
3964
3965         /**
3966          * Return the text to be used for a given extension tag.
3967          * This is the ghost of strip().
3968          *
3969          * @param array $params Associative array of parameters:
3970          *     name       PPNode for the tag name
3971          *     attr       PPNode for unparsed text where tag attributes are thought to be
3972          *     attributes Optional associative array of parsed attributes
3973          *     inner      Contents of extension element
3974          *     noClose    Original text did not have a close tag
3975          * @param $frame PPFrame
3976          *
3977          * @throws MWException
3978          * @return string
3979          */
3980         function extensionSubstitution( $params, $frame ) {
3981                 $name = $frame->expand( $params['name'] );
3982                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3983                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3984                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3985
3986                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
3987                         ( $this->ot['html'] || $this->ot['pre'] );
3988                 if ( $isFunctionTag ) {
3989                         $markerType = 'none';
3990                 } else {
3991                         $markerType = 'general';
3992                 }
3993                 if ( $this->ot['html'] || $isFunctionTag ) {
3994                         $name = strtolower( $name );
3995                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3996                         if ( isset( $params['attributes'] ) ) {
3997                                 $attributes = $attributes + $params['attributes'];
3998                         }
3999
4000                         if ( isset( $this->mTagHooks[$name] ) ) {
4001                                 # Workaround for PHP bug 35229 and similar
4002                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
4003                                         throw new MWException( "Tag hook for $name is not callable\n" );
4004                                 }
4005                                 $output = call_user_func_array( $this->mTagHooks[$name],
4006                                         array( $content, $attributes, $this, $frame ) );
4007                         } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
4008                                 list( $callback, ) = $this->mFunctionTagHooks[$name];
4009                                 if ( !is_callable( $callback ) ) {
4010                                         throw new MWException( "Tag hook for $name is not callable\n" );
4011                                 }
4012
4013                                 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
4014                         } else {
4015                                 $output = '<span class="error">Invalid tag extension name: ' .
4016                                         htmlspecialchars( $name ) . '</span>';
4017                         }
4018
4019                         if ( is_array( $output ) ) {
4020                                 # Extract flags to local scope (to override $markerType)
4021                                 $flags = $output;
4022                                 $output = $flags[0];
4023                                 unset( $flags[0] );
4024                                 extract( $flags );
4025                         }
4026                 } else {
4027                         if ( is_null( $attrText ) ) {
4028                                 $attrText = '';
4029                         }
4030                         if ( isset( $params['attributes'] ) ) {
4031                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
4032                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
4033                                                 htmlspecialchars( $attrValue ) . '"';
4034                                 }
4035                         }
4036                         if ( $content === null ) {
4037                                 $output = "<$name$attrText/>";
4038                         } else {
4039                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
4040                                 $output = "<$name$attrText>$content$close";
4041                         }
4042                 }
4043
4044                 if ( $markerType === 'none' ) {
4045                         return $output;
4046                 } elseif ( $markerType === 'nowiki' ) {
4047                         $this->mStripState->addNoWiki( $marker, $output );
4048                 } elseif ( $markerType === 'general' ) {
4049                         $this->mStripState->addGeneral( $marker, $output );
4050                 } else {
4051                         throw new MWException( __METHOD__ . ': invalid marker type' );
4052                 }
4053                 return $marker;
4054         }
4055
4056         /**
4057          * Increment an include size counter
4058          *
4059          * @param string $type the type of expansion
4060          * @param $size Integer: the size of the text
4061          * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
4062          */
4063         function incrementIncludeSize( $type, $size ) {
4064                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
4065                         return false;
4066                 } else {
4067                         $this->mIncludeSizes[$type] += $size;
4068                         return true;
4069                 }
4070         }
4071
4072         /**
4073          * Increment the expensive function count
4074          *
4075          * @return Boolean: false if the limit has been exceeded
4076          */
4077         function incrementExpensiveFunctionCount() {
4078                 $this->mExpensiveFunctionCount++;
4079                 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
4080         }
4081
4082         /**
4083          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
4084          * Fills $this->mDoubleUnderscores, returns the modified text
4085          *
4086          * @param $text string
4087          *
4088          * @return string
4089          */
4090         function doDoubleUnderscore( $text ) {
4091                 wfProfileIn( __METHOD__ );
4092
4093                 # The position of __TOC__ needs to be recorded
4094                 $mw = MagicWord::get( 'toc' );
4095                 if ( $mw->match( $text ) ) {
4096                         $this->mShowToc = true;
4097                         $this->mForceTocPosition = true;
4098
4099                         # Set a placeholder. At the end we'll fill it in with the TOC.
4100                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
4101
4102                         # Only keep the first one.
4103                         $text = $mw->replace( '', $text );
4104                 }
4105
4106                 # Now match and remove the rest of them
4107                 $mwa = MagicWord::getDoubleUnderscoreArray();
4108                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
4109
4110                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
4111                         $this->mOutput->mNoGallery = true;
4112                 }
4113                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
4114                         $this->mShowToc = false;
4115                 }
4116                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
4117                         $this->addTrackingCategory( 'hidden-category-category' );
4118                 }
4119                 # (bug 8068) Allow control over whether robots index a page.
4120                 #
4121                 # @todo FIXME: Bug 14899: __INDEX__ always overrides __NOINDEX__ here!  This
4122                 # is not desirable, the last one on the page should win.
4123                 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
4124                         $this->mOutput->setIndexPolicy( 'noindex' );
4125                         $this->addTrackingCategory( 'noindex-category' );
4126                 }
4127                 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
4128                         $this->mOutput->setIndexPolicy( 'index' );
4129                         $this->addTrackingCategory( 'index-category' );
4130                 }
4131
4132                 # Cache all double underscores in the database
4133                 foreach ( $this->mDoubleUnderscores as $key => $val ) {
4134                         $this->mOutput->setProperty( $key, '' );
4135                 }
4136
4137                 wfProfileOut( __METHOD__ );
4138                 return $text;
4139         }
4140
4141         /**
4142          * Add a tracking category, getting the title from a system message,
4143          * or print a debug message if the title is invalid.
4144          *
4145          * @param string $msg message key
4146          * @return Boolean: whether the addition was successful
4147          */
4148         public function addTrackingCategory( $msg ) {
4149                 if ( $this->mTitle->getNamespace() === NS_SPECIAL ) {
4150                         wfDebug( __METHOD__ . ": Not adding tracking category $msg to special page!\n" );
4151                         return false;
4152                 }
4153                 // Important to parse with correct title (bug 31469)
4154                 $cat = wfMessage( $msg )
4155                         ->title( $this->getTitle() )
4156                         ->inContentLanguage()
4157                         ->text();
4158
4159                 # Allow tracking categories to be disabled by setting them to "-"
4160                 if ( $cat === '-' ) {
4161                         return false;
4162                 }
4163
4164                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
4165                 if ( $containerCategory ) {
4166                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
4167                         return true;
4168                 } else {
4169                         wfDebug( __METHOD__ . ": [[MediaWiki:$msg]] is not a valid title!\n" );
4170                         return false;
4171                 }
4172         }
4173
4174         /**
4175          * This function accomplishes several tasks:
4176          * 1) Auto-number headings if that option is enabled
4177          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4178          * 3) Add a Table of contents on the top for users who have enabled the option
4179          * 4) Auto-anchor headings
4180          *
4181          * It loops through all headlines, collects the necessary data, then splits up the
4182          * string and re-inserts the newly formatted headlines.
4183          *
4184          * @param $text String
4185          * @param string $origText original, untouched wikitext
4186          * @param $isMain Boolean
4187          * @return mixed|string
4188          * @private
4189          */
4190         function formatHeadings( $text, $origText, $isMain = true ) {
4191                 global $wgMaxTocLevel, $wgExperimentalHtmlIds;
4192
4193                 # Inhibit editsection links if requested in the page
4194                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
4195                         $maybeShowEditLink = $showEditLink = false;
4196                 } else {
4197                         $maybeShowEditLink = true; /* Actual presence will depend on ParserOptions option */
4198                         $showEditLink = $this->mOptions->getEditSection();
4199                 }
4200                 if ( $showEditLink ) {
4201                         $this->mOutput->setEditSectionTokens( true );
4202                 }
4203
4204                 # Get all headlines for numbering them and adding funky stuff like [edit]
4205                 # links - this is for later, but we need the number of headlines right now
4206                 $matches = array();
4207                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?' . '>)\s*(?P<header>[\s\S]*?)\s*<\/H[1-6] *>/i', $text, $matches );
4208
4209                 # if there are fewer than 4 headlines in the article, do not show TOC
4210                 # unless it's been explicitly enabled.
4211                 $enoughToc = $this->mShowToc &&
4212                         ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
4213
4214                 # Allow user to stipulate that a page should have a "new section"
4215                 # link added via __NEWSECTIONLINK__
4216                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
4217                         $this->mOutput->setNewSection( true );
4218                 }
4219
4220                 # Allow user to remove the "new section"
4221                 # link via __NONEWSECTIONLINK__
4222                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
4223                         $this->mOutput->hideNewSection( true );
4224                 }
4225
4226                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4227                 # override above conditions and always show TOC above first header
4228                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4229                         $this->mShowToc = true;
4230                         $enoughToc = true;
4231                 }
4232
4233                 # headline counter
4234                 $headlineCount = 0;
4235                 $numVisible = 0;
4236
4237                 # Ugh .. the TOC should have neat indentation levels which can be
4238                 # passed to the skin functions. These are determined here
4239                 $toc = '';
4240                 $full = '';
4241                 $head = array();
4242                 $sublevelCount = array();
4243                 $levelCount = array();
4244                 $level = 0;
4245                 $prevlevel = 0;
4246                 $toclevel = 0;
4247                 $prevtoclevel = 0;
4248                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
4249                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4250                 $oldType = $this->mOutputType;
4251                 $this->setOutputType( self::OT_WIKI );
4252                 $frame = $this->getPreprocessor()->newFrame();
4253                 $root = $this->preprocessToDom( $origText );
4254                 $node = $root->getFirstChild();
4255                 $byteOffset = 0;
4256                 $tocraw = array();
4257                 $refers = array();
4258
4259                 foreach ( $matches[3] as $headline ) {
4260                         $isTemplate = false;
4261                         $titleText = false;
4262                         $sectionIndex = false;
4263                         $numbering = '';
4264                         $markerMatches = array();
4265                         if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4266                                 $serial = $markerMatches[1];
4267                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4268                                 $isTemplate = ( $titleText != $baseTitleText );
4269                                 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4270                         }
4271
4272                         if ( $toclevel ) {
4273                                 $prevlevel = $level;
4274                         }
4275                         $level = $matches[1][$headlineCount];
4276
4277                         if ( $level > $prevlevel ) {
4278                                 # Increase TOC level
4279                                 $toclevel++;
4280                                 $sublevelCount[$toclevel] = 0;
4281                                 if ( $toclevel < $wgMaxTocLevel ) {
4282                                         $prevtoclevel = $toclevel;
4283                                         $toc .= Linker::tocIndent();
4284                                         $numVisible++;
4285                                 }
4286                         } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4287                                 # Decrease TOC level, find level to jump to
4288
4289                                 for ( $i = $toclevel; $i > 0; $i-- ) {
4290                                         if ( $levelCount[$i] == $level ) {
4291                                                 # Found last matching level
4292                                                 $toclevel = $i;
4293                                                 break;
4294                                         } elseif ( $levelCount[$i] < $level ) {
4295                                                 # Found first matching level below current level
4296                                                 $toclevel = $i + 1;
4297                                                 break;
4298                                         }
4299                                 }
4300                                 if ( $i == 0 ) {
4301                                         $toclevel = 1;
4302                                 }
4303                                 if ( $toclevel < $wgMaxTocLevel ) {
4304                                         if ( $prevtoclevel < $wgMaxTocLevel ) {
4305                                                 # Unindent only if the previous toc level was shown :p
4306                                                 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4307                                                 $prevtoclevel = $toclevel;
4308                                         } else {
4309                                                 $toc .= Linker::tocLineEnd();
4310                                         }
4311                                 }
4312                         } else {
4313                                 # No change in level, end TOC line
4314                                 if ( $toclevel < $wgMaxTocLevel ) {
4315                                         $toc .= Linker::tocLineEnd();
4316                                 }
4317                         }
4318
4319                         $levelCount[$toclevel] = $level;
4320
4321                         # count number of headlines for each level
4322                         $sublevelCount[$toclevel]++;
4323                         $dot = 0;
4324                         for ( $i = 1; $i <= $toclevel; $i++ ) {
4325                                 if ( !empty( $sublevelCount[$i] ) ) {
4326                                         if ( $dot ) {
4327                                                 $numbering .= '.';
4328                                         }
4329                                         $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4330                                         $dot = 1;
4331                                 }
4332                         }
4333
4334                         # The safe header is a version of the header text safe to use for links
4335
4336                         # Remove link placeholders by the link text.
4337                         #     <!--LINK number-->
4338                         # turns into
4339                         #     link text with suffix
4340                         # Do this before unstrip since link text can contain strip markers
4341                         $safeHeadline = $this->replaceLinkHoldersText( $headline );
4342
4343                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4344                         $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4345
4346                         # Strip out HTML (first regex removes any tag not allowed)
4347                         # Allowed tags are:
4348                         # * <sup> and <sub> (bug 8393)
4349                         # * <i> (bug 26375)
4350                         # * <b> (r105284)
4351                         # * <span dir="rtl"> and <span dir="ltr"> (bug 35167)
4352                         #
4353                         # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4354                         # to allow setting directionality in toc items.
4355                         $tocline = preg_replace(
4356                                 array( '#<(?!/?(span|sup|sub|i|b)(?: [^>]*)?>).*?' . '>#', '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|i|b))(?: .*?)?' . '>#' ),
4357                                 array( '', '<$1>' ),
4358                                 $safeHeadline
4359                         );
4360                         $tocline = trim( $tocline );
4361
4362                         # For the anchor, strip out HTML-y stuff period
4363                         $safeHeadline = preg_replace( '/<.*?' . '>/', '', $safeHeadline );
4364                         $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4365
4366                         # Save headline for section edit hint before it's escaped
4367                         $headlineHint = $safeHeadline;
4368
4369                         if ( $wgExperimentalHtmlIds ) {
4370                                 # For reverse compatibility, provide an id that's
4371                                 # HTML4-compatible, like we used to.
4372                                 #
4373                                 # It may be worth noting, academically, that it's possible for
4374                                 # the legacy anchor to conflict with a non-legacy headline
4375                                 # anchor on the page.  In this case likely the "correct" thing
4376                                 # would be to either drop the legacy anchors or make sure
4377                                 # they're numbered first.  However, this would require people
4378                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4379                                 # manually, so let's not bother worrying about it.
4380                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
4381                                         array( 'noninitial', 'legacy' ) );
4382                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4383
4384                                 if ( $legacyHeadline == $safeHeadline ) {
4385                                         # No reason to have both (in fact, we can't)
4386                                         $legacyHeadline = false;
4387                                 }
4388                         } else {
4389                                 $legacyHeadline = false;
4390                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
4391                                         'noninitial' );
4392                         }
4393
4394                         # HTML names must be case-insensitively unique (bug 10721).
4395                         # This does not apply to Unicode characters per
4396                         # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
4397                         # @todo FIXME: We may be changing them depending on the current locale.
4398                         $arrayKey = strtolower( $safeHeadline );
4399                         if ( $legacyHeadline === false ) {
4400                                 $legacyArrayKey = false;
4401                         } else {
4402                                 $legacyArrayKey = strtolower( $legacyHeadline );
4403                         }
4404
4405                         # count how many in assoc. array so we can track dupes in anchors
4406                         if ( isset( $refers[$arrayKey] ) ) {
4407                                 $refers[$arrayKey]++;
4408                         } else {
4409                                 $refers[$arrayKey] = 1;
4410                         }
4411                         if ( isset( $refers[$legacyArrayKey] ) ) {
4412                                 $refers[$legacyArrayKey]++;
4413                         } else {
4414                                 $refers[$legacyArrayKey] = 1;
4415                         }
4416
4417                         # Don't number the heading if it is the only one (looks silly)
4418                         if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4419                                 # the two are different if the line contains a link
4420                                 $headline = Html::element( 'span', array( 'class' => 'mw-headline-number' ), $numbering ) . ' ' . $headline;
4421                         }
4422
4423                         # Create the anchor for linking from the TOC to the section
4424                         $anchor = $safeHeadline;
4425                         $legacyAnchor = $legacyHeadline;
4426                         if ( $refers[$arrayKey] > 1 ) {
4427                                 $anchor .= '_' . $refers[$arrayKey];
4428                         }
4429                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
4430                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
4431                         }
4432                         if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4433                                 $toc .= Linker::tocLine( $anchor, $tocline,
4434                                         $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4435                         }
4436
4437                         # Add the section to the section tree
4438                         # Find the DOM node for this header
4439                         while ( $node && !$isTemplate ) {
4440                                 if ( $node->getName() === 'h' ) {
4441                                         $bits = $node->splitHeading();
4442                                         if ( $bits['i'] == $sectionIndex ) {
4443                                                 break;
4444                                         }
4445                                 }
4446                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4447                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4448                                 $node = $node->getNextSibling();
4449                         }
4450                         $tocraw[] = array(
4451                                 'toclevel' => $toclevel,
4452                                 'level' => $level,
4453                                 'line' => $tocline,
4454                                 'number' => $numbering,
4455                                 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4456                                 'fromtitle' => $titleText,
4457                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
4458                                 'anchor' => $anchor,
4459                         );
4460
4461                         # give headline the correct <h#> tag
4462                         if ( $maybeShowEditLink && $sectionIndex !== false ) {
4463                                 // Output edit section links as markers with styles that can be customized by skins
4464                                 if ( $isTemplate ) {
4465                                         # Put a T flag in the section identifier, to indicate to extractSections()
4466                                         # that sections inside <includeonly> should be counted.
4467                                         $editlinkArgs = array( $titleText, "T-$sectionIndex"/*, null */ );
4468                                 } else {
4469                                         $editlinkArgs = array( $this->mTitle->getPrefixedText(), $sectionIndex, $headlineHint );
4470                                 }
4471                                 // We use a bit of pesudo-xml for editsection markers. The language converter is run later on
4472                                 // Using a UNIQ style marker leads to the converter screwing up the tokens when it converts stuff
4473                                 // And trying to insert strip tags fails too. At this point all real inputted tags have already been escaped
4474                                 // so we don't have to worry about a user trying to input one of these markers directly.
4475                                 // We use a page and section attribute to stop the language converter from converting these important bits
4476                                 // of data, but put the headline hint inside a content block because the language converter is supposed to
4477                                 // be able to convert that piece of data.
4478                                 $editlink = '<mw:editsection page="' . htmlspecialchars( $editlinkArgs[0] );
4479                                 $editlink .= '" section="' . htmlspecialchars( $editlinkArgs[1] ) . '"';
4480                                 if ( isset( $editlinkArgs[2] ) ) {
4481                                         $editlink .= '>' . $editlinkArgs[2] . '</mw:editsection>';
4482                                 } else {
4483                                         $editlink .= '/>';
4484                                 }
4485                         } else {
4486                                 $editlink = '';
4487                         }
4488                         $head[$headlineCount] = Linker::makeHeadline( $level,
4489                                 $matches['attrib'][$headlineCount], $anchor, $headline,
4490                                 $editlink, $legacyAnchor );
4491
4492                         $headlineCount++;
4493                 }
4494
4495                 $this->setOutputType( $oldType );
4496
4497                 # Never ever show TOC if no headers
4498                 if ( $numVisible < 1 ) {
4499                         $enoughToc = false;
4500                 }
4501
4502                 if ( $enoughToc ) {
4503                         if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4504                                 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4505                         }
4506                         $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4507                         $this->mOutput->setTOCHTML( $toc );
4508                 }
4509
4510                 if ( $isMain ) {
4511                         $this->mOutput->setSections( $tocraw );
4512                 }
4513
4514                 # split up and insert constructed headlines
4515                 $blocks = preg_split( '/<H[1-6].*?' . '>[\s\S]*?<\/H[1-6]>/i', $text );
4516                 $i = 0;
4517
4518                 // build an array of document sections
4519                 $sections = array();
4520                 foreach ( $blocks as $block ) {
4521                         // $head is zero-based, sections aren't.
4522                         if ( empty( $head[$i - 1] ) ) {
4523                                 $sections[$i] = $block;
4524                         } else {
4525                                 $sections[$i] = $head[$i - 1] . $block;
4526                         }
4527
4528                         /**
4529                          * Send a hook, one per section.
4530                          * The idea here is to be able to make section-level DIVs, but to do so in a
4531                          * lower-impact, more correct way than r50769
4532                          *
4533                          * $this : caller
4534                          * $section : the section number
4535                          * &$sectionContent : ref to the content of the section
4536                          * $showEditLinks : boolean describing whether this section has an edit link
4537                          */
4538                         wfRunHooks( 'ParserSectionCreate', array( $this, $i, &$sections[$i], $showEditLink ) );
4539
4540                         $i++;
4541                 }
4542
4543                 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4544                         // append the TOC at the beginning
4545                         // Top anchor now in skin
4546                         $sections[0] = $sections[0] . $toc . "\n";
4547                 }
4548
4549                 $full .= join( '', $sections );
4550
4551                 if ( $this->mForceTocPosition ) {
4552                         return str_replace( '<!--MWTOC-->', $toc, $full );
4553                 } else {
4554                         return $full;
4555                 }
4556         }
4557
4558         /**
4559          * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4560          * conversion, substitting signatures, {{subst:}} templates, etc.
4561          *
4562          * @param string $text the text to transform
4563          * @param $title Title: the Title object for the current article
4564          * @param $user User: the User object describing the current user
4565          * @param $options ParserOptions: parsing options
4566          * @param $clearState Boolean: whether to clear the parser state first
4567          * @return String: the altered wiki markup
4568          */
4569         public function preSaveTransform( $text, Title $title, User $user, ParserOptions $options, $clearState = true ) {
4570                 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4571                 $this->setUser( $user );
4572
4573                 $pairs = array(
4574                         "\r\n" => "\n",
4575                 );
4576                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4577                 if ( $options->getPreSaveTransform() ) {
4578                         $text = $this->pstPass2( $text, $user );
4579                 }
4580                 $text = $this->mStripState->unstripBoth( $text );
4581
4582                 $this->setUser( null ); #Reset
4583
4584                 return $text;
4585         }
4586
4587         /**
4588          * Pre-save transform helper function
4589          * @private
4590          *
4591          * @param $text string
4592          * @param $user User
4593          *
4594          * @return string
4595          */
4596         function pstPass2( $text, $user ) {
4597                 global $wgContLang;
4598
4599                 # Note: This is the timestamp saved as hardcoded wikitext to
4600                 # the database, we use $wgContLang here in order to give
4601                 # everyone the same signature and use the default one rather
4602                 # than the one selected in each user's preferences.
4603                 # (see also bug 12815)
4604                 $ts = $this->mOptions->getTimestamp();
4605                 $timestamp = MWTimestamp::getLocalInstance( $ts );
4606                 $ts = $timestamp->format( 'YmdHis' );
4607                 $tzMsg = $timestamp->format( 'T' );  # might vary on DST changeover!
4608
4609                 # Allow translation of timezones through wiki. format() can return
4610                 # whatever crap the system uses, localised or not, so we cannot
4611                 # ship premade translations.
4612                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4613                 $msg = wfMessage( $key )->inContentLanguage();
4614                 if ( $msg->exists() ) {
4615                         $tzMsg = $msg->text();
4616                 }
4617
4618                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4619
4620                 # Variable replacement
4621                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4622                 $text = $this->replaceVariables( $text );
4623
4624                 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4625                 # which may corrupt this parser instance via its wfMessage()->text() call-
4626
4627                 # Signatures
4628                 $sigText = $this->getUserSig( $user );
4629                 $text = strtr( $text, array(
4630                         '~~~~~' => $d,
4631                         '~~~~' => "$sigText $d",
4632                         '~~~' => $sigText
4633                 ) );
4634
4635                 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4636                 $tc = '[' . Title::legalChars() . ']';
4637                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4638
4639                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";                   # [[ns:page (context)|]]
4640                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";                           # [[ns:page（context）|]] (double-width brackets, added in r40257)
4641                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/";         # [[ns:page (context), context|]] (using either single or double-width comma)
4642                 $p2 = "/\[\[\\|($tc+)]]/";                                              # [[|page]] (reverse pipe trick: add context from page title)
4643
4644                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4645                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4646                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4647                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4648
4649                 $t = $this->mTitle->getText();
4650                 $m = array();
4651                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4652                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4653                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4654                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4655                 } else {
4656                         # if there's no context, don't bother duplicating the title
4657                         $text = preg_replace( $p2, '[[\\1]]', $text );
4658                 }
4659
4660                 # Trim trailing whitespace
4661                 $text = rtrim( $text );
4662
4663                 return $text;
4664         }
4665
4666         /**
4667          * Fetch the user's signature text, if any, and normalize to
4668          * validated, ready-to-insert wikitext.
4669          * If you have pre-fetched the nickname or the fancySig option, you can
4670          * specify them here to save a database query.
4671          * Do not reuse this parser instance after calling getUserSig(),
4672          * as it may have changed if it's the $wgParser.
4673          *
4674          * @param $user User
4675          * @param string|bool $nickname nickname to use or false to use user's default nickname
4676          * @param $fancySig Boolean|null whether the nicknname is the complete signature
4677          *                  or null to use default value
4678          * @return string
4679          */
4680         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4681                 global $wgMaxSigChars;
4682
4683                 $username = $user->getName();
4684
4685                 # If not given, retrieve from the user object.
4686                 if ( $nickname === false ) {
4687                         $nickname = $user->getOption( 'nickname' );
4688                 }
4689
4690                 if ( is_null( $fancySig ) ) {
4691                         $fancySig = $user->getBoolOption( 'fancysig' );
4692                 }
4693
4694                 $nickname = $nickname == null ? $username : $nickname;
4695
4696                 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4697                         $nickname = $username;
4698                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4699                 } elseif ( $fancySig !== false ) {
4700                         # Sig. might contain markup; validate this
4701                         if ( $this->validateSig( $nickname ) !== false ) {
4702                                 # Validated; clean up (if needed) and return it
4703                                 return $this->cleanSig( $nickname, true );
4704                         } else {
4705                                 # Failed to validate; fall back to the default
4706                                 $nickname = $username;
4707                                 wfDebug( __METHOD__ . ": $username has bad XML tags in signature.\n" );
4708                         }
4709                 }
4710
4711                 # Make sure nickname doesnt get a sig in a sig
4712                 $nickname = self::cleanSigInSig( $nickname );
4713
4714                 # If we're still here, make it a link to the user page
4715                 $userText = wfEscapeWikiText( $username );
4716                 $nickText = wfEscapeWikiText( $nickname );
4717                 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4718
4719                 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()->title( $this->getTitle() )->text();
4720         }
4721
4722         /**
4723          * Check that the user's signature contains no bad XML
4724          *
4725          * @param $text String
4726          * @return mixed An expanded string, or false if invalid.
4727          */
4728         function validateSig( $text ) {
4729                 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
4730         }
4731
4732         /**
4733          * Clean up signature text
4734          *
4735          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4736          * 2) Substitute all transclusions
4737          *
4738          * @param $text String
4739          * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4740          * @return String: signature text
4741          */
4742         public function cleanSig( $text, $parsing = false ) {
4743                 if ( !$parsing ) {
4744                         global $wgTitle;
4745                         $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4746                 }
4747
4748                 # Option to disable this feature
4749                 if ( !$this->mOptions->getCleanSignatures() ) {
4750                         return $text;
4751                 }
4752
4753                 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4754                 #  => Move this logic to braceSubstitution()
4755                 $substWord = MagicWord::get( 'subst' );
4756                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4757                 $substText = '{{' . $substWord->getSynonym( 0 );
4758
4759                 $text = preg_replace( $substRegex, $substText, $text );
4760                 $text = self::cleanSigInSig( $text );
4761                 $dom = $this->preprocessToDom( $text );
4762                 $frame = $this->getPreprocessor()->newFrame();
4763                 $text = $frame->expand( $dom );
4764
4765                 if ( !$parsing ) {
4766                         $text = $this->mStripState->unstripBoth( $text );
4767                 }
4768
4769                 return $text;
4770         }
4771
4772         /**
4773          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4774          *
4775          * @param $text String
4776          * @return String: signature text with /~{3,5}/ removed
4777          */
4778         public static function cleanSigInSig( $text ) {
4779                 $text = preg_replace( '/~{3,5}/', '', $text );
4780                 return $text;
4781         }
4782
4783         /**
4784          * Set up some variables which are usually set up in parse()
4785          * so that an external function can call some class members with confidence
4786          *
4787          * @param $title Title|null
4788          * @param $options ParserOptions
4789          * @param $outputType
4790          * @param $clearState bool
4791          */
4792         public function startExternalParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4793                 $this->startParse( $title, $options, $outputType, $clearState );
4794         }
4795
4796         /**
4797          * @param $title Title|null
4798          * @param $options ParserOptions
4799          * @param $outputType
4800          * @param $clearState bool
4801          */
4802         private function startParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4803                 $this->setTitle( $title );
4804                 $this->mOptions = $options;
4805                 $this->setOutputType( $outputType );
4806                 if ( $clearState ) {
4807                         $this->clearState();
4808                 }
4809         }
4810
4811         /**
4812          * Wrapper for preprocess()
4813          *
4814          * @param string $text the text to preprocess
4815          * @param $options ParserOptions: options
4816          * @param $title Title object or null to use $wgTitle
4817          * @return String
4818          */
4819         public function transformMsg( $text, $options, $title = null ) {
4820                 static $executing = false;
4821
4822                 # Guard against infinite recursion
4823                 if ( $executing ) {
4824                         return $text;
4825                 }
4826                 $executing = true;
4827
4828                 wfProfileIn( __METHOD__ );
4829                 if ( !$title ) {
4830                         global $wgTitle;
4831                         $title = $wgTitle;
4832                 }
4833
4834                 $text = $this->preprocess( $text, $title, $options );
4835
4836                 $executing = false;
4837                 wfProfileOut( __METHOD__ );
4838                 return $text;
4839         }
4840
4841         /**
4842          * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4843          * The callback should have the following form:
4844          *    function myParserHook( $text, $params, $parser, $frame ) { ... }
4845          *
4846          * Transform and return $text. Use $parser for any required context, e.g. use
4847          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4848          *
4849          * Hooks may return extended information by returning an array, of which the
4850          * first numbered element (index 0) must be the return string, and all other
4851          * entries are extracted into local variables within an internal function
4852          * in the Parser class.
4853          *
4854          * This interface (introduced r61913) appears to be undocumented, but
4855          * 'markerName' is used by some core tag hooks to override which strip
4856          * array their results are placed in. **Use great caution if attempting
4857          * this interface, as it is not documented and injudicious use could smash
4858          * private variables.**
4859          *
4860          * @param $tag Mixed: the tag to use, e.g. 'hook' for "<hook>"
4861          * @param $callback Mixed: the callback function (and object) to use for the tag
4862          * @throws MWException
4863          * @return Mixed|null The old value of the mTagHooks array associated with the hook
4864          */
4865         public function setHook( $tag, $callback ) {
4866                 $tag = strtolower( $tag );
4867                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4868                         throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4869                 }
4870                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4871                 $this->mTagHooks[$tag] = $callback;
4872                 if ( !in_array( $tag, $this->mStripList ) ) {
4873                         $this->mStripList[] = $tag;
4874                 }
4875
4876                 return $oldVal;
4877         }
4878
4879         /**
4880          * As setHook(), but letting the contents be parsed.
4881          *
4882          * Transparent tag hooks are like regular XML-style tag hooks, except they
4883          * operate late in the transformation sequence, on HTML instead of wikitext.
4884          *
4885          * This is probably obsoleted by things dealing with parser frames?
4886          * The only extension currently using it is geoserver.
4887          *
4888          * @since 1.10
4889          * @todo better document or deprecate this
4890          *
4891          * @param $tag Mixed: the tag to use, e.g. 'hook' for "<hook>"
4892          * @param $callback Mixed: the callback function (and object) to use for the tag
4893          * @throws MWException
4894          * @return Mixed|null The old value of the mTagHooks array associated with the hook
4895          */
4896         function setTransparentTagHook( $tag, $callback ) {
4897                 $tag = strtolower( $tag );
4898                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4899                         throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4900                 }
4901                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4902                 $this->mTransparentTagHooks[$tag] = $callback;
4903
4904                 return $oldVal;
4905         }
4906
4907         /**
4908          * Remove all tag hooks
4909          */
4910         function clearTagHooks() {
4911                 $this->mTagHooks = array();
4912                 $this->mFunctionTagHooks = array();
4913                 $this->mStripList = $this->mDefaultStripList;
4914         }
4915
4916         /**
4917          * Create a function, e.g. {{sum:1|2|3}}
4918          * The callback function should have the form:
4919          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4920          *
4921          * Or with SFH_OBJECT_ARGS:
4922          *    function myParserFunction( $parser, $frame, $args ) { ... }
4923          *
4924          * The callback may either return the text result of the function, or an array with the text
4925          * in element 0, and a number of flags in the other elements. The names of the flags are
4926          * specified in the keys. Valid flags are:
4927          *   found                     The text returned is valid, stop processing the template. This
4928          *                             is on by default.
4929          *   nowiki                    Wiki markup in the return value should be escaped
4930          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4931          *
4932          * @param string $id The magic word ID
4933          * @param $callback Mixed: the callback function (and object) to use
4934          * @param $flags Integer: a combination of the following flags:
4935          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4936          *
4937          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4938          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4939          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4940          *     the arguments, and to control the way they are expanded.
4941          *
4942          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4943          *     arguments, for instance:
4944          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4945          *
4946          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4947          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4948          *     working if/when this is changed.
4949          *
4950          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4951          *     expansion.
4952          *
4953          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4954          *     about the methods available in PPFrame and PPNode.
4955          *
4956          * @throws MWException
4957          * @return string|callback The old callback function for this name, if any
4958          */
4959         public function setFunctionHook( $id, $callback, $flags = 0 ) {
4960                 global $wgContLang;
4961
4962                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4963                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4964
4965                 # Add to function cache
4966                 $mw = MagicWord::get( $id );
4967                 if ( !$mw ) {
4968                         throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
4969                 }
4970
4971                 $synonyms = $mw->getSynonyms();
4972                 $sensitive = intval( $mw->isCaseSensitive() );
4973
4974                 foreach ( $synonyms as $syn ) {
4975                         # Case
4976                         if ( !$sensitive ) {
4977                                 $syn = $wgContLang->lc( $syn );
4978                         }
4979                         # Add leading hash
4980                         if ( !( $flags & SFH_NO_HASH ) ) {
4981                                 $syn = '#' . $syn;
4982                         }
4983                         # Remove trailing colon
4984                         if ( substr( $syn, -1, 1 ) === ':' ) {
4985                                 $syn = substr( $syn, 0, -1 );
4986                         }
4987                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4988                 }
4989                 return $oldVal;
4990         }
4991
4992         /**
4993          * Get all registered function hook identifiers
4994          *
4995          * @return Array
4996          */
4997         function getFunctionHooks() {
4998                 return array_keys( $this->mFunctionHooks );
4999         }
5000
5001         /**
5002          * Create a tag function, e.g. "<test>some stuff</test>".
5003          * Unlike tag hooks, tag functions are parsed at preprocessor level.
5004          * Unlike parser functions, their content is not preprocessed.
5005          * @param $tag
5006          * @param $callback
5007          * @param $flags
5008          * @throws MWException
5009          * @return null
5010          */
5011         function setFunctionTagHook( $tag, $callback, $flags ) {
5012                 $tag = strtolower( $tag );
5013                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
5014                         throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
5015                 }
5016                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
5017                         $this->mFunctionTagHooks[$tag] : null;
5018                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
5019
5020                 if ( !in_array( $tag, $this->mStripList ) ) {
5021                         $this->mStripList[] = $tag;
5022                 }
5023
5024                 return $old;
5025         }
5026
5027         /**
5028          * @todo FIXME: Update documentation. makeLinkObj() is deprecated.
5029          * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
5030          * Placeholders created in Skin::makeLinkObj()
5031          *
5032          * @param $text string
5033          * @param $options int
5034          *
5035          * @return array of link CSS classes, indexed by PDBK.
5036          */
5037         function replaceLinkHolders( &$text, $options = 0 ) {
5038                 return $this->mLinkHolders->replace( $text );
5039         }
5040
5041         /**
5042          * Replace "<!--LINK-->" link placeholders with plain text of links
5043          * (not HTML-formatted).
5044          *
5045          * @param $text String
5046          * @return String
5047          */
5048         function replaceLinkHoldersText( $text ) {
5049                 return $this->mLinkHolders->replaceText( $text );
5050         }
5051
5052         /**
5053          * Renders an image gallery from a text with one line per image.
5054          * text labels may be given by using |-style alternative text. E.g.
5055          *   Image:one.jpg|The number "1"
5056          *   Image:tree.jpg|A tree
5057          * given as text will return the HTML of a gallery with two images,
5058          * labeled 'The number "1"' and
5059          * 'A tree'.
5060          *
5061          * @param string $text
5062          * @param array $params
5063          * @return string HTML
5064          */
5065         function renderImageGallery( $text, $params ) {
5066                 wfProfileIn( __METHOD__ );
5067
5068                 $mode = false;
5069                 if ( isset( $params['mode'] ) ) {
5070                         $mode = $params['mode'];
5071                 }
5072
5073                 try {
5074                         $ig = ImageGalleryBase::factory( $mode );
5075                 } catch ( MWException $e ) {
5076                         // If invalid type set, fallback to default.
5077                         $ig = ImageGalleryBase::factory( false );
5078                 }
5079
5080                 $ig->setContextTitle( $this->mTitle );
5081                 $ig->setShowBytes( false );
5082                 $ig->setShowFilename( false );
5083                 $ig->setParser( $this );
5084                 $ig->setHideBadImages();
5085                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
5086
5087                 if ( isset( $params['showfilename'] ) ) {
5088                         $ig->setShowFilename( true );
5089                 } else {
5090                         $ig->setShowFilename( false );
5091                 }
5092                 if ( isset( $params['caption'] ) ) {
5093                         $caption = $params['caption'];
5094                         $caption = htmlspecialchars( $caption );
5095                         $caption = $this->replaceInternalLinks( $caption );
5096                         $ig->setCaptionHtml( $caption );
5097                 }
5098                 if ( isset( $params['perrow'] ) ) {
5099                         $ig->setPerRow( $params['perrow'] );
5100                 }
5101                 if ( isset( $params['widths'] ) ) {
5102                         $ig->setWidths( $params['widths'] );
5103                 }
5104                 if ( isset( $params['heights'] ) ) {
5105                         $ig->setHeights( $params['heights'] );
5106                 }
5107                 $ig->setAdditionalOptions( $params );
5108
5109                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
5110
5111                 $lines = StringUtils::explode( "\n", $text );
5112                 foreach ( $lines as $line ) {
5113                         # match lines like these:
5114                         # Image:someimage.jpg|This is some image
5115                         $matches = array();
5116                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
5117                         # Skip empty lines
5118                         if ( count( $matches ) == 0 ) {
5119                                 continue;
5120                         }
5121
5122                         if ( strpos( $matches[0], '%' ) !== false ) {
5123                                 $matches[1] = rawurldecode( $matches[1] );
5124                         }
5125                         $title = Title::newFromText( $matches[1], NS_FILE );
5126                         if ( is_null( $title ) ) {
5127                                 # Bogus title. Ignore these so we don't bomb out later.
5128                                 continue;
5129                         }
5130
5131                         # We need to get what handler the file uses, to figure out parameters.
5132                         # Note, a hook can overide the file name, and chose an entirely different
5133                         # file (which potentially could be of a different type and have different handler).
5134                         $options = array();
5135                         $descQuery = false;
5136                         wfRunHooks( 'BeforeParserFetchFileAndTitle',
5137                                 array( $this, $title, &$options, &$descQuery ) );
5138                         # Don't register it now, as ImageGallery does that later.
5139                         $file = $this->fetchFileNoRegister( $title, $options );
5140                         $handler = $file ? $file->getHandler() : false;
5141
5142                         wfProfileIn( __METHOD__ . '-getMagicWord' );
5143                         $paramMap = array(
5144                                 'img_alt' => 'gallery-internal-alt',
5145                                 'img_link' => 'gallery-internal-link',
5146                         );
5147                         if ( $handler ) {
5148                                 $paramMap = $paramMap + $handler->getParamMap();
5149                                 // We don't want people to specify per-image widths.
5150                                 // Additionally the width parameter would need special casing anyhow.
5151                                 unset( $paramMap['img_width'] );
5152                         }
5153
5154                         $mwArray = new MagicWordArray( array_keys( $paramMap ) );
5155                         wfProfileOut( __METHOD__ . '-getMagicWord' );
5156
5157                         $label = '';
5158                         $alt = '';
5159                         $link = '';
5160                         $handlerOptions = array();
5161                         if ( isset( $matches[3] ) ) {
5162                                 // look for an |alt= definition while trying not to break existing
5163                                 // captions with multiple pipes (|) in it, until a more sensible grammar
5164                                 // is defined for images in galleries
5165
5166                                 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5167                                 // splitting on '|' is a bit odd, and different from makeImage.
5168                                 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5169                                 $parameterMatches = StringUtils::explode( '|', $matches[3] );
5170
5171                                 foreach ( $parameterMatches as $parameterMatch ) {
5172                                         list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5173                                         if ( $magicName ) {
5174                                                 $paramName = $paramMap[$magicName];
5175
5176                                                 switch ( $paramName ) {
5177                                                 case 'gallery-internal-alt':
5178                                                         $alt = $this->stripAltText( $match, false );
5179                                                         break;
5180                                                 case 'gallery-internal-link':
5181                                                         $linkValue = strip_tags( $this->replaceLinkHoldersText( $match ) );
5182                                                         $chars = self::EXT_LINK_URL_CLASS;
5183                                                         $prots = $this->mUrlProtocols;
5184                                                         //check to see if link matches an absolute url, if not then it must be a wiki link.
5185                                                         if ( preg_match( "/^($prots)$chars+$/u", $linkValue ) ) {
5186                                                                 $link = $linkValue;
5187                                                         } else {
5188                                                                 $localLinkTitle = Title::newFromText( $linkValue );
5189                                                                 if ( $localLinkTitle !== null ) {
5190                                                                         $link = $localLinkTitle->getLocalURL();
5191                                                                 }
5192                                                         }
5193                                                         break;
5194                                                 default:
5195                                                         // Must be a handler specific parameter.
5196                                                         if ( $handler->validateParam( $paramName, $match ) ) {
5197                                                                 $handlerOptions[$paramName] = $match;
5198                                                         } else {
5199                                                                 // Guess not. Append it to the caption.
5200                                                                 wfDebug( "$parameterMatch failed parameter validation" );
5201                                                                 $label .= '|' . $parameterMatch;
5202                                                         }
5203                                                 }
5204
5205                                         } else {
5206                                                 // concatenate all other pipes
5207                                                 $label .= '|' . $parameterMatch;
5208                                         }
5209                                 }
5210                                 // remove the first pipe
5211                                 $label = substr( $label, 1 );
5212                         }
5213
5214                         $ig->add( $title, $label, $alt, $link, $handlerOptions );
5215                 }
5216                 $html = $ig->toHTML();
5217                 wfProfileOut( __METHOD__ );
5218                 return $html;
5219         }
5220
5221         /**
5222          * @param $handler
5223          * @return array
5224          */
5225         function getImageParams( $handler ) {
5226                 if ( $handler ) {
5227                         $handlerClass = get_class( $handler );
5228                 } else {
5229                         $handlerClass = '';
5230                 }
5231                 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
5232                         # Initialise static lists
5233                         static $internalParamNames = array(
5234                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
5235                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5236                                         'bottom', 'text-bottom' ),
5237                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
5238                                         'upright', 'border', 'link', 'alt', 'class' ),
5239                         );
5240                         static $internalParamMap;
5241                         if ( !$internalParamMap ) {
5242                                 $internalParamMap = array();
5243                                 foreach ( $internalParamNames as $type => $names ) {
5244                                         foreach ( $names as $name ) {
5245                                                 $magicName = str_replace( '-', '_', "img_$name" );
5246                                                 $internalParamMap[$magicName] = array( $type, $name );
5247                                         }
5248                                 }
5249                         }
5250
5251                         # Add handler params
5252                         $paramMap = $internalParamMap;
5253                         if ( $handler ) {
5254                                 $handlerParamMap = $handler->getParamMap();
5255                                 foreach ( $handlerParamMap as $magic => $paramName ) {
5256                                         $paramMap[$magic] = array( 'handler', $paramName );
5257                                 }
5258                         }
5259                         $this->mImageParams[$handlerClass] = $paramMap;
5260                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
5261                 }
5262                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
5263         }
5264
5265         /**
5266          * Parse image options text and use it to make an image
5267          *
5268          * @param $title Title
5269          * @param $options String
5270          * @param $holders LinkHolderArray|bool
5271          * @return string HTML
5272          */
5273         function makeImage( $title, $options, $holders = false ) {
5274                 # Check if the options text is of the form "options|alt text"
5275                 # Options are:
5276                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
5277                 #  * left       no resizing, just left align. label is used for alt= only
5278                 #  * right      same, but right aligned
5279                 #  * none       same, but not aligned
5280                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
5281                 #  * center     center the image
5282                 #  * frame      Keep original image size, no magnify-button.
5283                 #  * framed     Same as "frame"
5284                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
5285                 #  * upright    reduce width for upright images, rounded to full __0 px
5286                 #  * border     draw a 1px border around the image
5287                 #  * alt        Text for HTML alt attribute (defaults to empty)
5288                 #  * class      Set a class for img node
5289                 #  * link       Set the target of the image link. Can be external, interwiki, or local
5290                 # vertical-align values (no % or length right now):
5291                 #  * baseline
5292                 #  * sub
5293                 #  * super
5294                 #  * top
5295                 #  * text-top
5296                 #  * middle
5297                 #  * bottom
5298                 #  * text-bottom
5299
5300                 $parts = StringUtils::explode( "|", $options );
5301
5302                 # Give extensions a chance to select the file revision for us
5303                 $options = array();
5304                 $descQuery = false;
5305                 wfRunHooks( 'BeforeParserFetchFileAndTitle',
5306                         array( $this, $title, &$options, &$descQuery ) );
5307                 # Fetch and register the file (file title may be different via hooks)
5308                 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5309
5310                 # Get parameter map
5311                 $handler = $file ? $file->getHandler() : false;
5312
5313                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5314
5315                 if ( !$file ) {
5316                         $this->addTrackingCategory( 'broken-file-category' );
5317                 }
5318
5319                 # Process the input parameters
5320                 $caption = '';
5321                 $params = array( 'frame' => array(), 'handler' => array(),
5322                         'horizAlign' => array(), 'vertAlign' => array() );
5323                 foreach ( $parts as $part ) {
5324                         $part = trim( $part );
5325                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5326                         $validated = false;
5327                         if ( isset( $paramMap[$magicName] ) ) {
5328                                 list( $type, $paramName ) = $paramMap[$magicName];
5329
5330                                 # Special case; width and height come in one variable together
5331                                 if ( $type === 'handler' && $paramName === 'width' ) {
5332                                         $parsedWidthParam = $this->parseWidthParam( $value );
5333                                         if ( isset( $parsedWidthParam['width'] ) ) {
5334                                                 $width = $parsedWidthParam['width'];
5335                                                 if ( $handler->validateParam( 'width', $width ) ) {
5336                                                         $params[$type]['width'] = $width;
5337                                                         $validated = true;
5338                                                 }
5339                                         }
5340                                         if ( isset( $parsedWidthParam['height'] ) ) {
5341                                                 $height = $parsedWidthParam['height'];
5342                                                 if ( $handler->validateParam( 'height', $height ) ) {
5343                                                         $params[$type]['height'] = $height;
5344                                                         $validated = true;
5345                                                 }
5346                                         }
5347                                         # else no validation -- bug 13436
5348                                 } else {
5349                                         if ( $type === 'handler' ) {
5350                                                 # Validate handler parameter
5351                                                 $validated = $handler->validateParam( $paramName, $value );
5352                                         } else {
5353                                                 # Validate internal parameters
5354                                                 switch ( $paramName ) {
5355                                                 case 'manualthumb':
5356                                                 case 'alt':
5357                                                 case 'class':
5358                                                         # @todo FIXME: Possibly check validity here for
5359                                                         # manualthumb? downstream behavior seems odd with
5360                                                         # missing manual thumbs.
5361                                                         $validated = true;
5362                                                         $value = $this->stripAltText( $value, $holders );
5363                                                         break;
5364                                                 case 'link':
5365                                                         $chars = self::EXT_LINK_URL_CLASS;
5366                                                         $prots = $this->mUrlProtocols;
5367                                                         if ( $value === '' ) {
5368                                                                 $paramName = 'no-link';
5369                                                                 $value = true;
5370                                                                 $validated = true;
5371                                                         } elseif ( preg_match( "/^(?i)$prots/", $value ) ) {
5372                                                                 if ( preg_match( "/^((?i)$prots)$chars+$/u", $value, $m ) ) {
5373                                                                         $paramName = 'link-url';
5374                                                                         $this->mOutput->addExternalLink( $value );
5375                                                                         if ( $this->mOptions->getExternalLinkTarget() ) {
5376                                                                                 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5377                                                                         }
5378                                                                         $validated = true;
5379                                                                 }
5380                                                         } else {
5381                                                                 $linkTitle = Title::newFromText( $value );
5382                                                                 if ( $linkTitle ) {
5383                                                                         $paramName = 'link-title';
5384                                                                         $value = $linkTitle;
5385                                                                         $this->mOutput->addLink( $linkTitle );
5386                                                                         $validated = true;
5387                                                                 }
5388                                                         }
5389                                                         break;
5390                                                 default:
5391                                                         # Most other things appear to be empty or numeric...
5392                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
5393                                                 }
5394                                         }
5395
5396                                         if ( $validated ) {
5397                                                 $params[$type][$paramName] = $value;
5398                                         }
5399                                 }
5400                         }
5401                         if ( !$validated ) {
5402                                 $caption = $part;
5403                         }
5404                 }
5405
5406                 # Process alignment parameters
5407                 if ( $params['horizAlign'] ) {
5408                         $params['frame']['align'] = key( $params['horizAlign'] );
5409                 }
5410                 if ( $params['vertAlign'] ) {
5411                         $params['frame']['valign'] = key( $params['vertAlign'] );
5412                 }
5413
5414                 $params['frame']['caption'] = $caption;
5415
5416                 # Will the image be presented in a frame, with the caption below?
5417                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
5418                                                  isset( $params['frame']['framed'] ) ||
5419                                                  isset( $params['frame']['thumbnail'] ) ||
5420                                                  isset( $params['frame']['manualthumb'] );
5421
5422                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
5423                 # came to also set the caption, ordinary text after the image -- which
5424                 # makes no sense, because that just repeats the text multiple times in
5425                 # screen readers.  It *also* came to set the title attribute.
5426                 #
5427                 # Now that we have an alt attribute, we should not set the alt text to
5428                 # equal the caption: that's worse than useless, it just repeats the
5429                 # text.  This is the framed/thumbnail case.  If there's no caption, we
5430                 # use the unnamed parameter for alt text as well, just for the time be-
5431                 # ing, if the unnamed param is set and the alt param is not.
5432                 #
5433                 # For the future, we need to figure out if we want to tweak this more,
5434                 # e.g., introducing a title= parameter for the title; ignoring the un-
5435                 # named parameter entirely for images without a caption; adding an ex-
5436                 # plicit caption= parameter and preserving the old magic unnamed para-
5437                 # meter for BC; ...
5438                 if ( $imageIsFramed ) { # Framed image
5439                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5440                                 # No caption or alt text, add the filename as the alt text so
5441                                 # that screen readers at least get some description of the image
5442                                 $params['frame']['alt'] = $title->getText();
5443                         }
5444                         # Do not set $params['frame']['title'] because tooltips don't make sense
5445                         # for framed images
5446                 } else { # Inline image
5447                         if ( !isset( $params['frame']['alt'] ) ) {
5448                                 # No alt text, use the "caption" for the alt text
5449                                 if ( $caption !== '' ) {
5450                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5451                                 } else {
5452                                         # No caption, fall back to using the filename for the
5453                                         # alt text
5454                                         $params['frame']['alt'] = $title->getText();
5455                                 }
5456                         }
5457                         # Use the "caption" for the tooltip text
5458                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5459                 }
5460
5461                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params, $this ) );
5462
5463                 # Linker does the rest
5464                 $time = isset( $options['time'] ) ? $options['time'] : false;
5465                 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5466                         $time, $descQuery, $this->mOptions->getThumbSize() );
5467
5468                 # Give the handler a chance to modify the parser object
5469                 if ( $handler ) {
5470                         $handler->parserTransformHook( $this, $file );
5471                 }
5472
5473                 return $ret;
5474         }
5475
5476         /**
5477          * @param $caption
5478          * @param $holders LinkHolderArray
5479          * @return mixed|String
5480          */
5481         protected function stripAltText( $caption, $holders ) {
5482                 # Strip bad stuff out of the title (tooltip).  We can't just use
5483                 # replaceLinkHoldersText() here, because if this function is called
5484                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5485                 if ( $holders ) {
5486                         $tooltip = $holders->replaceText( $caption );
5487                 } else {
5488                         $tooltip = $this->replaceLinkHoldersText( $caption );
5489                 }
5490
5491                 # make sure there are no placeholders in thumbnail attributes
5492                 # that are later expanded to html- so expand them now and
5493                 # remove the tags
5494                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5495                 $tooltip = Sanitizer::stripAllTags( $tooltip );
5496
5497                 return $tooltip;
5498         }
5499
5500         /**
5501          * Set a flag in the output object indicating that the content is dynamic and
5502          * shouldn't be cached.
5503          */
5504         function disableCache() {
5505                 wfDebug( "Parser output marked as uncacheable.\n" );
5506                 if ( !$this->mOutput ) {
5507                         throw new MWException( __METHOD__ .
5508                                 " can only be called when actually parsing something" );
5509                 }
5510                 $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
5511                 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5512         }
5513
5514         /**
5515          * Callback from the Sanitizer for expanding items found in HTML attribute
5516          * values, so they can be safely tested and escaped.
5517          *
5518          * @param $text String
5519          * @param $frame PPFrame
5520          * @return String
5521          */
5522         function attributeStripCallback( &$text, $frame = false ) {
5523                 $text = $this->replaceVariables( $text, $frame );
5524                 $text = $this->mStripState->unstripBoth( $text );
5525                 return $text;
5526         }
5527
5528         /**
5529          * Accessor
5530          *
5531          * @return array
5532          */
5533         function getTags() {
5534                 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ), array_keys( $this->mFunctionTagHooks ) );
5535         }
5536
5537         /**
5538          * Replace transparent tags in $text with the values given by the callbacks.
5539          *
5540          * Transparent tag hooks are like regular XML-style tag hooks, except they
5541          * operate late in the transformation sequence, on HTML instead of wikitext.
5542          *
5543          * @param $text string
5544          *
5545          * @return string
5546          */
5547         function replaceTransparentTags( $text ) {
5548                 $matches = array();
5549                 $elements = array_keys( $this->mTransparentTagHooks );
5550                 $text = self::extractTagsAndParams( $elements, $text, $matches, $this->mUniqPrefix );
5551                 $replacements = array();
5552
5553                 foreach ( $matches as $marker => $data ) {
5554                         list( $element, $content, $params, $tag ) = $data;
5555                         $tagName = strtolower( $element );
5556                         if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5557                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
5558                         } else {
5559                                 $output = $tag;
5560                         }
5561                         $replacements[$marker] = $output;
5562                 }
5563                 return strtr( $text, $replacements );
5564         }
5565
5566         /**
5567          * Break wikitext input into sections, and either pull or replace
5568          * some particular section's text.
5569          *
5570          * External callers should use the getSection and replaceSection methods.
5571          *
5572          * @param string $text Page wikitext
5573          * @param string $section a section identifier string of the form:
5574          *   "<flag1> - <flag2> - ... - <section number>"
5575          *
5576          * Currently the only recognised flag is "T", which means the target section number
5577          * was derived during a template inclusion parse, in other words this is a template
5578          * section edit link. If no flags are given, it was an ordinary section edit link.
5579          * This flag is required to avoid a section numbering mismatch when a section is
5580          * enclosed by "<includeonly>" (bug 6563).
5581          *
5582          * The section number 0 pulls the text before the first heading; other numbers will
5583          * pull the given section along with its lower-level subsections. If the section is
5584          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5585          *
5586          * Section 0 is always considered to exist, even if it only contains the empty
5587          * string. If $text is the empty string and section 0 is replaced, $newText is
5588          * returned.
5589          *
5590          * @param string $mode one of "get" or "replace"
5591          * @param string $newText replacement text for section data.
5592          * @return String: for "get", the extracted section text.
5593          *                 for "replace", the whole page with the section replaced.
5594          */
5595         private function extractSections( $text, $section, $mode, $newText = '' ) {
5596                 global $wgTitle; # not generally used but removes an ugly failure mode
5597                 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5598                 $outText = '';
5599                 $frame = $this->getPreprocessor()->newFrame();
5600
5601                 # Process section extraction flags
5602                 $flags = 0;
5603                 $sectionParts = explode( '-', $section );
5604                 $sectionIndex = array_pop( $sectionParts );
5605                 foreach ( $sectionParts as $part ) {
5606                         if ( $part === 'T' ) {
5607                                 $flags |= self::PTD_FOR_INCLUSION;
5608                         }
5609                 }
5610
5611                 # Check for empty input
5612                 if ( strval( $text ) === '' ) {
5613                         # Only sections 0 and T-0 exist in an empty document
5614                         if ( $sectionIndex == 0 ) {
5615                                 if ( $mode === 'get' ) {
5616                                         return '';
5617                                 } else {
5618                                         return $newText;
5619                                 }
5620                         } else {
5621                                 if ( $mode === 'get' ) {
5622                                         return $newText;
5623                                 } else {
5624                                         return $text;
5625                                 }
5626                         }
5627                 }
5628
5629                 # Preprocess the text
5630                 $root = $this->preprocessToDom( $text, $flags );
5631
5632                 # <h> nodes indicate section breaks
5633                 # They can only occur at the top level, so we can find them by iterating the root's children
5634                 $node = $root->getFirstChild();
5635
5636                 # Find the target section
5637                 if ( $sectionIndex == 0 ) {
5638                         # Section zero doesn't nest, level=big
5639                         $targetLevel = 1000;
5640                 } else {
5641                         while ( $node ) {
5642                                 if ( $node->getName() === 'h' ) {
5643                                         $bits = $node->splitHeading();
5644                                         if ( $bits['i'] == $sectionIndex ) {
5645                                                 $targetLevel = $bits['level'];
5646                                                 break;
5647                                         }
5648                                 }
5649                                 if ( $mode === 'replace' ) {
5650                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5651                                 }
5652                                 $node = $node->getNextSibling();
5653                         }
5654                 }
5655
5656                 if ( !$node ) {
5657                         # Not found
5658                         if ( $mode === 'get' ) {
5659                                 return $newText;
5660                         } else {
5661                                 return $text;
5662                         }
5663                 }
5664
5665                 # Find the end of the section, including nested sections
5666                 do {
5667                         if ( $node->getName() === 'h' ) {
5668                                 $bits = $node->splitHeading();
5669                                 $curLevel = $bits['level'];
5670                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5671                                         break;
5672                                 }
5673                         }
5674                         if ( $mode === 'get' ) {
5675                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5676                         }
5677                         $node = $node->getNextSibling();
5678                 } while ( $node );
5679
5680                 # Write out the remainder (in replace mode only)
5681                 if ( $mode === 'replace' ) {
5682                         # Output the replacement text
5683                         # Add two newlines on -- trailing whitespace in $newText is conventionally
5684                         # stripped by the editor, so we need both newlines to restore the paragraph gap
5685                         # Only add trailing whitespace if there is newText
5686                         if ( $newText != "" ) {
5687                                 $outText .= $newText . "\n\n";
5688                         }
5689
5690                         while ( $node ) {
5691                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5692                                 $node = $node->getNextSibling();
5693                         }
5694                 }
5695
5696                 if ( is_string( $outText ) ) {
5697                         # Re-insert stripped tags
5698                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5699                 }
5700
5701                 return $outText;
5702         }
5703
5704         /**
5705          * This function returns the text of a section, specified by a number ($section).
5706          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5707          * the first section before any such heading (section 0).
5708          *
5709          * If a section contains subsections, these are also returned.
5710          *
5711          * @param string $text text to look in
5712          * @param string $section section identifier
5713          * @param string $deftext default to return if section is not found
5714          * @return string text of the requested section
5715          */
5716         public function getSection( $text, $section, $deftext = '' ) {
5717                 return $this->extractSections( $text, $section, "get", $deftext );
5718         }
5719
5720         /**
5721          * This function returns $oldtext after the content of the section
5722          * specified by $section has been replaced with $text. If the target
5723          * section does not exist, $oldtext is returned unchanged.
5724          *
5725          * @param string $oldtext former text of the article
5726          * @param int $section section identifier
5727          * @param string $text replacing text
5728          * @return String: modified text
5729          */
5730         public function replaceSection( $oldtext, $section, $text ) {
5731                 return $this->extractSections( $oldtext, $section, "replace", $text );
5732         }
5733
5734         /**
5735          * Get the ID of the revision we are parsing
5736          *
5737          * @return Mixed: integer or null
5738          */
5739         function getRevisionId() {
5740                 return $this->mRevisionId;
5741         }
5742
5743         /**
5744          * Get the revision object for $this->mRevisionId
5745          *
5746          * @return Revision|null either a Revision object or null
5747          */
5748         protected function getRevisionObject() {
5749                 if ( !is_null( $this->mRevisionObject ) ) {
5750                         return $this->mRevisionObject;
5751                 }
5752                 if ( is_null( $this->mRevisionId ) ) {
5753                         return null;
5754                 }
5755
5756                 $this->mRevisionObject = Revision::newFromId( $this->mRevisionId );
5757                 return $this->mRevisionObject;
5758         }
5759
5760         /**
5761          * Get the timestamp associated with the current revision, adjusted for
5762          * the default server-local timestamp
5763          */
5764         function getRevisionTimestamp() {
5765                 if ( is_null( $this->mRevisionTimestamp ) ) {
5766                         wfProfileIn( __METHOD__ );
5767
5768                         global $wgContLang;
5769
5770                         $revObject = $this->getRevisionObject();
5771                         $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5772
5773                         # The cryptic '' timezone parameter tells to use the site-default
5774                         # timezone offset instead of the user settings.
5775                         #
5776                         # Since this value will be saved into the parser cache, served
5777                         # to other users, and potentially even used inside links and such,
5778                         # it needs to be consistent for all visitors.
5779                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5780
5781                         wfProfileOut( __METHOD__ );
5782                 }
5783                 return $this->mRevisionTimestamp;
5784         }
5785
5786         /**
5787          * Get the name of the user that edited the last revision
5788          *
5789          * @return String: user name
5790          */
5791         function getRevisionUser() {
5792                 if ( is_null( $this->mRevisionUser ) ) {
5793                         $revObject = $this->getRevisionObject();
5794
5795                         # if this template is subst: the revision id will be blank,
5796                         # so just use the current user's name
5797                         if ( $revObject ) {
5798                                 $this->mRevisionUser = $revObject->getUserText();
5799                         } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5800                                 $this->mRevisionUser = $this->getUser()->getName();
5801                         }
5802                 }
5803                 return $this->mRevisionUser;
5804         }
5805
5806         /**
5807          * Mutator for $mDefaultSort
5808          *
5809          * @param string $sort New value
5810          */
5811         public function setDefaultSort( $sort ) {
5812                 $this->mDefaultSort = $sort;
5813                 $this->mOutput->setProperty( 'defaultsort', $sort );
5814         }
5815
5816         /**
5817          * Accessor for $mDefaultSort
5818          * Will use the empty string if none is set.
5819          *
5820          * This value is treated as a prefix, so the
5821          * empty string is equivalent to sorting by
5822          * page name.
5823          *
5824          * @return string
5825          */
5826         public function getDefaultSort() {
5827                 if ( $this->mDefaultSort !== false ) {
5828                         return $this->mDefaultSort;
5829                 } else {
5830                         return '';
5831                 }
5832         }
5833
5834         /**
5835          * Accessor for $mDefaultSort
5836          * Unlike getDefaultSort(), will return false if none is set
5837          *
5838          * @return string or false
5839          */
5840         public function getCustomDefaultSort() {
5841                 return $this->mDefaultSort;
5842         }
5843
5844         /**
5845          * Try to guess the section anchor name based on a wikitext fragment
5846          * presumably extracted from a heading, for example "Header" from
5847          * "== Header ==".
5848          *
5849          * @param $text string
5850          *
5851          * @return string
5852          */
5853         public function guessSectionNameFromWikiText( $text ) {
5854                 # Strip out wikitext links(they break the anchor)
5855                 $text = $this->stripSectionName( $text );
5856                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5857                 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5858         }
5859
5860         /**
5861          * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5862          * instead.  For use in redirects, since IE6 interprets Redirect: headers
5863          * as something other than UTF-8 (apparently?), resulting in breakage.
5864          *
5865          * @param string $text The section name
5866          * @return string An anchor
5867          */
5868         public function guessLegacySectionNameFromWikiText( $text ) {
5869                 # Strip out wikitext links(they break the anchor)
5870                 $text = $this->stripSectionName( $text );
5871                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5872                 return '#' . Sanitizer::escapeId( $text, array( 'noninitial', 'legacy' ) );
5873         }
5874
5875         /**
5876          * Strips a text string of wikitext for use in a section anchor
5877          *
5878          * Accepts a text string and then removes all wikitext from the
5879          * string and leaves only the resultant text (i.e. the result of
5880          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5881          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5882          * to create valid section anchors by mimicing the output of the
5883          * parser when headings are parsed.
5884          *
5885          * @param string $text text string to be stripped of wikitext
5886          * for use in a Section anchor
5887          * @return string Filtered text string
5888          */
5889         public function stripSectionName( $text ) {
5890                 # Strip internal link markup
5891                 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5892                 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5893
5894                 # Strip external link markup
5895                 # @todo FIXME: Not tolerant to blank link text
5896                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5897                 # on how many empty links there are on the page - need to figure that out.
5898                 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5899
5900                 # Parse wikitext quotes (italics & bold)
5901                 $text = $this->doQuotes( $text );
5902
5903                 # Strip HTML tags
5904                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5905                 return $text;
5906         }
5907
5908         /**
5909          * strip/replaceVariables/unstrip for preprocessor regression testing
5910          *
5911          * @param $text string
5912          * @param $title Title
5913          * @param $options ParserOptions
5914          * @param $outputType int
5915          *
5916          * @return string
5917          */
5918         function testSrvus( $text, Title $title, ParserOptions $options, $outputType = self::OT_HTML ) {
5919                 $this->startParse( $title, $options, $outputType, true );
5920
5921                 $text = $this->replaceVariables( $text );
5922                 $text = $this->mStripState->unstripBoth( $text );
5923                 $text = Sanitizer::removeHTMLtags( $text );
5924                 return $text;
5925         }
5926
5927         /**
5928          * @param $text string
5929          * @param $title Title
5930          * @param $options ParserOptions
5931          * @return string
5932          */
5933         function testPst( $text, Title $title, ParserOptions $options ) {
5934                 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
5935         }
5936
5937         /**
5938          * @param $text
5939          * @param $title Title
5940          * @param $options ParserOptions
5941          * @return string
5942          */
5943         function testPreprocess( $text, Title $title, ParserOptions $options ) {
5944                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5945         }
5946
5947         /**
5948          * Call a callback function on all regions of the given text that are not
5949          * inside strip markers, and replace those regions with the return value
5950          * of the callback. For example, with input:
5951          *
5952          *  aaa<MARKER>bbb
5953          *
5954          * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5955          * two strings will be replaced with the value returned by the callback in
5956          * each case.
5957          *
5958          * @param $s string
5959          * @param $callback
5960          *
5961          * @return string
5962          */
5963         function markerSkipCallback( $s, $callback ) {
5964                 $i = 0;
5965                 $out = '';
5966                 while ( $i < strlen( $s ) ) {
5967                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5968                         if ( $markerStart === false ) {
5969                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5970                                 break;
5971                         } else {
5972                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5973                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5974                                 if ( $markerEnd === false ) {
5975                                         $out .= substr( $s, $markerStart );
5976                                         break;
5977                                 } else {
5978                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5979                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5980                                         $i = $markerEnd;
5981                                 }
5982                         }
5983                 }
5984                 return $out;
5985         }
5986
5987         /**
5988          * Remove any strip markers found in the given text.
5989          *
5990          * @param $text Input string
5991          * @return string
5992          */
5993         function killMarkers( $text ) {
5994                 return $this->mStripState->killMarkers( $text );
5995         }
5996
5997         /**
5998          * Save the parser state required to convert the given half-parsed text to
5999          * HTML. "Half-parsed" in this context means the output of
6000          * recursiveTagParse() or internalParse(). This output has strip markers
6001          * from replaceVariables (extensionSubstitution() etc.), and link
6002          * placeholders from replaceLinkHolders().
6003          *
6004          * Returns an array which can be serialized and stored persistently. This
6005          * array can later be loaded into another parser instance with
6006          * unserializeHalfParsedText(). The text can then be safely incorporated into
6007          * the return value of a parser hook.
6008          *
6009          * @param $text string
6010          *
6011          * @return array
6012          */
6013         function serializeHalfParsedText( $text ) {
6014                 wfProfileIn( __METHOD__ );
6015                 $data = array(
6016                         'text' => $text,
6017                         'version' => self::HALF_PARSED_VERSION,
6018                         'stripState' => $this->mStripState->getSubState( $text ),
6019                         'linkHolders' => $this->mLinkHolders->getSubArray( $text )
6020                 );
6021                 wfProfileOut( __METHOD__ );
6022                 return $data;
6023         }
6024
6025         /**
6026          * Load the parser state given in the $data array, which is assumed to
6027          * have been generated by serializeHalfParsedText(). The text contents is
6028          * extracted from the array, and its markers are transformed into markers
6029          * appropriate for the current Parser instance. This transformed text is
6030          * returned, and can be safely included in the return value of a parser
6031          * hook.
6032          *
6033          * If the $data array has been stored persistently, the caller should first
6034          * check whether it is still valid, by calling isValidHalfParsedText().
6035          *
6036          * @param array $data Serialized data
6037          * @throws MWException
6038          * @return String
6039          */
6040         function unserializeHalfParsedText( $data ) {
6041                 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
6042                         throw new MWException( __METHOD__ . ': invalid version' );
6043                 }
6044
6045                 # First, extract the strip state.
6046                 $texts = array( $data['text'] );
6047                 $texts = $this->mStripState->merge( $data['stripState'], $texts );
6048
6049                 # Now renumber links
6050                 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
6051
6052                 # Should be good to go.
6053                 return $texts[0];
6054         }
6055
6056         /**
6057          * Returns true if the given array, presumed to be generated by
6058          * serializeHalfParsedText(), is compatible with the current version of the
6059          * parser.
6060          *
6061          * @param $data Array
6062          *
6063          * @return bool
6064          */
6065         function isValidHalfParsedText( $data ) {
6066                 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
6067         }
6068
6069         /**
6070          * Parsed a width param of imagelink like 300px or 200x300px
6071          *
6072          * @param $value String
6073          *
6074          * @return array
6075          * @since 1.20
6076          */
6077         public function parseWidthParam( $value ) {
6078                 $parsedWidthParam = array();
6079                 if ( $value === '' ) {
6080                         return $parsedWidthParam;
6081                 }
6082                 $m = array();
6083                 # (bug 13500) In both cases (width/height and width only),
6084                 # permit trailing "px" for backward compatibility.
6085                 if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6086                         $width = intval( $m[1] );
6087                         $height = intval( $m[2] );
6088                         $parsedWidthParam['width'] = $width;
6089                         $parsedWidthParam['height'] = $height;
6090                 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6091                         $width = intval( $value );
6092                         $parsedWidthParam['width'] = $width;
6093                 }
6094                 return $parsedWidthParam;
6095         }
6096 }