includes/parser/Parser.php

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