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 $mRevisionSize; # Size to display in {{REVISIONSIZE}} variable
 195         var $mRevIdForTs;   # The revision ID which was used to fetch the timestamp
 196         var $mInputSize = false; # For {{PAGESIZE}} on current page.
 197
 198         /**
 199          * @var string
 200          */
 201         var $mUniqPrefix;
 202
 203         /**
 204          * @var Array with the language name of each language link (i.e. the
 205          * interwiki prefix) in the key, value arbitrary. Used to avoid sending
 206          * duplicate language links to the ParserOutput.
 207          */
 208         var $mLangLinkLanguages;
 209
 210         /**
 211          * Constructor
 212          *
 213          * @param $conf array
 214          */
 215         public function __construct( $conf = array() ) {
 216                 $this->mConf = $conf;
 217                 $this->mUrlProtocols = wfUrlProtocols();
 218                 $this->mExtLinkBracketedRegex = '/\[(((?i)' . $this->mUrlProtocols . ')' .
 219                         self::EXT_LINK_URL_CLASS . '+)\p{Zs}*([^\]\\x00-\\x08\\x0a-\\x1F]*?)\]/Su';
 220                 if ( isset( $conf['preprocessorClass'] ) ) {
 221                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 222                 } elseif ( defined( 'HPHP_VERSION' ) ) {
 223                         # Preprocessor_Hash is much faster than Preprocessor_DOM under HipHop
 224                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 225                 } elseif ( extension_loaded( 'domxml' ) ) {
 226                         # PECL extension that conflicts with the core DOM extension (bug 13770)
 227                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 228                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 229                 } elseif ( extension_loaded( 'dom' ) ) {
 230                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 231                 } else {
 232                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 233                 }
 234                 wfDebug( __CLASS__ . ": using preprocessor: {$this->mPreprocessorClass}\n" );
 235         }
 236
 237         /**
 238          * Reduce memory usage to reduce the impact of circular references
 239          */
 240         function __destruct() {
 241                 if ( isset( $this->mLinkHolders ) ) {
 242                         unset( $this->mLinkHolders );
 243                 }
 244                 foreach ( $this as $name => $value ) {
 245                         unset( $this->$name );
 246                 }
 247         }
 248
 249         /**
 250          * Allow extensions to clean up when the parser is cloned
 251          */
 252         function __clone() {
 253                 wfRunHooks( 'ParserCloned', array( $this ) );
 254         }
 255
 256         /**
 257          * Do various kinds of initialisation on the first call of the parser
 258          */
 259         function firstCallInit() {
 260                 if ( !$this->mFirstCall ) {
 261                         return;
 262                 }
 263                 $this->mFirstCall = false;
 264
 265                 wfProfileIn( __METHOD__ );
 266
 267                 CoreParserFunctions::register( $this );
 268                 CoreTagHooks::register( $this );
 269                 $this->initialiseVariables();
 270
 271                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 272                 wfProfileOut( __METHOD__ );
 273         }
 274
 275         /**
 276          * Clear Parser state
 277          *
 278          * @private
 279          */
 280         function clearState() {
 281                 wfProfileIn( __METHOD__ );
 282                 if ( $this->mFirstCall ) {
 283                         $this->firstCallInit();
 284                 }
 285                 $this->mOutput = new ParserOutput;
 286                 $this->mOptions->registerWatcher( array( $this->mOutput, 'recordOption' ) );
 287                 $this->mAutonumber = 0;
 288                 $this->mLastSection = '';
 289                 $this->mDTopen = false;
 290                 $this->mIncludeCount = array();
 291                 $this->mArgStack = false;
 292                 $this->mInPre = false;
 293                 $this->mLinkHolders = new LinkHolderArray( $this );
 294                 $this->mLinkID = 0;
 295                 $this->mRevisionObject = $this->mRevisionTimestamp =
 296                         $this->mRevisionId = $this->mRevisionUser = $this->mRevisionSize = null;
 297                 $this->mVarCache = array();
 298                 $this->mUser = null;
 299                 $this->mLangLinkLanguages = array();
 300
 301                 /**
 302                  * Prefix for temporary replacement strings for the multipass parser.
 303                  * \x07 should never appear in input as it's disallowed in XML.
 304                  * Using it at the front also gives us a little extra robustness
 305                  * since it shouldn't match when butted up against identifier-like
 306                  * string constructs.
 307                  *
 308                  * Must not consist of all title characters, or else it will change
 309                  * the behavior of <nowiki> in a link.
 310                  */
 311                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 312                 $this->mStripState = new StripState( $this->mUniqPrefix );
 313
 314                 # Clear these on every parse, bug 4549
 315                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 316
 317                 $this->mShowToc = true;
 318                 $this->mForceTocPosition = false;
 319                 $this->mIncludeSizes = array(
 320                         'post-expand' => 0,
 321                         'arg' => 0,
 322                 );
 323                 $this->mPPNodeCount = 0;
 324                 $this->mGeneratedPPNodeCount = 0;
 325                 $this->mHighestExpansionDepth = 0;
 326                 $this->mDefaultSort = false;
 327                 $this->mHeadings = array();
 328                 $this->mDoubleUnderscores = array();
 329                 $this->mExpensiveFunctionCount = 0;
 330
 331                 # Fix cloning
 332                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 333                         $this->mPreprocessor = null;
 334                 }
 335
 336                 wfRunHooks( 'ParserClearState', array( &$this ) );
 337                 wfProfileOut( __METHOD__ );
 338         }
 339
 340         /**
 341          * Convert wikitext to HTML
 342          * Do not call this function recursively.
 343          *
 344          * @param string $text text we want to parse
 345          * @param $title Title object
 346          * @param $options ParserOptions
 347          * @param $linestart boolean
 348          * @param $clearState boolean
 349          * @param int $revid number to pass in {{REVISIONID}}
 350          * @return ParserOutput a ParserOutput
 351          */
 352         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 353                 /**
 354                  * First pass--just handle <nowiki> sections, pass the rest off
 355                  * to internalParse() which does all the real work.
 356                  */
 357
 358                 global $wgUseTidy, $wgAlwaysUseTidy, $wgShowHostnames;
 359                 $fname = __METHOD__ . '-' . wfGetCaller();
 360                 wfProfileIn( __METHOD__ );
 361                 wfProfileIn( $fname );
 362
 363                 $this->startParse( $title, $options, self::OT_HTML, $clearState );
 364
 365                 $this->mInputSize = strlen( $text );
 366                 if ( $this->mOptions->getEnableLimitReport() ) {
 367                         $this->mOutput->resetParseStartTime();
 368                 }
 369
 370                 # Remove the strip marker tag prefix from the input, if present.
 371                 if ( $clearState ) {
 372                         $text = str_replace( $this->mUniqPrefix, '', $text );
 373                 }
 374
 375                 $oldRevisionId = $this->mRevisionId;
 376                 $oldRevisionObject = $this->mRevisionObject;
 377                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 378                 $oldRevisionUser = $this->mRevisionUser;
 379                 $oldRevisionSize = $this->mRevisionSize;
 380                 if ( $revid !== null ) {
 381                         $this->mRevisionId = $revid;
 382                         $this->mRevisionObject = null;
 383                         $this->mRevisionTimestamp = null;
 384                         $this->mRevisionUser = null;
 385                         $this->mRevisionSize = null;
 386                 }
 387
 388                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 389                 # No more strip!
 390                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 391                 $text = $this->internalParse( $text );
 392                 wfRunHooks( 'ParserAfterParse', array( &$this, &$text, &$this->mStripState ) );
 393
 394                 $text = $this->mStripState->unstripGeneral( $text );
 395
 396                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 397                 $fixtags = array(
 398                         # french spaces, last one Guillemet-left
 399                         # only if there is something before the space
 400                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&#160;',
 401                         # french spaces, Guillemet-right
 402                         '/(\\302\\253) /' => '\\1&#160;',
 403                         '/&#160;(!\s*important)/' => ' \\1', # Beware of CSS magic word !important, bug #11874.
 404                 );
 405                 $text = preg_replace( array_keys( $fixtags ), array_values( $fixtags ), $text );
 406
 407                 $text = $this->doBlockLevels( $text, $linestart );
 408
 409                 $this->replaceLinkHolders( $text );
 410
 411                 /**
 412                  * The input doesn't get language converted if
 413                  * a) It's disabled
 414                  * b) Content isn't converted
 415                  * c) It's a conversion table
 416                  * d) it is an interface message (which is in the user language)
 417                  */
 418                 if ( !( $options->getDisableContentConversion()
 419                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] ) ) )
 420                 {
 421                         if ( !$this->mOptions->getInterfaceMessage() ) {
 422                                 # The position of the convert() call should not be changed. it
 423                                 # assumes that the links are all replaced and the only thing left
 424                                 # is the <nowiki> mark.
 425                                 $text = $this->getConverterLanguage()->convert( $text );
 426                         }
 427                 }
 428
 429                 /**
 430                  * A converted title will be provided in the output object if title and
 431                  * content conversion are enabled, the article text does not contain
 432                  * a conversion-suppressing double-underscore tag, and no
 433                  * {{DISPLAYTITLE:...}} is present. DISPLAYTITLE takes precedence over
 434                  * automatic link conversion.
 435                  */
 436                 if ( !( $options->getDisableTitleConversion()
 437                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 438                                 || isset( $this->mDoubleUnderscores['notitleconvert'] )
 439                                 || $this->mOutput->getDisplayTitle() !== false ) )
 440                 {
 441                         $convruletitle = $this->getConverterLanguage()->getConvRuleTitle();
 442                         if ( $convruletitle ) {
 443                                 $this->mOutput->setTitleText( $convruletitle );
 444                         } else {
 445                                 $titleText = $this->getConverterLanguage()->convertTitle( $title );
 446                                 $this->mOutput->setTitleText( $titleText );
 447                         }
 448                 }
 449
 450                 $text = $this->mStripState->unstripNoWiki( $text );
 451
 452                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 453
 454                 $text = $this->replaceTransparentTags( $text );
 455                 $text = $this->mStripState->unstripGeneral( $text );
 456
 457                 $text = Sanitizer::normalizeCharReferences( $text );
 458
 459                 if ( ( $wgUseTidy && $this->mOptions->getTidy() ) || $wgAlwaysUseTidy ) {
 460                         $text = MWTidy::tidy( $text );
 461                 } else {
 462                         # attempt to sanitize at least some nesting problems
 463                         # (bug #2702 and quite a few others)
 464                         $tidyregs = array(
 465                                 # ''Something [http://www.cool.com cool''] -->
 466                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 467                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 468                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 469                                 # fix up an anchor inside another anchor, only
 470                                 # at least for a single single nested link (bug 3695)
 471                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 472                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 473                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 474                                 # contains a div, so fix it up here; replace
 475                                 # div with escaped text
 476                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 477                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 478                                 # remove empty italic or bold tag pairs, some
 479                                 # introduced by rules above
 480                                 '/<([bi])><\/\\1>/' => '',
 481                         );
 482
 483                         $text = preg_replace(
 484                                 array_keys( $tidyregs ),
 485                                 array_values( $tidyregs ),
 486                                 $text );
 487                 }
 488
 489                 if ( $this->mExpensiveFunctionCount > $this->mOptions->getExpensiveParserFunctionLimit() ) {
 490                         $this->limitationWarn( 'expensive-parserfunction',
 491                                 $this->mExpensiveFunctionCount,
 492                                 $this->mOptions->getExpensiveParserFunctionLimit()
 493                         );
 494                 }
 495
 496                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 497
 498                 # Information on include size limits, for the benefit of users who try to skirt them
 499                 if ( $this->mOptions->getEnableLimitReport() ) {
 500                         $max = $this->mOptions->getMaxIncludeSize();
 501
 502                         $cpuTime = $this->mOutput->getTimeSinceStart( 'cpu' );
 503                         if ( $cpuTime !== null ) {
 504                                 $this->mOutput->setLimitReportData( 'limitreport-cputime',
 505                                         sprintf( "%.3f", $cpuTime )
 506                                 );
 507                         }
 508
 509                         $wallTime = $this->mOutput->getTimeSinceStart( 'wall' );
 510                         $this->mOutput->setLimitReportData( 'limitreport-walltime',
 511                                 sprintf( "%.3f", $wallTime )
 512                         );
 513
 514                         $this->mOutput->setLimitReportData( 'limitreport-ppvisitednodes',
 515                                 array( $this->mPPNodeCount, $this->mOptions->getMaxPPNodeCount() )
 516                         );
 517                         $this->mOutput->setLimitReportData( 'limitreport-ppgeneratednodes',
 518                                 array( $this->mGeneratedPPNodeCount, $this->mOptions->getMaxGeneratedPPNodeCount() )
 519                         );
 520                         $this->mOutput->setLimitReportData( 'limitreport-postexpandincludesize',
 521                                 array( $this->mIncludeSizes['post-expand'], $max )
 522                         );
 523                         $this->mOutput->setLimitReportData( 'limitreport-templateargumentsize',
 524                                 array( $this->mIncludeSizes['arg'], $max )
 525                         );
 526                         $this->mOutput->setLimitReportData( 'limitreport-expansiondepth',
 527                                 array( $this->mHighestExpansionDepth, $this->mOptions->getMaxPPExpandDepth() )
 528                         );
 529                         $this->mOutput->setLimitReportData( 'limitreport-expensivefunctioncount',
 530                                 array( $this->mExpensiveFunctionCount, $this->mOptions->getExpensiveParserFunctionLimit() )
 531                         );
 532                         wfRunHooks( 'ParserLimitReportPrepare', array( $this, $this->mOutput ) );
 533
 534                         $limitReport = "NewPP limit report\n";
 535                         if ( $wgShowHostnames ) {
 536                                 $limitReport .= 'Parsed by ' . wfHostname() . "\n";
 537                         }
 538                         foreach ( $this->mOutput->getLimitReportData() as $key => $value ) {
 539                                 if ( wfRunHooks( 'ParserLimitReportFormat',
 540                                         array( $key, $value, &$limitReport, false, false )
 541                                 ) ) {
 542                                         $keyMsg = wfMessage( $key )->inLanguage( 'en' )->useDatabase( false );
 543                                         $valueMsg = wfMessage( array( "$key-value-text", "$key-value" ) )
 544                                                 ->inLanguage( 'en' )->useDatabase( false );
 545                                         if ( !$valueMsg->exists() ) {
 546                                                 $valueMsg = new RawMessage( '$1' );
 547                                         }
 548                                         if ( !$keyMsg->isDisabled() && !$valueMsg->isDisabled() ) {
 549                                                 $valueMsg->params( $value );
 550                                                 $limitReport .= "{$keyMsg->text()}: {$valueMsg->text()}\n";
 551                                         }
 552                                 }
 553                         }
 554                         // Since we're not really outputting HTML, decode the entities and
 555                         // then re-encode the things that need hiding inside HTML comments.
 556                         $limitReport = htmlspecialchars_decode( $limitReport );
 557                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 558
 559                         // Sanitize for comment. Note '‐' in the replacement is U+2010,
 560                         // which looks much like the problematic '-'.
 561                         $limitReport = str_replace( array( '-', '&' ), array( '‐', '&amp;' ), $limitReport );
 562                         $text .= "\n<!-- \n$limitReport-->\n";
 563
 564                         if ( $this->mGeneratedPPNodeCount > $this->mOptions->getMaxGeneratedPPNodeCount() / 10 ) {
 565                                 wfDebugLog( 'generated-pp-node-count', $this->mGeneratedPPNodeCount . ' ' .
 566                                         $this->mTitle->getPrefixedDBkey() );
 567                         }
 568                 }
 569                 $this->mOutput->setText( $text );
 570
 571                 $this->mRevisionId = $oldRevisionId;
 572                 $this->mRevisionObject = $oldRevisionObject;
 573                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 574                 $this->mRevisionUser = $oldRevisionUser;
 575                 $this->mRevisionSize = $oldRevisionSize;
 576                 $this->mInputSize = false;
 577                 wfProfileOut( $fname );
 578                 wfProfileOut( __METHOD__ );
 579
 580                 return $this->mOutput;
 581         }
 582
 583         /**
 584          * Recursive parser entry point that can be called from an extension tag
 585          * hook.
 586          *
 587          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 588          *
 589          * @param string $text text extension wants to have parsed
 590          * @param $frame PPFrame: The frame to use for expanding any template variables
 591          *
 592          * @return string
 593          */
 594         function recursiveTagParse( $text, $frame = false ) {
 595                 wfProfileIn( __METHOD__ );
 596                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 597                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 598                 $text = $this->internalParse( $text, false, $frame );
 599                 wfProfileOut( __METHOD__ );
 600                 return $text;
 601         }
 602
 603         /**
 604          * Expand templates and variables in the text, producing valid, static wikitext.
 605          * Also removes comments.
 606          * @return mixed|string
 607          */
 608         function preprocess( $text, Title $title = null, ParserOptions $options, $revid = null ) {
 609                 wfProfileIn( __METHOD__ );
 610                 $this->startParse( $title, $options, self::OT_PREPROCESS, true );
 611                 if ( $revid !== null ) {
 612                         $this->mRevisionId = $revid;
 613                 }
 614                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 615                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 616                 $text = $this->replaceVariables( $text );
 617                 $text = $this->mStripState->unstripBoth( $text );
 618                 wfProfileOut( __METHOD__ );
 619                 return $text;
 620         }
 621
 622         /**
 623          * Recursive parser entry point that can be called from an extension tag
 624          * hook.
 625          *
 626          * @param string $text text to be expanded
 627          * @param $frame PPFrame: The frame to use for expanding any template variables
 628          * @return String
 629          * @since 1.19
 630          */
 631         public function recursivePreprocess( $text, $frame = false ) {
 632                 wfProfileIn( __METHOD__ );
 633                 $text = $this->replaceVariables( $text, $frame );
 634                 $text = $this->mStripState->unstripBoth( $text );
 635                 wfProfileOut( __METHOD__ );
 636                 return $text;
 637         }
 638
 639         /**
 640          * Process the wikitext for the "?preload=" feature. (bug 5210)
 641          *
 642          * "<noinclude>", "<includeonly>" etc. are parsed as for template
 643          * transclusion, comments, templates, arguments, tags hooks and parser
 644          * functions are untouched.
 645          *
 646          * @param $text String
 647          * @param $title Title
 648          * @param $options ParserOptions
 649          * @return String
 650          */
 651         public function getPreloadText( $text, Title $title, ParserOptions $options ) {
 652                 # Parser (re)initialisation
 653                 $this->startParse( $title, $options, self::OT_PLAIN, true );
 654
 655                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 656                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
 657                 $text = $this->getPreprocessor()->newFrame()->expand( $dom, $flags );
 658                 $text = $this->mStripState->unstripBoth( $text );
 659                 return $text;
 660         }
 661
 662         /**
 663          * Get a random string
 664          *
 665          * @return string
 666          */
 667         public static function getRandomString() {
 668                 return wfRandomString( 16 );
 669         }
 670
 671         /**
 672          * Set the current user.
 673          * Should only be used when doing pre-save transform.
 674          *
 675          * @param $user Mixed: User object or null (to reset)
 676          */
 677         function setUser( $user ) {
 678                 $this->mUser = $user;
 679         }
 680
 681         /**
 682          * Accessor for mUniqPrefix.
 683          *
 684          * @return String
 685          */
 686         public function uniqPrefix() {
 687                 if ( !isset( $this->mUniqPrefix ) ) {
 688                         # @todo FIXME: This is probably *horribly wrong*
 689                         # LanguageConverter seems to want $wgParser's uniqPrefix, however
 690                         # if this is called for a parser cache hit, the parser may not
 691                         # have ever been initialized in the first place.
 692                         # Not really sure what the heck is supposed to be going on here.
 693                         return '';
 694                         # throw new MWException( "Accessing uninitialized mUniqPrefix" );
 695                 }
 696                 return $this->mUniqPrefix;
 697         }
 698
 699         /**
 700          * Set the context title
 701          *
 702          * @param $t Title
 703          */
 704         function setTitle( $t ) {
 705                 if ( !$t || $t instanceof FakeTitle ) {
 706                         $t = Title::newFromText( 'NO TITLE' );
 707                 }
 708
 709                 if ( strval( $t->getFragment() ) !== '' ) {
 710                         # Strip the fragment to avoid various odd effects
 711                         $this->mTitle = clone $t;
 712                         $this->mTitle->setFragment( '' );
 713                 } else {
 714                         $this->mTitle = $t;
 715                 }
 716         }
 717
 718         /**
 719          * Accessor for the Title object
 720          *
 721          * @return Title object
 722          */
 723         function getTitle() {
 724                 return $this->mTitle;
 725         }
 726
 727         /**
 728          * Accessor/mutator for the Title object
 729          *
 730          * @param $x Title object or null to just get the current one
 731          * @return Title object
 732          */
 733         function Title( $x = null ) {
 734                 return wfSetVar( $this->mTitle, $x );
 735         }
 736
 737         /**
 738          * Set the output type
 739          *
 740          * @param $ot Integer: new value
 741          */
 742         function setOutputType( $ot ) {
 743                 $this->mOutputType = $ot;
 744                 # Shortcut alias
 745                 $this->ot = array(
 746                         'html' => $ot == self::OT_HTML,
 747                         'wiki' => $ot == self::OT_WIKI,
 748                         'pre' => $ot == self::OT_PREPROCESS,
 749                         'plain' => $ot == self::OT_PLAIN,
 750                 );
 751         }
 752
 753         /**
 754          * Accessor/mutator for the output type
 755          *
 756          * @param int|null $x New value or null to just get the current one
 757          * @return Integer
 758          */
 759         function OutputType( $x = null ) {
 760                 return wfSetVar( $this->mOutputType, $x );
 761         }
 762
 763         /**
 764          * Get the ParserOutput object
 765          *
 766          * @return ParserOutput object
 767          */
 768         function getOutput() {
 769                 return $this->mOutput;
 770         }
 771
 772         /**
 773          * Get the ParserOptions object
 774          *
 775          * @return ParserOptions object
 776          */
 777         function getOptions() {
 778                 return $this->mOptions;
 779         }
 780
 781         /**
 782          * Accessor/mutator for the ParserOptions object
 783          *
 784          * @param $x ParserOptions New value or null to just get the current one
 785          * @return ParserOptions Current ParserOptions object
 786          */
 787         function Options( $x = null ) {
 788                 return wfSetVar( $this->mOptions, $x );
 789         }
 790
 791         /**
 792          * @return int
 793          */
 794         function nextLinkID() {
 795                 return $this->mLinkID++;
 796         }
 797
 798         /**
 799          * @param $id int
 800          */
 801         function setLinkID( $id ) {
 802                 $this->mLinkID = $id;
 803         }
 804
 805         /**
 806          * Get a language object for use in parser functions such as {{FORMATNUM:}}
 807          * @return Language
 808          */
 809         function getFunctionLang() {
 810                 return $this->getTargetLanguage();
 811         }
 812
 813         /**
 814          * Get the target language for the content being parsed. This is usually the
 815          * language that the content is in.
 816          *
 817          * @since 1.19
 818          *
 819          * @throws MWException
 820          * @return Language|null
 821          */
 822         public function getTargetLanguage() {
 823                 $target = $this->mOptions->getTargetLanguage();
 824
 825                 if ( $target !== null ) {
 826                         return $target;
 827                 } elseif ( $this->mOptions->getInterfaceMessage() ) {
 828                         return $this->mOptions->getUserLangObj();
 829                 } elseif ( is_null( $this->mTitle ) ) {
 830                         throw new MWException( __METHOD__ . ': $this->mTitle is null' );
 831                 }
 832
 833                 return $this->mTitle->getPageLanguage();
 834         }
 835
 836         /**
 837          * Get the language object for language conversion
 838          */
 839         function getConverterLanguage() {
 840                 return $this->getTargetLanguage();
 841         }
 842
 843         /**
 844          * Get a User object either from $this->mUser, if set, or from the
 845          * ParserOptions object otherwise
 846          *
 847          * @return User object
 848          */
 849         function getUser() {
 850                 if ( !is_null( $this->mUser ) ) {
 851                         return $this->mUser;
 852                 }
 853                 return $this->mOptions->getUser();
 854         }
 855
 856         /**
 857          * Get a preprocessor object
 858          *
 859          * @return Preprocessor instance
 860          */
 861         function getPreprocessor() {
 862                 if ( !isset( $this->mPreprocessor ) ) {
 863                         $class = $this->mPreprocessorClass;
 864                         $this->mPreprocessor = new $class( $this );
 865                 }
 866                 return $this->mPreprocessor;
 867         }
 868
 869         /**
 870          * Replaces all occurrences of HTML-style comments and the given tags
 871          * in the text with a random marker and returns the next text. The output
 872          * parameter $matches will be an associative array filled with data in
 873          * the form:
 874          *
 875          * @code
 876          *   'UNIQ-xxxxx' => array(
 877          *     'element',
 878          *     'tag content',
 879          *     array( 'param' => 'x' ),
 880          *     '<element param="x">tag content</element>' ) )
 881          * @endcode
 882          *
 883          * @param array $elements list of element names. Comments are always extracted.
 884          * @param string $text Source text string.
 885          * @param array $matches Out parameter, Array: extracted tags
 886          * @param $uniq_prefix string
 887          * @return String: stripped text
 888          */
 889         public static function extractTagsAndParams( $elements, $text, &$matches, $uniq_prefix = '' ) {
 890                 static $n = 1;
 891                 $stripped = '';
 892                 $matches = array();
 893
 894                 $taglist = implode( '|', $elements );
 895                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 896
 897                 while ( $text != '' ) {
 898                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 899                         $stripped .= $p[0];
 900                         if ( count( $p ) < 5 ) {
 901                                 break;
 902                         }
 903                         if ( count( $p ) > 5 ) {
 904                                 # comment
 905                                 $element = $p[4];
 906                                 $attributes = '';
 907                                 $close = '';
 908                                 $inside = $p[5];
 909                         } else {
 910                                 # tag
 911                                 $element = $p[1];
 912                                 $attributes = $p[2];
 913                                 $close = $p[3];
 914                                 $inside = $p[4];
 915                         }
 916
 917                         $marker = "$uniq_prefix-$element-" . sprintf( '%08X', $n++ ) . self::MARKER_SUFFIX;
 918                         $stripped .= $marker;
 919
 920                         if ( $close === '/>' ) {
 921                                 # Empty element tag, <tag />
 922                                 $content = null;
 923                                 $text = $inside;
 924                                 $tail = null;
 925                         } else {
 926                                 if ( $element === '!--' ) {
 927                                         $end = '/(-->)/';
 928                                 } else {
 929                                         $end = "/(<\\/$element\\s*>)/i";
 930                                 }
 931                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 932                                 $content = $q[0];
 933                                 if ( count( $q ) < 3 ) {
 934                                         # No end tag -- let it run out to the end of the text.
 935                                         $tail = '';
 936                                         $text = '';
 937                                 } else {
 938                                         $tail = $q[1];
 939                                         $text = $q[2];
 940                                 }
 941                         }
 942
 943                         $matches[$marker] = array( $element,
 944                                 $content,
 945                                 Sanitizer::decodeTagAttributes( $attributes ),
 946                                 "<$element$attributes$close$content$tail" );
 947                 }
 948                 return $stripped;
 949         }
 950
 951         /**
 952          * Get a list of strippable XML-like elements
 953          *
 954          * @return array
 955          */
 956         function getStripList() {
 957                 return $this->mStripList;
 958         }
 959
 960         /**
 961          * Add an item to the strip state
 962          * Returns the unique tag which must be inserted into the stripped text
 963          * The tag will be replaced with the original text in unstrip()
 964          *
 965          * @param $text string
 966          *
 967          * @return string
 968          */
 969         function insertStripItem( $text ) {
 970                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 971                 $this->mMarkerIndex++;
 972                 $this->mStripState->addGeneral( $rnd, $text );
 973                 return $rnd;
 974         }
 975
 976         /**
 977          * parse the wiki syntax used to render tables
 978          *
 979          * @private
 980          * @return string
 981          */
 982         function doTableStuff( $text ) {
 983                 wfProfileIn( __METHOD__ );
 984
 985                 $lines = StringUtils::explode( "\n", $text );
 986                 $out = '';
 987                 $td_history = array(); # Is currently a td tag open?
 988                 $last_tag_history = array(); # Save history of last lag activated (td, th or caption)
 989                 $tr_history = array(); # Is currently a tr tag open?
 990                 $tr_attributes = array(); # history of tr attributes
 991                 $has_opened_tr = array(); # Did this table open a <tr> element?
 992                 $indent_level = 0; # indent level of the table
 993
 994                 foreach ( $lines as $outLine ) {
 995                         $line = trim( $outLine );
 996
 997                         if ( $line === '' ) { # empty line, go to next line
 998                                 $out .= $outLine . "\n";
 999                                 continue;
1000                         }
1001
1002                         $first_character = $line[0];
1003                         $matches = array();
1004
1005                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line, $matches ) ) {
1006                                 # First check if we are starting a new table
1007                                 $indent_level = strlen( $matches[1] );
1008
1009                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
1010                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'table' );
1011
1012                                 $outLine = str_repeat( '<dl><dd>', $indent_level ) . "<table{$attributes}>";
1013                                 array_push( $td_history, false );
1014                                 array_push( $last_tag_history, '' );
1015                                 array_push( $tr_history, false );
1016                                 array_push( $tr_attributes, '' );
1017                                 array_push( $has_opened_tr, false );
1018                         } elseif ( count( $td_history ) == 0 ) {
1019                                 # Don't do any of the following
1020                                 $out .= $outLine . "\n";
1021                                 continue;
1022                         } elseif ( substr( $line, 0, 2 ) === '|}' ) {
1023                                 # We are ending a table
1024                                 $line = '</table>' . substr( $line, 2 );
1025                                 $last_tag = array_pop( $last_tag_history );
1026
1027                                 if ( !array_pop( $has_opened_tr ) ) {
1028                                         $line = "<tr><td></td></tr>{$line}";
1029                                 }
1030
1031                                 if ( array_pop( $tr_history ) ) {
1032                                         $line = "</tr>{$line}";
1033                                 }
1034
1035                                 if ( array_pop( $td_history ) ) {
1036                                         $line = "</{$last_tag}>{$line}";
1037                                 }
1038                                 array_pop( $tr_attributes );
1039                                 $outLine = $line . str_repeat( '</dd></dl>', $indent_level );
1040                         } elseif ( substr( $line, 0, 2 ) === '|-' ) {
1041                                 # Now we have a table row
1042                                 $line = preg_replace( '#^\|-+#', '', $line );
1043
1044                                 # Whats after the tag is now only attributes
1045                                 $attributes = $this->mStripState->unstripBoth( $line );
1046                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
1047                                 array_pop( $tr_attributes );
1048                                 array_push( $tr_attributes, $attributes );
1049
1050                                 $line = '';
1051                                 $last_tag = array_pop( $last_tag_history );
1052                                 array_pop( $has_opened_tr );
1053                                 array_push( $has_opened_tr, true );
1054
1055                                 if ( array_pop( $tr_history ) ) {
1056                                         $line = '</tr>';
1057                                 }
1058
1059                                 if ( array_pop( $td_history ) ) {
1060                                         $line = "</{$last_tag}>{$line}";
1061                                 }
1062
1063                                 $outLine = $line;
1064                                 array_push( $tr_history, false );
1065                                 array_push( $td_history, false );
1066                                 array_push( $last_tag_history, '' );
1067                         } elseif ( $first_character === '|' || $first_character === '!' || substr( $line, 0, 2 ) === '|+' ) {
1068                                 # This might be cell elements, td, th or captions
1069                                 if ( substr( $line, 0, 2 ) === '|+' ) {
1070                                         $first_character = '+';
1071                                         $line = substr( $line, 1 );
1072                                 }
1073
1074                                 $line = substr( $line, 1 );
1075
1076                                 if ( $first_character === '!' ) {
1077                                         $line = str_replace( '!!', '||', $line );
1078                                 }
1079
1080                                 # Split up multiple cells on the same line.
1081                                 # FIXME : This can result in improper nesting of tags processed
1082                                 # by earlier parser steps, but should avoid splitting up eg
1083                                 # attribute values containing literal "||".
1084                                 $cells = StringUtils::explodeMarkup( '||', $line );
1085
1086                                 $outLine = '';
1087
1088                                 # Loop through each table cell
1089                                 foreach ( $cells as $cell ) {
1090                                         $previous = '';
1091                                         if ( $first_character !== '+' ) {
1092                                                 $tr_after = array_pop( $tr_attributes );
1093                                                 if ( !array_pop( $tr_history ) ) {
1094                                                         $previous = "<tr{$tr_after}>\n";
1095                                                 }
1096                                                 array_push( $tr_history, true );
1097                                                 array_push( $tr_attributes, '' );
1098                                                 array_pop( $has_opened_tr );
1099                                                 array_push( $has_opened_tr, true );
1100                                         }
1101
1102                                         $last_tag = array_pop( $last_tag_history );
1103
1104                                         if ( array_pop( $td_history ) ) {
1105                                                 $previous = "</{$last_tag}>\n{$previous}";
1106                                         }
1107
1108                                         if ( $first_character === '|' ) {
1109                                                 $last_tag = 'td';
1110                                         } elseif ( $first_character === '!' ) {
1111                                                 $last_tag = 'th';
1112                                         } elseif ( $first_character === '+' ) {
1113                                                 $last_tag = 'caption';
1114                                         } else {
1115                                                 $last_tag = '';
1116                                         }
1117
1118                                         array_push( $last_tag_history, $last_tag );
1119
1120                                         # A cell could contain both parameters and data
1121                                         $cell_data = explode( '|', $cell, 2 );
1122
1123                                         # Bug 553: Note that a '|' inside an invalid link should not
1124                                         # be mistaken as delimiting cell parameters
1125                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
1126                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
1127                                         } elseif ( count( $cell_data ) == 1 ) {
1128                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
1129                                         } else {
1130                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
1131                                                 $attributes = Sanitizer::fixTagAttributes( $attributes, $last_tag );
1132                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
1133                                         }
1134
1135                                         $outLine .= $cell;
1136                                         array_push( $td_history, true );
1137                                 }
1138                         }
1139                         $out .= $outLine . "\n";
1140                 }
1141
1142                 # Closing open td, tr && table
1143                 while ( count( $td_history ) > 0 ) {
1144                         if ( array_pop( $td_history ) ) {
1145                                 $out .= "</td>\n";
1146                         }
1147                         if ( array_pop( $tr_history ) ) {
1148                                 $out .= "</tr>\n";
1149                         }
1150                         if ( !array_pop( $has_opened_tr ) ) {
1151                                 $out .= "<tr><td></td></tr>\n";
1152                         }
1153
1154                         $out .= "</table>\n";
1155                 }
1156
1157                 # Remove trailing line-ending (b/c)
1158                 if ( substr( $out, -1 ) === "\n" ) {
1159                         $out = substr( $out, 0, -1 );
1160                 }
1161
1162                 # special case: don't return empty table
1163                 if ( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
1164                         $out = '';
1165                 }
1166
1167                 wfProfileOut( __METHOD__ );
1168
1169                 return $out;
1170         }
1171
1172         /**
1173          * Helper function for parse() that transforms wiki markup into
1174          * HTML. Only called for $mOutputType == self::OT_HTML.
1175          *
1176          * @private
1177          *
1178          * @param $text string
1179          * @param $isMain bool
1180          * @param $frame bool
1181          *
1182          * @return string
1183          */
1184         function internalParse( $text, $isMain = true, $frame = false ) {
1185                 wfProfileIn( __METHOD__ );
1186
1187                 $origText = $text;
1188
1189                 # Hook to suspend the parser in this state
1190                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
1191                         wfProfileOut( __METHOD__ );
1192                         return $text;
1193                 }
1194
1195                 # if $frame is provided, then use $frame for replacing any variables
1196                 if ( $frame ) {
1197                         # use frame depth to infer how include/noinclude tags should be handled
1198                         # depth=0 means this is the top-level document; otherwise it's an included document
1199                         if ( !$frame->depth ) {
1200                                 $flag = 0;
1201                         } else {
1202                                 $flag = Parser::PTD_FOR_INCLUSION;
1203                         }
1204                         $dom = $this->preprocessToDom( $text, $flag );
1205                         $text = $frame->expand( $dom );
1206                 } else {
1207                         # if $frame is not provided, then use old-style replaceVariables
1208                         $text = $this->replaceVariables( $text );
1209                 }
1210
1211                 wfRunHooks( 'InternalParseBeforeSanitize', array( &$this, &$text, &$this->mStripState ) );
1212                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
1213                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
1214
1215                 # Tables need to come after variable replacement for things to work
1216                 # properly; putting them before other transformations should keep
1217                 # exciting things like link expansions from showing up in surprising
1218                 # places.
1219                 $text = $this->doTableStuff( $text );
1220
1221                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1222
1223                 $text = $this->doDoubleUnderscore( $text );
1224
1225                 $text = $this->doHeadings( $text );
1226                 $text = $this->replaceInternalLinks( $text );
1227                 $text = $this->doAllQuotes( $text );
1228                 $text = $this->replaceExternalLinks( $text );
1229
1230                 # replaceInternalLinks may sometimes leave behind
1231                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1232                 $text = str_replace( $this->mUniqPrefix . 'NOPARSE', '', $text );
1233
1234                 $text = $this->doMagicLinks( $text );
1235                 $text = $this->formatHeadings( $text, $origText, $isMain );
1236
1237                 wfProfileOut( __METHOD__ );
1238                 return $text;
1239         }
1240
1241         /**
1242          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1243          * magic external links.
1244          *
1245          * DML
1246          * @private
1247          *
1248          * @param $text string
1249          *
1250          * @return string
1251          */
1252         function doMagicLinks( $text ) {
1253                 wfProfileIn( __METHOD__ );
1254                 $prots = wfUrlProtocolsWithoutProtRel();
1255                 $urlChar = self::EXT_LINK_URL_CLASS;
1256                 $text = preg_replace_callback(
1257                         '!(?:                           # Start cases
1258                                 (<a[ \t\r\n>].*?</a>) |     # m[1]: Skip link text
1259                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1260                                 (\\b(?i:$prots)$urlChar+) |  # m[3]: Free external links" . '
1261                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1262                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1263                                         (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1264                                         (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1265                                         [0-9Xx]                 # check digit
1266                                         \b)
1267                         )!xu', array( &$this, 'magicLinkCallback' ), $text );
1268                 wfProfileOut( __METHOD__ );
1269                 return $text;
1270         }
1271
1272         /**
1273          * @throws MWException
1274          * @param $m array
1275          * @return HTML|string
1276          */
1277         function magicLinkCallback( $m ) {
1278                 if ( isset( $m[1] ) && $m[1] !== '' ) {
1279                         # Skip anchor
1280                         return $m[0];
1281                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1282                         # Skip HTML element
1283                         return $m[0];
1284                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1285                         # Free external link
1286                         return $this->makeFreeExternalLink( $m[0] );
1287                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1288                         # RFC or PMID
1289                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1290                                 $keyword = 'RFC';
1291                                 $urlmsg = 'rfcurl';
1292                                 $CssClass = 'mw-magiclink-rfc';
1293                                 $id = $m[4];
1294                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1295                                 $keyword = 'PMID';
1296                                 $urlmsg = 'pubmedurl';
1297                                 $CssClass = 'mw-magiclink-pmid';
1298                                 $id = $m[4];
1299                         } else {
1300                                 throw new MWException( __METHOD__ . ': unrecognised match type "' .
1301                                         substr( $m[0], 0, 20 ) . '"' );
1302                         }
1303                         $url = wfMessage( $urlmsg, $id )->inContentLanguage()->text();
1304                         return Linker::makeExternalLink( $url, "{$keyword} {$id}", true, $CssClass );
1305                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1306                         # ISBN
1307                         $isbn = $m[5];
1308                         $num = strtr( $isbn, array(
1309                                 '-' => '',
1310                                 ' ' => '',
1311                                 'x' => 'X',
1312                         ));
1313                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1314                         return '<a href="' .
1315                                 htmlspecialchars( $titleObj->getLocalURL() ) .
1316                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1317                 } else {
1318                         return $m[0];
1319                 }
1320         }
1321
1322         /**
1323          * Make a free external link, given a user-supplied URL
1324          *
1325          * @param $url string
1326          *
1327          * @return string HTML
1328          * @private
1329          */
1330         function makeFreeExternalLink( $url ) {
1331                 wfProfileIn( __METHOD__ );
1332
1333                 $trail = '';
1334
1335                 # The characters '<' and '>' (which were escaped by
1336                 # removeHTMLtags()) should not be included in
1337                 # URLs, per RFC 2396.
1338                 $m2 = array();
1339                 if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1340                         $trail = substr( $url, $m2[0][1] ) . $trail;
1341                         $url = substr( $url, 0, $m2[0][1] );
1342                 }
1343
1344                 # Move trailing punctuation to $trail
1345                 $sep = ',;\.:!?';
1346                 # If there is no left bracket, then consider right brackets fair game too
1347                 if ( strpos( $url, '(' ) === false ) {
1348                         $sep .= ')';
1349                 }
1350
1351                 $numSepChars = strspn( strrev( $url ), $sep );
1352                 if ( $numSepChars ) {
1353                         $trail = substr( $url, -$numSepChars ) . $trail;
1354                         $url = substr( $url, 0, -$numSepChars );
1355                 }
1356
1357                 $url = Sanitizer::cleanUrl( $url );
1358
1359                 # Is this an external image?
1360                 $text = $this->maybeMakeExternalImage( $url );
1361                 if ( $text === false ) {
1362                         # Not an image, make a link
1363                         $text = Linker::makeExternalLink( $url,
1364                                 $this->getConverterLanguage()->markNoConversion( $url, true ),
1365                                 true, 'free',
1366                                 $this->getExternalLinkAttribs( $url ) );
1367                         # Register it in the output object...
1368                         # Replace unnecessary URL escape codes with their equivalent characters
1369                         $pasteurized = self::replaceUnusualEscapes( $url );
1370                         $this->mOutput->addExternalLink( $pasteurized );
1371                 }
1372                 wfProfileOut( __METHOD__ );
1373                 return $text . $trail;
1374         }
1375
1376         /**
1377          * Parse headers and return html
1378          *
1379          * @private
1380          *
1381          * @param $text string
1382          *
1383          * @return string
1384          */
1385         function doHeadings( $text ) {
1386                 wfProfileIn( __METHOD__ );
1387                 for ( $i = 6; $i >= 1; --$i ) {
1388                         $h = str_repeat( '=', $i );
1389                         $text = preg_replace( "/^$h(.+)$h\\s*$/m", "<h$i>\\1</h$i>", $text );
1390                 }
1391                 wfProfileOut( __METHOD__ );
1392                 return $text;
1393         }
1394
1395         /**
1396          * Replace single quotes with HTML markup
1397          * @private
1398          *
1399          * @param $text string
1400          *
1401          * @return string the altered text
1402          */
1403         function doAllQuotes( $text ) {
1404                 wfProfileIn( __METHOD__ );
1405                 $outtext = '';
1406                 $lines = StringUtils::explode( "\n", $text );
1407                 foreach ( $lines as $line ) {
1408                         $outtext .= $this->doQuotes( $line ) . "\n";
1409                 }
1410                 $outtext = substr( $outtext, 0, -1 );
1411                 wfProfileOut( __METHOD__ );
1412                 return $outtext;
1413         }
1414
1415         /**
1416          * Helper function for doAllQuotes()
1417          *
1418          * @param $text string
1419          *
1420          * @return string
1421          */
1422         public function doQuotes( $text ) {
1423                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1424                 $countarr = count( $arr );
1425                 if ( $countarr == 1 ) {
1426                         return $text;
1427                 }
1428
1429                 // First, do some preliminary work. This may shift some apostrophes from
1430                 // being mark-up to being text. It also counts the number of occurrences
1431                 // of bold and italics mark-ups.
1432                 $numbold = 0;
1433                 $numitalics = 0;
1434                 for ( $i = 1; $i < $countarr; $i += 2 ) {
1435                         $thislen = strlen( $arr[$i] );
1436                         // If there are ever four apostrophes, assume the first is supposed to
1437                         // be text, and the remaining three constitute mark-up for bold text.
1438                         // (bug 13227: ''''foo'''' turns into ' ''' foo ' ''')
1439                         if ( $thislen == 4 ) {
1440                                 $arr[$i - 1] .= "'";
1441                                 $arr[$i] = "'''";
1442                                 $thislen = 3;
1443                         } elseif ( $thislen > 5 ) {
1444                                 // If there are more than 5 apostrophes in a row, assume they're all
1445                                 // text except for the last 5.
1446                                 // (bug 13227: ''''''foo'''''' turns into ' ''''' foo ' ''''')
1447                                 $arr[$i - 1] .= str_repeat( "'", $thislen - 5 );
1448                                 $arr[$i] = "'''''";
1449                                 $thislen = 5;
1450                         }
1451                         // Count the number of occurrences of bold and italics mark-ups.
1452                         if ( $thislen == 2 ) {
1453                                 $numitalics++;
1454                         } elseif ( $thislen == 3 ) {
1455                                 $numbold++;
1456                         } elseif ( $thislen == 5 ) {
1457                                 $numitalics++;
1458                                 $numbold++;
1459                         }
1460                 }
1461
1462                 // If there is an odd number of both bold and italics, it is likely
1463                 // that one of the bold ones was meant to be an apostrophe followed
1464                 // by italics. Which one we cannot know for certain, but it is more
1465                 // likely to be one that has a single-letter word before it.
1466                 if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) ) {
1467                         $firstsingleletterword = -1;
1468                         $firstmultiletterword = -1;
1469                         $firstspace = -1;
1470                         for ( $i = 1; $i < $countarr; $i += 2 ) {
1471                                 if ( strlen( $arr[$i] ) == 3 ) {
1472                                         $x1 = substr( $arr[$i - 1], -1 );
1473                                         $x2 = substr( $arr[$i - 1], -2, 1 );
1474                                         if ( $x1 === ' ' ) {
1475                                                 if ( $firstspace == -1 ) {
1476                                                         $firstspace = $i;
1477                                                 }
1478                                         } elseif ( $x2 === ' ' ) {
1479                                                 if ( $firstsingleletterword == -1 ) {
1480                                                         $firstsingleletterword = $i;
1481                                                         // if $firstsingleletterword is set, we don't
1482                                                         // look at the other options, so we can bail early.
1483                                                         break;
1484                                                 }
1485                                         } else {
1486                                                 if ( $firstmultiletterword == -1 ) {
1487                                                         $firstmultiletterword = $i;
1488                                                 }
1489                                         }
1490                                 }
1491                         }
1492
1493                         // If there is a single-letter word, use it!
1494                         if ( $firstsingleletterword > -1 ) {
1495                                 $arr[$firstsingleletterword] = "''";
1496                                 $arr[$firstsingleletterword - 1] .= "'";
1497                         } elseif ( $firstmultiletterword > -1 ) {
1498                                 // If not, but there's a multi-letter word, use that one.
1499                                 $arr[$firstmultiletterword] = "''";
1500                                 $arr[$firstmultiletterword - 1] .= "'";
1501                         } elseif ( $firstspace > -1 ) {
1502                                 // ... otherwise use the first one that has neither.
1503                                 // (notice that it is possible for all three to be -1 if, for example,
1504                                 // there is only one pentuple-apostrophe in the line)
1505                                 $arr[$firstspace] = "''";
1506                                 $arr[$firstspace - 1] .= "'";
1507                         }
1508                 }
1509
1510                 // Now let's actually convert our apostrophic mush to HTML!
1511                 $output = '';
1512                 $buffer = '';
1513                 $state = '';
1514                 $i = 0;
1515                 foreach ( $arr as $r ) {
1516                         if ( ( $i % 2 ) == 0 ) {
1517                                 if ( $state === 'both' ) {
1518                                         $buffer .= $r;
1519                                 } else {
1520                                         $output .= $r;
1521                                 }
1522                         } else {
1523                                 $thislen = strlen( $r );
1524                                 if ( $thislen == 2 ) {
1525                                         if ( $state === 'i' ) {
1526                                                 $output .= '</i>';
1527                                                 $state = '';
1528                                         } elseif ( $state === 'bi' ) {
1529                                                 $output .= '</i>';
1530                                                 $state = 'b';
1531                                         } elseif ( $state === 'ib' ) {
1532                                                 $output .= '</b></i><b>';
1533                                                 $state = 'b';
1534                                         } elseif ( $state === 'both' ) {
1535                                                 $output .= '<b><i>' . $buffer . '</i>';
1536                                                 $state = 'b';
1537                                         } else { // $state can be 'b' or ''
1538                                                 $output .= '<i>';
1539                                                 $state .= 'i';
1540                                         }
1541                                 } elseif ( $thislen == 3 ) {
1542                                         if ( $state === 'b' ) {
1543                                                 $output .= '</b>';
1544                                                 $state = '';
1545                                         } elseif ( $state === 'bi' ) {
1546                                                 $output .= '</i></b><i>';
1547                                                 $state = 'i';
1548                                         } elseif ( $state === 'ib' ) {
1549                                                 $output .= '</b>';
1550                                                 $state = 'i';
1551                                         } elseif ( $state === 'both' ) {
1552                                                 $output .= '<i><b>' . $buffer . '</b>';
1553                                                 $state = 'i';
1554                                         } else { // $state can be 'i' or ''
1555                                                 $output .= '<b>';
1556                                                 $state .= 'b';
1557                                         }
1558                                 } elseif ( $thislen == 5 ) {
1559                                         if ( $state === 'b' ) {
1560                                                 $output .= '</b><i>';
1561                                                 $state = 'i';
1562                                         } elseif ( $state === 'i' ) {
1563                                                 $output .= '</i><b>';
1564                                                 $state = 'b';
1565                                         } elseif ( $state === 'bi' ) {
1566                                                 $output .= '</i></b>';
1567                                                 $state = '';
1568                                         } elseif ( $state === 'ib' ) {
1569                                                 $output .= '</b></i>';
1570                                                 $state = '';
1571                                         } elseif ( $state === 'both' ) {
1572                                                 $output .= '<i><b>' . $buffer . '</b></i>';
1573                                                 $state = '';
1574                                         } else { // ($state == '')
1575                                                 $buffer = '';
1576                                                 $state = 'both';
1577                                         }
1578                                 }
1579                         }
1580                         $i++;
1581                 }
1582                 // Now close all remaining tags.  Notice that the order is important.
1583                 if ( $state === 'b' || $state === 'ib' ) {
1584                         $output .= '</b>';
1585                 }
1586                 if ( $state === 'i' || $state === 'bi' || $state === 'ib' ) {
1587                         $output .= '</i>';
1588                 }
1589                 if ( $state === 'bi' ) {
1590                         $output .= '</b>';
1591                 }
1592                 // There might be lonely ''''', so make sure we have a buffer
1593                 if ( $state === 'both' && $buffer ) {
1594                         $output .= '<b><i>' . $buffer . '</i></b>';
1595                 }
1596                 return $output;
1597         }
1598
1599         /**
1600          * Replace external links (REL)
1601          *
1602          * Note: this is all very hackish and the order of execution matters a lot.
1603          * Make sure to run tests/parserTests.php if you change this code.
1604          *
1605          * @private
1606          *
1607          * @param $text string
1608          *
1609          * @throws MWException
1610          * @return string
1611          */
1612         function replaceExternalLinks( $text ) {
1613                 wfProfileIn( __METHOD__ );
1614
1615                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1616                 if ( $bits === false ) {
1617                         wfProfileOut( __METHOD__ );
1618                         throw new MWException( "PCRE needs to be compiled with --enable-unicode-properties in order for MediaWiki to function" );
1619                 }
1620                 $s = array_shift( $bits );
1621
1622                 $i = 0;
1623                 while ( $i < count( $bits ) ) {
1624                         $url = $bits[$i++];
1625                         $i++; // protocol
1626                         $text = $bits[$i++];
1627                         $trail = $bits[$i++];
1628
1629                         # The characters '<' and '>' (which were escaped by
1630                         # removeHTMLtags()) should not be included in
1631                         # URLs, per RFC 2396.
1632                         $m2 = array();
1633                         if ( preg_match( '/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE ) ) {
1634                                 $text = substr( $url, $m2[0][1] ) . ' ' . $text;
1635                                 $url = substr( $url, 0, $m2[0][1] );
1636                         }
1637
1638                         # If the link text is an image URL, replace it with an <img> tag
1639                         # This happened by accident in the original parser, but some people used it extensively
1640                         $img = $this->maybeMakeExternalImage( $text );
1641                         if ( $img !== false ) {
1642                                 $text = $img;
1643                         }
1644
1645                         $dtrail = '';
1646
1647                         # Set linktype for CSS - if URL==text, link is essentially free
1648                         $linktype = ( $text === $url ) ? 'free' : 'text';
1649
1650                         # No link text, e.g. [http://domain.tld/some.link]
1651                         if ( $text == '' ) {
1652                                 # Autonumber
1653                                 $langObj = $this->getTargetLanguage();
1654                                 $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1655                                 $linktype = 'autonumber';
1656                         } else {
1657                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1658                                 # Check for trail
1659                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1660                         }
1661
1662                         $text = $this->getConverterLanguage()->markNoConversion( $text );
1663
1664                         $url = Sanitizer::cleanUrl( $url );
1665
1666                         # Use the encoded URL
1667                         # This means that users can paste URLs directly into the text
1668                         # Funny characters like ö aren't valid in URLs anyway
1669                         # This was changed in August 2004
1670                         $s .= Linker::makeExternalLink( $url, $text, false, $linktype,
1671                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1672
1673                         # Register link in the output object.
1674                         # Replace unnecessary URL escape codes with the referenced character
1675                         # This prevents spammers from hiding links from the filters
1676                         $pasteurized = self::replaceUnusualEscapes( $url );
1677                         $this->mOutput->addExternalLink( $pasteurized );
1678                 }
1679
1680                 wfProfileOut( __METHOD__ );
1681                 return $s;
1682         }
1683         /**
1684          * Get the rel attribute for a particular external link.
1685          *
1686          * @since 1.21
1687          * @param string|bool $url optional URL, to extract the domain from for rel =>
1688          *   nofollow if appropriate
1689          * @param $title Title optional Title, for wgNoFollowNsExceptions lookups
1690          * @return string|null rel attribute for $url
1691          */
1692         public static function getExternalLinkRel( $url = false, $title = null ) {
1693                 global $wgNoFollowLinks, $wgNoFollowNsExceptions, $wgNoFollowDomainExceptions;
1694                 $ns = $title ? $title->getNamespace() : false;
1695                 if ( $wgNoFollowLinks && !in_array( $ns, $wgNoFollowNsExceptions ) &&
1696                                 !wfMatchesDomainList( $url, $wgNoFollowDomainExceptions ) )
1697                 {
1698                         return 'nofollow';
1699                 }
1700                 return null;
1701         }
1702         /**
1703          * Get an associative array of additional HTML attributes appropriate for a
1704          * particular external link.  This currently may include rel => nofollow
1705          * (depending on configuration, namespace, and the URL's domain) and/or a
1706          * target attribute (depending on configuration).
1707          *
1708          * @param string|bool $url optional URL, to extract the domain from for rel =>
1709          *   nofollow if appropriate
1710          * @return Array associative array of HTML attributes
1711          */
1712         function getExternalLinkAttribs( $url = false ) {
1713                 $attribs = array();
1714                 $attribs['rel'] = self::getExternalLinkRel( $url, $this->mTitle );
1715
1716                 if ( $this->mOptions->getExternalLinkTarget() ) {
1717                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1718                 }
1719                 return $attribs;
1720         }
1721
1722         /**
1723          * Replace unusual URL escape codes with their equivalent characters
1724          *
1725          * @param $url String
1726          * @return String
1727          *
1728          * @todo  This can merge genuinely required bits in the path or query string,
1729          *        breaking legit URLs. A proper fix would treat the various parts of
1730          *        the URL differently; as a workaround, just use the output for
1731          *        statistical records, not for actual linking/output.
1732          */
1733         static function replaceUnusualEscapes( $url ) {
1734                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1735                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1736         }
1737
1738         /**
1739          * Callback function used in replaceUnusualEscapes().
1740          * Replaces unusual URL escape codes with their equivalent character
1741          *
1742          * @param $matches array
1743          *
1744          * @return string
1745          */
1746         private static function replaceUnusualEscapesCallback( $matches ) {
1747                 $char = urldecode( $matches[0] );
1748                 $ord = ord( $char );
1749                 # Is it an unsafe or HTTP reserved character according to RFC 1738?
1750                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1751                         # No, shouldn't be escaped
1752                         return $char;
1753                 } else {
1754                         # Yes, leave it escaped
1755                         return $matches[0];
1756                 }
1757         }
1758
1759         /**
1760          * make an image if it's allowed, either through the global
1761          * option, through the exception, or through the on-wiki whitelist
1762          * @private
1763          *
1764          * $param $url string
1765          *
1766          * @return string
1767          */
1768         function maybeMakeExternalImage( $url ) {
1769                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1770                 $imagesexception = !empty( $imagesfrom );
1771                 $text = false;
1772                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1773                 if ( $imagesexception && is_array( $imagesfrom ) ) {
1774                         $imagematch = false;
1775                         foreach ( $imagesfrom as $match ) {
1776                                 if ( strpos( $url, $match ) === 0 ) {
1777                                         $imagematch = true;
1778                                         break;
1779                                 }
1780                         }
1781                 } elseif ( $imagesexception ) {
1782                         $imagematch = ( strpos( $url, $imagesfrom ) === 0 );
1783                 } else {
1784                         $imagematch = false;
1785                 }
1786                 if ( $this->mOptions->getAllowExternalImages()
1787                          || ( $imagesexception && $imagematch ) ) {
1788                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1789                                 # Image found
1790                                 $text = Linker::makeExternalImage( $url );
1791                         }
1792                 }
1793                 if ( !$text && $this->mOptions->getEnableImageWhitelist()
1794                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1795                         $whitelist = explode( "\n", wfMessage( 'external_image_whitelist' )->inContentLanguage()->text() );
1796                         foreach ( $whitelist as $entry ) {
1797                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1798                                 if ( strpos( $entry, '#' ) === 0 || $entry === '' ) {
1799                                         continue;
1800                                 }
1801                                 if ( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1802                                         # Image matches a whitelist entry
1803                                         $text = Linker::makeExternalImage( $url );
1804                                         break;
1805                                 }
1806                         }
1807                 }
1808                 return $text;
1809         }
1810
1811         /**
1812          * Process [[ ]] wikilinks
1813          *
1814          * @param $s string
1815          *
1816          * @return String: processed text
1817          *
1818          * @private
1819          */
1820         function replaceInternalLinks( $s ) {
1821                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1822                 return $s;
1823         }
1824
1825         /**
1826          * Process [[ ]] wikilinks (RIL)
1827          * @param $s
1828          * @throws MWException
1829          * @return LinkHolderArray
1830          *
1831          * @private
1832          */
1833         function replaceInternalLinks2( &$s ) {
1834                 wfProfileIn( __METHOD__ );
1835
1836                 wfProfileIn( __METHOD__ . '-setup' );
1837                 static $tc = false, $e1, $e1_img;
1838                 # the % is needed to support urlencoded titles as well
1839                 if ( !$tc ) {
1840                         $tc = Title::legalChars() . '#%';
1841                         # Match a link having the form [[namespace:link|alternate]]trail
1842                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1843                         # Match cases where there is no "]]", which might still be images
1844                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1845                 }
1846
1847                 $holders = new LinkHolderArray( $this );
1848
1849                 # split the entire text string on occurrences of [[
1850                 $a = StringUtils::explode( '[[', ' ' . $s );
1851                 # get the first element (all text up to first [[), and remove the space we added
1852                 $s = $a->current();
1853                 $a->next();
1854                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1855                 $s = substr( $s, 1 );
1856
1857                 $useLinkPrefixExtension = $this->getTargetLanguage()->linkPrefixExtension();
1858                 $e2 = null;
1859                 if ( $useLinkPrefixExtension ) {
1860                         # Match the end of a line for a word that's not followed by whitespace,
1861                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1862                         $e2 = wfMessage( 'linkprefix' )->inContentLanguage()->text();
1863                 }
1864
1865                 if ( is_null( $this->mTitle ) ) {
1866                         wfProfileOut( __METHOD__ . '-setup' );
1867                         wfProfileOut( __METHOD__ );
1868                         throw new MWException( __METHOD__ . ": \$this->mTitle is null\n" );
1869                 }
1870                 $nottalk = !$this->mTitle->isTalkPage();
1871
1872                 if ( $useLinkPrefixExtension ) {
1873                         $m = array();
1874                         if ( preg_match( $e2, $s, $m ) ) {
1875                                 $first_prefix = $m[2];
1876                         } else {
1877                                 $first_prefix = false;
1878                         }
1879                 } else {
1880                         $prefix = '';
1881                 }
1882
1883                 $useSubpages = $this->areSubpagesAllowed();
1884                 wfProfileOut( __METHOD__ . '-setup' );
1885
1886                 # Loop for each link
1887                 for ( ; $line !== false && $line !== null; $a->next(), $line = $a->current() ) {
1888                         # Check for excessive memory usage
1889                         if ( $holders->isBig() ) {
1890                                 # Too big
1891                                 # Do the existence check, replace the link holders and clear the array
1892                                 $holders->replace( $s );
1893                                 $holders->clear();
1894                         }
1895
1896                         if ( $useLinkPrefixExtension ) {
1897                                 wfProfileIn( __METHOD__ . '-prefixhandling' );
1898                                 if ( preg_match( $e2, $s, $m ) ) {
1899                                         $prefix = $m[2];
1900                                         $s = $m[1];
1901                                 } else {
1902                                         $prefix = '';
1903                                 }
1904                                 # first link
1905                                 if ( $first_prefix ) {
1906                                         $prefix = $first_prefix;
1907                                         $first_prefix = false;
1908                                 }
1909                                 wfProfileOut( __METHOD__ . '-prefixhandling' );
1910                         }
1911
1912                         $might_be_img = false;
1913
1914                         wfProfileIn( __METHOD__ . "-e1" );
1915                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1916                                 $text = $m[2];
1917                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1918                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1919                                 # the real problem is with the $e1 regex
1920                                 # See bug 1300.
1921                                 #
1922                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1923                                 # and no image is in sight. See bug 2095.
1924                                 #
1925                                 if ( $text !== '' &&
1926                                         substr( $m[3], 0, 1 ) === ']' &&
1927                                         strpos( $text, '[' ) !== false
1928                                 )
1929                                 {
1930                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1931                                         $m[3] = substr( $m[3], 1 );
1932                                 }
1933                                 # fix up urlencoded title texts
1934                                 if ( strpos( $m[1], '%' ) !== false ) {
1935                                         # Should anchors '#' also be rejected?
1936                                         $m[1] = str_replace( array( '<', '>' ), array( '&lt;', '&gt;' ), rawurldecode( $m[1] ) );
1937                                 }
1938                                 $trail = $m[3];
1939                         } elseif ( preg_match( $e1_img, $line, $m ) ) { # Invalid, but might be an image with a link in its caption
1940                                 $might_be_img = true;
1941                                 $text = $m[2];
1942                                 if ( strpos( $m[1], '%' ) !== false ) {
1943                                         $m[1] = rawurldecode( $m[1] );
1944                                 }
1945                                 $trail = "";
1946                         } else { # Invalid form; output directly
1947                                 $s .= $prefix . '[[' . $line;
1948                                 wfProfileOut( __METHOD__ . "-e1" );
1949                                 continue;
1950                         }
1951                         wfProfileOut( __METHOD__ . "-e1" );
1952                         wfProfileIn( __METHOD__ . "-misc" );
1953
1954                         # Don't allow internal links to pages containing
1955                         # PROTO: where PROTO is a valid URL protocol; these
1956                         # should be external links.
1957                         if ( preg_match( '/^(?i:' . $this->mUrlProtocols . ')/', $m[1] ) ) {
1958                                 $s .= $prefix . '[[' . $line;
1959                                 wfProfileOut( __METHOD__ . "-misc" );
1960                                 continue;
1961                         }
1962
1963                         # Make subpage if necessary
1964                         if ( $useSubpages ) {
1965                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1966                         } else {
1967                                 $link = $m[1];
1968                         }
1969
1970                         $noforce = ( substr( $m[1], 0, 1 ) !== ':' );
1971                         if ( !$noforce ) {
1972                                 # Strip off leading ':'
1973                                 $link = substr( $link, 1 );
1974                         }
1975
1976                         wfProfileOut( __METHOD__ . "-misc" );
1977                         wfProfileIn( __METHOD__ . "-title" );
1978                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1979                         if ( $nt === null ) {
1980                                 $s .= $prefix . '[[' . $line;
1981                                 wfProfileOut( __METHOD__ . "-title" );
1982                                 continue;
1983                         }
1984
1985                         $ns = $nt->getNamespace();
1986                         $iw = $nt->getInterWiki();
1987                         wfProfileOut( __METHOD__ . "-title" );
1988
1989                         if ( $might_be_img ) { # if this is actually an invalid link
1990                                 wfProfileIn( __METHOD__ . "-might_be_img" );
1991                                 if ( $ns == NS_FILE && $noforce ) { # but might be an image
1992                                         $found = false;
1993                                         while ( true ) {
1994                                                 # look at the next 'line' to see if we can close it there
1995                                                 $a->next();
1996                                                 $next_line = $a->current();
1997                                                 if ( $next_line === false || $next_line === null ) {
1998                                                         break;
1999                                                 }
2000                                                 $m = explode( ']]', $next_line, 3 );
2001                                                 if ( count( $m ) == 3 ) {
2002                                                         # the first ]] closes the inner link, the second the image
2003                                                         $found = true;
2004                                                         $text .= "[[{$m[0]}]]{$m[1]}";
2005                                                         $trail = $m[2];
2006                                                         break;
2007                                                 } elseif ( count( $m ) == 2 ) {
2008                                                         # if there's exactly one ]] that's fine, we'll keep looking
2009                                                         $text .= "[[{$m[0]}]]{$m[1]}";
2010                                                 } else {
2011                                                         # if $next_line is invalid too, we need look no further
2012                                                         $text .= '[[' . $next_line;
2013                                                         break;
2014                                                 }
2015                                         }
2016                                         if ( !$found ) {
2017                                                 # we couldn't find the end of this imageLink, so output it raw
2018                                                 # but don't ignore what might be perfectly normal links in the text we've examined
2019                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
2020                                                 $s .= "{$prefix}[[$link|$text";
2021                                                 # note: no $trail, because without an end, there *is* no trail
2022                                                 wfProfileOut( __METHOD__ . "-might_be_img" );
2023                                                 continue;
2024                                         }
2025                                 } else { # it's not an image, so output it raw
2026                                         $s .= "{$prefix}[[$link|$text";
2027                                         # note: no $trail, because without an end, there *is* no trail
2028                                         wfProfileOut( __METHOD__ . "-might_be_img" );
2029                                         continue;
2030                                 }
2031                                 wfProfileOut( __METHOD__ . "-might_be_img" );
2032                         }
2033
2034                         $wasblank = ( $text == '' );
2035                         if ( $wasblank ) {
2036                                 $text = $link;
2037                         } else {
2038                                 # Bug 4598 madness. Handle the quotes only if they come from the alternate part
2039                                 # [[Lista d''e paise d''o munno]] -> <a href="...">Lista d''e paise d''o munno</a>
2040                                 # [[Criticism of Harry Potter|Criticism of ''Harry Potter'']]
2041                                 #    -> <a href="Criticism of Harry Potter">Criticism of <i>Harry Potter</i></a>
2042                                 $text = $this->doQuotes( $text );
2043                         }
2044
2045                         # Link not escaped by : , create the various objects
2046                         if ( $noforce ) {
2047                                 # Interwikis
2048                                 wfProfileIn( __METHOD__ . "-interwiki" );
2049                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && Language::fetchLanguageName( $iw, null, 'mw' ) ) {
2050                                         // XXX: the above check prevents links to sites with identifiers that are not language codes
2051
2052                                         # Bug 24502: filter duplicates
2053                                         if ( !isset( $this->mLangLinkLanguages[$iw] ) ) {
2054                                                 $this->mLangLinkLanguages[$iw] = true;
2055                                                 $this->mOutput->addLanguageLink( $nt->getFullText() );
2056                                         }
2057
2058                                         $s = rtrim( $s . $prefix );
2059                                         $s .= trim( $trail, "\n" ) == '' ? '': $prefix . $trail;
2060                                         wfProfileOut( __METHOD__ . "-interwiki" );
2061                                         continue;
2062                                 }
2063                                 wfProfileOut( __METHOD__ . "-interwiki" );
2064
2065                                 if ( $ns == NS_FILE ) {
2066                                         wfProfileIn( __METHOD__ . "-image" );
2067                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
2068                                                 if ( $wasblank ) {
2069                                                         # if no parameters were passed, $text
2070                                                         # becomes something like "File:Foo.png",
2071                                                         # which we don't want to pass on to the
2072                                                         # image generator
2073                                                         $text = '';
2074                                                 } else {
2075                                                         # recursively parse links inside the image caption
2076                                                         # actually, this will parse them in any other parameters, too,
2077                                                         # but it might be hard to fix that, and it doesn't matter ATM
2078                                                         $text = $this->replaceExternalLinks( $text );
2079                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
2080                                                 }
2081                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
2082                                                 $s .= $prefix . $this->armorLinks(
2083                                                         $this->makeImage( $nt, $text, $holders ) ) . $trail;
2084                                         } else {
2085                                                 $s .= $prefix . $trail;
2086                                         }
2087                                         wfProfileOut( __METHOD__ . "-image" );
2088                                         continue;
2089                                 }
2090
2091                                 if ( $ns == NS_CATEGORY ) {
2092                                         wfProfileIn( __METHOD__ . "-category" );
2093                                         $s = rtrim( $s . "\n" ); # bug 87
2094
2095                                         if ( $wasblank ) {
2096                                                 $sortkey = $this->getDefaultSort();
2097                                         } else {
2098                                                 $sortkey = $text;
2099                                         }
2100                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
2101                                         $sortkey = str_replace( "\n", '', $sortkey );
2102                                         $sortkey = $this->getConverterLanguage()->convertCategoryKey( $sortkey );
2103                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
2104
2105                                         /**
2106                                          * Strip the whitespace Category links produce, see bug 87
2107                                          * @todo We might want to use trim($tmp, "\n") here.
2108                                          */
2109                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
2110
2111                                         wfProfileOut( __METHOD__ . "-category" );
2112                                         continue;
2113                                 }
2114                         }
2115
2116                         # Self-link checking
2117                         if ( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
2118                                 if ( $nt->equals( $this->mTitle ) || ( !$nt->isKnown() && in_array(
2119                                         $this->mTitle->getPrefixedText(),
2120                                         $this->getConverterLanguage()->autoConvertToAllVariants( $nt->getPrefixedText() ),
2121                                         true
2122                                 ) ) ) {
2123                                         $s .= $prefix . Linker::makeSelfLinkObj( $nt, $text, '', $trail );
2124                                         continue;
2125                                 }
2126                         }
2127
2128                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
2129                         # @todo FIXME: Should do batch file existence checks, see comment below
2130                         if ( $ns == NS_MEDIA ) {
2131                                 wfProfileIn( __METHOD__ . "-media" );
2132                                 # Give extensions a chance to select the file revision for us
2133                                 $options = array();
2134                                 $descQuery = false;
2135                                 wfRunHooks( 'BeforeParserFetchFileAndTitle',
2136                                         array( $this, $nt, &$options, &$descQuery ) );
2137                                 # Fetch and register the file (file title may be different via hooks)
2138                                 list( $file, $nt ) = $this->fetchFileAndTitle( $nt, $options );
2139                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
2140                                 $s .= $prefix . $this->armorLinks(
2141                                         Linker::makeMediaLinkFile( $nt, $file, $text ) ) . $trail;
2142                                 wfProfileOut( __METHOD__ . "-media" );
2143                                 continue;
2144                         }
2145
2146                         wfProfileIn( __METHOD__ . "-always_known" );
2147                         # Some titles, such as valid special pages or files in foreign repos, should
2148                         # be shown as bluelinks even though they're not included in the page table
2149                         #
2150                         # @todo FIXME: isAlwaysKnown() can be expensive for file links; we should really do
2151                         # batch file existence checks for NS_FILE and NS_MEDIA
2152                         if ( $iw == '' && $nt->isAlwaysKnown() ) {
2153                                 $this->mOutput->addLink( $nt );
2154                                 $s .= $this->makeKnownLinkHolder( $nt, $text, array(), $trail, $prefix );
2155                         } else {
2156                                 # Links will be added to the output link list after checking
2157                                 $s .= $holders->makeHolder( $nt, $text, array(), $trail, $prefix );
2158                         }
2159                         wfProfileOut( __METHOD__ . "-always_known" );
2160                 }
2161                 wfProfileOut( __METHOD__ );
2162                 return $holders;
2163         }
2164
2165         /**
2166          * Render a forced-blue link inline; protect against double expansion of
2167          * URLs if we're in a mode that prepends full URL prefixes to internal links.
2168          * Since this little disaster has to split off the trail text to avoid
2169          * breaking URLs in the following text without breaking trails on the
2170          * wiki links, it's been made into a horrible function.
2171          *
2172          * @param $nt Title
2173          * @param $text String
2174          * @param array $query or String
2175          * @param $trail String
2176          * @param $prefix String
2177          * @return String: HTML-wikitext mix oh yuck
2178          */
2179         function makeKnownLinkHolder( $nt, $text = '', $query = array(), $trail = '', $prefix = '' ) {
2180                 list( $inside, $trail ) = Linker::splitTrail( $trail );
2181
2182                 if ( is_string( $query ) ) {
2183                         $query = wfCgiToArray( $query );
2184                 }
2185                 if ( $text == '' ) {
2186                         $text = htmlspecialchars( $nt->getPrefixedText() );
2187                 }
2188
2189                 $link = Linker::linkKnown( $nt, "$prefix$text$inside", array(), $query );
2190
2191                 return $this->armorLinks( $link ) . $trail;
2192         }
2193
2194         /**
2195          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
2196          * going to go through further parsing steps before inline URL expansion.
2197          *
2198          * Not needed quite as much as it used to be since free links are a bit
2199          * more sensible these days. But bracketed links are still an issue.
2200          *
2201          * @param string $text more-or-less HTML
2202          * @return String: less-or-more HTML with NOPARSE bits
2203          */
2204         function armorLinks( $text ) {
2205                 return preg_replace( '/\b((?i)' . $this->mUrlProtocols . ')/',
2206                         "{$this->mUniqPrefix}NOPARSE$1", $text );
2207         }
2208
2209         /**
2210          * Return true if subpage links should be expanded on this page.
2211          * @return Boolean
2212          */
2213         function areSubpagesAllowed() {
2214                 # Some namespaces don't allow subpages
2215                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
2216         }
2217
2218         /**
2219          * Handle link to subpage if necessary
2220          *
2221          * @param string $target the source of the link
2222          * @param &$text String: the link text, modified as necessary
2223          * @return string the full name of the link
2224          * @private
2225          */
2226         function maybeDoSubpageLink( $target, &$text ) {
2227                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
2228         }
2229
2230         /**#@+
2231          * Used by doBlockLevels()
2232          * @private
2233          *
2234          * @return string
2235          */
2236         function closeParagraph() {
2237                 $result = '';
2238                 if ( $this->mLastSection != '' ) {
2239                         $result = '</' . $this->mLastSection . ">\n";
2240                 }
2241                 $this->mInPre = false;
2242                 $this->mLastSection = '';
2243                 return $result;
2244         }
2245
2246         /**
2247          * getCommon() returns the length of the longest common substring
2248          * of both arguments, starting at the beginning of both.
2249          * @private
2250          *
2251          * @param $st1 string
2252          * @param $st2 string
2253          *
2254          * @return int
2255          */
2256         function getCommon( $st1, $st2 ) {
2257                 $fl = strlen( $st1 );
2258                 $shorter = strlen( $st2 );
2259                 if ( $fl < $shorter ) {
2260                         $shorter = $fl;
2261                 }
2262
2263                 for ( $i = 0; $i < $shorter; ++$i ) {
2264                         if ( $st1[$i] != $st2[$i] ) {
2265                                 break;
2266                         }
2267                 }
2268                 return $i;
2269         }
2270
2271         /**
2272          * These next three functions open, continue, and close the list
2273          * element appropriate to the prefix character passed into them.
2274          * @private
2275          *
2276          * @param $char string
2277          *
2278          * @return string
2279          */
2280         function openList( $char ) {
2281                 $result = $this->closeParagraph();
2282
2283                 if ( '*' === $char ) {
2284                         $result .= "<ul>\n<li>";
2285                 } elseif ( '#' === $char ) {
2286                         $result .= "<ol>\n<li>";
2287                 } elseif ( ':' === $char ) {
2288                         $result .= "<dl>\n<dd>";
2289                 } elseif ( ';' === $char ) {
2290                         $result .= "<dl>\n<dt>";
2291                         $this->mDTopen = true;
2292                 } else {
2293                         $result = '<!-- ERR 1 -->';
2294                 }
2295
2296                 return $result;
2297         }
2298
2299         /**
2300          * TODO: document
2301          * @param $char String
2302          * @private
2303          *
2304          * @return string
2305          */
2306         function nextItem( $char ) {
2307                 if ( '*' === $char || '#' === $char ) {
2308                         return "</li>\n<li>";
2309                 } elseif ( ':' === $char || ';' === $char ) {
2310                         $close = "</dd>\n";
2311                         if ( $this->mDTopen ) {
2312                                 $close = "</dt>\n";
2313                         }
2314                         if ( ';' === $char ) {
2315                                 $this->mDTopen = true;
2316                                 return $close . '<dt>';
2317                         } else {
2318                                 $this->mDTopen = false;
2319                                 return $close . '<dd>';
2320                         }
2321                 }
2322                 return '<!-- ERR 2 -->';
2323         }
2324
2325         /**
2326          * TODO: document
2327          * @param $char String
2328          * @private
2329          *
2330          * @return string
2331          */
2332         function closeList( $char ) {
2333                 if ( '*' === $char ) {
2334                         $text = "</li>\n</ul>";
2335                 } elseif ( '#' === $char ) {
2336                         $text = "</li>\n</ol>";
2337                 } elseif ( ':' === $char ) {
2338                         if ( $this->mDTopen ) {
2339                                 $this->mDTopen = false;
2340                                 $text = "</dt>\n</dl>";
2341                         } else {
2342                                 $text = "</dd>\n</dl>";
2343                         }
2344                 } else {
2345                         return '<!-- ERR 3 -->';
2346                 }
2347                 return $text . "\n";
2348         }
2349         /**#@-*/
2350
2351         /**
2352          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2353          *
2354          * @param $text String
2355          * @param $linestart Boolean: whether or not this is at the start of a line.
2356          * @private
2357          * @return string the lists rendered as HTML
2358          */
2359         function doBlockLevels( $text, $linestart ) {
2360                 wfProfileIn( __METHOD__ );
2361
2362                 # Parsing through the text line by line.  The main thing
2363                 # happening here is handling of block-level elements p, pre,
2364                 # and making lists from lines starting with * # : etc.
2365                 #
2366                 $textLines = StringUtils::explode( "\n", $text );
2367
2368                 $lastPrefix = $output = '';
2369                 $this->mDTopen = $inBlockElem = false;
2370                 $prefixLength = 0;
2371                 $paragraphStack = false;
2372                 $inBlockquote = false;
2373
2374                 foreach ( $textLines as $oLine ) {
2375                         # Fix up $linestart
2376                         if ( !$linestart ) {
2377                                 $output .= $oLine;
2378                                 $linestart = true;
2379                                 continue;
2380                         }
2381                         # * = ul
2382                         # # = ol
2383                         # ; = dt
2384                         # : = dd
2385
2386                         $lastPrefixLength = strlen( $lastPrefix );
2387                         $preCloseMatch = preg_match( '/<\\/pre/i', $oLine );
2388                         $preOpenMatch = preg_match( '/<pre/i', $oLine );
2389                         # If not in a <pre> element, scan for and figure out what prefixes are there.
2390                         if ( !$this->mInPre ) {
2391                                 # Multiple prefixes may abut each other for nested lists.
2392                                 $prefixLength = strspn( $oLine, '*#:;' );
2393                                 $prefix = substr( $oLine, 0, $prefixLength );
2394
2395                                 # eh?
2396                                 # ; and : are both from definition-lists, so they're equivalent
2397                                 #  for the purposes of determining whether or not we need to open/close
2398                                 #  elements.
2399                                 $prefix2 = str_replace( ';', ':', $prefix );
2400                                 $t = substr( $oLine, $prefixLength );
2401                                 $this->mInPre = (bool)$preOpenMatch;
2402                         } else {
2403                                 # Don't interpret any other prefixes in preformatted text
2404                                 $prefixLength = 0;
2405                                 $prefix = $prefix2 = '';
2406                                 $t = $oLine;
2407                         }
2408
2409                         # List generation
2410                         if ( $prefixLength && $lastPrefix === $prefix2 ) {
2411                                 # Same as the last item, so no need to deal with nesting or opening stuff
2412                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2413                                 $paragraphStack = false;
2414
2415                                 if ( substr( $prefix, -1 ) === ';' ) {
2416                                         # The one nasty exception: definition lists work like this:
2417                                         # ; title : definition text
2418                                         # So we check for : in the remainder text to split up the
2419                                         # title and definition, without b0rking links.
2420                                         $term = $t2 = '';
2421                                         if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2422                                                 $t = $t2;
2423                                                 $output .= $term . $this->nextItem( ':' );
2424                                         }
2425                                 }
2426                         } elseif ( $prefixLength || $lastPrefixLength ) {
2427                                 # We need to open or close prefixes, or both.
2428
2429                                 # Either open or close a level...
2430                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2431                                 $paragraphStack = false;
2432
2433                                 # Close all the prefixes which aren't shared.
2434                                 while ( $commonPrefixLength < $lastPrefixLength ) {
2435                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength - 1] );
2436                                         --$lastPrefixLength;
2437                                 }
2438
2439                                 # Continue the current prefix if appropriate.
2440                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2441                                         $output .= $this->nextItem( $prefix[$commonPrefixLength - 1] );
2442                                 }
2443
2444                                 # Open prefixes where appropriate.
2445                                 while ( $prefixLength > $commonPrefixLength ) {
2446                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2447                                         $output .= $this->openList( $char );
2448
2449                                         if ( ';' === $char ) {
2450                                                 # @todo FIXME: This is dupe of code above
2451                                                 if ( $this->findColonNoLinks( $t, $term, $t2 ) !== false ) {
2452                                                         $t = $t2;
2453                                                         $output .= $term . $this->nextItem( ':' );
2454                                                 }
2455                                         }
2456                                         ++$commonPrefixLength;
2457                                 }
2458                                 $lastPrefix = $prefix2;
2459                         }
2460
2461                         # If we have no prefixes, go to paragraph mode.
2462                         if ( 0 == $prefixLength ) {
2463                                 wfProfileIn( __METHOD__ . "-paragraph" );
2464                                 # No prefix (not in list)--go to paragraph mode
2465                                 # XXX: use a stack for nestable elements like span, table and div
2466                                 $openmatch = preg_match( '/(?:<table|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<dl|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2467                                 $closematch = preg_match(
2468                                         '/(?:<\\/table|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|' .
2469                                         '<td|<th|<\\/?blockquote|<\\/?div|<hr|<\\/pre|<\\/p|' . $this->mUniqPrefix . '-pre|<\\/li|<\\/ul|<\\/ol|<\\/dl|<\\/?center)/iS', $t );
2470                                 if ( $openmatch or $closematch ) {
2471                                         $paragraphStack = false;
2472                                         # TODO bug 5718: paragraph closed
2473                                         $output .= $this->closeParagraph();
2474                                         if ( $preOpenMatch and !$preCloseMatch ) {
2475                                                 $this->mInPre = true;
2476                                         }
2477                                         $bqOffset = 0;
2478                                         while ( preg_match( '/<(\\/?)blockquote[\s>]/i', $t, $bqMatch, PREG_OFFSET_CAPTURE, $bqOffset ) ) {
2479                                                 $inBlockquote = !$bqMatch[1][0]; // is this a close tag?
2480                                                 $bqOffset = $bqMatch[0][1] + strlen( $bqMatch[0][0] );
2481                                         }
2482                                         $inBlockElem = !$closematch;
2483                                 } elseif ( !$inBlockElem && !$this->mInPre ) {
2484                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' || trim( $t ) != '' ) and !$inBlockquote ) {
2485                                                 # pre
2486                                                 if ( $this->mLastSection !== 'pre' ) {
2487                                                         $paragraphStack = false;
2488                                                         $output .= $this->closeParagraph() . '<pre>';
2489                                                         $this->mLastSection = 'pre';
2490                                                 }
2491                                                 $t = substr( $t, 1 );
2492                                         } else {
2493                                                 # paragraph
2494                                                 if ( trim( $t ) === '' ) {
2495                                                         if ( $paragraphStack ) {
2496                                                                 $output .= $paragraphStack . '<br />';
2497                                                                 $paragraphStack = false;
2498                                                                 $this->mLastSection = 'p';
2499                                                         } else {
2500                                                                 if ( $this->mLastSection !== 'p' ) {
2501                                                                         $output .= $this->closeParagraph();
2502                                                                         $this->mLastSection = '';
2503                                                                         $paragraphStack = '<p>';
2504                                                                 } else {
2505                                                                         $paragraphStack = '</p><p>';
2506                                                                 }
2507                                                         }
2508                                                 } else {
2509                                                         if ( $paragraphStack ) {
2510                                                                 $output .= $paragraphStack;
2511                                                                 $paragraphStack = false;
2512                                                                 $this->mLastSection = 'p';
2513                                                         } elseif ( $this->mLastSection !== 'p' ) {
2514                                                                 $output .= $this->closeParagraph() . '<p>';
2515                                                                 $this->mLastSection = 'p';
2516                                                         }
2517                                                 }
2518                                         }
2519                                 }
2520                                 wfProfileOut( __METHOD__ . "-paragraph" );
2521                         }
2522                         # somewhere above we forget to get out of pre block (bug 785)
2523                         if ( $preCloseMatch && $this->mInPre ) {
2524                                 $this->mInPre = false;
2525                         }
2526                         if ( $paragraphStack === false ) {
2527                                 $output .= $t . "\n";
2528                         }
2529                 }
2530                 while ( $prefixLength ) {
2531                         $output .= $this->closeList( $prefix2[$prefixLength - 1] );
2532                         --$prefixLength;
2533                 }
2534                 if ( $this->mLastSection != '' ) {
2535                         $output .= '</' . $this->mLastSection . '>';
2536                         $this->mLastSection = '';
2537                 }
2538
2539                 wfProfileOut( __METHOD__ );
2540                 return $output;
2541         }
2542
2543         /**
2544          * Split up a string on ':', ignoring any occurrences inside tags
2545          * to prevent illegal overlapping.
2546          *
2547          * @param string $str the string to split
2548          * @param &$before String set to everything before the ':'
2549          * @param &$after String set to everything after the ':'
2550          * @throws MWException
2551          * @return String the position of the ':', or false if none found
2552          */
2553         function findColonNoLinks( $str, &$before, &$after ) {
2554                 wfProfileIn( __METHOD__ );
2555
2556                 $pos = strpos( $str, ':' );
2557                 if ( $pos === false ) {
2558                         # Nothing to find!
2559                         wfProfileOut( __METHOD__ );
2560                         return false;
2561                 }
2562
2563                 $lt = strpos( $str, '<' );
2564                 if ( $lt === false || $lt > $pos ) {
2565                         # Easy; no tag nesting to worry about
2566                         $before = substr( $str, 0, $pos );
2567                         $after = substr( $str, $pos + 1 );
2568                         wfProfileOut( __METHOD__ );
2569                         return $pos;
2570                 }
2571
2572                 # Ugly state machine to walk through avoiding tags.
2573                 $state = self::COLON_STATE_TEXT;
2574                 $stack = 0;
2575                 $len = strlen( $str );
2576                 for ( $i = 0; $i < $len; $i++ ) {
2577                         $c = $str[$i];
2578
2579                         switch ( $state ) {
2580                         # (Using the number is a performance hack for common cases)
2581                         case 0: # self::COLON_STATE_TEXT:
2582                                 switch ( $c ) {
2583                                 case "<":
2584                                         # Could be either a <start> tag or an </end> tag
2585                                         $state = self::COLON_STATE_TAGSTART;
2586                                         break;
2587                                 case ":":
2588                                         if ( $stack == 0 ) {
2589                                                 # We found it!
2590                                                 $before = substr( $str, 0, $i );
2591                                                 $after = substr( $str, $i + 1 );
2592                                                 wfProfileOut( __METHOD__ );
2593                                                 return $i;
2594                                         }
2595                                         # Embedded in a tag; don't break it.
2596                                         break;
2597                                 default:
2598                                         # Skip ahead looking for something interesting
2599                                         $colon = strpos( $str, ':', $i );
2600                                         if ( $colon === false ) {
2601                                                 # Nothing else interesting
2602                                                 wfProfileOut( __METHOD__ );
2603                                                 return false;
2604                                         }
2605                                         $lt = strpos( $str, '<', $i );
2606                                         if ( $stack === 0 ) {
2607                                                 if ( $lt === false || $colon < $lt ) {
2608                                                         # We found it!
2609                                                         $before = substr( $str, 0, $colon );
2610                                                         $after = substr( $str, $colon + 1 );
2611                                                         wfProfileOut( __METHOD__ );
2612                                                         return $i;
2613                                                 }
2614                                         }
2615                                         if ( $lt === false ) {
2616                                                 # Nothing else interesting to find; abort!
2617                                                 # We're nested, but there's no close tags left. Abort!
2618                                                 break 2;
2619                                         }
2620                                         # Skip ahead to next tag start
2621                                         $i = $lt;
2622                                         $state = self::COLON_STATE_TAGSTART;
2623                                 }
2624                                 break;
2625                         case 1: # self::COLON_STATE_TAG:
2626                                 # In a <tag>
2627                                 switch ( $c ) {
2628                                 case ">":
2629                                         $stack++;
2630                                         $state = self::COLON_STATE_TEXT;
2631                                         break;
2632                                 case "/":
2633                                         # Slash may be followed by >?
2634                                         $state = self::COLON_STATE_TAGSLASH;
2635                                         break;
2636                                 default:
2637                                         # ignore
2638                                 }
2639                                 break;
2640                         case 2: # self::COLON_STATE_TAGSTART:
2641                                 switch ( $c ) {
2642                                 case "/":
2643                                         $state = self::COLON_STATE_CLOSETAG;
2644                                         break;
2645                                 case "!":
2646                                         $state = self::COLON_STATE_COMMENT;
2647                                         break;
2648                                 case ">":
2649                                         # Illegal early close? This shouldn't happen D:
2650                                         $state = self::COLON_STATE_TEXT;
2651                                         break;
2652                                 default:
2653                                         $state = self::COLON_STATE_TAG;
2654                                 }
2655                                 break;
2656                         case 3: # self::COLON_STATE_CLOSETAG:
2657                                 # In a </tag>
2658                                 if ( $c === ">" ) {
2659                                         $stack--;
2660                                         if ( $stack < 0 ) {
2661                                                 wfDebug( __METHOD__ . ": Invalid input; too many close tags\n" );
2662                                                 wfProfileOut( __METHOD__ );
2663                                                 return false;
2664                                         }
2665                                         $state = self::COLON_STATE_TEXT;
2666                                 }
2667                                 break;
2668                         case self::COLON_STATE_TAGSLASH:
2669                                 if ( $c === ">" ) {
2670                                         # Yes, a self-closed tag <blah/>
2671                                         $state = self::COLON_STATE_TEXT;
2672                                 } else {
2673                                         # Probably we're jumping the gun, and this is an attribute
2674                                         $state = self::COLON_STATE_TAG;
2675                                 }
2676                                 break;
2677                         case 5: # self::COLON_STATE_COMMENT:
2678                                 if ( $c === "-" ) {
2679                                         $state = self::COLON_STATE_COMMENTDASH;
2680                                 }
2681                                 break;
2682                         case self::COLON_STATE_COMMENTDASH:
2683                                 if ( $c === "-" ) {
2684                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2685                                 } else {
2686                                         $state = self::COLON_STATE_COMMENT;
2687                                 }
2688                                 break;
2689                         case self::COLON_STATE_COMMENTDASHDASH:
2690                                 if ( $c === ">" ) {
2691                                         $state = self::COLON_STATE_TEXT;
2692                                 } else {
2693                                         $state = self::COLON_STATE_COMMENT;
2694                                 }
2695                                 break;
2696                         default:
2697                                 wfProfileOut( __METHOD__ );
2698                                 throw new MWException( "State machine error in " . __METHOD__ );
2699                         }
2700                 }
2701                 if ( $stack > 0 ) {
2702                         wfDebug( __METHOD__ . ": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2703                         wfProfileOut( __METHOD__ );
2704                         return false;
2705                 }
2706                 wfProfileOut( __METHOD__ );
2707                 return false;
2708         }
2709
2710         /**
2711          * Return value of a magic variable (like PAGENAME)
2712          *
2713          * @private
2714          *
2715          * @param $index integer
2716          * @param bool|\PPFrame $frame
2717          *
2718          * @throws MWException
2719          * @return string
2720          */
2721         function getVariableValue( $index, $frame = false ) {
2722                 global $wgContLang, $wgSitename, $wgServer;
2723                 global $wgArticlePath, $wgScriptPath, $wgStylePath;
2724
2725                 if ( is_null( $this->mTitle ) ) {
2726                         // If no title set, bad things are going to happen
2727                         // later. Title should always be set since this
2728                         // should only be called in the middle of a parse
2729                         // operation (but the unit-tests do funky stuff)
2730                         throw new MWException( __METHOD__ . ' Should only be '
2731                                 . ' called while parsing (no title set)' );
2732                 }
2733
2734                 /**
2735                  * Some of these require message or data lookups and can be
2736                  * expensive to check many times.
2737                  */
2738                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2739                         if ( isset( $this->mVarCache[$index] ) ) {
2740                                 return $this->mVarCache[$index];
2741                         }
2742                 }
2743
2744                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2745                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2746
2747                 $pageLang = $this->getFunctionLang();
2748
2749                 switch ( $index ) {
2750                         case 'currentmonth':
2751                                 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'm' ) );
2752                                 break;
2753                         case 'currentmonth1':
2754                                 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2755                                 break;
2756                         case 'currentmonthname':
2757                                 $value = $pageLang->getMonthName( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2758                                 break;
2759                         case 'currentmonthnamegen':
2760                                 $value = $pageLang->getMonthNameGen( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2761                                 break;
2762                         case 'currentmonthabbrev':
2763                                 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getInstance( $ts )->format( 'n' ) );
2764                                 break;
2765                         case 'currentday':
2766                                 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'j' ) );
2767                                 break;
2768                         case 'currentday2':
2769                                 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'd' ) );
2770                                 break;
2771                         case 'localmonth':
2772                                 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'm' ) );
2773                                 break;
2774                         case 'localmonth1':
2775                                 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2776                                 break;
2777                         case 'localmonthname':
2778                                 $value = $pageLang->getMonthName( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2779                                 break;
2780                         case 'localmonthnamegen':
2781                                 $value = $pageLang->getMonthNameGen( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2782                                 break;
2783                         case 'localmonthabbrev':
2784                                 $value = $pageLang->getMonthAbbreviation( MWTimestamp::getLocalInstance( $ts )->format( 'n' ) );
2785                                 break;
2786                         case 'localday':
2787                                 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'j' ) );
2788                                 break;
2789                         case 'localday2':
2790                                 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'd' ) );
2791                                 break;
2792                         case 'pagename':
2793                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2794                                 break;
2795                         case 'pagenamee':
2796                                 $value = wfEscapeWikiText( $this->mTitle->getPartialURL() );
2797                                 break;
2798                         case 'fullpagename':
2799                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2800                                 break;
2801                         case 'fullpagenamee':
2802                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedURL() );
2803                                 break;
2804                         case 'subpagename':
2805                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2806                                 break;
2807                         case 'subpagenamee':
2808                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageUrlForm() );
2809                                 break;
2810                         case 'rootpagename':
2811                                 $value = wfEscapeWikiText( $this->mTitle->getRootText() );
2812                                 break;
2813                         case 'rootpagenamee':
2814                                 $value = wfEscapeWikiText( wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getRootText() ) ) );
2815                                 break;
2816                         case 'basepagename':
2817                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2818                                 break;
2819                         case 'basepagenamee':
2820                                 $value = wfEscapeWikiText( wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) ) );
2821                                 break;
2822                         case 'talkpagename':
2823                                 if ( $this->mTitle->canTalk() ) {
2824                                         $talkPage = $this->mTitle->getTalkPage();
2825                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2826                                 } else {
2827                                         $value = '';
2828                                 }
2829                                 break;
2830                         case 'talkpagenamee':
2831                                 if ( $this->mTitle->canTalk() ) {
2832                                         $talkPage = $this->mTitle->getTalkPage();
2833                                         $value = wfEscapeWikiText( $talkPage->getPrefixedURL() );
2834                                 } else {
2835                                         $value = '';
2836                                 }
2837                                 break;
2838                         case 'subjectpagename':
2839                                 $subjPage = $this->mTitle->getSubjectPage();
2840                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2841                                 break;
2842                         case 'subjectpagenamee':
2843                                 $subjPage = $this->mTitle->getSubjectPage();
2844                                 $value = wfEscapeWikiText( $subjPage->getPrefixedURL() );
2845                                 break;
2846                         case 'pageid': // requested in bug 23427
2847                                 $pageid = $this->getTitle()->getArticleID();
2848                                 if ( $pageid == 0 ) {
2849                                         # 0 means the page doesn't exist in the database,
2850                                         # which means the user is previewing a new page.
2851                                         # The vary-revision flag must be set, because the magic word
2852                                         # will have a different value once the page is saved.
2853                                         $this->mOutput->setFlag( 'vary-revision' );
2854                                         wfDebug( __METHOD__ . ": {{PAGEID}} used in a new page, setting vary-revision...\n" );
2855                                 }
2856                                 $value = $pageid ? $pageid : null;
2857                                 break;
2858                         case 'revisionid':
2859                                 # Let the edit saving system know we should parse the page
2860                                 # *after* a revision ID has been assigned.
2861                                 $this->mOutput->setFlag( 'vary-revision' );
2862                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2863                                 $value = $this->mRevisionId;
2864                                 break;
2865                         case 'revisionday':
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__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2870                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2871                                 break;
2872                         case 'revisionday2':
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__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2877                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2878                                 break;
2879                         case 'revisionmonth':
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__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2884                                 $value = substr( $this->getRevisionTimestamp(), 4, 2 );
2885                                 break;
2886                         case 'revisionmonth1':
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__ . ": {{REVISIONMONTH1}} used, setting vary-revision...\n" );
2891                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2892                                 break;
2893                         case 'revisionyear':
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__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2898                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2899                                 break;
2900                         case 'revisiontimestamp':
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__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2905                                 $value = $this->getRevisionTimestamp();
2906                                 break;
2907                         case 'revisionuser':
2908                                 # Let the edit saving system know we should parse the page
2909                                 # *after* a revision ID has been assigned. This is for null edits.
2910                                 $this->mOutput->setFlag( 'vary-revision' );
2911                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2912                                 $value = $this->getRevisionUser();
2913                                 break;
2914                         case 'revisionsize':
2915                                 # Let the edit saving system know we should parse the page
2916                                 # *after* a revision ID has been assigned. This is for null edits.
2917                                 $this->mOutput->setFlag( 'vary-revision' );
2918                                 wfDebug( __METHOD__ . ": {{REVISIONSIZE}} used, setting vary-revision...\n" );
2919                                 $value = $this->getRevisionSize();
2920                                 break;
2921                         case 'namespace':
2922                                 $value = str_replace( '_', ' ', $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2923                                 break;
2924                         case 'namespacee':
2925                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2926                                 break;
2927                         case 'namespacenumber':
2928                                 $value = $this->mTitle->getNamespace();
2929                                 break;
2930                         case 'talkspace':
2931                                 $value = $this->mTitle->canTalk() ? str_replace( '_', ' ', $this->mTitle->getTalkNsText() ) : '';
2932                                 break;
2933                         case 'talkspacee':
2934                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2935                                 break;
2936                         case 'subjectspace':
2937                                 $value = str_replace( '_', ' ', $this->mTitle->getSubjectNsText() );
2938                                 break;
2939                         case 'subjectspacee':
2940                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2941                                 break;
2942                         case 'currentdayname':
2943                                 $value = $pageLang->getWeekdayName( MWTimestamp::getInstance( $ts )->format( 'w' ) + 1 );
2944                                 break;
2945                         case 'currentyear':
2946                                 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'Y' ), true );
2947                                 break;
2948                         case 'currenttime':
2949                                 $value = $pageLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2950                                 break;
2951                         case 'currenthour':
2952                                 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'H' ), true );
2953                                 break;
2954                         case 'currentweek':
2955                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2956                                 # int to remove the padding
2957                                 $value = $pageLang->formatNum( (int)MWTimestamp::getInstance( $ts )->format( 'W' ) );
2958                                 break;
2959                         case 'currentdow':
2960                                 $value = $pageLang->formatNum( MWTimestamp::getInstance( $ts )->format( 'w' ) );
2961                                 break;
2962                         case 'localdayname':
2963                                 $value = $pageLang->getWeekdayName( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) + 1 );
2964                                 break;
2965                         case 'localyear':
2966                                 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'Y' ), true );
2967                                 break;
2968                         case 'localtime':
2969                                 $value = $pageLang->time( MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' ), false, false );
2970                                 break;
2971                         case 'localhour':
2972                                 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'H' ), true );
2973                                 break;
2974                         case 'localweek':
2975                                 # @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2976                                 # int to remove the padding
2977                                 $value = $pageLang->formatNum( (int)MWTimestamp::getLocalInstance( $ts )->format( 'W' ) );
2978                                 break;
2979                         case 'localdow':
2980                                 $value = $pageLang->formatNum( MWTimestamp::getLocalInstance( $ts )->format( 'w' ) );
2981                                 break;
2982                         case 'numberofarticles':
2983                                 $value = $pageLang->formatNum( SiteStats::articles() );
2984                                 break;
2985                         case 'numberoffiles':
2986                                 $value = $pageLang->formatNum( SiteStats::images() );
2987                                 break;
2988                         case 'numberofusers':
2989                                 $value = $pageLang->formatNum( SiteStats::users() );
2990                                 break;
2991                         case 'numberofactiveusers':
2992                                 $value = $pageLang->formatNum( SiteStats::activeUsers() );
2993                                 break;
2994                         case 'numberofpages':
2995                                 $value = $pageLang->formatNum( SiteStats::pages() );
2996                                 break;
2997                         case 'numberofadmins':
2998                                 $value = $pageLang->formatNum( SiteStats::numberingroup( 'sysop' ) );
2999                                 break;
3000                         case 'numberofedits':
3001                                 $value = $pageLang->formatNum( SiteStats::edits() );
3002                                 break;
3003                         case 'numberofviews':
3004                                 global $wgDisableCounters;
3005                                 $value = !$wgDisableCounters ? $pageLang->formatNum( SiteStats::views() ) : '';
3006                                 break;
3007                         case 'currenttimestamp':
3008                                 $value = wfTimestamp( TS_MW, $ts );
3009                                 break;
3010                         case 'localtimestamp':
3011                                 $value = MWTimestamp::getLocalInstance( $ts )->format( 'YmdHis' );
3012                                 break;
3013                         case 'currentversion':
3014                                 $value = SpecialVersion::getVersion();
3015                                 break;
3016                         case 'articlepath':
3017                                 return $wgArticlePath;
3018                         case 'sitename':
3019                                 return $wgSitename;
3020                         case 'server':
3021                                 return $wgServer;
3022                         case 'servername':
3023                                 $serverParts = wfParseUrl( $wgServer );
3024                                 return $serverParts && isset( $serverParts['host'] ) ? $serverParts['host'] : $wgServer;
3025                         case 'scriptpath':
3026                                 return $wgScriptPath;
3027                         case 'stylepath':
3028                                 return $wgStylePath;
3029                         case 'directionmark':
3030                                 return $pageLang->getDirMark();
3031                         case 'contentlanguage':
3032                                 global $wgLanguageCode;
3033                                 return $wgLanguageCode;
3034                         default:
3035                                 $ret = null;
3036                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) ) {
3037                                         return $ret;
3038                                 } else {
3039                                         return null;
3040                                 }
3041                 }
3042
3043                 if ( $index ) {
3044                         $this->mVarCache[$index] = $value;
3045                 }
3046
3047                 return $value;
3048         }
3049
3050         /**
3051          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
3052          *
3053          * @private
3054          */
3055         function initialiseVariables() {
3056                 wfProfileIn( __METHOD__ );
3057                 $variableIDs = MagicWord::getVariableIDs();
3058                 $substIDs = MagicWord::getSubstIDs();
3059
3060                 $this->mVariables = new MagicWordArray( $variableIDs );
3061                 $this->mSubstWords = new MagicWordArray( $substIDs );
3062                 wfProfileOut( __METHOD__ );
3063         }
3064
3065         /**
3066          * Preprocess some wikitext and return the document tree.
3067          * This is the ghost of replace_variables().
3068          *
3069          * @param string $text The text to parse
3070          * @param $flags Integer: bitwise combination of:
3071          *          self::PTD_FOR_INCLUSION    Handle "<noinclude>" and "<includeonly>" as if the text is being
3072          *                                     included. Default is to assume a direct page view.
3073          *
3074          * The generated DOM tree must depend only on the input text and the flags.
3075          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
3076          *
3077          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
3078          * change in the DOM tree for a given text, must be passed through the section identifier
3079          * in the section edit link and thus back to extractSections().
3080          *
3081          * The output of this function is currently only cached in process memory, but a persistent
3082          * cache may be implemented at a later date which takes further advantage of these strict
3083          * dependency requirements.
3084          *
3085          * @private
3086          *
3087          * @return PPNode
3088          */
3089         function preprocessToDom( $text, $flags = 0 ) {
3090                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
3091                 return $dom;
3092         }
3093
3094         /**
3095          * Return a three-element array: leading whitespace, string contents, trailing whitespace
3096          *
3097          * @param $s string
3098          *
3099          * @return array
3100          */
3101         public static function splitWhitespace( $s ) {
3102                 $ltrimmed = ltrim( $s );
3103                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
3104                 $trimmed = rtrim( $ltrimmed );
3105                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
3106                 if ( $diff > 0 ) {
3107                         $w2 = substr( $ltrimmed, -$diff );
3108                 } else {
3109                         $w2 = '';
3110                 }
3111                 return array( $w1, $trimmed, $w2 );
3112         }
3113
3114         /**
3115          * Replace magic variables, templates, and template arguments
3116          * with the appropriate text. Templates are substituted recursively,
3117          * taking care to avoid infinite loops.
3118          *
3119          * Note that the substitution depends on value of $mOutputType:
3120          *  self::OT_WIKI: only {{subst:}} templates
3121          *  self::OT_PREPROCESS: templates but not extension tags
3122          *  self::OT_HTML: all templates and extension tags
3123          *
3124          * @param string $text the text to transform
3125          * @param $frame PPFrame Object describing the arguments passed to the template.
3126          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
3127          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
3128          * @param $argsOnly Boolean only do argument (triple-brace) expansion, not double-brace expansion
3129          * @private
3130          *
3131          * @return string
3132          */
3133         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
3134                 # Is there any text? Also, Prevent too big inclusions!
3135                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
3136                         return $text;
3137                 }
3138                 wfProfileIn( __METHOD__ );
3139
3140                 if ( $frame === false ) {
3141                         $frame = $this->getPreprocessor()->newFrame();
3142                 } elseif ( !( $frame instanceof PPFrame ) ) {
3143                         wfDebug( __METHOD__ . " called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
3144                         $frame = $this->getPreprocessor()->newCustomFrame( $frame );
3145                 }
3146
3147                 $dom = $this->preprocessToDom( $text );
3148                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
3149                 $text = $frame->expand( $dom, $flags );
3150
3151                 wfProfileOut( __METHOD__ );
3152                 return $text;
3153         }
3154
3155         /**
3156          * Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
3157          *
3158          * @param $args array
3159          *
3160          * @return array
3161          */
3162         static function createAssocArgs( $args ) {
3163                 $assocArgs = array();
3164                 $index = 1;
3165                 foreach ( $args as $arg ) {
3166                         $eqpos = strpos( $arg, '=' );
3167                         if ( $eqpos === false ) {
3168                                 $assocArgs[$index++] = $arg;
3169                         } else {
3170                                 $name = trim( substr( $arg, 0, $eqpos ) );
3171                                 $value = trim( substr( $arg, $eqpos + 1 ) );
3172                                 if ( $value === false ) {
3173                                         $value = '';
3174                                 }
3175                                 if ( $name !== false ) {
3176                                         $assocArgs[$name] = $value;
3177                                 }
3178                         }
3179                 }
3180
3181                 return $assocArgs;
3182         }
3183
3184         /**
3185          * Warn the user when a parser limitation is reached
3186          * Will warn at most once the user per limitation type
3187          *
3188          * @param string $limitationType should be one of:
3189          *   'expensive-parserfunction' (corresponding messages:
3190          *       'expensive-parserfunction-warning',
3191          *       'expensive-parserfunction-category')
3192          *   'post-expand-template-argument' (corresponding messages:
3193          *       'post-expand-template-argument-warning',
3194          *       'post-expand-template-argument-category')
3195          *   'post-expand-template-inclusion' (corresponding messages:
3196          *       'post-expand-template-inclusion-warning',
3197          *       'post-expand-template-inclusion-category')
3198          *   'node-count-exceeded' (corresponding messages:
3199          *       'node-count-exceeded-warning',
3200          *       'node-count-exceeded-category')
3201          *   'expansion-depth-exceeded' (corresponding messages:
3202          *       'expansion-depth-exceeded-warning',
3203          *       'expansion-depth-exceeded-category')
3204          * @param int|null $current Current value
3205          * @param int|null $max Maximum allowed, when an explicit limit has been
3206          *       exceeded, provide the values (optional)
3207          */
3208         function limitationWarn( $limitationType, $current = '', $max = '' ) {
3209                 # does no harm if $current and $max are present but are unnecessary for the message
3210                 $warning = wfMessage( "$limitationType-warning" )->numParams( $current, $max )
3211                         ->inLanguage( $this->mOptions->getUserLangObj() )->text();
3212                 $this->mOutput->addWarning( $warning );
3213                 $this->addTrackingCategory( "$limitationType-category" );
3214         }
3215
3216         /**
3217          * Return the text of a template, after recursively
3218          * replacing any variables or templates within the template.
3219          *
3220          * @param array $piece the parts of the template
3221          *  $piece['title']: the title, i.e. the part before the |
3222          *  $piece['parts']: the parameter array
3223          *  $piece['lineStart']: whether the brace was at the start of a line
3224          * @param $frame PPFrame The current frame, contains template arguments
3225          * @throws MWException
3226          * @return String: the text of the template
3227          * @private
3228          */
3229         function braceSubstitution( $piece, $frame ) {
3230                 wfProfileIn( __METHOD__ );
3231                 wfProfileIn( __METHOD__ . '-setup' );
3232
3233                 # Flags
3234                 $found = false;             # $text has been filled
3235                 $nowiki = false;            # wiki markup in $text should be escaped
3236                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
3237                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
3238                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
3239                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
3240
3241                 # Title object, where $text came from
3242                 $title = false;
3243
3244                 # $part1 is the bit before the first |, and must contain only title characters.
3245                 # Various prefixes will be stripped from it later.
3246                 $titleWithSpaces = $frame->expand( $piece['title'] );
3247                 $part1 = trim( $titleWithSpaces );
3248                 $titleText = false;
3249
3250                 # Original title text preserved for various purposes
3251                 $originalTitle = $part1;
3252
3253                 # $args is a list of argument nodes, starting from index 0, not including $part1
3254                 # @todo FIXME: If piece['parts'] is null then the call to getLength() below won't work b/c this $args isn't an object
3255                 $args = ( null == $piece['parts'] ) ? array() : $piece['parts'];
3256                 wfProfileOut( __METHOD__ . '-setup' );
3257
3258                 $titleProfileIn = null; // profile templates
3259
3260                 # SUBST
3261                 wfProfileIn( __METHOD__ . '-modifiers' );
3262                 if ( !$found ) {
3263
3264                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
3265
3266                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
3267                         # Decide whether to expand template or keep wikitext as-is.
3268                         if ( $this->ot['wiki'] ) {
3269                                 if ( $substMatch === false ) {
3270                                         $literal = true;  # literal when in PST with no prefix
3271                                 } else {
3272                                         $literal = false; # expand when in PST with subst: or safesubst:
3273                                 }
3274                         } else {
3275                                 if ( $substMatch == 'subst' ) {
3276                                         $literal = true;  # literal when not in PST with plain subst:
3277                                 } else {
3278                                         $literal = false; # expand when not in PST with safesubst: or no prefix
3279                                 }
3280                         }
3281                         if ( $literal ) {
3282                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3283                                 $isLocalObj = true;
3284                                 $found = true;
3285                         }
3286                 }
3287
3288                 # Variables
3289                 if ( !$found && $args->getLength() == 0 ) {
3290                         $id = $this->mVariables->matchStartToEnd( $part1 );
3291                         if ( $id !== false ) {
3292                                 $text = $this->getVariableValue( $id, $frame );
3293                                 if ( MagicWord::getCacheTTL( $id ) > -1 ) {
3294                                         $this->mOutput->updateCacheExpiry( MagicWord::getCacheTTL( $id ) );
3295                                 }
3296                                 $found = true;
3297                         }
3298                 }
3299
3300                 # MSG, MSGNW and RAW
3301                 if ( !$found ) {
3302                         # Check for MSGNW:
3303                         $mwMsgnw = MagicWord::get( 'msgnw' );
3304                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
3305                                 $nowiki = true;
3306                         } else {
3307                                 # Remove obsolete MSG:
3308                                 $mwMsg = MagicWord::get( 'msg' );
3309                                 $mwMsg->matchStartAndRemove( $part1 );
3310                         }
3311
3312                         # Check for RAW:
3313                         $mwRaw = MagicWord::get( 'raw' );
3314                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
3315                                 $forceRawInterwiki = true;
3316                         }
3317                 }
3318                 wfProfileOut( __METHOD__ . '-modifiers' );
3319
3320                 # Parser functions
3321                 if ( !$found ) {
3322                         wfProfileIn( __METHOD__ . '-pfunc' );
3323
3324                         $colonPos = strpos( $part1, ':' );
3325                         if ( $colonPos !== false ) {
3326                                 $func = substr( $part1, 0, $colonPos );
3327                                 $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
3328                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
3329                                         $funcArgs[] = $args->item( $i );
3330                                 }
3331                                 try {
3332                                         $result = $this->callParserFunction( $frame, $func, $funcArgs );
3333                                 } catch ( Exception $ex ) {
3334                                         wfProfileOut( __METHOD__ . '-pfunc' );
3335                                         wfProfileOut( __METHOD__ );
3336                                         throw $ex;
3337                                 }
3338
3339                                 # The interface for parser functions allows for extracting
3340                                 # flags into the local scope. Extract any forwarded flags
3341                                 # here.
3342                                 extract( $result );
3343                         }
3344                         wfProfileOut( __METHOD__ . '-pfunc' );
3345                 }
3346
3347                 # Finish mangling title and then check for loops.
3348                 # Set $title to a Title object and $titleText to the PDBK
3349                 if ( !$found ) {
3350                         $ns = NS_TEMPLATE;
3351                         # Split the title into page and subpage
3352                         $subpage = '';
3353                         $relative = $this->maybeDoSubpageLink( $part1, $subpage );
3354                         if ( $part1 !== $relative ) {
3355                                 $part1 = $relative;
3356                                 $ns = $this->mTitle->getNamespace();
3357                         }
3358                         $title = Title::newFromText( $part1, $ns );
3359                         if ( $title ) {
3360                                 $titleText = $title->getPrefixedText();
3361                                 # Check for language variants if the template is not found
3362                                 if ( $this->getConverterLanguage()->hasVariants() && $title->getArticleID() == 0 ) {
3363                                         $this->getConverterLanguage()->findVariantLink( $part1, $title, true );
3364                                 }
3365                                 # Do recursion depth check
3366                                 $limit = $this->mOptions->getMaxTemplateDepth();
3367                                 if ( $frame->depth >= $limit ) {
3368                                         $found = true;
3369                                         $text = '<span class="error">'
3370                                                 . wfMessage( 'parser-template-recursion-depth-warning' )
3371                                                         ->numParams( $limit )->inContentLanguage()->text()
3372                                                 . '</span>';
3373                                 }
3374                         }
3375                 }
3376
3377                 # Load from database
3378                 if ( !$found && $title ) {
3379                         if ( !Profiler::instance()->isPersistent() ) {
3380                                 # Too many unique items can kill profiling DBs/collectors
3381                                 $titleProfileIn = __METHOD__ . "-title-" . $title->getPrefixedDBkey();
3382                                 wfProfileIn( $titleProfileIn ); // template in
3383                         }
3384                         wfProfileIn( __METHOD__ . '-loadtpl' );
3385                         if ( !$title->isExternal() ) {
3386                                 if ( $title->isSpecialPage()
3387                                         && $this->mOptions->getAllowSpecialInclusion()
3388                                         && $this->ot['html'] )
3389                                 {
3390                                         // Pass the template arguments as URL parameters.
3391                                         // "uselang" will have no effect since the Language object
3392                                         // is forced to the one defined in ParserOptions.
3393                                         $pageArgs = array();
3394                                         for ( $i = 0; $i < $args->getLength(); $i++ ) {
3395                                                 $bits = $args->item( $i )->splitArg();
3396                                                 if ( strval( $bits['index'] ) === '' ) {
3397                                                         $name = trim( $frame->expand( $bits['name'], PPFrame::STRIP_COMMENTS ) );
3398                                                         $value = trim( $frame->expand( $bits['value'] ) );
3399                                                         $pageArgs[$name] = $value;
3400                                                 }
3401                                         }
3402
3403                                         // Create a new context to execute the special page
3404                                         $context = new RequestContext;
3405                                         $context->setTitle( $title );
3406                                         $context->setRequest( new FauxRequest( $pageArgs ) );
3407                                         $context->setUser( $this->getUser() );
3408                                         $context->setLanguage( $this->mOptions->getUserLangObj() );
3409                                         $ret = SpecialPageFactory::capturePath( $title, $context );
3410                                         if ( $ret ) {
3411                                                 $text = $context->getOutput()->getHTML();
3412                                                 $this->mOutput->addOutputPageMetadata( $context->getOutput() );
3413                                                 $found = true;
3414                                                 $isHTML = true;
3415                                                 $this->disableCache();
3416                                         }
3417                                 } elseif ( MWNamespace::isNonincludable( $title->getNamespace() ) ) {
3418                                         $found = false; # access denied
3419                                         wfDebug( __METHOD__ . ": template inclusion denied for " . $title->getPrefixedDBkey() );
3420                                 } else {
3421                                         list( $text, $title ) = $this->getTemplateDom( $title );
3422                                         if ( $text !== false ) {
3423                                                 $found = true;
3424                                                 $isChildObj = true;
3425                                         }
3426                                 }
3427
3428                                 # If the title is valid but undisplayable, make a link to it
3429                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3430                                         $text = "[[:$titleText]]";
3431                                         $found = true;
3432                                 }
3433                         } elseif ( $title->isTrans() ) {
3434                                 # Interwiki transclusion
3435                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3436                                         $text = $this->interwikiTransclude( $title, 'render' );
3437                                         $isHTML = true;
3438                                 } else {
3439                                         $text = $this->interwikiTransclude( $title, 'raw' );
3440                                         # Preprocess it like a template
3441                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3442                                         $isChildObj = true;
3443                                 }
3444                                 $found = true;
3445                         }
3446
3447                         # Do infinite loop check
3448                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3449                         if ( !$frame->loopCheck( $title ) ) {
3450                                 $found = true;
3451                                 $text = '<span class="error">'
3452                                         . wfMessage( 'parser-template-loop-warning', $titleText )->inContentLanguage()->text()
3453                                         . '</span>';
3454                                 wfDebug( __METHOD__ . ": template loop broken at '$titleText'\n" );
3455                         }
3456                         wfProfileOut( __METHOD__ . '-loadtpl' );
3457                 }
3458
3459                 # If we haven't found text to substitute by now, we're done
3460                 # Recover the source wikitext and return it
3461                 if ( !$found ) {
3462                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3463                         if ( $titleProfileIn ) {
3464                                 wfProfileOut( $titleProfileIn ); // template out
3465                         }
3466                         wfProfileOut( __METHOD__ );
3467                         return array( 'object' => $text );
3468                 }
3469
3470                 # Expand DOM-style return values in a child frame
3471                 if ( $isChildObj ) {
3472                         # Clean up argument array
3473                         $newFrame = $frame->newChild( $args, $title );
3474
3475                         if ( $nowiki ) {
3476                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3477                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3478                                 # Expansion is eligible for the empty-frame cache
3479                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3480                                         $text = $this->mTplExpandCache[$titleText];
3481                                 } else {
3482                                         $text = $newFrame->expand( $text );
3483                                         $this->mTplExpandCache[$titleText] = $text;
3484                                 }
3485                         } else {
3486                                 # Uncached expansion
3487                                 $text = $newFrame->expand( $text );
3488                         }
3489                 }
3490                 if ( $isLocalObj && $nowiki ) {
3491                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3492                         $isLocalObj = false;
3493                 }
3494
3495                 if ( $titleProfileIn ) {
3496                         wfProfileOut( $titleProfileIn ); // template out
3497                 }
3498
3499                 # Replace raw HTML by a placeholder
3500                 if ( $isHTML ) {
3501                         $text = $this->insertStripItem( $text );
3502                 } elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3503                         # Escape nowiki-style return values
3504                         $text = wfEscapeWikiText( $text );
3505                 } elseif ( is_string( $text )
3506                         && !$piece['lineStart']
3507                         && preg_match( '/^(?:{\\||:|;|#|\*)/', $text ) )
3508                 {
3509                         # Bug 529: if the template begins with a table or block-level
3510                         # element, it should be treated as beginning a new line.
3511                         # This behavior is somewhat controversial.
3512                         $text = "\n" . $text;
3513                 }
3514
3515                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3516                         # Error, oversize inclusion
3517                         if ( $titleText !== false ) {
3518                                 # Make a working, properly escaped link if possible (bug 23588)
3519                                 $text = "[[:$titleText]]";
3520                         } else {
3521                                 # This will probably not be a working link, but at least it may
3522                                 # provide some hint of where the problem is
3523                                 preg_replace( '/^:/', '', $originalTitle );
3524                                 $text = "[[:$originalTitle]]";
3525                         }
3526                         $text .= $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3527                         $this->limitationWarn( 'post-expand-template-inclusion' );
3528                 }
3529
3530                 if ( $isLocalObj ) {
3531                         $ret = array( 'object' => $text );
3532                 } else {
3533                         $ret = array( 'text' => $text );
3534                 }
3535
3536                 wfProfileOut( __METHOD__ );
3537                 return $ret;
3538         }
3539
3540         /**
3541          * Call a parser function and return an array with text and flags.
3542          *
3543          * The returned array will always contain a boolean 'found', indicating
3544          * whether the parser function was found or not. It may also contain the
3545          * following:
3546          *  text: string|object, resulting wikitext or PP DOM object
3547          *  isHTML: bool, $text is HTML, armour it against wikitext transformation
3548          *  isChildObj: bool, $text is a DOM node needing expansion in a child frame
3549          *  isLocalObj: bool, $text is a DOM node needing expansion in the current frame
3550          *  nowiki: bool, wiki markup in $text should be escaped
3551          *
3552          * @since 1.21
3553          * @param $frame PPFrame The current frame, contains template arguments
3554          * @param $function string Function name
3555          * @param $args array Arguments to the function
3556          * @return array
3557          */
3558         public function callParserFunction( $frame, $function, array $args = array() ) {
3559                 global $wgContLang;
3560
3561                 wfProfileIn( __METHOD__ );
3562
3563                 # Case sensitive functions
3564                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
3565                         $function = $this->mFunctionSynonyms[1][$function];
3566                 } else {
3567                         # Case insensitive functions
3568                         $function = $wgContLang->lc( $function );
3569                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
3570                                 $function = $this->mFunctionSynonyms[0][$function];
3571                         } else {
3572                                 wfProfileOut( __METHOD__ );
3573                                 return array( 'found' => false );
3574                         }
3575                 }
3576
3577                 wfProfileIn( __METHOD__ . '-pfunc-' . $function );
3578                 list( $callback, $flags ) = $this->mFunctionHooks[$function];
3579
3580                 # Workaround for PHP bug 35229 and similar
3581                 if ( !is_callable( $callback ) ) {
3582                         wfProfileOut( __METHOD__ . '-pfunc-' . $function );
3583                         wfProfileOut( __METHOD__ );
3584                         throw new MWException( "Tag hook for $function is not callable\n" );
3585                 }
3586
3587                 $allArgs = array( &$this );
3588                 if ( $flags & SFH_OBJECT_ARGS ) {
3589                         # Convert arguments to PPNodes and collect for appending to $allArgs
3590                         $funcArgs = array();
3591                         foreach ( $args as $k => $v ) {
3592                                 if ( $v instanceof PPNode || $k === 0 ) {
3593                                         $funcArgs[] = $v;
3594                                 } else {
3595                                         $funcArgs[] = $this->mPreprocessor->newPartNodeArray( array( $k => $v ) )->item( 0 );
3596                                 }
3597                         }
3598
3599                         # Add a frame parameter, and pass the arguments as an array
3600                         $allArgs[] = $frame;
3601                         $allArgs[] = $funcArgs;
3602                 } else {
3603                         # Convert arguments to plain text and append to $allArgs
3604                         foreach ( $args as $k => $v ) {
3605                                 if ( $v instanceof PPNode ) {
3606                                         $allArgs[] = trim( $frame->expand( $v ) );
3607                                 } elseif ( is_int( $k ) && $k >= 0 ) {
3608                                         $allArgs[] = trim( $v );
3609                                 } else {
3610                                         $allArgs[] = trim( "$k=$v" );
3611                                 }
3612                         }
3613                 }
3614
3615                 $result = call_user_func_array( $callback, $allArgs );
3616
3617                 # The interface for function hooks allows them to return a wikitext
3618                 # string or an array containing the string and any flags. This mungs
3619                 # things around to match what this method should return.
3620                 if ( !is_array( $result ) ) {
3621                         $result = array(
3622                                 'found' => true,
3623                                 'text' => $result,
3624                         );
3625                 } else {
3626                         if ( isset( $result[0] ) && !isset( $result['text'] ) ) {
3627                                 $result['text'] = $result[0];
3628                         }
3629                         unset( $result[0] );
3630                         $result += array(
3631                                 'found' => true,
3632                         );
3633                 }
3634
3635                 $noparse = true;
3636                 $preprocessFlags = 0;
3637                 if ( isset( $result['noparse'] ) ) {
3638                         $noparse = $result['noparse'];
3639                 }
3640                 if ( isset( $result['preprocessFlags'] ) ) {
3641                         $preprocessFlags = $result['preprocessFlags'];
3642                 }
3643
3644                 if ( !$noparse ) {
3645                         $result['text'] = $this->preprocessToDom( $result['text'], $preprocessFlags );
3646                         $result['isChildObj'] = true;
3647                 }
3648                 wfProfileOut( __METHOD__ . '-pfunc-' . $function );
3649                 wfProfileOut( __METHOD__ );
3650
3651                 return $result;
3652         }
3653
3654         /**
3655          * Get the semi-parsed DOM representation of a template with a given title,
3656          * and its redirect destination title. Cached.
3657          *
3658          * @param $title Title
3659          *
3660          * @return array
3661          */
3662         function getTemplateDom( $title ) {
3663                 $cacheTitle = $title;
3664                 $titleText = $title->getPrefixedDBkey();
3665
3666                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3667                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3668                         $title = Title::makeTitle( $ns, $dbk );
3669                         $titleText = $title->getPrefixedDBkey();
3670                 }
3671                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3672                         return array( $this->mTplDomCache[$titleText], $title );
3673                 }
3674
3675                 # Cache miss, go to the database
3676                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3677
3678                 if ( $text === false ) {
3679                         $this->mTplDomCache[$titleText] = false;
3680                         return array( false, $title );
3681                 }
3682
3683                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3684                 $this->mTplDomCache[$titleText] = $dom;
3685
3686                 if ( !$title->equals( $cacheTitle ) ) {
3687                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3688                                 array( $title->getNamespace(), $cdb = $title->getDBkey() );
3689                 }
3690
3691                 return array( $dom, $title );
3692         }
3693
3694         /**
3695          * Fetch the unparsed text of a template and register a reference to it.
3696          * @param Title $title
3697          * @return Array ( string or false, Title )
3698          */
3699         function fetchTemplateAndTitle( $title ) {
3700                 $templateCb = $this->mOptions->getTemplateCallback(); # Defaults to Parser::statelessFetchTemplate()
3701                 $stuff = call_user_func( $templateCb, $title, $this );
3702                 $text = $stuff['text'];
3703                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3704                 if ( isset( $stuff['deps'] ) ) {
3705                         foreach ( $stuff['deps'] as $dep ) {
3706                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3707                                 if ( $dep['title']->equals( $this->getTitle() ) ) {
3708                                         // If we transclude ourselves, the final result
3709                                         // will change based on the new version of the page
3710                                         $this->mOutput->setFlag( 'vary-revision' );
3711                                 }
3712                         }
3713                 }
3714                 return array( $text, $finalTitle );
3715         }
3716
3717         /**
3718          * Fetch the unparsed text of a template and register a reference to it.
3719          * @param Title $title
3720          * @return mixed string or false
3721          */
3722         function fetchTemplate( $title ) {
3723                 $rv = $this->fetchTemplateAndTitle( $title );
3724                 return $rv[0];
3725         }
3726
3727         /**
3728          * Static function to get a template
3729          * Can be overridden via ParserOptions::setTemplateCallback().
3730          *
3731          * @param $title  Title
3732          * @param $parser Parser
3733          *
3734          * @return array
3735          */
3736         static function statelessFetchTemplate( $title, $parser = false ) {
3737                 $text = $skip = false;
3738                 $finalTitle = $title;
3739                 $deps = array();
3740
3741                 # Loop to fetch the article, with up to 1 redirect
3742                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3743                         # Give extensions a chance to select the revision instead
3744                         $id = false; # Assume current
3745                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle',
3746                                 array( $parser, $title, &$skip, &$id ) );
3747
3748                         if ( $skip ) {
3749                                 $text = false;
3750                                 $deps[] = array(
3751                                         'title' => $title,
3752                                         'page_id' => $title->getArticleID(),
3753                                         'rev_id' => null
3754                                 );
3755                                 break;
3756                         }
3757                         # Get the revision
3758                         $rev = $id
3759                                 ? Revision::newFromId( $id )
3760                                 : Revision::newFromTitle( $title, false, Revision::READ_NORMAL );
3761                         $rev_id = $rev ? $rev->getId() : 0;
3762                         # If there is no current revision, there is no page
3763                         if ( $id === false && !$rev ) {
3764                                 $linkCache = LinkCache::singleton();
3765                                 $linkCache->addBadLinkObj( $title );
3766                         }
3767
3768                         $deps[] = array(
3769                                 'title' => $title,
3770                                 'page_id' => $title->getArticleID(),
3771                                 'rev_id' => $rev_id );
3772                         if ( $rev && !$title->equals( $rev->getTitle() ) ) {
3773                                 # We fetched a rev from a different title; register it too...
3774                                 $deps[] = array(
3775                                         'title' => $rev->getTitle(),
3776                                         'page_id' => $rev->getPage(),
3777                                         'rev_id' => $rev_id );
3778                         }
3779
3780                         if ( $rev ) {
3781                                 $content = $rev->getContent();
3782                                 $text = $content ? $content->getWikitextForTransclusion() : null;
3783
3784                                 if ( $text === false || $text === null ) {
3785                                         $text = false;
3786                                         break;
3787                                 }
3788                         } elseif ( $title->getNamespace() == NS_MEDIAWIKI ) {
3789                                 global $wgContLang;
3790                                 $message = wfMessage( $wgContLang->lcfirst( $title->getText() ) )->inContentLanguage();
3791                                 if ( !$message->exists() ) {
3792                                         $text = false;
3793                                         break;
3794                                 }
3795                                 $content = $message->content();
3796                                 $text = $message->plain();
3797                         } else {
3798                                 break;
3799                         }
3800                         if ( !$content ) {
3801                                 break;
3802                         }
3803                         # Redirect?
3804                         $finalTitle = $title;
3805                         $title = $content->getRedirectTarget();
3806                 }
3807                 return array(
3808                         'text' => $text,
3809                         'finalTitle' => $finalTitle,
3810                         'deps' => $deps );
3811         }
3812
3813         /**
3814          * Fetch a file and its title and register a reference to it.
3815          * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3816          * @param Title $title
3817          * @param array $options Array of options to RepoGroup::findFile
3818          * @return File|bool
3819          */
3820         function fetchFile( $title, $options = array() ) {
3821                 $res = $this->fetchFileAndTitle( $title, $options );
3822                 return $res[0];
3823         }
3824
3825         /**
3826          * Fetch a file and its title and register a reference to it.
3827          * If 'broken' is a key in $options then the file will appear as a broken thumbnail.
3828          * @param Title $title
3829          * @param array $options Array of options to RepoGroup::findFile
3830          * @return Array ( File or false, Title of file )
3831          */
3832         function fetchFileAndTitle( $title, $options = array() ) {
3833                 $file = $this->fetchFileNoRegister( $title, $options );
3834
3835                 $time = $file ? $file->getTimestamp() : false;
3836                 $sha1 = $file ? $file->getSha1() : false;
3837                 # Register the file as a dependency...
3838                 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3839                 if ( $file && !$title->equals( $file->getTitle() ) ) {
3840                         # Update fetched file title
3841                         $title = $file->getTitle();
3842                         if ( is_null( $file->getRedirectedTitle() ) ) {
3843                                 # This file was not a redirect, but the title does not match.
3844                                 # Register under the new name because otherwise the link will
3845                                 # get lost.
3846                                 $this->mOutput->addImage( $title->getDBkey(), $time, $sha1 );
3847                         }
3848                 }
3849                 return array( $file, $title );
3850         }
3851
3852         /**
3853          * Helper function for fetchFileAndTitle.
3854          *
3855          * Also useful if you need to fetch a file but not use it yet,
3856          * for example to get the file's handler.
3857          *
3858          * @param Title $title
3859          * @param array $options Array of options to RepoGroup::findFile
3860          * @return File or false
3861          */
3862         protected function fetchFileNoRegister( $title, $options = array() ) {
3863                 if ( isset( $options['broken'] ) ) {
3864                         $file = false; // broken thumbnail forced by hook
3865                 } elseif ( isset( $options['sha1'] ) ) { // get by (sha1,timestamp)
3866                         $file = RepoGroup::singleton()->findFileFromKey( $options['sha1'], $options );
3867                 } else { // get by (name,timestamp)
3868                         $file = wfFindFile( $title, $options );
3869                 }
3870                 return $file;
3871         }
3872
3873         /**
3874          * Transclude an interwiki link.
3875          *
3876          * @param $title Title
3877          * @param $action
3878          *
3879          * @return string
3880          */
3881         function interwikiTransclude( $title, $action ) {
3882                 global $wgEnableScaryTranscluding;
3883
3884                 if ( !$wgEnableScaryTranscluding ) {
3885                         return wfMessage( 'scarytranscludedisabled' )->inContentLanguage()->text();
3886                 }
3887
3888                 $url = $title->getFullURL( array( 'action' => $action ) );
3889
3890                 if ( strlen( $url ) > 255 ) {
3891                         return wfMessage( 'scarytranscludetoolong' )->inContentLanguage()->text();
3892                 }
3893                 return $this->fetchScaryTemplateMaybeFromCache( $url );
3894         }
3895
3896         /**
3897          * @param $url string
3898          * @return Mixed|String
3899          */
3900         function fetchScaryTemplateMaybeFromCache( $url ) {
3901                 global $wgTranscludeCacheExpiry;
3902                 $dbr = wfGetDB( DB_SLAVE );
3903                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3904                 $obj = $dbr->selectRow( 'transcache', array( 'tc_time', 'tc_contents' ),
3905                                 array( 'tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3906                 if ( $obj ) {
3907                         return $obj->tc_contents;
3908                 }
3909
3910                 $req = MWHttpRequest::factory( $url );
3911                 $status = $req->execute(); // Status object
3912                 if ( $status->isOK() ) {
3913                         $text = $req->getContent();
3914                 } elseif ( $req->getStatus() != 200 ) { // Though we failed to fetch the content, this status is useless.
3915                         return wfMessage( 'scarytranscludefailed-httpstatus', $url, $req->getStatus() /* HTTP status */ )->inContentLanguage()->text();
3916                 } else {
3917                         return wfMessage( 'scarytranscludefailed', $url )->inContentLanguage()->text();
3918                 }
3919
3920                 $dbw = wfGetDB( DB_MASTER );
3921                 $dbw->replace( 'transcache', array( 'tc_url' ), array(
3922                         'tc_url' => $url,
3923                         'tc_time' => $dbw->timestamp( time() ),
3924                         'tc_contents' => $text
3925                 ) );
3926                 return $text;
3927         }
3928
3929         /**
3930          * Triple brace replacement -- used for template arguments
3931          * @private
3932          *
3933          * @param $piece array
3934          * @param $frame PPFrame
3935          *
3936          * @return array
3937          */
3938         function argSubstitution( $piece, $frame ) {
3939                 wfProfileIn( __METHOD__ );
3940
3941                 $error = false;
3942                 $parts = $piece['parts'];
3943                 $nameWithSpaces = $frame->expand( $piece['title'] );
3944                 $argName = trim( $nameWithSpaces );
3945                 $object = false;
3946                 $text = $frame->getArgument( $argName );
3947                 if ( $text === false && $parts->getLength() > 0
3948                   && (
3949                         $this->ot['html']
3950                         || $this->ot['pre']
3951                         || ( $this->ot['wiki'] && $frame->isTemplate() )
3952                   )
3953                 ) {
3954                         # No match in frame, use the supplied default
3955                         $object = $parts->item( 0 )->getChildren();
3956                 }
3957                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3958                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3959                         $this->limitationWarn( 'post-expand-template-argument' );
3960                 }
3961
3962                 if ( $text === false && $object === false ) {
3963                         # No match anywhere
3964                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3965                 }
3966                 if ( $error !== false ) {
3967                         $text .= $error;
3968                 }
3969                 if ( $object !== false ) {
3970                         $ret = array( 'object' => $object );
3971                 } else {
3972                         $ret = array( 'text' => $text );
3973                 }
3974
3975                 wfProfileOut( __METHOD__ );
3976                 return $ret;
3977         }
3978
3979         /**
3980          * Return the text to be used for a given extension tag.
3981          * This is the ghost of strip().
3982          *
3983          * @param array $params Associative array of parameters:
3984          *     name       PPNode for the tag name
3985          *     attr       PPNode for unparsed text where tag attributes are thought to be
3986          *     attributes Optional associative array of parsed attributes
3987          *     inner      Contents of extension element
3988          *     noClose    Original text did not have a close tag
3989          * @param $frame PPFrame
3990          *
3991          * @throws MWException
3992          * @return string
3993          */
3994         function extensionSubstitution( $params, $frame ) {
3995                 $name = $frame->expand( $params['name'] );
3996                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3997                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3998                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf( '%08X', $this->mMarkerIndex++ ) . self::MARKER_SUFFIX;
3999
4000                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower( $name )] ) &&
4001                         ( $this->ot['html'] || $this->ot['pre'] );
4002                 if ( $isFunctionTag ) {
4003                         $markerType = 'none';
4004                 } else {
4005                         $markerType = 'general';
4006                 }
4007                 if ( $this->ot['html'] || $isFunctionTag ) {
4008                         $name = strtolower( $name );
4009                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
4010                         if ( isset( $params['attributes'] ) ) {
4011                                 $attributes = $attributes + $params['attributes'];
4012                         }
4013
4014                         if ( isset( $this->mTagHooks[$name] ) ) {
4015                                 # Workaround for PHP bug 35229 and similar
4016                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
4017                                         throw new MWException( "Tag hook for $name is not callable\n" );
4018                                 }
4019                                 $output = call_user_func_array( $this->mTagHooks[$name],
4020                                         array( $content, $attributes, $this, $frame ) );
4021                         } elseif ( isset( $this->mFunctionTagHooks[$name] ) ) {
4022                                 list( $callback, ) = $this->mFunctionTagHooks[$name];
4023                                 if ( !is_callable( $callback ) ) {
4024                                         throw new MWException( "Tag hook for $name is not callable\n" );
4025                                 }
4026
4027                                 $output = call_user_func_array( $callback, array( &$this, $frame, $content, $attributes ) );
4028                         } else {
4029                                 $output = '<span class="error">Invalid tag extension name: ' .
4030                                         htmlspecialchars( $name ) . '</span>';
4031                         }
4032
4033                         if ( is_array( $output ) ) {
4034                                 # Extract flags to local scope (to override $markerType)
4035                                 $flags = $output;
4036                                 $output = $flags[0];
4037                                 unset( $flags[0] );
4038                                 extract( $flags );
4039                         }
4040                 } else {
4041                         if ( is_null( $attrText ) ) {
4042                                 $attrText = '';
4043                         }
4044                         if ( isset( $params['attributes'] ) ) {
4045                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
4046                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
4047                                                 htmlspecialchars( $attrValue ) . '"';
4048                                 }
4049                         }
4050                         if ( $content === null ) {
4051                                 $output = "<$name$attrText/>";
4052                         } else {
4053                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
4054                                 $output = "<$name$attrText>$content$close";
4055                         }
4056                 }
4057
4058                 if ( $markerType === 'none' ) {
4059                         return $output;
4060                 } elseif ( $markerType === 'nowiki' ) {
4061                         $this->mStripState->addNoWiki( $marker, $output );
4062                 } elseif ( $markerType === 'general' ) {
4063                         $this->mStripState->addGeneral( $marker, $output );
4064                 } else {
4065                         throw new MWException( __METHOD__ . ': invalid marker type' );
4066                 }
4067                 return $marker;
4068         }
4069
4070         /**
4071          * Increment an include size counter
4072          *
4073          * @param string $type the type of expansion
4074          * @param $size Integer: the size of the text
4075          * @return Boolean: false if this inclusion would take it over the maximum, true otherwise
4076          */
4077         function incrementIncludeSize( $type, $size ) {
4078                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
4079                         return false;
4080                 } else {
4081                         $this->mIncludeSizes[$type] += $size;
4082                         return true;
4083                 }
4084         }
4085
4086         /**
4087          * Increment the expensive function count
4088          *
4089          * @return Boolean: false if the limit has been exceeded
4090          */
4091         function incrementExpensiveFunctionCount() {
4092                 $this->mExpensiveFunctionCount++;
4093                 return $this->mExpensiveFunctionCount <= $this->mOptions->getExpensiveParserFunctionLimit();
4094         }
4095
4096         /**
4097          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
4098          * Fills $this->mDoubleUnderscores, returns the modified text
4099          *
4100          * @param $text string
4101          *
4102          * @return string
4103          */
4104         function doDoubleUnderscore( $text ) {
4105                 wfProfileIn( __METHOD__ );
4106
4107                 # The position of __TOC__ needs to be recorded
4108                 $mw = MagicWord::get( 'toc' );
4109                 if ( $mw->match( $text ) ) {
4110                         $this->mShowToc = true;
4111                         $this->mForceTocPosition = true;
4112
4113                         # Set a placeholder. At the end we'll fill it in with the TOC.
4114                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
4115
4116                         # Only keep the first one.
4117                         $text = $mw->replace( '', $text );
4118                 }
4119
4120                 # Now match and remove the rest of them
4121                 $mwa = MagicWord::getDoubleUnderscoreArray();
4122                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
4123
4124                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
4125                         $this->mOutput->mNoGallery = true;
4126                 }
4127                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
4128                         $this->mShowToc = false;
4129                 }
4130                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
4131                         $this->addTrackingCategory( 'hidden-category-category' );
4132                 }
4133                 # (bug 8068) Allow control over whether robots index a page.
4134                 #
4135                 # @todo FIXME: Bug 14899: __INDEX__ always overrides __NOINDEX__ here!  This
4136                 # is not desirable, the last one on the page should win.
4137                 if ( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
4138                         $this->mOutput->setIndexPolicy( 'noindex' );
4139                         $this->addTrackingCategory( 'noindex-category' );
4140                 }
4141                 if ( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ) {
4142                         $this->mOutput->setIndexPolicy( 'index' );
4143                         $this->addTrackingCategory( 'index-category' );
4144                 }
4145
4146                 # Cache all double underscores in the database
4147                 foreach ( $this->mDoubleUnderscores as $key => $val ) {
4148                         $this->mOutput->setProperty( $key, '' );
4149                 }
4150
4151                 wfProfileOut( __METHOD__ );
4152                 return $text;
4153         }
4154
4155         /**
4156          * Add a tracking category, getting the title from a system message,
4157          * or print a debug message if the title is invalid.
4158          *
4159          * @param string $msg message key
4160          * @return Boolean: whether the addition was successful
4161          */
4162         public function addTrackingCategory( $msg ) {
4163                 if ( $this->mTitle->getNamespace() === NS_SPECIAL ) {
4164                         wfDebug( __METHOD__ . ": Not adding tracking category $msg to special page!\n" );
4165                         return false;
4166                 }
4167                 // Important to parse with correct title (bug 31469)
4168                 $cat = wfMessage( $msg )
4169                         ->title( $this->getTitle() )
4170                         ->inContentLanguage()
4171                         ->text();
4172
4173                 # Allow tracking categories to be disabled by setting them to "-"
4174                 if ( $cat === '-' ) {
4175                         return false;
4176                 }
4177
4178                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
4179                 if ( $containerCategory ) {
4180                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
4181                         return true;
4182                 } else {
4183                         wfDebug( __METHOD__ . ": [[MediaWiki:$msg]] is not a valid title!\n" );
4184                         return false;
4185                 }
4186         }
4187
4188         /**
4189          * This function accomplishes several tasks:
4190          * 1) Auto-number headings if that option is enabled
4191          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
4192          * 3) Add a Table of contents on the top for users who have enabled the option
4193          * 4) Auto-anchor headings
4194          *
4195          * It loops through all headlines, collects the necessary data, then splits up the
4196          * string and re-inserts the newly formatted headlines.
4197          *
4198          * @param $text String
4199          * @param string $origText original, untouched wikitext
4200          * @param $isMain Boolean
4201          * @return mixed|string
4202          * @private
4203          */
4204         function formatHeadings( $text, $origText, $isMain = true ) {
4205                 global $wgMaxTocLevel, $wgExperimentalHtmlIds;
4206
4207                 # Inhibit editsection links if requested in the page
4208                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
4209                         $maybeShowEditLink = $showEditLink = false;
4210                 } else {
4211                         $maybeShowEditLink = true; /* Actual presence will depend on ParserOptions option */
4212                         $showEditLink = $this->mOptions->getEditSection();
4213                 }
4214                 if ( $showEditLink ) {
4215                         $this->mOutput->setEditSectionTokens( true );
4216                 }
4217
4218                 # Get all headlines for numbering them and adding funky stuff like [edit]
4219                 # links - this is for later, but we need the number of headlines right now
4220                 $matches = array();
4221                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?' . '>)\s*(?P<header>[\s\S]*?)\s*<\/H[1-6] *>/i', $text, $matches );
4222
4223                 # if there are fewer than 4 headlines in the article, do not show TOC
4224                 # unless it's been explicitly enabled.
4225                 $enoughToc = $this->mShowToc &&
4226                         ( ( $numMatches >= 4 ) || $this->mForceTocPosition );
4227
4228                 # Allow user to stipulate that a page should have a "new section"
4229                 # link added via __NEWSECTIONLINK__
4230                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
4231                         $this->mOutput->setNewSection( true );
4232                 }
4233
4234                 # Allow user to remove the "new section"
4235                 # link via __NONEWSECTIONLINK__
4236                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
4237                         $this->mOutput->hideNewSection( true );
4238                 }
4239
4240                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
4241                 # override above conditions and always show TOC above first header
4242                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
4243                         $this->mShowToc = true;
4244                         $enoughToc = true;
4245                 }
4246
4247                 # headline counter
4248                 $headlineCount = 0;
4249                 $numVisible = 0;
4250
4251                 # Ugh .. the TOC should have neat indentation levels which can be
4252                 # passed to the skin functions. These are determined here
4253                 $toc = '';
4254                 $full = '';
4255                 $head = array();
4256                 $sublevelCount = array();
4257                 $levelCount = array();
4258                 $level = 0;
4259                 $prevlevel = 0;
4260                 $toclevel = 0;
4261                 $prevtoclevel = 0;
4262                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
4263                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
4264                 $oldType = $this->mOutputType;
4265                 $this->setOutputType( self::OT_WIKI );
4266                 $frame = $this->getPreprocessor()->newFrame();
4267                 $root = $this->preprocessToDom( $origText );
4268                 $node = $root->getFirstChild();
4269                 $byteOffset = 0;
4270                 $tocraw = array();
4271                 $refers = array();
4272
4273                 foreach ( $matches[3] as $headline ) {
4274                         $isTemplate = false;
4275                         $titleText = false;
4276                         $sectionIndex = false;
4277                         $numbering = '';
4278                         $markerMatches = array();
4279                         if ( preg_match( "/^$markerRegex/", $headline, $markerMatches ) ) {
4280                                 $serial = $markerMatches[1];
4281                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
4282                                 $isTemplate = ( $titleText != $baseTitleText );
4283                                 $headline = preg_replace( "/^$markerRegex\\s*/", "", $headline );
4284                         }
4285
4286                         if ( $toclevel ) {
4287                                 $prevlevel = $level;
4288                         }
4289                         $level = $matches[1][$headlineCount];
4290
4291                         if ( $level > $prevlevel ) {
4292                                 # Increase TOC level
4293                                 $toclevel++;
4294                                 $sublevelCount[$toclevel] = 0;
4295                                 if ( $toclevel < $wgMaxTocLevel ) {
4296                                         $prevtoclevel = $toclevel;
4297                                         $toc .= Linker::tocIndent();
4298                                         $numVisible++;
4299                                 }
4300                         } elseif ( $level < $prevlevel && $toclevel > 1 ) {
4301                                 # Decrease TOC level, find level to jump to
4302
4303                                 for ( $i = $toclevel; $i > 0; $i-- ) {
4304                                         if ( $levelCount[$i] == $level ) {
4305                                                 # Found last matching level
4306                                                 $toclevel = $i;
4307                                                 break;
4308                                         } elseif ( $levelCount[$i] < $level ) {
4309                                                 # Found first matching level below current level
4310                                                 $toclevel = $i + 1;
4311                                                 break;
4312                                         }
4313                                 }
4314                                 if ( $i == 0 ) {
4315                                         $toclevel = 1;
4316                                 }
4317                                 if ( $toclevel < $wgMaxTocLevel ) {
4318                                         if ( $prevtoclevel < $wgMaxTocLevel ) {
4319                                                 # Unindent only if the previous toc level was shown :p
4320                                                 $toc .= Linker::tocUnindent( $prevtoclevel - $toclevel );
4321                                                 $prevtoclevel = $toclevel;
4322                                         } else {
4323                                                 $toc .= Linker::tocLineEnd();
4324                                         }
4325                                 }
4326                         } else {
4327                                 # No change in level, end TOC line
4328                                 if ( $toclevel < $wgMaxTocLevel ) {
4329                                         $toc .= Linker::tocLineEnd();
4330                                 }
4331                         }
4332
4333                         $levelCount[$toclevel] = $level;
4334
4335                         # count number of headlines for each level
4336                         $sublevelCount[$toclevel]++;
4337                         $dot = 0;
4338                         for ( $i = 1; $i <= $toclevel; $i++ ) {
4339                                 if ( !empty( $sublevelCount[$i] ) ) {
4340                                         if ( $dot ) {
4341                                                 $numbering .= '.';
4342                                         }
4343                                         $numbering .= $this->getTargetLanguage()->formatNum( $sublevelCount[$i] );
4344                                         $dot = 1;
4345                                 }
4346                         }
4347
4348                         # The safe header is a version of the header text safe to use for links
4349
4350                         # Remove link placeholders by the link text.
4351                         #     <!--LINK number-->
4352                         # turns into
4353                         #     link text with suffix
4354                         # Do this before unstrip since link text can contain strip markers
4355                         $safeHeadline = $this->replaceLinkHoldersText( $headline );
4356
4357                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
4358                         $safeHeadline = $this->mStripState->unstripBoth( $safeHeadline );
4359
4360                         # Strip out HTML (first regex removes any tag not allowed)
4361                         # Allowed tags are:
4362                         # * <sup> and <sub> (bug 8393)
4363                         # * <i> (bug 26375)
4364                         # * <b> (r105284)
4365                         # * <span dir="rtl"> and <span dir="ltr"> (bug 35167)
4366                         #
4367                         # We strip any parameter from accepted tags (second regex), except dir="rtl|ltr" from <span>,
4368                         # to allow setting directionality in toc items.
4369                         $tocline = preg_replace(
4370                                 array( '#<(?!/?(span|sup|sub|i|b)(?: [^>]*)?>).*?' . '>#', '#<(/?(?:span(?: dir="(?:rtl|ltr)")?|sup|sub|i|b))(?: .*?)?' . '>#' ),
4371                                 array( '', '<$1>' ),
4372                                 $safeHeadline
4373                         );
4374                         $tocline = trim( $tocline );
4375
4376                         # For the anchor, strip out HTML-y stuff period
4377                         $safeHeadline = preg_replace( '/<.*?' . '>/', '', $safeHeadline );
4378                         $safeHeadline = Sanitizer::normalizeSectionNameWhitespace( $safeHeadline );
4379
4380                         # Save headline for section edit hint before it's escaped
4381                         $headlineHint = $safeHeadline;
4382
4383                         if ( $wgExperimentalHtmlIds ) {
4384                                 # For reverse compatibility, provide an id that's
4385                                 # HTML4-compatible, like we used to.
4386                                 #
4387                                 # It may be worth noting, academically, that it's possible for
4388                                 # the legacy anchor to conflict with a non-legacy headline
4389                                 # anchor on the page.  In this case likely the "correct" thing
4390                                 # would be to either drop the legacy anchors or make sure
4391                                 # they're numbered first.  However, this would require people
4392                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
4393                                 # manually, so let's not bother worrying about it.
4394                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
4395                                         array( 'noninitial', 'legacy' ) );
4396                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
4397
4398                                 if ( $legacyHeadline == $safeHeadline ) {
4399                                         # No reason to have both (in fact, we can't)
4400                                         $legacyHeadline = false;
4401                                 }
4402                         } else {
4403                                 $legacyHeadline = false;
4404                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
4405                                         'noninitial' );
4406                         }
4407
4408                         # HTML names must be case-insensitively unique (bug 10721).
4409                         # This does not apply to Unicode characters per
4410                         # http://dev.w3.org/html5/spec/infrastructure.html#case-sensitivity-and-string-comparison
4411                         # @todo FIXME: We may be changing them depending on the current locale.
4412                         $arrayKey = strtolower( $safeHeadline );
4413                         if ( $legacyHeadline === false ) {
4414                                 $legacyArrayKey = false;
4415                         } else {
4416                                 $legacyArrayKey = strtolower( $legacyHeadline );
4417                         }
4418
4419                         # count how many in assoc. array so we can track dupes in anchors
4420                         if ( isset( $refers[$arrayKey] ) ) {
4421                                 $refers[$arrayKey]++;
4422                         } else {
4423                                 $refers[$arrayKey] = 1;
4424                         }
4425                         if ( isset( $refers[$legacyArrayKey] ) ) {
4426                                 $refers[$legacyArrayKey]++;
4427                         } else {
4428                                 $refers[$legacyArrayKey] = 1;
4429                         }
4430
4431                         # Don't number the heading if it is the only one (looks silly)
4432                         if ( count( $matches[3] ) > 1 && $this->mOptions->getNumberHeadings() ) {
4433                                 # the two are different if the line contains a link
4434                                 $headline = Html::element( 'span', array( 'class' => 'mw-headline-number' ), $numbering ) . ' ' . $headline;
4435                         }
4436
4437                         # Create the anchor for linking from the TOC to the section
4438                         $anchor = $safeHeadline;
4439                         $legacyAnchor = $legacyHeadline;
4440                         if ( $refers[$arrayKey] > 1 ) {
4441                                 $anchor .= '_' . $refers[$arrayKey];
4442                         }
4443                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
4444                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
4445                         }
4446                         if ( $enoughToc && ( !isset( $wgMaxTocLevel ) || $toclevel < $wgMaxTocLevel ) ) {
4447                                 $toc .= Linker::tocLine( $anchor, $tocline,
4448                                         $numbering, $toclevel, ( $isTemplate ? false : $sectionIndex ) );
4449                         }
4450
4451                         # Add the section to the section tree
4452                         # Find the DOM node for this header
4453                         $noOffset = ( $isTemplate || $sectionIndex === false );
4454                         while ( $node && !$noOffset ) {
4455                                 if ( $node->getName() === 'h' ) {
4456                                         $bits = $node->splitHeading();
4457                                         if ( $bits['i'] == $sectionIndex ) {
4458                                                 break;
4459                                         }
4460                                 }
4461                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
4462                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
4463                                 $node = $node->getNextSibling();
4464                         }
4465                         $tocraw[] = array(
4466                                 'toclevel' => $toclevel,
4467                                 'level' => $level,
4468                                 'line' => $tocline,
4469                                 'number' => $numbering,
4470                                 'index' => ( $isTemplate ? 'T-' : '' ) . $sectionIndex,
4471                                 'fromtitle' => $titleText,
4472                                 'byteoffset' => ( $noOffset ? null : $byteOffset ),
4473                                 'anchor' => $anchor,
4474                         );
4475
4476                         # give headline the correct <h#> tag
4477                         if ( $maybeShowEditLink && $sectionIndex !== false ) {
4478                                 // Output edit section links as markers with styles that can be customized by skins
4479                                 if ( $isTemplate ) {
4480                                         # Put a T flag in the section identifier, to indicate to extractSections()
4481                                         # that sections inside <includeonly> should be counted.
4482                                         $editlinkArgs = array( $titleText, "T-$sectionIndex"/*, null */ );
4483                                 } else {
4484                                         $editlinkArgs = array( $this->mTitle->getPrefixedText(), $sectionIndex, $headlineHint );
4485                                 }
4486                                 // We use a bit of pesudo-xml for editsection markers. The language converter is run later on
4487                                 // Using a UNIQ style marker leads to the converter screwing up the tokens when it converts stuff
4488                                 // And trying to insert strip tags fails too. At this point all real inputted tags have already been escaped
4489                                 // so we don't have to worry about a user trying to input one of these markers directly.
4490                                 // We use a page and section attribute to stop the language converter from converting these important bits
4491                                 // of data, but put the headline hint inside a content block because the language converter is supposed to
4492                                 // be able to convert that piece of data.
4493                                 $editlink = '<mw:editsection page="' . htmlspecialchars( $editlinkArgs[0] );
4494                                 $editlink .= '" section="' . htmlspecialchars( $editlinkArgs[1] ) . '"';
4495                                 if ( isset( $editlinkArgs[2] ) ) {
4496                                         $editlink .= '>' . $editlinkArgs[2] . '</mw:editsection>';
4497                                 } else {
4498                                         $editlink .= '/>';
4499                                 }
4500                         } else {
4501                                 $editlink = '';
4502                         }
4503                         $head[$headlineCount] = Linker::makeHeadline( $level,
4504                                 $matches['attrib'][$headlineCount], $anchor, $headline,
4505                                 $editlink, $legacyAnchor );
4506
4507                         $headlineCount++;
4508                 }
4509
4510                 $this->setOutputType( $oldType );
4511
4512                 # Never ever show TOC if no headers
4513                 if ( $numVisible < 1 ) {
4514                         $enoughToc = false;
4515                 }
4516
4517                 if ( $enoughToc ) {
4518                         if ( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
4519                                 $toc .= Linker::tocUnindent( $prevtoclevel - 1 );
4520                         }
4521                         $toc = Linker::tocList( $toc, $this->mOptions->getUserLangObj() );
4522                         $this->mOutput->setTOCHTML( $toc );
4523                 }
4524
4525                 if ( $isMain ) {
4526                         $this->mOutput->setSections( $tocraw );
4527                 }
4528
4529                 # split up and insert constructed headlines
4530                 $blocks = preg_split( '/<H[1-6].*?' . '>[\s\S]*?<\/H[1-6]>/i', $text );
4531                 $i = 0;
4532
4533                 // build an array of document sections
4534                 $sections = array();
4535                 foreach ( $blocks as $block ) {
4536                         // $head is zero-based, sections aren't.
4537                         if ( empty( $head[$i - 1] ) ) {
4538                                 $sections[$i] = $block;
4539                         } else {
4540                                 $sections[$i] = $head[$i - 1] . $block;
4541                         }
4542
4543                         /**
4544                          * Send a hook, one per section.
4545                          * The idea here is to be able to make section-level DIVs, but to do so in a
4546                          * lower-impact, more correct way than r50769
4547                          *
4548                          * $this : caller
4549                          * $section : the section number
4550                          * &$sectionContent : ref to the content of the section
4551                          * $showEditLinks : boolean describing whether this section has an edit link
4552                          */
4553                         wfRunHooks( 'ParserSectionCreate', array( $this, $i, &$sections[$i], $showEditLink ) );
4554
4555                         $i++;
4556                 }
4557
4558                 if ( $enoughToc && $isMain && !$this->mForceTocPosition ) {
4559                         // append the TOC at the beginning
4560                         // Top anchor now in skin
4561                         $sections[0] = $sections[0] . $toc . "\n";
4562                 }
4563
4564                 $full .= join( '', $sections );
4565
4566                 if ( $this->mForceTocPosition ) {
4567                         return str_replace( '<!--MWTOC-->', $toc, $full );
4568                 } else {
4569                         return $full;
4570                 }
4571         }
4572
4573         /**
4574          * Transform wiki markup when saving a page by doing "\r\n" -> "\n"
4575          * conversion, substitting signatures, {{subst:}} templates, etc.
4576          *
4577          * @param string $text the text to transform
4578          * @param $title Title: the Title object for the current article
4579          * @param $user User: the User object describing the current user
4580          * @param $options ParserOptions: parsing options
4581          * @param $clearState Boolean: whether to clear the parser state first
4582          * @return String: the altered wiki markup
4583          */
4584         public function preSaveTransform( $text, Title $title, User $user, ParserOptions $options, $clearState = true ) {
4585                 $this->startParse( $title, $options, self::OT_WIKI, $clearState );
4586                 $this->setUser( $user );
4587
4588                 $pairs = array(
4589                         "\r\n" => "\n",
4590                 );
4591                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
4592                 if ( $options->getPreSaveTransform() ) {
4593                         $text = $this->pstPass2( $text, $user );
4594                 }
4595                 $text = $this->mStripState->unstripBoth( $text );
4596
4597                 $this->setUser( null ); #Reset
4598
4599                 return $text;
4600         }
4601
4602         /**
4603          * Pre-save transform helper function
4604          * @private
4605          *
4606          * @param $text string
4607          * @param $user User
4608          *
4609          * @return string
4610          */
4611         function pstPass2( $text, $user ) {
4612                 global $wgContLang;
4613
4614                 # Note: This is the timestamp saved as hardcoded wikitext to
4615                 # the database, we use $wgContLang here in order to give
4616                 # everyone the same signature and use the default one rather
4617                 # than the one selected in each user's preferences.
4618                 # (see also bug 12815)
4619                 $ts = $this->mOptions->getTimestamp();
4620                 $timestamp = MWTimestamp::getLocalInstance( $ts );
4621                 $ts = $timestamp->format( 'YmdHis' );
4622                 $tzMsg = $timestamp->format( 'T' );  # might vary on DST changeover!
4623
4624                 # Allow translation of timezones through wiki. format() can return
4625                 # whatever crap the system uses, localised or not, so we cannot
4626                 # ship premade translations.
4627                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4628                 $msg = wfMessage( $key )->inContentLanguage();
4629                 if ( $msg->exists() ) {
4630                         $tzMsg = $msg->text();
4631                 }
4632
4633                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4634
4635                 # Variable replacement
4636                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4637                 $text = $this->replaceVariables( $text );
4638
4639                 # This works almost by chance, as the replaceVariables are done before the getUserSig(),
4640                 # which may corrupt this parser instance via its wfMessage()->text() call-
4641
4642                 # Signatures
4643                 $sigText = $this->getUserSig( $user );
4644                 $text = strtr( $text, array(
4645                         '~~~~~' => $d,
4646                         '~~~~' => "$sigText $d",
4647                         '~~~' => $sigText
4648                 ) );
4649
4650                 # Context links ("pipe tricks"): [[|name]] and [[name (context)|]]
4651                 $tc = '[' . Title::legalChars() . ']';
4652                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
4653
4654                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\))\\|]]/";                   # [[ns:page (context)|]]
4655                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)( ?（$tc+）)\\|]]/";                           # [[ns:page（context）|]] (double-width brackets, added in r40257)
4656                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( ?\\($tc+\\)|)((?:, |，)$tc+|)\\|]]/";         # [[ns:page (context), context|]] (using either single or double-width comma)
4657                 $p2 = "/\[\[\\|($tc+)]]/";                                              # [[|page]] (reverse pipe trick: add context from page title)
4658
4659                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4660                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4661                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4662                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4663
4664                 $t = $this->mTitle->getText();
4665                 $m = array();
4666                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4667                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4668                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4669                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4670                 } else {
4671                         # if there's no context, don't bother duplicating the title
4672                         $text = preg_replace( $p2, '[[\\1]]', $text );
4673                 }
4674
4675                 # Trim trailing whitespace
4676                 $text = rtrim( $text );
4677
4678                 return $text;
4679         }
4680
4681         /**
4682          * Fetch the user's signature text, if any, and normalize to
4683          * validated, ready-to-insert wikitext.
4684          * If you have pre-fetched the nickname or the fancySig option, you can
4685          * specify them here to save a database query.
4686          * Do not reuse this parser instance after calling getUserSig(),
4687          * as it may have changed if it's the $wgParser.
4688          *
4689          * @param $user User
4690          * @param string|bool $nickname nickname to use or false to use user's default nickname
4691          * @param $fancySig Boolean|null whether the nicknname is the complete signature
4692          *                  or null to use default value
4693          * @return string
4694          */
4695         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4696                 global $wgMaxSigChars;
4697
4698                 $username = $user->getName();
4699
4700                 # If not given, retrieve from the user object.
4701                 if ( $nickname === false ) {
4702                         $nickname = $user->getOption( 'nickname' );
4703                 }
4704
4705                 if ( is_null( $fancySig ) ) {
4706                         $fancySig = $user->getBoolOption( 'fancysig' );
4707                 }
4708
4709                 $nickname = $nickname == null ? $username : $nickname;
4710
4711                 if ( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4712                         $nickname = $username;
4713                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4714                 } elseif ( $fancySig !== false ) {
4715                         # Sig. might contain markup; validate this
4716                         if ( $this->validateSig( $nickname ) !== false ) {
4717                                 # Validated; clean up (if needed) and return it
4718                                 return $this->cleanSig( $nickname, true );
4719                         } else {
4720                                 # Failed to validate; fall back to the default
4721                                 $nickname = $username;
4722                                 wfDebug( __METHOD__ . ": $username has bad XML tags in signature.\n" );
4723                         }
4724                 }
4725
4726                 # Make sure nickname doesnt get a sig in a sig
4727                 $nickname = self::cleanSigInSig( $nickname );
4728
4729                 # If we're still here, make it a link to the user page
4730                 $userText = wfEscapeWikiText( $username );
4731                 $nickText = wfEscapeWikiText( $nickname );
4732                 $msgName = $user->isAnon() ? 'signature-anon' : 'signature';
4733
4734                 return wfMessage( $msgName, $userText, $nickText )->inContentLanguage()->title( $this->getTitle() )->text();
4735         }
4736
4737         /**
4738          * Check that the user's signature contains no bad XML
4739          *
4740          * @param $text String
4741          * @return mixed An expanded string, or false if invalid.
4742          */
4743         function validateSig( $text ) {
4744                 return Xml::isWellFormedXmlFragment( $text ) ? $text : false;
4745         }
4746
4747         /**
4748          * Clean up signature text
4749          *
4750          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4751          * 2) Substitute all transclusions
4752          *
4753          * @param $text String
4754          * @param bool $parsing Whether we're cleaning (preferences save) or parsing
4755          * @return String: signature text
4756          */
4757         public function cleanSig( $text, $parsing = false ) {
4758                 if ( !$parsing ) {
4759                         global $wgTitle;
4760                         $this->startParse( $wgTitle, new ParserOptions, self::OT_PREPROCESS, true );
4761                 }
4762
4763                 # Option to disable this feature
4764                 if ( !$this->mOptions->getCleanSignatures() ) {
4765                         return $text;
4766                 }
4767
4768                 # @todo FIXME: Regex doesn't respect extension tags or nowiki
4769                 #  => Move this logic to braceSubstitution()
4770                 $substWord = MagicWord::get( 'subst' );
4771                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4772                 $substText = '{{' . $substWord->getSynonym( 0 );
4773
4774                 $text = preg_replace( $substRegex, $substText, $text );
4775                 $text = self::cleanSigInSig( $text );
4776                 $dom = $this->preprocessToDom( $text );
4777                 $frame = $this->getPreprocessor()->newFrame();
4778                 $text = $frame->expand( $dom );
4779
4780                 if ( !$parsing ) {
4781                         $text = $this->mStripState->unstripBoth( $text );
4782                 }
4783
4784                 return $text;
4785         }
4786
4787         /**
4788          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4789          *
4790          * @param $text String
4791          * @return String: signature text with /~{3,5}/ removed
4792          */
4793         public static function cleanSigInSig( $text ) {
4794                 $text = preg_replace( '/~{3,5}/', '', $text );
4795                 return $text;
4796         }
4797
4798         /**
4799          * Set up some variables which are usually set up in parse()
4800          * so that an external function can call some class members with confidence
4801          *
4802          * @param $title Title|null
4803          * @param $options ParserOptions
4804          * @param $outputType
4805          * @param $clearState bool
4806          */
4807         public function startExternalParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4808                 $this->startParse( $title, $options, $outputType, $clearState );
4809         }
4810
4811         /**
4812          * @param $title Title|null
4813          * @param $options ParserOptions
4814          * @param $outputType
4815          * @param $clearState bool
4816          */
4817         private function startParse( Title $title = null, ParserOptions $options, $outputType, $clearState = true ) {
4818                 $this->setTitle( $title );
4819                 $this->mOptions = $options;
4820                 $this->setOutputType( $outputType );
4821                 if ( $clearState ) {
4822                         $this->clearState();
4823                 }
4824         }
4825
4826         /**
4827          * Wrapper for preprocess()
4828          *
4829          * @param string $text the text to preprocess
4830          * @param $options ParserOptions: options
4831          * @param $title Title object or null to use $wgTitle
4832          * @return String
4833          */
4834         public function transformMsg( $text, $options, $title = null ) {
4835                 static $executing = false;
4836
4837                 # Guard against infinite recursion
4838                 if ( $executing ) {
4839                         return $text;
4840                 }
4841                 $executing = true;
4842
4843                 wfProfileIn( __METHOD__ );
4844                 if ( !$title ) {
4845                         global $wgTitle;
4846                         $title = $wgTitle;
4847                 }
4848
4849                 $text = $this->preprocess( $text, $title, $options );
4850
4851                 $executing = false;
4852                 wfProfileOut( __METHOD__ );
4853                 return $text;
4854         }
4855
4856         /**
4857          * Create an HTML-style tag, e.g. "<yourtag>special text</yourtag>"
4858          * The callback should have the following form:
4859          *    function myParserHook( $text, $params, $parser, $frame ) { ... }
4860          *
4861          * Transform and return $text. Use $parser for any required context, e.g. use
4862          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4863          *
4864          * Hooks may return extended information by returning an array, of which the
4865          * first numbered element (index 0) must be the return string, and all other
4866          * entries are extracted into local variables within an internal function
4867          * in the Parser class.
4868          *
4869          * This interface (introduced r61913) appears to be undocumented, but
4870          * 'markerName' is used by some core tag hooks to override which strip
4871          * array their results are placed in. **Use great caution if attempting
4872          * this interface, as it is not documented and injudicious use could smash
4873          * private variables.**
4874          *
4875          * @param $tag Mixed: the tag to use, e.g. 'hook' for "<hook>"
4876          * @param $callback Mixed: the callback function (and object) to use for the tag
4877          * @throws MWException
4878          * @return Mixed|null The old value of the mTagHooks array associated with the hook
4879          */
4880         public function setHook( $tag, $callback ) {
4881                 $tag = strtolower( $tag );
4882                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4883                         throw new MWException( "Invalid character {$m[0]} in setHook('$tag', ...) call" );
4884                 }
4885                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4886                 $this->mTagHooks[$tag] = $callback;
4887                 if ( !in_array( $tag, $this->mStripList ) ) {
4888                         $this->mStripList[] = $tag;
4889                 }
4890
4891                 return $oldVal;
4892         }
4893
4894         /**
4895          * As setHook(), but letting the contents be parsed.
4896          *
4897          * Transparent tag hooks are like regular XML-style tag hooks, except they
4898          * operate late in the transformation sequence, on HTML instead of wikitext.
4899          *
4900          * This is probably obsoleted by things dealing with parser frames?
4901          * The only extension currently using it is geoserver.
4902          *
4903          * @since 1.10
4904          * @todo better document or deprecate this
4905          *
4906          * @param $tag Mixed: the tag to use, e.g. 'hook' for "<hook>"
4907          * @param $callback Mixed: the callback function (and object) to use for the tag
4908          * @throws MWException
4909          * @return Mixed|null The old value of the mTagHooks array associated with the hook
4910          */
4911         function setTransparentTagHook( $tag, $callback ) {
4912                 $tag = strtolower( $tag );
4913                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
4914                         throw new MWException( "Invalid character {$m[0]} in setTransparentHook('$tag', ...) call" );
4915                 }
4916                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4917                 $this->mTransparentTagHooks[$tag] = $callback;
4918
4919                 return $oldVal;
4920         }
4921
4922         /**
4923          * Remove all tag hooks
4924          */
4925         function clearTagHooks() {
4926                 $this->mTagHooks = array();
4927                 $this->mFunctionTagHooks = array();
4928                 $this->mStripList = $this->mDefaultStripList;
4929         }
4930
4931         /**
4932          * Create a function, e.g. {{sum:1|2|3}}
4933          * The callback function should have the form:
4934          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4935          *
4936          * Or with SFH_OBJECT_ARGS:
4937          *    function myParserFunction( $parser, $frame, $args ) { ... }
4938          *
4939          * The callback may either return the text result of the function, or an array with the text
4940          * in element 0, and a number of flags in the other elements. The names of the flags are
4941          * specified in the keys. Valid flags are:
4942          *   found                     The text returned is valid, stop processing the template. This
4943          *                             is on by default.
4944          *   nowiki                    Wiki markup in the return value should be escaped
4945          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4946          *
4947          * @param string $id The magic word ID
4948          * @param $callback Mixed: the callback function (and object) to use
4949          * @param $flags Integer: a combination of the following flags:
4950          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4951          *
4952          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4953          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4954          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4955          *     the arguments, and to control the way they are expanded.
4956          *
4957          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4958          *     arguments, for instance:
4959          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4960          *
4961          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4962          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4963          *     working if/when this is changed.
4964          *
4965          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4966          *     expansion.
4967          *
4968          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4969          *     about the methods available in PPFrame and PPNode.
4970          *
4971          * @throws MWException
4972          * @return string|callback The old callback function for this name, if any
4973          */
4974         public function setFunctionHook( $id, $callback, $flags = 0 ) {
4975                 global $wgContLang;
4976
4977                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4978                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4979
4980                 # Add to function cache
4981                 $mw = MagicWord::get( $id );
4982                 if ( !$mw ) {
4983                         throw new MWException( __METHOD__ . '() expecting a magic word identifier.' );
4984                 }
4985
4986                 $synonyms = $mw->getSynonyms();
4987                 $sensitive = intval( $mw->isCaseSensitive() );
4988
4989                 foreach ( $synonyms as $syn ) {
4990                         # Case
4991                         if ( !$sensitive ) {
4992                                 $syn = $wgContLang->lc( $syn );
4993                         }
4994                         # Add leading hash
4995                         if ( !( $flags & SFH_NO_HASH ) ) {
4996                                 $syn = '#' . $syn;
4997                         }
4998                         # Remove trailing colon
4999                         if ( substr( $syn, -1, 1 ) === ':' ) {
5000                                 $syn = substr( $syn, 0, -1 );
5001                         }
5002                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
5003                 }
5004                 return $oldVal;
5005         }
5006
5007         /**
5008          * Get all registered function hook identifiers
5009          *
5010          * @return Array
5011          */
5012         function getFunctionHooks() {
5013                 return array_keys( $this->mFunctionHooks );
5014         }
5015
5016         /**
5017          * Create a tag function, e.g. "<test>some stuff</test>".
5018          * Unlike tag hooks, tag functions are parsed at preprocessor level.
5019          * Unlike parser functions, their content is not preprocessed.
5020          * @param $tag
5021          * @param $callback
5022          * @param $flags
5023          * @throws MWException
5024          * @return null
5025          */
5026         function setFunctionTagHook( $tag, $callback, $flags ) {
5027                 $tag = strtolower( $tag );
5028                 if ( preg_match( '/[<>\r\n]/', $tag, $m ) ) {
5029                         throw new MWException( "Invalid character {$m[0]} in setFunctionTagHook('$tag', ...) call" );
5030                 }
5031                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
5032                         $this->mFunctionTagHooks[$tag] : null;
5033                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
5034
5035                 if ( !in_array( $tag, $this->mStripList ) ) {
5036                         $this->mStripList[] = $tag;
5037                 }
5038
5039                 return $old;
5040         }
5041
5042         /**
5043          * @todo FIXME: Update documentation. makeLinkObj() is deprecated.
5044          * Replace "<!--LINK-->" link placeholders with actual links, in the buffer
5045          * Placeholders created in Skin::makeLinkObj()
5046          *
5047          * @param $text string
5048          * @param $options int
5049          *
5050          * @return array of link CSS classes, indexed by PDBK.
5051          */
5052         function replaceLinkHolders( &$text, $options = 0 ) {
5053                 return $this->mLinkHolders->replace( $text );
5054         }
5055
5056         /**
5057          * Replace "<!--LINK-->" link placeholders with plain text of links
5058          * (not HTML-formatted).
5059          *
5060          * @param $text String
5061          * @return String
5062          */
5063         function replaceLinkHoldersText( $text ) {
5064                 return $this->mLinkHolders->replaceText( $text );
5065         }
5066
5067         /**
5068          * Renders an image gallery from a text with one line per image.
5069          * text labels may be given by using |-style alternative text. E.g.
5070          *   Image:one.jpg|The number "1"
5071          *   Image:tree.jpg|A tree
5072          * given as text will return the HTML of a gallery with two images,
5073          * labeled 'The number "1"' and
5074          * 'A tree'.
5075          *
5076          * @param string $text
5077          * @param array $params
5078          * @return string HTML
5079          */
5080         function renderImageGallery( $text, $params ) {
5081                 wfProfileIn( __METHOD__ );
5082
5083                 $mode = false;
5084                 if ( isset( $params['mode'] ) ) {
5085                         $mode = $params['mode'];
5086                 }
5087
5088                 try {
5089                         $ig = ImageGalleryBase::factory( $mode );
5090                 } catch ( MWException $e ) {
5091                         // If invalid type set, fallback to default.
5092                         $ig = ImageGalleryBase::factory( false );
5093                 }
5094
5095                 $ig->setContextTitle( $this->mTitle );
5096                 $ig->setShowBytes( false );
5097                 $ig->setShowFilename( false );
5098                 $ig->setParser( $this );
5099                 $ig->setHideBadImages();
5100                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
5101
5102                 if ( isset( $params['showfilename'] ) ) {
5103                         $ig->setShowFilename( true );
5104                 } else {
5105                         $ig->setShowFilename( false );
5106                 }
5107                 if ( isset( $params['caption'] ) ) {
5108                         $caption = $params['caption'];
5109                         $caption = htmlspecialchars( $caption );
5110                         $caption = $this->replaceInternalLinks( $caption );
5111                         $ig->setCaptionHtml( $caption );
5112                 }
5113                 if ( isset( $params['perrow'] ) ) {
5114                         $ig->setPerRow( $params['perrow'] );
5115                 }
5116                 if ( isset( $params['widths'] ) ) {
5117                         $ig->setWidths( $params['widths'] );
5118                 }
5119                 if ( isset( $params['heights'] ) ) {
5120                         $ig->setHeights( $params['heights'] );
5121                 }
5122                 $ig->setAdditionalOptions( $params );
5123
5124                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
5125
5126                 $lines = StringUtils::explode( "\n", $text );
5127                 foreach ( $lines as $line ) {
5128                         # match lines like these:
5129                         # Image:someimage.jpg|This is some image
5130                         $matches = array();
5131                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
5132                         # Skip empty lines
5133                         if ( count( $matches ) == 0 ) {
5134                                 continue;
5135                         }
5136
5137                         if ( strpos( $matches[0], '%' ) !== false ) {
5138                                 $matches[1] = rawurldecode( $matches[1] );
5139                         }
5140                         $title = Title::newFromText( $matches[1], NS_FILE );
5141                         if ( is_null( $title ) ) {
5142                                 # Bogus title. Ignore these so we don't bomb out later.
5143                                 continue;
5144                         }
5145
5146                         # We need to get what handler the file uses, to figure out parameters.
5147                         # Note, a hook can overide the file name, and chose an entirely different
5148                         # file (which potentially could be of a different type and have different handler).
5149                         $options = array();
5150                         $descQuery = false;
5151                         wfRunHooks( 'BeforeParserFetchFileAndTitle',
5152                                 array( $this, $title, &$options, &$descQuery ) );
5153                         # Don't register it now, as ImageGallery does that later.
5154                         $file = $this->fetchFileNoRegister( $title, $options );
5155                         $handler = $file ? $file->getHandler() : false;
5156
5157                         wfProfileIn( __METHOD__ . '-getMagicWord' );
5158                         $paramMap = array(
5159                                 'img_alt' => 'gallery-internal-alt',
5160                                 'img_link' => 'gallery-internal-link',
5161                         );
5162                         if ( $handler ) {
5163                                 $paramMap = $paramMap + $handler->getParamMap();
5164                                 // We don't want people to specify per-image widths.
5165                                 // Additionally the width parameter would need special casing anyhow.
5166                                 unset( $paramMap['img_width'] );
5167                         }
5168
5169                         $mwArray = new MagicWordArray( array_keys( $paramMap ) );
5170                         wfProfileOut( __METHOD__ . '-getMagicWord' );
5171
5172                         $label = '';
5173                         $alt = '';
5174                         $link = '';
5175                         $handlerOptions = array();
5176                         if ( isset( $matches[3] ) ) {
5177                                 // look for an |alt= definition while trying not to break existing
5178                                 // captions with multiple pipes (|) in it, until a more sensible grammar
5179                                 // is defined for images in galleries
5180
5181                                 // FIXME: Doing recursiveTagParse at this stage, and the trim before
5182                                 // splitting on '|' is a bit odd, and different from makeImage.
5183                                 $matches[3] = $this->recursiveTagParse( trim( $matches[3] ) );
5184                                 $parameterMatches = StringUtils::explode( '|', $matches[3] );
5185
5186                                 foreach ( $parameterMatches as $parameterMatch ) {
5187                                         list( $magicName, $match ) = $mwArray->matchVariableStartToEnd( $parameterMatch );
5188                                         if ( $magicName ) {
5189                                                 $paramName = $paramMap[$magicName];
5190
5191                                                 switch ( $paramName ) {
5192                                                 case 'gallery-internal-alt':
5193                                                         $alt = $this->stripAltText( $match, false );
5194                                                         break;
5195                                                 case 'gallery-internal-link':
5196                                                         $linkValue = strip_tags( $this->replaceLinkHoldersText( $match ) );
5197                                                         $chars = self::EXT_LINK_URL_CLASS;
5198                                                         $prots = $this->mUrlProtocols;
5199                                                         //check to see if link matches an absolute url, if not then it must be a wiki link.
5200                                                         if ( preg_match( "/^($prots)$chars+$/u", $linkValue ) ) {
5201                                                                 $link = $linkValue;
5202                                                         } else {
5203                                                                 $localLinkTitle = Title::newFromText( $linkValue );
5204                                                                 if ( $localLinkTitle !== null ) {
5205                                                                         $link = $localLinkTitle->getLocalURL();
5206                                                                 }
5207                                                         }
5208                                                         break;
5209                                                 default:
5210                                                         // Must be a handler specific parameter.
5211                                                         if ( $handler->validateParam( $paramName, $match ) ) {
5212                                                                 $handlerOptions[$paramName] = $match;
5213                                                         } else {
5214                                                                 // Guess not. Append it to the caption.
5215                                                                 wfDebug( "$parameterMatch failed parameter validation" );
5216                                                                 $label .= '|' . $parameterMatch;
5217                                                         }
5218                                                 }
5219
5220                                         } else {
5221                                                 // concatenate all other pipes
5222                                                 $label .= '|' . $parameterMatch;
5223                                         }
5224                                 }
5225                                 // remove the first pipe
5226                                 $label = substr( $label, 1 );
5227                         }
5228
5229                         $ig->add( $title, $label, $alt, $link, $handlerOptions );
5230                 }
5231                 $html = $ig->toHTML();
5232                 wfProfileOut( __METHOD__ );
5233                 return $html;
5234         }
5235
5236         /**
5237          * @param $handler
5238          * @return array
5239          */
5240         function getImageParams( $handler ) {
5241                 if ( $handler ) {
5242                         $handlerClass = get_class( $handler );
5243                 } else {
5244                         $handlerClass = '';
5245                 }
5246                 if ( !isset( $this->mImageParams[$handlerClass] ) ) {
5247                         # Initialise static lists
5248                         static $internalParamNames = array(
5249                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
5250                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
5251                                         'bottom', 'text-bottom' ),
5252                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
5253                                         'upright', 'border', 'link', 'alt', 'class' ),
5254                         );
5255                         static $internalParamMap;
5256                         if ( !$internalParamMap ) {
5257                                 $internalParamMap = array();
5258                                 foreach ( $internalParamNames as $type => $names ) {
5259                                         foreach ( $names as $name ) {
5260                                                 $magicName = str_replace( '-', '_', "img_$name" );
5261                                                 $internalParamMap[$magicName] = array( $type, $name );
5262                                         }
5263                                 }
5264                         }
5265
5266                         # Add handler params
5267                         $paramMap = $internalParamMap;
5268                         if ( $handler ) {
5269                                 $handlerParamMap = $handler->getParamMap();
5270                                 foreach ( $handlerParamMap as $magic => $paramName ) {
5271                                         $paramMap[$magic] = array( 'handler', $paramName );
5272                                 }
5273                         }
5274                         $this->mImageParams[$handlerClass] = $paramMap;
5275                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
5276                 }
5277                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
5278         }
5279
5280         /**
5281          * Parse image options text and use it to make an image
5282          *
5283          * @param $title Title
5284          * @param $options String
5285          * @param $holders LinkHolderArray|bool
5286          * @return string HTML
5287          */
5288         function makeImage( $title, $options, $holders = false ) {
5289                 # Check if the options text is of the form "options|alt text"
5290                 # Options are:
5291                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
5292                 #  * left       no resizing, just left align. label is used for alt= only
5293                 #  * right      same, but right aligned
5294                 #  * none       same, but not aligned
5295                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
5296                 #  * center     center the image
5297                 #  * frame      Keep original image size, no magnify-button.
5298                 #  * framed     Same as "frame"
5299                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
5300                 #  * upright    reduce width for upright images, rounded to full __0 px
5301                 #  * border     draw a 1px border around the image
5302                 #  * alt        Text for HTML alt attribute (defaults to empty)
5303                 #  * class      Set a class for img node
5304                 #  * link       Set the target of the image link. Can be external, interwiki, or local
5305                 # vertical-align values (no % or length right now):
5306                 #  * baseline
5307                 #  * sub
5308                 #  * super
5309                 #  * top
5310                 #  * text-top
5311                 #  * middle
5312                 #  * bottom
5313                 #  * text-bottom
5314
5315                 $parts = StringUtils::explode( "|", $options );
5316
5317                 # Give extensions a chance to select the file revision for us
5318                 $options = array();
5319                 $descQuery = false;
5320                 wfRunHooks( 'BeforeParserFetchFileAndTitle',
5321                         array( $this, $title, &$options, &$descQuery ) );
5322                 # Fetch and register the file (file title may be different via hooks)
5323                 list( $file, $title ) = $this->fetchFileAndTitle( $title, $options );
5324
5325                 # Get parameter map
5326                 $handler = $file ? $file->getHandler() : false;
5327
5328                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
5329
5330                 if ( !$file ) {
5331                         $this->addTrackingCategory( 'broken-file-category' );
5332                 }
5333
5334                 # Process the input parameters
5335                 $caption = '';
5336                 $params = array( 'frame' => array(), 'handler' => array(),
5337                         'horizAlign' => array(), 'vertAlign' => array() );
5338                 foreach ( $parts as $part ) {
5339                         $part = trim( $part );
5340                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
5341                         $validated = false;
5342                         if ( isset( $paramMap[$magicName] ) ) {
5343                                 list( $type, $paramName ) = $paramMap[$magicName];
5344
5345                                 # Special case; width and height come in one variable together
5346                                 if ( $type === 'handler' && $paramName === 'width' ) {
5347                                         $parsedWidthParam = $this->parseWidthParam( $value );
5348                                         if ( isset( $parsedWidthParam['width'] ) ) {
5349                                                 $width = $parsedWidthParam['width'];
5350                                                 if ( $handler->validateParam( 'width', $width ) ) {
5351                                                         $params[$type]['width'] = $width;
5352                                                         $validated = true;
5353                                                 }
5354                                         }
5355                                         if ( isset( $parsedWidthParam['height'] ) ) {
5356                                                 $height = $parsedWidthParam['height'];
5357                                                 if ( $handler->validateParam( 'height', $height ) ) {
5358                                                         $params[$type]['height'] = $height;
5359                                                         $validated = true;
5360                                                 }
5361                                         }
5362                                         # else no validation -- bug 13436
5363                                 } else {
5364                                         if ( $type === 'handler' ) {
5365                                                 # Validate handler parameter
5366                                                 $validated = $handler->validateParam( $paramName, $value );
5367                                         } else {
5368                                                 # Validate internal parameters
5369                                                 switch ( $paramName ) {
5370                                                 case 'manualthumb':
5371                                                 case 'alt':
5372                                                 case 'class':
5373                                                         # @todo FIXME: Possibly check validity here for
5374                                                         # manualthumb? downstream behavior seems odd with
5375                                                         # missing manual thumbs.
5376                                                         $validated = true;
5377                                                         $value = $this->stripAltText( $value, $holders );
5378                                                         break;
5379                                                 case 'link':
5380                                                         $chars = self::EXT_LINK_URL_CLASS;
5381                                                         $prots = $this->mUrlProtocols;
5382                                                         if ( $value === '' ) {
5383                                                                 $paramName = 'no-link';
5384                                                                 $value = true;
5385                                                                 $validated = true;
5386                                                         } elseif ( preg_match( "/^(?i)$prots/", $value ) ) {
5387                                                                 if ( preg_match( "/^((?i)$prots)$chars+$/u", $value, $m ) ) {
5388                                                                         $paramName = 'link-url';
5389                                                                         $this->mOutput->addExternalLink( $value );
5390                                                                         if ( $this->mOptions->getExternalLinkTarget() ) {
5391                                                                                 $params[$type]['link-target'] = $this->mOptions->getExternalLinkTarget();
5392                                                                         }
5393                                                                         $validated = true;
5394                                                                 }
5395                                                         } else {
5396                                                                 $linkTitle = Title::newFromText( $value );
5397                                                                 if ( $linkTitle ) {
5398                                                                         $paramName = 'link-title';
5399                                                                         $value = $linkTitle;
5400                                                                         $this->mOutput->addLink( $linkTitle );
5401                                                                         $validated = true;
5402                                                                 }
5403                                                         }
5404                                                         break;
5405                                                 default:
5406                                                         # Most other things appear to be empty or numeric...
5407                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
5408                                                 }
5409                                         }
5410
5411                                         if ( $validated ) {
5412                                                 $params[$type][$paramName] = $value;
5413                                         }
5414                                 }
5415                         }
5416                         if ( !$validated ) {
5417                                 $caption = $part;
5418                         }
5419                 }
5420
5421                 # Process alignment parameters
5422                 if ( $params['horizAlign'] ) {
5423                         $params['frame']['align'] = key( $params['horizAlign'] );
5424                 }
5425                 if ( $params['vertAlign'] ) {
5426                         $params['frame']['valign'] = key( $params['vertAlign'] );
5427                 }
5428
5429                 $params['frame']['caption'] = $caption;
5430
5431                 # Will the image be presented in a frame, with the caption below?
5432                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
5433                                                  isset( $params['frame']['framed'] ) ||
5434                                                  isset( $params['frame']['thumbnail'] ) ||
5435                                                  isset( $params['frame']['manualthumb'] );
5436
5437                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
5438                 # came to also set the caption, ordinary text after the image -- which
5439                 # makes no sense, because that just repeats the text multiple times in
5440                 # screen readers.  It *also* came to set the title attribute.
5441                 #
5442                 # Now that we have an alt attribute, we should not set the alt text to
5443                 # equal the caption: that's worse than useless, it just repeats the
5444                 # text.  This is the framed/thumbnail case.  If there's no caption, we
5445                 # use the unnamed parameter for alt text as well, just for the time be-
5446                 # ing, if the unnamed param is set and the alt param is not.
5447                 #
5448                 # For the future, we need to figure out if we want to tweak this more,
5449                 # e.g., introducing a title= parameter for the title; ignoring the un-
5450                 # named parameter entirely for images without a caption; adding an ex-
5451                 # plicit caption= parameter and preserving the old magic unnamed para-
5452                 # meter for BC; ...
5453                 if ( $imageIsFramed ) { # Framed image
5454                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
5455                                 # No caption or alt text, add the filename as the alt text so
5456                                 # that screen readers at least get some description of the image
5457                                 $params['frame']['alt'] = $title->getText();
5458                         }
5459                         # Do not set $params['frame']['title'] because tooltips don't make sense
5460                         # for framed images
5461                 } else { # Inline image
5462                         if ( !isset( $params['frame']['alt'] ) ) {
5463                                 # No alt text, use the "caption" for the alt text
5464                                 if ( $caption !== '' ) {
5465                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
5466                                 } else {
5467                                         # No caption, fall back to using the filename for the
5468                                         # alt text
5469                                         $params['frame']['alt'] = $title->getText();
5470                                 }
5471                         }
5472                         # Use the "caption" for the tooltip text
5473                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
5474                 }
5475
5476                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params, $this ) );
5477
5478                 # Linker does the rest
5479                 $time = isset( $options['time'] ) ? $options['time'] : false;
5480                 $ret = Linker::makeImageLink( $this, $title, $file, $params['frame'], $params['handler'],
5481                         $time, $descQuery, $this->mOptions->getThumbSize() );
5482
5483                 # Give the handler a chance to modify the parser object
5484                 if ( $handler ) {
5485                         $handler->parserTransformHook( $this, $file );
5486                 }
5487
5488                 return $ret;
5489         }
5490
5491         /**
5492          * @param $caption
5493          * @param $holders LinkHolderArray
5494          * @return mixed|String
5495          */
5496         protected function stripAltText( $caption, $holders ) {
5497                 # Strip bad stuff out of the title (tooltip).  We can't just use
5498                 # replaceLinkHoldersText() here, because if this function is called
5499                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
5500                 if ( $holders ) {
5501                         $tooltip = $holders->replaceText( $caption );
5502                 } else {
5503                         $tooltip = $this->replaceLinkHoldersText( $caption );
5504                 }
5505
5506                 # make sure there are no placeholders in thumbnail attributes
5507                 # that are later expanded to html- so expand them now and
5508                 # remove the tags
5509                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
5510                 $tooltip = Sanitizer::stripAllTags( $tooltip );
5511
5512                 return $tooltip;
5513         }
5514
5515         /**
5516          * Set a flag in the output object indicating that the content is dynamic and
5517          * shouldn't be cached.
5518          */
5519         function disableCache() {
5520                 wfDebug( "Parser output marked as uncacheable.\n" );
5521                 if ( !$this->mOutput ) {
5522                         throw new MWException( __METHOD__ .
5523                                 " can only be called when actually parsing something" );
5524                 }
5525                 $this->mOutput->setCacheTime( -1 ); // old style, for compatibility
5526                 $this->mOutput->updateCacheExpiry( 0 ); // new style, for consistency
5527         }
5528
5529         /**
5530          * Callback from the Sanitizer for expanding items found in HTML attribute
5531          * values, so they can be safely tested and escaped.
5532          *
5533          * @param $text String
5534          * @param $frame PPFrame
5535          * @return String
5536          */
5537         function attributeStripCallback( &$text, $frame = false ) {
5538                 $text = $this->replaceVariables( $text, $frame );
5539                 $text = $this->mStripState->unstripBoth( $text );
5540                 return $text;
5541         }
5542
5543         /**
5544          * Accessor
5545          *
5546          * @return array
5547          */
5548         function getTags() {
5549                 return array_merge( array_keys( $this->mTransparentTagHooks ), array_keys( $this->mTagHooks ), array_keys( $this->mFunctionTagHooks ) );
5550         }
5551
5552         /**
5553          * Replace transparent tags in $text with the values given by the callbacks.
5554          *
5555          * Transparent tag hooks are like regular XML-style tag hooks, except they
5556          * operate late in the transformation sequence, on HTML instead of wikitext.
5557          *
5558          * @param $text string
5559          *
5560          * @return string
5561          */
5562         function replaceTransparentTags( $text ) {
5563                 $matches = array();
5564                 $elements = array_keys( $this->mTransparentTagHooks );
5565                 $text = self::extractTagsAndParams( $elements, $text, $matches, $this->mUniqPrefix );
5566                 $replacements = array();
5567
5568                 foreach ( $matches as $marker => $data ) {
5569                         list( $element, $content, $params, $tag ) = $data;
5570                         $tagName = strtolower( $element );
5571                         if ( isset( $this->mTransparentTagHooks[$tagName] ) ) {
5572                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName], array( $content, $params, $this ) );
5573                         } else {
5574                                 $output = $tag;
5575                         }
5576                         $replacements[$marker] = $output;
5577                 }
5578                 return strtr( $text, $replacements );
5579         }
5580
5581         /**
5582          * Break wikitext input into sections, and either pull or replace
5583          * some particular section's text.
5584          *
5585          * External callers should use the getSection and replaceSection methods.
5586          *
5587          * @param string $text Page wikitext
5588          * @param string $section a section identifier string of the form:
5589          *   "<flag1> - <flag2> - ... - <section number>"
5590          *
5591          * Currently the only recognised flag is "T", which means the target section number
5592          * was derived during a template inclusion parse, in other words this is a template
5593          * section edit link. If no flags are given, it was an ordinary section edit link.
5594          * This flag is required to avoid a section numbering mismatch when a section is
5595          * enclosed by "<includeonly>" (bug 6563).
5596          *
5597          * The section number 0 pulls the text before the first heading; other numbers will
5598          * pull the given section along with its lower-level subsections. If the section is
5599          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
5600          *
5601          * Section 0 is always considered to exist, even if it only contains the empty
5602          * string. If $text is the empty string and section 0 is replaced, $newText is
5603          * returned.
5604          *
5605          * @param string $mode one of "get" or "replace"
5606          * @param string $newText replacement text for section data.
5607          * @return String: for "get", the extracted section text.
5608          *                 for "replace", the whole page with the section replaced.
5609          */
5610         private function extractSections( $text, $section, $mode, $newText = '' ) {
5611                 global $wgTitle; # not generally used but removes an ugly failure mode
5612                 $this->startParse( $wgTitle, new ParserOptions, self::OT_PLAIN, true );
5613                 $outText = '';
5614                 $frame = $this->getPreprocessor()->newFrame();
5615
5616                 # Process section extraction flags
5617                 $flags = 0;
5618                 $sectionParts = explode( '-', $section );
5619                 $sectionIndex = array_pop( $sectionParts );
5620                 foreach ( $sectionParts as $part ) {
5621                         if ( $part === 'T' ) {
5622                                 $flags |= self::PTD_FOR_INCLUSION;
5623                         }
5624                 }
5625
5626                 # Check for empty input
5627                 if ( strval( $text ) === '' ) {
5628                         # Only sections 0 and T-0 exist in an empty document
5629                         if ( $sectionIndex == 0 ) {
5630                                 if ( $mode === 'get' ) {
5631                                         return '';
5632                                 } else {
5633                                         return $newText;
5634                                 }
5635                         } else {
5636                                 if ( $mode === 'get' ) {
5637                                         return $newText;
5638                                 } else {
5639                                         return $text;
5640                                 }
5641                         }
5642                 }
5643
5644                 # Preprocess the text
5645                 $root = $this->preprocessToDom( $text, $flags );
5646
5647                 # <h> nodes indicate section breaks
5648                 # They can only occur at the top level, so we can find them by iterating the root's children
5649                 $node = $root->getFirstChild();
5650
5651                 # Find the target section
5652                 if ( $sectionIndex == 0 ) {
5653                         # Section zero doesn't nest, level=big
5654                         $targetLevel = 1000;
5655                 } else {
5656                         while ( $node ) {
5657                                 if ( $node->getName() === 'h' ) {
5658                                         $bits = $node->splitHeading();
5659                                         if ( $bits['i'] == $sectionIndex ) {
5660                                                 $targetLevel = $bits['level'];
5661                                                 break;
5662                                         }
5663                                 }
5664                                 if ( $mode === 'replace' ) {
5665                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5666                                 }
5667                                 $node = $node->getNextSibling();
5668                         }
5669                 }
5670
5671                 if ( !$node ) {
5672                         # Not found
5673                         if ( $mode === 'get' ) {
5674                                 return $newText;
5675                         } else {
5676                                 return $text;
5677                         }
5678                 }
5679
5680                 # Find the end of the section, including nested sections
5681                 do {
5682                         if ( $node->getName() === 'h' ) {
5683                                 $bits = $node->splitHeading();
5684                                 $curLevel = $bits['level'];
5685                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
5686                                         break;
5687                                 }
5688                         }
5689                         if ( $mode === 'get' ) {
5690                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5691                         }
5692                         $node = $node->getNextSibling();
5693                 } while ( $node );
5694
5695                 # Write out the remainder (in replace mode only)
5696                 if ( $mode === 'replace' ) {
5697                         # Output the replacement text
5698                         # Add two newlines on -- trailing whitespace in $newText is conventionally
5699                         # stripped by the editor, so we need both newlines to restore the paragraph gap
5700                         # Only add trailing whitespace if there is newText
5701                         if ( $newText != "" ) {
5702                                 $outText .= $newText . "\n\n";
5703                         }
5704
5705                         while ( $node ) {
5706                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
5707                                 $node = $node->getNextSibling();
5708                         }
5709                 }
5710
5711                 if ( is_string( $outText ) ) {
5712                         # Re-insert stripped tags
5713                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
5714                 }
5715
5716                 return $outText;
5717         }
5718
5719         /**
5720          * This function returns the text of a section, specified by a number ($section).
5721          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
5722          * the first section before any such heading (section 0).
5723          *
5724          * If a section contains subsections, these are also returned.
5725          *
5726          * @param string $text text to look in
5727          * @param string $section section identifier
5728          * @param string $deftext default to return if section is not found
5729          * @return string text of the requested section
5730          */
5731         public function getSection( $text, $section, $deftext = '' ) {
5732                 return $this->extractSections( $text, $section, "get", $deftext );
5733         }
5734
5735         /**
5736          * This function returns $oldtext after the content of the section
5737          * specified by $section has been replaced with $text. If the target
5738          * section does not exist, $oldtext is returned unchanged.
5739          *
5740          * @param string $oldtext former text of the article
5741          * @param int $section section identifier
5742          * @param string $text replacing text
5743          * @return String: modified text
5744          */
5745         public function replaceSection( $oldtext, $section, $text ) {
5746                 return $this->extractSections( $oldtext, $section, "replace", $text );
5747         }
5748
5749         /**
5750          * Get the ID of the revision we are parsing
5751          *
5752          * @return Mixed: integer or null
5753          */
5754         function getRevisionId() {
5755                 return $this->mRevisionId;
5756         }
5757
5758         /**
5759          * Get the revision object for $this->mRevisionId
5760          *
5761          * @return Revision|null either a Revision object or null
5762          */
5763         protected function getRevisionObject() {
5764                 if ( !is_null( $this->mRevisionObject ) ) {
5765                         return $this->mRevisionObject;
5766                 }
5767                 if ( is_null( $this->mRevisionId ) ) {
5768                         return null;
5769                 }
5770
5771                 $this->mRevisionObject = Revision::newFromId( $this->mRevisionId );
5772                 return $this->mRevisionObject;
5773         }
5774
5775         /**
5776          * Get the timestamp associated with the current revision, adjusted for
5777          * the default server-local timestamp
5778          */
5779         function getRevisionTimestamp() {
5780                 if ( is_null( $this->mRevisionTimestamp ) ) {
5781                         wfProfileIn( __METHOD__ );
5782
5783                         global $wgContLang;
5784
5785                         $revObject = $this->getRevisionObject();
5786                         $timestamp = $revObject ? $revObject->getTimestamp() : wfTimestampNow();
5787
5788                         # The cryptic '' timezone parameter tells to use the site-default
5789                         # timezone offset instead of the user settings.
5790                         #
5791                         # Since this value will be saved into the parser cache, served
5792                         # to other users, and potentially even used inside links and such,
5793                         # it needs to be consistent for all visitors.
5794                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
5795
5796                         wfProfileOut( __METHOD__ );
5797                 }
5798                 return $this->mRevisionTimestamp;
5799         }
5800
5801         /**
5802          * Get the name of the user that edited the last revision
5803          *
5804          * @return String: user name
5805          */
5806         function getRevisionUser() {
5807                 if ( is_null( $this->mRevisionUser ) ) {
5808                         $revObject = $this->getRevisionObject();
5809
5810                         # if this template is subst: the revision id will be blank,
5811                         # so just use the current user's name
5812                         if ( $revObject ) {
5813                                 $this->mRevisionUser = $revObject->getUserText();
5814                         } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5815                                 $this->mRevisionUser = $this->getUser()->getName();
5816                         }
5817                 }
5818                 return $this->mRevisionUser;
5819         }
5820
5821         /**
5822          * Get the size of the revision
5823          *
5824          * @return int|null revision size
5825          */
5826         function getRevisionSize() {
5827                 if ( is_null( $this->mRevisionSize ) ) {
5828                         $revObject = $this->getRevisionObject();
5829
5830                         # if this variable is subst: the revision id will be blank,
5831                         # so just use the parser input size, because the own substituation
5832                         # will change the size.
5833                         if ( $revObject ) {
5834                                 $this->mRevisionSize = $revObject->getSize();
5835                         } elseif ( $this->ot['wiki'] || $this->mOptions->getIsPreview() ) {
5836                                 $this->mRevisionSize = $this->mInputSize;
5837                         }
5838                 }
5839                 return $this->mRevisionSize;
5840         }
5841
5842         /**
5843          * Mutator for $mDefaultSort
5844          *
5845          * @param string $sort New value
5846          */
5847         public function setDefaultSort( $sort ) {
5848                 $this->mDefaultSort = $sort;
5849                 $this->mOutput->setProperty( 'defaultsort', $sort );
5850         }
5851
5852         /**
5853          * Accessor for $mDefaultSort
5854          * Will use the empty string if none is set.
5855          *
5856          * This value is treated as a prefix, so the
5857          * empty string is equivalent to sorting by
5858          * page name.
5859          *
5860          * @return string
5861          */
5862         public function getDefaultSort() {
5863                 if ( $this->mDefaultSort !== false ) {
5864                         return $this->mDefaultSort;
5865                 } else {
5866                         return '';
5867                 }
5868         }
5869
5870         /**
5871          * Accessor for $mDefaultSort
5872          * Unlike getDefaultSort(), will return false if none is set
5873          *
5874          * @return string or false
5875          */
5876         public function getCustomDefaultSort() {
5877                 return $this->mDefaultSort;
5878         }
5879
5880         /**
5881          * Try to guess the section anchor name based on a wikitext fragment
5882          * presumably extracted from a heading, for example "Header" from
5883          * "== Header ==".
5884          *
5885          * @param $text string
5886          *
5887          * @return string
5888          */
5889         public function guessSectionNameFromWikiText( $text ) {
5890                 # Strip out wikitext links(they break the anchor)
5891                 $text = $this->stripSectionName( $text );
5892                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5893                 return '#' . Sanitizer::escapeId( $text, 'noninitial' );
5894         }
5895
5896         /**
5897          * Same as guessSectionNameFromWikiText(), but produces legacy anchors
5898          * instead.  For use in redirects, since IE6 interprets Redirect: headers
5899          * as something other than UTF-8 (apparently?), resulting in breakage.
5900          *
5901          * @param string $text The section name
5902          * @return string An anchor
5903          */
5904         public function guessLegacySectionNameFromWikiText( $text ) {
5905                 # Strip out wikitext links(they break the anchor)
5906                 $text = $this->stripSectionName( $text );
5907                 $text = Sanitizer::normalizeSectionNameWhitespace( $text );
5908                 return '#' . Sanitizer::escapeId( $text, array( 'noninitial', 'legacy' ) );
5909         }
5910
5911         /**
5912          * Strips a text string of wikitext for use in a section anchor
5913          *
5914          * Accepts a text string and then removes all wikitext from the
5915          * string and leaves only the resultant text (i.e. the result of
5916          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5917          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5918          * to create valid section anchors by mimicing the output of the
5919          * parser when headings are parsed.
5920          *
5921          * @param string $text text string to be stripped of wikitext
5922          * for use in a Section anchor
5923          * @return string Filtered text string
5924          */
5925         public function stripSectionName( $text ) {
5926                 # Strip internal link markup
5927                 $text = preg_replace( '/\[\[:?([^[|]+)\|([^[]+)\]\]/', '$2', $text );
5928                 $text = preg_replace( '/\[\[:?([^[]+)\|?\]\]/', '$1', $text );
5929
5930                 # Strip external link markup
5931                 # @todo FIXME: Not tolerant to blank link text
5932                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5933                 # on how many empty links there are on the page - need to figure that out.
5934                 $text = preg_replace( '/\[(?i:' . $this->mUrlProtocols . ')([^ ]+?) ([^[]+)\]/', '$2', $text );
5935
5936                 # Parse wikitext quotes (italics & bold)
5937                 $text = $this->doQuotes( $text );
5938
5939                 # Strip HTML tags
5940                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5941                 return $text;
5942         }
5943
5944         /**
5945          * strip/replaceVariables/unstrip for preprocessor regression testing
5946          *
5947          * @param $text string
5948          * @param $title Title
5949          * @param $options ParserOptions
5950          * @param $outputType int
5951          *
5952          * @return string
5953          */
5954         function testSrvus( $text, Title $title, ParserOptions $options, $outputType = self::OT_HTML ) {
5955                 $this->startParse( $title, $options, $outputType, true );
5956
5957                 $text = $this->replaceVariables( $text );
5958                 $text = $this->mStripState->unstripBoth( $text );
5959                 $text = Sanitizer::removeHTMLtags( $text );
5960                 return $text;
5961         }
5962
5963         /**
5964          * @param $text string
5965          * @param $title Title
5966          * @param $options ParserOptions
5967          * @return string
5968          */
5969         function testPst( $text, Title $title, ParserOptions $options ) {
5970                 return $this->preSaveTransform( $text, $title, $options->getUser(), $options );
5971         }
5972
5973         /**
5974          * @param $text
5975          * @param $title Title
5976          * @param $options ParserOptions
5977          * @return string
5978          */
5979         function testPreprocess( $text, Title $title, ParserOptions $options ) {
5980                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5981         }
5982
5983         /**
5984          * Call a callback function on all regions of the given text that are not
5985          * inside strip markers, and replace those regions with the return value
5986          * of the callback. For example, with input:
5987          *
5988          *  aaa<MARKER>bbb
5989          *
5990          * This will call the callback function twice, with 'aaa' and 'bbb'. Those
5991          * two strings will be replaced with the value returned by the callback in
5992          * each case.
5993          *
5994          * @param $s string
5995          * @param $callback
5996          *
5997          * @return string
5998          */
5999         function markerSkipCallback( $s, $callback ) {
6000                 $i = 0;
6001                 $out = '';
6002                 while ( $i < strlen( $s ) ) {
6003                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
6004                         if ( $markerStart === false ) {
6005                                 $out .= call_user_func( $callback, substr( $s, $i ) );
6006                                 break;
6007                         } else {
6008                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
6009                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
6010                                 if ( $markerEnd === false ) {
6011                                         $out .= substr( $s, $markerStart );
6012                                         break;
6013                                 } else {
6014                                         $markerEnd += strlen( self::MARKER_SUFFIX );
6015                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
6016                                         $i = $markerEnd;
6017                                 }
6018                         }
6019                 }
6020                 return $out;
6021         }
6022
6023         /**
6024          * Remove any strip markers found in the given text.
6025          *
6026          * @param $text Input string
6027          * @return string
6028          */
6029         function killMarkers( $text ) {
6030                 return $this->mStripState->killMarkers( $text );
6031         }
6032
6033         /**
6034          * Save the parser state required to convert the given half-parsed text to
6035          * HTML. "Half-parsed" in this context means the output of
6036          * recursiveTagParse() or internalParse(). This output has strip markers
6037          * from replaceVariables (extensionSubstitution() etc.), and link
6038          * placeholders from replaceLinkHolders().
6039          *
6040          * Returns an array which can be serialized and stored persistently. This
6041          * array can later be loaded into another parser instance with
6042          * unserializeHalfParsedText(). The text can then be safely incorporated into
6043          * the return value of a parser hook.
6044          *
6045          * @param $text string
6046          *
6047          * @return array
6048          */
6049         function serializeHalfParsedText( $text ) {
6050                 wfProfileIn( __METHOD__ );
6051                 $data = array(
6052                         'text' => $text,
6053                         'version' => self::HALF_PARSED_VERSION,
6054                         'stripState' => $this->mStripState->getSubState( $text ),
6055                         'linkHolders' => $this->mLinkHolders->getSubArray( $text )
6056                 );
6057                 wfProfileOut( __METHOD__ );
6058                 return $data;
6059         }
6060
6061         /**
6062          * Load the parser state given in the $data array, which is assumed to
6063          * have been generated by serializeHalfParsedText(). The text contents is
6064          * extracted from the array, and its markers are transformed into markers
6065          * appropriate for the current Parser instance. This transformed text is
6066          * returned, and can be safely included in the return value of a parser
6067          * hook.
6068          *
6069          * If the $data array has been stored persistently, the caller should first
6070          * check whether it is still valid, by calling isValidHalfParsedText().
6071          *
6072          * @param array $data Serialized data
6073          * @throws MWException
6074          * @return String
6075          */
6076         function unserializeHalfParsedText( $data ) {
6077                 if ( !isset( $data['version'] ) || $data['version'] != self::HALF_PARSED_VERSION ) {
6078                         throw new MWException( __METHOD__ . ': invalid version' );
6079                 }
6080
6081                 # First, extract the strip state.
6082                 $texts = array( $data['text'] );
6083                 $texts = $this->mStripState->merge( $data['stripState'], $texts );
6084
6085                 # Now renumber links
6086                 $texts = $this->mLinkHolders->mergeForeign( $data['linkHolders'], $texts );
6087
6088                 # Should be good to go.
6089                 return $texts[0];
6090         }
6091
6092         /**
6093          * Returns true if the given array, presumed to be generated by
6094          * serializeHalfParsedText(), is compatible with the current version of the
6095          * parser.
6096          *
6097          * @param $data Array
6098          *
6099          * @return bool
6100          */
6101         function isValidHalfParsedText( $data ) {
6102                 return isset( $data['version'] ) && $data['version'] == self::HALF_PARSED_VERSION;
6103         }
6104
6105         /**
6106          * Parsed a width param of imagelink like 300px or 200x300px
6107          *
6108          * @param $value String
6109          *
6110          * @return array
6111          * @since 1.20
6112          */
6113         public function parseWidthParam( $value ) {
6114                 $parsedWidthParam = array();
6115                 if ( $value === '' ) {
6116                         return $parsedWidthParam;
6117                 }
6118                 $m = array();
6119                 # (bug 13500) In both cases (width/height and width only),
6120                 # permit trailing "px" for backward compatibility.
6121                 if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
6122                         $width = intval( $m[1] );
6123                         $height = intval( $m[2] );
6124                         $parsedWidthParam['width'] = $width;
6125                         $parsedWidthParam['height'] = $height;
6126                 } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
6127                         $width = intval( $value );
6128                         $parsedWidthParam['width'] = $width;
6129                 }
6130                 return $parsedWidthParam;
6131         }
6132 }