includes/parser/Parser.php

   1 <?php
   2 /**
   3  * @defgroup Parser Parser
   4  *
   5  * @file
   6  * @ingroup Parser
   7  * File for Parser and related classes
   8  */
   9
  10
  11 /**
  12  * PHP Parser - Processes wiki markup (which uses a more user-friendly
  13  * syntax, such as "[[link]]" for making links), and provides a one-way
  14  * transformation of that wiki markup it into XHTML output / markup
  15  * (which in turn the browser understands, and can display).
  16  *
  17  * <pre>
  18  * There are six main entry points into the Parser class:
  19  * parse()
  20  *   produces HTML output
  21  * preSaveTransform().
  22  *   produces altered wiki markup.
  23  * preprocess()
  24  *   removes HTML comments and expands templates
  25  * cleanSig()
  26  *   Cleans a signature before saving it to preferences
  27  * extractSections()
  28  *   Extracts sections from an article for section editing
  29  * getTransclusionText()
  30  *   Extracts the text of a template with only <includeonly>, etc., parsed
  31  *
  32  * Globals used:
  33  *    objects:   $wgLang, $wgContLang
  34  *
  35  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  36  *
  37  * settings:
  38  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  39  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  40  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  41  *  $wgMaxArticleSize*
  42  *
  43  *  * only within ParserOptions
  44  * </pre>
  45  *
  46  * @ingroup Parser
  47  */
  48 class Parser
  49 {
  50         /**
  51          * Update this version number when the ParserOutput format
  52          * changes in an incompatible way, so the parser cache
  53          * can automatically discard old data.
  54          */
  55         const VERSION = '1.6.4';
  56
  57         # Flags for Parser::setFunctionHook
  58         # Also available as global constants from Defines.php
  59         const SFH_NO_HASH = 1;
  60         const SFH_OBJECT_ARGS = 2;
  61
  62         # Constants needed for external link processing
  63         # Everything except bracket, space, or control characters
  64         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  65         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  66                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  67
  68         // State constants for the definition list colon extraction
  69         const COLON_STATE_TEXT = 0;
  70         const COLON_STATE_TAG = 1;
  71         const COLON_STATE_TAGSTART = 2;
  72         const COLON_STATE_CLOSETAG = 3;
  73         const COLON_STATE_TAGSLASH = 4;
  74         const COLON_STATE_COMMENT = 5;
  75         const COLON_STATE_COMMENTDASH = 6;
  76         const COLON_STATE_COMMENTDASHDASH = 7;
  77
  78         // Flags for preprocessToDom
  79         const PTD_FOR_INCLUSION = 1;
  80
  81         // Allowed values for $this->mOutputType
  82         // Parameter to startExternalParse().
  83         const OT_HTML = 1;
  84         const OT_WIKI = 2;
  85         const OT_PREPROCESS = 3;
  86         const OT_MSG = 3;
  87         const OT_INCLUDES = 4;
  88
  89         // Marker Suffix needs to be accessible staticly.
  90         const MARKER_SUFFIX = "-QINU\x7f";
  91
  92         /**#@+
  93          * @private
  94          */
  95         # Persistent:
  96         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
  97                 $mSubsts, $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerIndex,
  98                 $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols, $mDefaultStripList,
  99                 $mVarCache, $mConf, $mFunctionTagHooks;
 100
 101
 102         # Cleared with clearState():
 103         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 104         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 105         var $mLinkHolders, $mLinkID;
 106         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 107         var $mTplExpandCache; // empty-frame expansion cache
 108         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 109         var $mExpensiveFunctionCount; // number of expensive parser function calls
 110
 111         # Temporary
 112         # These are variables reset at least once per parse regardless of $clearState
 113         var $mOptions,      // ParserOptions object
 114                 $mTitle,        // Title context, used for self-link rendering and similar things
 115                 $mOutputType,   // Output type, one of the OT_xxx constants
 116                 $ot,            // Shortcut alias, see setOutputType()
 117                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 118                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 119                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 120
 121         /**#@-*/
 122
 123         /**
 124          * Constructor
 125          *
 126          * @public
 127          */
 128         function __construct( $conf = array() ) {
 129                 $this->mConf = $conf;
 130                 $this->mTagHooks = array();
 131                 $this->mTransparentTagHooks = array();
 132                 $this->mFunctionHooks = array();
 133                 $this->mFunctionTagHooks = array();
 134                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 135                 $this->mDefaultStripList = $this->mStripList = array();
 136                 $this->mUrlProtocols = wfUrlProtocols();
 137                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 138                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 139                 $this->mVarCache = array();
 140                 if ( isset( $conf['preprocessorClass'] ) ) {
 141                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 142                 } elseif ( extension_loaded( 'domxml' ) ) {
 143                         // PECL extension that conflicts with the core DOM extension (bug 13770)
 144                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 145                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 146                 } elseif ( extension_loaded( 'dom' ) ) {
 147                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 148                 } else {
 149                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 150                 }
 151                 $this->mMarkerIndex = 0;
 152                 $this->mFirstCall = true;
 153         }
 154
 155         /**
 156          * Reduce memory usage to reduce the impact of circular references
 157          */
 158         function __destruct() {
 159                 if ( isset( $this->mLinkHolders ) ) {
 160                         $this->mLinkHolders->__destruct();
 161                 }
 162                 foreach ( $this as $name => $value ) {
 163                         unset( $this->$name );
 164                 }
 165         }
 166
 167         /**
 168          * Do various kinds of initialisation on the first call of the parser
 169          */
 170         function firstCallInit() {
 171                 if ( !$this->mFirstCall ) {
 172                         return;
 173                 }
 174                 $this->mFirstCall = false;
 175
 176                 wfProfileIn( __METHOD__ );
 177
 178                 CoreParserFunctions::register( $this );
 179                 CoreTagHooks::register( $this );
 180                 $this->initialiseVariables();
 181
 182                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 183                 wfProfileOut( __METHOD__ );
 184         }
 185
 186         /**
 187          * Clear Parser state
 188          *
 189          * @private
 190          */
 191         function clearState() {
 192                 wfProfileIn( __METHOD__ );
 193                 if ( $this->mFirstCall ) {
 194                         $this->firstCallInit();
 195                 }
 196                 $this->mOutput = new ParserOutput;
 197                 $this->mAutonumber = 0;
 198                 $this->mLastSection = '';
 199                 $this->mDTopen = false;
 200                 $this->mIncludeCount = array();
 201                 $this->mStripState = new StripState;
 202                 $this->mArgStack = false;
 203                 $this->mInPre = false;
 204                 $this->mLinkHolders = new LinkHolderArray( $this );
 205                 $this->mLinkID = 0;
 206                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 207                 $this->mVarCache = array();
 208
 209                 /**
 210                  * Prefix for temporary replacement strings for the multipass parser.
 211                  * \x07 should never appear in input as it's disallowed in XML.
 212                  * Using it at the front also gives us a little extra robustness
 213                  * since it shouldn't match when butted up against identifier-like
 214                  * string constructs.
 215                  *
 216                  * Must not consist of all title characters, or else it will change
 217                  * the behaviour of <nowiki> in a link.
 218                  */
 219                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 220                 # Changed to \x7f to allow XML double-parsing -- TS
 221                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 222
 223
 224                 # Clear these on every parse, bug 4549
 225                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 226
 227                 $this->mShowToc = true;
 228                 $this->mForceTocPosition = false;
 229                 $this->mIncludeSizes = array(
 230                         'post-expand' => 0,
 231                         'arg' => 0,
 232                 );
 233                 $this->mPPNodeCount = 0;
 234                 $this->mDefaultSort = false;
 235                 $this->mHeadings = array();
 236                 $this->mDoubleUnderscores = array();
 237                 $this->mExpensiveFunctionCount = 0;
 238
 239                 # Fix cloning
 240                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 241                         $this->mPreprocessor = null;
 242                 }
 243
 244                 wfRunHooks( 'ParserClearState', array( &$this ) );
 245                 wfProfileOut( __METHOD__ );
 246         }
 247
 248         function setOutputType( $ot ) {
 249                 $this->mOutputType = $ot;
 250                 // Shortcut alias
 251                 $this->ot = array(
 252                         'html' => $ot == self::OT_HTML,
 253                         'wiki' => $ot == self::OT_WIKI,
 254                         'pre' => $ot == self::OT_PREPROCESS,
 255                 );
 256         }
 257
 258         /**
 259          * Set the context title
 260          */
 261         function setTitle( $t ) {
 262                 if ( !$t || $t instanceof FakeTitle ) {
 263                         $t = Title::newFromText( 'NO TITLE' );
 264                 }
 265
 266                 if ( strval( $t->getFragment() ) !== '' ) {
 267                         # Strip the fragment to avoid various odd effects
 268                         $this->mTitle = clone $t;
 269                         $this->mTitle->setFragment( '' );
 270                 } else {
 271                         $this->mTitle = $t;
 272                 }
 273         }
 274
 275         /**
 276          * Accessor for mUniqPrefix.
 277          *
 278          * @public
 279          */
 280         function uniqPrefix() {
 281                 if( !isset( $this->mUniqPrefix ) ) {
 282                         // @todo Fixme: this is probably *horribly wrong*
 283                         // LanguageConverter seems to want $wgParser's uniqPrefix, however
 284                         // if this is called for a parser cache hit, the parser may not
 285                         // have ever been initialized in the first place.
 286                         // Not really sure what the heck is supposed to be going on here.
 287                         return '';
 288                         //throw new MWException( "Accessing uninitialized mUniqPrefix" );
 289                 }
 290                 return $this->mUniqPrefix;
 291         }
 292
 293         /**
 294          * Convert wikitext to HTML
 295          * Do not call this function recursively.
 296          *
 297          * @param $text String: text we want to parse
 298          * @param $title A title object
 299          * @param $options ParserOptions
 300          * @param $linestart boolean
 301          * @param $clearState boolean
 302          * @param $revid Int: number to pass in {{REVISIONID}}
 303          * @return ParserOutput a ParserOutput
 304          */
 305         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 306                 /**
 307                  * First pass--just handle <nowiki> sections, pass the rest off
 308                  * to internalParse() which does all the real work.
 309                  */
 310
 311                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang, $wgDisableLangConversion;
 312                 $fname = __METHOD__.'-' . wfGetCaller();
 313                 wfProfileIn( __METHOD__ );
 314                 wfProfileIn( $fname );
 315
 316                 if ( $clearState ) {
 317                         $this->clearState();
 318                 }
 319
 320                 $this->mOptions = $options;
 321                 $this->setTitle( $title ); // Page title has to be set for the pre-processor
 322
 323                 $oldRevisionId = $this->mRevisionId;
 324                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 325                 if( $revid !== null ) {
 326                         $this->mRevisionId = $revid;
 327                         $this->mRevisionTimestamp = null;
 328                 }
 329                 $this->setOutputType( self::OT_HTML );
 330                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 331                 # No more strip!
 332                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 333                 $text = $this->internalParse( $text );
 334
 335                 $text = $this->mStripState->unstripGeneral( $text );
 336
 337                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 338                 $fixtags = array(
 339                         # french spaces, last one Guillemet-left
 340                         # only if there is something before the space
 341                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 342                         # french spaces, Guillemet-right
 343                         '/(\\302\\253) /' => '\\1&nbsp;',
 344                         '/&nbsp;(!\s*important)/' => ' \\1', #Beware of CSS magic word !important, bug #11874.
 345                 );
 346                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 347
 348                 $text = $this->doBlockLevels( $text, $linestart );
 349
 350                 $this->replaceLinkHolders( $text );
 351
 352                 // The position of the convert() call should not be changed. it
 353                 // assumes that the links are all replaced and the only thing left
 354                 // is the <nowiki> mark.
 355                 if ( !( $wgDisableLangConversion
 356                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 357                                 || $this->mTitle->isTalkPage()
 358                                 || $this->mTitle->isConversionTable() ) ) {
 359                         $text = $wgContLang->convert( $text );
 360                 }
 361
 362                 // A title may have been set in a conversion rule.
 363                 // Note that if a user tries to set a title in a conversion
 364                 // rule but content conversion was not done, then the parser
 365                 // won't pick it up.  This is probably expected behavior.
 366                 if ( $wgContLang->getConvRuleTitle() ) {
 367                         $this->mOutput->setTitleText( $wgContLang->getConvRuleTitle() );
 368                 } elseif ( !( $wgDisableLangConversion
 369                                           || isset( $this->mDoubleUnderscores['notitleconvert'] ) ) ) {
 370                         $this->mOutput->setTitleText( $wgContLang->convert( $title->getPrefixedText() ) );
 371                 }
 372
 373                 $text = $this->mStripState->unstripNoWiki( $text );
 374
 375                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 376
 377 //!JF Move to its own function
 378
 379                 $uniq_prefix = $this->mUniqPrefix;
 380                 $matches = array();
 381                 $elements = array_keys( $this->mTransparentTagHooks );
 382                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 383
 384                 foreach( $matches as $marker => $data ) {
 385                         list( $element, $content, $params, $tag ) = $data;
 386                         $tagName = strtolower( $element );
 387                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 388                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 389                                         array( $content, $params, $this ) );
 390                         } else {
 391                                 $output = $tag;
 392                         }
 393                         $this->mStripState->general->setPair( $marker, $output );
 394                 }
 395                 $text = $this->mStripState->unstripGeneral( $text );
 396
 397                 $text = Sanitizer::normalizeCharReferences( $text );
 398
 399                 if ( ( $wgUseTidy && $this->mOptions->mTidy ) || $wgAlwaysUseTidy ) {
 400                         $text = MWTidy::tidy( $text );
 401                 } else {
 402                         # attempt to sanitize at least some nesting problems
 403                         # (bug #2702 and quite a few others)
 404                         $tidyregs = array(
 405                                 # ''Something [http://www.cool.com cool''] -->
 406                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 407                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 408                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 409                                 # fix up an anchor inside another anchor, only
 410                                 # at least for a single single nested link (bug 3695)
 411                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 412                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 413                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 414                                 # contains a div, so fix it up here; replace
 415                                 # div with escaped text
 416                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 417                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 418                                 # remove empty italic or bold tag pairs, some
 419                                 # introduced by rules above
 420                                 '/<([bi])><\/\\1>/' => '',
 421                         );
 422
 423                         $text = preg_replace(
 424                                 array_keys( $tidyregs ),
 425                                 array_values( $tidyregs ),
 426                                 $text );
 427                 }
 428                 global $wgExpensiveParserFunctionLimit;
 429                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 430                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 431                 }
 432
 433                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 434
 435                 # Information on include size limits, for the benefit of users who try to skirt them
 436                 if ( $this->mOptions->getEnableLimitReport() ) {
 437                         global $wgExpensiveParserFunctionLimit;
 438                         $max = $this->mOptions->getMaxIncludeSize();
 439                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 440                         $limitReport =
 441                                 "NewPP limit report\n" .
 442                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 443                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 444                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 445                                 $PFreport;
 446                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 447                         $text .= "\n<!-- \n$limitReport-->\n";
 448                 }
 449                 $this->mOutput->setText( $text );
 450
 451                 $this->mRevisionId = $oldRevisionId;
 452                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 453                 wfProfileOut( $fname );
 454                 wfProfileOut( __METHOD__ );
 455
 456                 return $this->mOutput;
 457         }
 458
 459         /**
 460          * Recursive parser entry point that can be called from an extension tag
 461          * hook.
 462          *
 463          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 464          *
 465          * @param $text String: text extension wants to have parsed
 466          * @param PPFrame $frame: The frame to use for expanding any template variables
 467          */
 468         function recursiveTagParse( $text, $frame=false ) {
 469                 wfProfileIn( __METHOD__ );
 470                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 471                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 472                 $text = $this->internalParse( $text, false, $frame );
 473                 wfProfileOut( __METHOD__ );
 474                 return $text;
 475         }
 476
 477         /**
 478          * Expand templates and variables in the text, producing valid, static wikitext.
 479          * Also removes comments.
 480          */
 481         function preprocess( $text, $title, $options, $revid = null ) {
 482                 wfProfileIn( __METHOD__ );
 483                 $this->clearState();
 484                 $this->setOutputType( self::OT_PREPROCESS );
 485                 $this->mOptions = $options;
 486                 $this->setTitle( $title );
 487                 if( $revid !== null ) {
 488                         $this->mRevisionId = $revid;
 489                 }
 490                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 491                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 492                 $text = $this->replaceVariables( $text );
 493                 $text = $this->mStripState->unstripBoth( $text );
 494                 wfProfileOut( __METHOD__ );
 495                 return $text;
 496         }
 497
 498         /**
 499          * Get the wikitext of a page as though it was transcluded.
 500          *
 501          * Specifically <includeonly> etc. are parsed, redirects are followed, comments
 502          * are removed, but templates arguments and parser functions are untouched.
 503          *
 504          * This is not called by the parser itself, see braceSubstitution for its transclusion.
 505          */
 506         public function getTransclusionText( $title, $options ) {
 507                 // Must initialize first
 508                 $this->clearState();
 509                 $this->setOutputType( self::OT_INCLUDES );
 510                 $this->mOptions = $options;
 511                 $this->setTitle( new FakeTitle );
 512
 513                 list( $text, $title ) = $this->getTemplateDom( $title );
 514                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 515                 return $this->getPreprocessor()->newFrame()->expand( $text, $flags );
 516         }
 517
 518         /**
 519          * Get a random string
 520          *
 521          * @private
 522          * @static
 523          */
 524         function getRandomString() {
 525                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 526         }
 527
 528         function &getTitle() { return $this->mTitle; }
 529         function getOptions() { return $this->mOptions; }
 530         function getRevisionId() { return $this->mRevisionId; }
 531         function getOutput() { return $this->mOutput; }
 532         function nextLinkID() { return $this->mLinkID++; }
 533
 534         function getFunctionLang() {
 535                 global $wgLang, $wgContLang;
 536
 537                 $target = $this->mOptions->getTargetLanguage();
 538                 if ( $target !== null ) {
 539                         return $target;
 540                 } else {
 541                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 542                 }
 543         }
 544
 545         /**
 546          * Get a preprocessor object
 547          */
 548         function getPreprocessor() {
 549                 if ( !isset( $this->mPreprocessor ) ) {
 550                         $class = $this->mPreprocessorClass;
 551                         $this->mPreprocessor = new $class( $this );
 552                 }
 553                 return $this->mPreprocessor;
 554         }
 555
 556         /**
 557          * Replaces all occurrences of HTML-style comments and the given tags
 558          * in the text with a random marker and returns the next text. The output
 559          * parameter $matches will be an associative array filled with data in
 560          * the form:
 561          *   'UNIQ-xxxxx' => array(
 562          *     'element',
 563          *     'tag content',
 564          *     array( 'param' => 'x' ),
 565          *     '<element param="x">tag content</element>' ) )
 566          *
 567          * @param $elements list of element names. Comments are always extracted.
 568          * @param $text Source text string.
 569          * @param $uniq_prefix
 570          *
 571          * @public
 572          * @static
 573          */
 574         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 575                 static $n = 1;
 576                 $stripped = '';
 577                 $matches = array();
 578
 579                 $taglist = implode( '|', $elements );
 580                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 581
 582                 while ( $text != '' ) {
 583                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 584                         $stripped .= $p[0];
 585                         if( count( $p ) < 5 ) {
 586                                 break;
 587                         }
 588                         if( count( $p ) > 5 ) {
 589                                 // comment
 590                                 $element    = $p[4];
 591                                 $attributes = '';
 592                                 $close      = '';
 593                                 $inside     = $p[5];
 594                         } else {
 595                                 // tag
 596                                 $element    = $p[1];
 597                                 $attributes = $p[2];
 598                                 $close      = $p[3];
 599                                 $inside     = $p[4];
 600                         }
 601
 602                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . self::MARKER_SUFFIX;
 603                         $stripped .= $marker;
 604
 605                         if ( $close === '/>' ) {
 606                                 // Empty element tag, <tag />
 607                                 $content = null;
 608                                 $text = $inside;
 609                                 $tail = null;
 610                         } else {
 611                                 if( $element === '!--' ) {
 612                                         $end = '/(-->)/';
 613                                 } else {
 614                                         $end = "/(<\\/$element\\s*>)/i";
 615                                 }
 616                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 617                                 $content = $q[0];
 618                                 if( count( $q ) < 3 ) {
 619                                         # No end tag -- let it run out to the end of the text.
 620                                         $tail = '';
 621                                         $text = '';
 622                                 } else {
 623                                         $tail = $q[1];
 624                                         $text = $q[2];
 625                                 }
 626                         }
 627
 628                         $matches[$marker] = array( $element,
 629                                 $content,
 630                                 Sanitizer::decodeTagAttributes( $attributes ),
 631                                 "<$element$attributes$close$content$tail" );
 632                 }
 633                 return $stripped;
 634         }
 635
 636         /**
 637          * Get a list of strippable XML-like elements
 638          */
 639         function getStripList() {
 640                 return $this->mStripList;
 641         }
 642
 643         /**
 644          * @deprecated use replaceVariables
 645          */
 646         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 647                 return $text;
 648         }
 649
 650         /**
 651          * Restores pre, math, and other extensions removed by strip()
 652          *
 653          * always call unstripNoWiki() after this one
 654          * @private
 655          * @deprecated use $this->mStripState->unstrip()
 656          */
 657         function unstrip( $text, $state ) {
 658                 return $state->unstripGeneral( $text );
 659         }
 660
 661         /**
 662          * Always call this after unstrip() to preserve the order
 663          *
 664          * @private
 665          * @deprecated use $this->mStripState->unstrip()
 666          */
 667         function unstripNoWiki( $text, $state ) {
 668                 return $state->unstripNoWiki( $text );
 669         }
 670
 671         /**
 672          * @deprecated use $this->mStripState->unstripBoth()
 673          */
 674         function unstripForHTML( $text ) {
 675                 return $this->mStripState->unstripBoth( $text );
 676         }
 677
 678         /**
 679          * Add an item to the strip state
 680          * Returns the unique tag which must be inserted into the stripped text
 681          * The tag will be replaced with the original text in unstrip()
 682          *
 683          * @private
 684          */
 685         function insertStripItem( $text ) {
 686                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 687                 $this->mMarkerIndex++;
 688                 $this->mStripState->general->setPair( $rnd, $text );
 689                 return $rnd;
 690         }
 691
 692         /**
 693          * Interface with html tidy
 694          * @deprecated Use MWTidy::tidy()
 695          */
 696         public static function tidy( $text ) {
 697                 wfDeprecated( __METHOD__ );
 698                 return MWTidy::tidy( $text );
 699         }
 700
 701         /**
 702          * parse the wiki syntax used to render tables
 703          *
 704          * @private
 705          */
 706         function doTableStuff ( $text ) {
 707                 wfProfileIn( __METHOD__ );
 708
 709                 $lines = StringUtils::explode( "\n", $text );
 710                 $out = '';
 711                 $td_history = array (); // Is currently a td tag open?
 712                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 713                 $tr_history = array (); // Is currently a tr tag open?
 714                 $tr_attributes = array (); // history of tr attributes
 715                 $has_opened_tr = array(); // Did this table open a <tr> element?
 716                 $indent_level = 0; // indent level of the table
 717
 718                 foreach ( $lines as $outLine ) {
 719                         $line = trim( $outLine );
 720
 721                         if( $line == '' ) { // empty line, go to next line
 722                                 $out .= $outLine."\n";
 723                                 continue;
 724                         }
 725                         $first_character = $line[0];
 726                         $matches = array();
 727
 728                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 729                                 // First check if we are starting a new table
 730                                 $indent_level = strlen( $matches[1] );
 731
 732                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 733                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 734
 735                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 736                                 array_push( $td_history , false );
 737                                 array_push( $last_tag_history , '' );
 738                                 array_push( $tr_history , false );
 739                                 array_push( $tr_attributes , '' );
 740                                 array_push( $has_opened_tr , false );
 741                         } else if ( count ( $td_history ) == 0 ) {
 742                                 // Don't do any of the following
 743                                 $out .= $outLine."\n";
 744                                 continue;
 745                         } else if ( substr ( $line , 0 , 2 ) === '|}' ) {
 746                                 // We are ending a table
 747                                 $line = '</table>' . substr ( $line , 2 );
 748                                 $last_tag = array_pop ( $last_tag_history );
 749
 750                                 if ( !array_pop ( $has_opened_tr ) ) {
 751                                         $line = "<tr><td></td></tr>{$line}";
 752                                 }
 753
 754                                 if ( array_pop ( $tr_history ) ) {
 755                                         $line = "</tr>{$line}";
 756                                 }
 757
 758                                 if ( array_pop ( $td_history ) ) {
 759                                         $line = "</{$last_tag}>{$line}";
 760                                 }
 761                                 array_pop ( $tr_attributes );
 762                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 763                         } else if ( substr ( $line , 0 , 2 ) === '|-' ) {
 764                                 // Now we have a table row
 765                                 $line = preg_replace( '#^\|-+#', '', $line );
 766
 767                                 // Whats after the tag is now only attributes
 768                                 $attributes = $this->mStripState->unstripBoth( $line );
 769                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 770                                 array_pop( $tr_attributes );
 771                                 array_push( $tr_attributes, $attributes );
 772
 773                                 $line = '';
 774                                 $last_tag = array_pop ( $last_tag_history );
 775                                 array_pop ( $has_opened_tr );
 776                                 array_push ( $has_opened_tr , true );
 777
 778                                 if ( array_pop ( $tr_history ) ) {
 779                                         $line = '</tr>';
 780                                 }
 781
 782                                 if ( array_pop ( $td_history ) ) {
 783                                         $line = "</{$last_tag}>{$line}";
 784                                 }
 785
 786                                 $outLine = $line;
 787                                 array_push ( $tr_history , false );
 788                                 array_push ( $td_history , false );
 789                                 array_push ( $last_tag_history , '' );
 790                         }
 791                         else if ( $first_character === '|' || $first_character === '!' || substr ( $line , 0 , 2 )  === '|+' ) {
 792                                 // This might be cell elements, td, th or captions
 793                                 if ( substr ( $line , 0 , 2 ) === '|+' ) {
 794                                         $first_character = '+';
 795                                         $line = substr ( $line , 1 );
 796                                 }
 797
 798                                 $line = substr ( $line , 1 );
 799
 800                                 if ( $first_character === '!' ) {
 801                                         $line = str_replace ( '!!' , '||' , $line );
 802                                 }
 803
 804                                 // Split up multiple cells on the same line.
 805                                 // FIXME : This can result in improper nesting of tags processed
 806                                 // by earlier parser steps, but should avoid splitting up eg
 807                                 // attribute values containing literal "||".
 808                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 809
 810                                 $outLine = '';
 811
 812                                 // Loop through each table cell
 813                                 foreach ( $cells as $cell )
 814                                 {
 815                                         $previous = '';
 816                                         if ( $first_character !== '+' )
 817                                         {
 818                                                 $tr_after = array_pop ( $tr_attributes );
 819                                                 if ( !array_pop ( $tr_history ) ) {
 820                                                         $previous = "<tr{$tr_after}>\n";
 821                                                 }
 822                                                 array_push ( $tr_history , true );
 823                                                 array_push ( $tr_attributes , '' );
 824                                                 array_pop ( $has_opened_tr );
 825                                                 array_push ( $has_opened_tr , true );
 826                                         }
 827
 828                                         $last_tag = array_pop ( $last_tag_history );
 829
 830                                         if ( array_pop ( $td_history ) ) {
 831                                                 $previous = "</{$last_tag}>{$previous}";
 832                                         }
 833
 834                                         if ( $first_character === '|' ) {
 835                                                 $last_tag = 'td';
 836                                         } else if ( $first_character === '!' ) {
 837                                                 $last_tag = 'th';
 838                                         } else if ( $first_character === '+' ) {
 839                                                 $last_tag = 'caption';
 840                                         } else {
 841                                                 $last_tag = '';
 842                                         }
 843
 844                                         array_push ( $last_tag_history , $last_tag );
 845
 846                                         // A cell could contain both parameters and data
 847                                         $cell_data = explode ( '|' , $cell , 2 );
 848
 849                                         // Bug 553: Note that a '|' inside an invalid link should not
 850                                         // be mistaken as delimiting cell parameters
 851                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 852                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 853                                         } else if ( count ( $cell_data ) == 1 )
 854                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 855                                         else {
 856                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 857                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 858                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 859                                         }
 860
 861                                         $outLine .= $cell;
 862                                         array_push ( $td_history , true );
 863                                 }
 864                         }
 865                         $out .= $outLine . "\n";
 866                 }
 867
 868                 // Closing open td, tr && table
 869                 while ( count ( $td_history ) > 0 )
 870                 {
 871                         if ( array_pop ( $td_history ) ) {
 872                                 $out .= "</td>\n";
 873                         }
 874                         if ( array_pop ( $tr_history ) ) {
 875                                 $out .= "</tr>\n";
 876                         }
 877                         if ( !array_pop ( $has_opened_tr ) ) {
 878                                 $out .= "<tr><td></td></tr>\n" ;
 879                         }
 880
 881                         $out .= "</table>\n";
 882                 }
 883
 884                 // Remove trailing line-ending (b/c)
 885                 if ( substr( $out, -1 ) === "\n" ) {
 886                         $out = substr( $out, 0, -1 );
 887                 }
 888
 889                 // special case: don't return empty table
 890                 if( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 891                         $out = '';
 892                 }
 893
 894                 wfProfileOut( __METHOD__ );
 895
 896                 return $out;
 897         }
 898
 899         /**
 900          * Helper function for parse() that transforms wiki markup into
 901          * HTML. Only called for $mOutputType == self::OT_HTML.
 902          *
 903          * @private
 904          */
 905         function internalParse( $text, $isMain = true, $frame=false ) {
 906                 wfProfileIn( __METHOD__ );
 907
 908                 $origText = $text;
 909
 910                 # Hook to suspend the parser in this state
 911                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 912                         wfProfileOut( __METHOD__ );
 913                         return $text ;
 914                 }
 915
 916                 // if $frame is provided, then use $frame for replacing any variables
 917                 if ($frame) {
 918                         // use frame depth to infer how include/noinclude tags should be handled
 919                         // depth=0 means this is the top-level document; otherwise it's an included document
 920                         if( !$frame->depth )
 921                                 $flag = 0;
 922                         else
 923                                 $flag = Parser::PTD_FOR_INCLUSION;
 924                         $dom = $this->preprocessToDom( $text, $flag );
 925                         $text = $frame->expand( $dom );
 926                 }
 927                 // if $frame is not provided, then use old-style replaceVariables
 928                 else {
 929                         $text = $this->replaceVariables( $text );
 930                 }
 931
 932                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 933                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 934
 935                 // Tables need to come after variable replacement for things to work
 936                 // properly; putting them before other transformations should keep
 937                 // exciting things like link expansions from showing up in surprising
 938                 // places.
 939                 $text = $this->doTableStuff( $text );
 940
 941                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 942
 943                 $text = $this->doDoubleUnderscore( $text );
 944
 945                 $text = $this->doHeadings( $text );
 946                 if( $this->mOptions->getUseDynamicDates() ) {
 947                         $df = DateFormatter::getInstance();
 948                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 949                 }
 950                 $text = $this->doAllQuotes( $text );
 951                 $text = $this->replaceInternalLinks( $text );
 952                 $text = $this->replaceExternalLinks( $text );
 953
 954                 # replaceInternalLinks may sometimes leave behind
 955                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 956                 $text = str_replace($this->mUniqPrefix.'NOPARSE', '', $text);
 957
 958                 $text = $this->doMagicLinks( $text );
 959                 $text = $this->formatHeadings( $text, $origText, $isMain );
 960
 961                 wfProfileOut( __METHOD__ );
 962                 return $text;
 963         }
 964
 965         /**
 966          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 967          * magic external links.
 968          *
 969          * DML
 970          * @private
 971          */
 972         function doMagicLinks( $text ) {
 973                 wfProfileIn( __METHOD__ );
 974                 $prots = $this->mUrlProtocols;
 975                 $urlChar = self::EXT_LINK_URL_CLASS;
 976                 $text = preg_replace_callback(
 977                         '!(?:                           # Start cases
 978                                 (<a.*?</a>) |               # m[1]: Skip link text
 979                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
 980                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
 981                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
 982                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
 983                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 984                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 985                                     [0-9Xx]                 # check digit
 986                                     \b)
 987                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 988                 wfProfileOut( __METHOD__ );
 989                 return $text;
 990         }
 991
 992         function magicLinkCallback( $m ) {
 993                 if ( isset( $m[1] ) && $m[1] !== '' ) {
 994                         # Skip anchor
 995                         return $m[0];
 996                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
 997                         # Skip HTML element
 998                         return $m[0];
 999                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1000                         # Free external link
1001                         return $this->makeFreeExternalLink( $m[0] );
1002                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1003                         # RFC or PMID
1004                         $CssClass = '';
1005                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1006                                 $keyword = 'RFC';
1007                                 $urlmsg = 'rfcurl';
1008                                 $CssClass = 'mw-magiclink-rfc';
1009                                 $id = $m[4];
1010                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1011                                 $keyword = 'PMID';
1012                                 $urlmsg = 'pubmedurl';
1013                                 $CssClass = 'mw-magiclink-pmid';
1014                                 $id = $m[4];
1015                         } else {
1016                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1017                                         substr($m[0], 0, 20 ) . '"' );
1018                         }
1019                         $url = wfMsg( $urlmsg, $id);
1020                         $sk = $this->mOptions->getSkin();
1021                         $la = $sk->getExternalLinkAttributes( "external $CssClass" );
1022                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1023                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1024                         # ISBN
1025                         $isbn = $m[5];
1026                         $num = strtr( $isbn, array(
1027                                 '-' => '',
1028                                 ' ' => '',
1029                                 'x' => 'X',
1030                         ));
1031                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1032                         return'<a href="' .
1033                                 $titleObj->escapeLocalUrl() .
1034                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1035                 } else {
1036                         return $m[0];
1037                 }
1038         }
1039
1040         /**
1041          * Make a free external link, given a user-supplied URL
1042          * @return HTML
1043          * @private
1044          */
1045         function makeFreeExternalLink( $url ) {
1046                 global $wgContLang;
1047                 wfProfileIn( __METHOD__ );
1048
1049                 $sk = $this->mOptions->getSkin();
1050                 $trail = '';
1051
1052                 # The characters '<' and '>' (which were escaped by
1053                 # removeHTMLtags()) should not be included in
1054                 # URLs, per RFC 2396.
1055                 $m2 = array();
1056                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1057                         $trail = substr($url, $m2[0][1]) . $trail;
1058                         $url = substr($url, 0, $m2[0][1]);
1059                 }
1060
1061                 # Move trailing punctuation to $trail
1062                 $sep = ',;\.:!?';
1063                 # If there is no left bracket, then consider right brackets fair game too
1064                 if ( strpos( $url, '(' ) === false ) {
1065                         $sep .= ')';
1066                 }
1067
1068                 $numSepChars = strspn( strrev( $url ), $sep );
1069                 if ( $numSepChars ) {
1070                         $trail = substr( $url, -$numSepChars ) . $trail;
1071                         $url = substr( $url, 0, -$numSepChars );
1072                 }
1073
1074                 $url = Sanitizer::cleanUrl( $url );
1075
1076                 # Is this an external image?
1077                 $text = $this->maybeMakeExternalImage( $url );
1078                 if ( $text === false ) {
1079                         # Not an image, make a link
1080                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1081                                 $this->getExternalLinkAttribs( $url ) );
1082                         # Register it in the output object...
1083                         # Replace unnecessary URL escape codes with their equivalent characters
1084                         $pasteurized = self::replaceUnusualEscapes( $url );
1085                         $this->mOutput->addExternalLink( $pasteurized );
1086                 }
1087                 wfProfileOut( __METHOD__ );
1088                 return $text . $trail;
1089         }
1090
1091
1092         /**
1093          * Parse headers and return html
1094          *
1095          * @private
1096          */
1097         function doHeadings( $text ) {
1098                 wfProfileIn( __METHOD__ );
1099                 for ( $i = 6; $i >= 1; --$i ) {
1100                         $h = str_repeat( '=', $i );
1101                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1102                           "<h$i>\\1</h$i>", $text );
1103                 }
1104                 wfProfileOut( __METHOD__ );
1105                 return $text;
1106         }
1107
1108         /**
1109          * Replace single quotes with HTML markup
1110          * @private
1111          * @return string the altered text
1112          */
1113         function doAllQuotes( $text ) {
1114                 wfProfileIn( __METHOD__ );
1115                 $outtext = '';
1116                 $lines = StringUtils::explode( "\n", $text );
1117                 foreach ( $lines as $line ) {
1118                         $outtext .= $this->doQuotes( $line ) . "\n";
1119                 }
1120                 $outtext = substr($outtext, 0,-1);
1121                 wfProfileOut( __METHOD__ );
1122                 return $outtext;
1123         }
1124
1125         /**
1126          * Helper function for doAllQuotes()
1127          */
1128         public function doQuotes( $text ) {
1129                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1130                 if ( count( $arr ) == 1 )
1131                         return $text;
1132                 else
1133                 {
1134                         # First, do some preliminary work. This may shift some apostrophes from
1135                         # being mark-up to being text. It also counts the number of occurrences
1136                         # of bold and italics mark-ups.
1137                         $i = 0;
1138                         $numbold = 0;
1139                         $numitalics = 0;
1140                         foreach ( $arr as $r )
1141                         {
1142                                 if ( ( $i % 2 ) == 1 )
1143                                 {
1144                                         # If there are ever four apostrophes, assume the first is supposed to
1145                                         # be text, and the remaining three constitute mark-up for bold text.
1146                                         if ( strlen( $arr[$i] ) == 4 )
1147                                         {
1148                                                 $arr[$i-1] .= "'";
1149                                                 $arr[$i] = "'''";
1150                                         }
1151                                         # If there are more than 5 apostrophes in a row, assume they're all
1152                                         # text except for the last 5.
1153                                         else if ( strlen( $arr[$i] ) > 5 )
1154                                         {
1155                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1156                                                 $arr[$i] = "'''''";
1157                                         }
1158                                         # Count the number of occurrences of bold and italics mark-ups.
1159                                         # We are not counting sequences of five apostrophes.
1160                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1161                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1162                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1163                                 }
1164                                 $i++;
1165                         }
1166
1167                         # If there is an odd number of both bold and italics, it is likely
1168                         # that one of the bold ones was meant to be an apostrophe followed
1169                         # by italics. Which one we cannot know for certain, but it is more
1170                         # likely to be one that has a single-letter word before it.
1171                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1172                         {
1173                                 $i = 0;
1174                                 $firstsingleletterword = -1;
1175                                 $firstmultiletterword = -1;
1176                                 $firstspace = -1;
1177                                 foreach ( $arr as $r )
1178                                 {
1179                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1180                                         {
1181                                                 $x1 = substr ($arr[$i-1], -1);
1182                                                 $x2 = substr ($arr[$i-1], -2, 1);
1183                                                 if ($x1 === ' ') {
1184                                                         if ($firstspace == -1) $firstspace = $i;
1185                                                 } else if ($x2 === ' ') {
1186                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1187                                                 } else {
1188                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1189                                                 }
1190                                         }
1191                                         $i++;
1192                                 }
1193
1194                                 # If there is a single-letter word, use it!
1195                                 if ($firstsingleletterword > -1)
1196                                 {
1197                                         $arr [ $firstsingleletterword ] = "''";
1198                                         $arr [ $firstsingleletterword-1 ] .= "'";
1199                                 }
1200                                 # If not, but there's a multi-letter word, use that one.
1201                                 else if ($firstmultiletterword > -1)
1202                                 {
1203                                         $arr [ $firstmultiletterword ] = "''";
1204                                         $arr [ $firstmultiletterword-1 ] .= "'";
1205                                 }
1206                                 # ... otherwise use the first one that has neither.
1207                                 # (notice that it is possible for all three to be -1 if, for example,
1208                                 # there is only one pentuple-apostrophe in the line)
1209                                 else if ($firstspace > -1)
1210                                 {
1211                                         $arr [ $firstspace ] = "''";
1212                                         $arr [ $firstspace-1 ] .= "'";
1213                                 }
1214                         }
1215
1216                         # Now let's actually convert our apostrophic mush to HTML!
1217                         $output = '';
1218                         $buffer = '';
1219                         $state = '';
1220                         $i = 0;
1221                         foreach ($arr as $r)
1222                         {
1223                                 if (($i % 2) == 0)
1224                                 {
1225                                         if ($state === 'both')
1226                                                 $buffer .= $r;
1227                                         else
1228                                                 $output .= $r;
1229                                 }
1230                                 else
1231                                 {
1232                                         if (strlen ($r) == 2)
1233                                         {
1234                                                 if ($state === 'i')
1235                                                 { $output .= '</i>'; $state = ''; }
1236                                                 else if ($state === 'bi')
1237                                                 { $output .= '</i>'; $state = 'b'; }
1238                                                 else if ($state === 'ib')
1239                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1240                                                 else if ($state === 'both')
1241                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1242                                                 else # $state can be 'b' or ''
1243                                                 { $output .= '<i>'; $state .= 'i'; }
1244                                         }
1245                                         else if (strlen ($r) == 3)
1246                                         {
1247                                                 if ($state === 'b')
1248                                                 { $output .= '</b>'; $state = ''; }
1249                                                 else if ($state === 'bi')
1250                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1251                                                 else if ($state === 'ib')
1252                                                 { $output .= '</b>'; $state = 'i'; }
1253                                                 else if ($state === 'both')
1254                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1255                                                 else # $state can be 'i' or ''
1256                                                 { $output .= '<b>'; $state .= 'b'; }
1257                                         }
1258                                         else if (strlen ($r) == 5)
1259                                         {
1260                                                 if ($state === 'b')
1261                                                 { $output .= '</b><i>'; $state = 'i'; }
1262                                                 else if ($state === 'i')
1263                                                 { $output .= '</i><b>'; $state = 'b'; }
1264                                                 else if ($state === 'bi')
1265                                                 { $output .= '</i></b>'; $state = ''; }
1266                                                 else if ($state === 'ib')
1267                                                 { $output .= '</b></i>'; $state = ''; }
1268                                                 else if ($state === 'both')
1269                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1270                                                 else # ($state == '')
1271                                                 { $buffer = ''; $state = 'both'; }
1272                                         }
1273                                 }
1274                                 $i++;
1275                         }
1276                         # Now close all remaining tags.  Notice that the order is important.
1277                         if ($state === 'b' || $state === 'ib')
1278                                 $output .= '</b>';
1279                         if ($state === 'i' || $state === 'bi' || $state === 'ib')
1280                                 $output .= '</i>';
1281                         if ($state === 'bi')
1282                                 $output .= '</b>';
1283                         # There might be lonely ''''', so make sure we have a buffer
1284                         if ($state === 'both' && $buffer)
1285                                 $output .= '<b><i>'.$buffer.'</i></b>';
1286                         return $output;
1287                 }
1288         }
1289
1290         /**
1291          * Replace external links (REL)
1292          *
1293          * Note: this is all very hackish and the order of execution matters a lot.
1294          * Make sure to run maintenance/parserTests.php if you change this code.
1295          *
1296          * @private
1297          */
1298         function replaceExternalLinks( $text ) {
1299                 global $wgContLang;
1300                 wfProfileIn( __METHOD__ );
1301
1302                 $sk = $this->mOptions->getSkin();
1303
1304                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1305                 $s = array_shift( $bits );
1306
1307                 $i = 0;
1308                 while ( $i<count( $bits ) ) {
1309                         $url = $bits[$i++];
1310                         $protocol = $bits[$i++];
1311                         $text = $bits[$i++];
1312                         $trail = $bits[$i++];
1313
1314                         # The characters '<' and '>' (which were escaped by
1315                         # removeHTMLtags()) should not be included in
1316                         # URLs, per RFC 2396.
1317                         $m2 = array();
1318                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1319                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1320                                 $url = substr($url, 0, $m2[0][1]);
1321                         }
1322
1323                         # If the link text is an image URL, replace it with an <img> tag
1324                         # This happened by accident in the original parser, but some people used it extensively
1325                         $img = $this->maybeMakeExternalImage( $text );
1326                         if ( $img !== false ) {
1327                                 $text = $img;
1328                         }
1329
1330                         $dtrail = '';
1331
1332                         # Set linktype for CSS - if URL==text, link is essentially free
1333                         $linktype = ($text === $url) ? 'free' : 'text';
1334
1335                         # No link text, e.g. [http://domain.tld/some.link]
1336                         if ( $text == '' ) {
1337                                 # Autonumber if allowed. See bug #5918
1338                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1339                                         $langObj = $this->getFunctionLang();
1340                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1341                                         $linktype = 'autonumber';
1342                                 } else {
1343                                         # Otherwise just use the URL
1344                                         $text = htmlspecialchars( $url );
1345                                         $linktype = 'free';
1346                                 }
1347                         } else {
1348                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1349                                 # Check for trail
1350                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1351                         }
1352
1353                         $text = $wgContLang->markNoConversion($text);
1354
1355                         $url = Sanitizer::cleanUrl( $url );
1356
1357                         # Use the encoded URL
1358                         # This means that users can paste URLs directly into the text
1359                         # Funny characters like &ouml; aren't valid in URLs anyway
1360                         # This was changed in August 2004
1361                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1362                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1363
1364                         # Register link in the output object.
1365                         # Replace unnecessary URL escape codes with the referenced character
1366                         # This prevents spammers from hiding links from the filters
1367                         $pasteurized = self::replaceUnusualEscapes( $url );
1368                         $this->mOutput->addExternalLink( $pasteurized );
1369                 }
1370
1371                 wfProfileOut( __METHOD__ );
1372                 return $s;
1373         }
1374
1375         /**
1376          * Get an associative array of additional HTML attributes appropriate for a
1377          * particular external link.  This currently may include rel => nofollow
1378          * (depending on configuration, namespace, and the URL's domain) and/or a
1379          * target attribute (depending on configuration).
1380          *
1381          * @param string $url Optional URL, to extract the domain from for rel =>
1382          *   nofollow if appropriate
1383          * @return array Associative array of HTML attributes
1384          */
1385         function getExternalLinkAttribs( $url = false ) {
1386                 $attribs = array();
1387                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1388                 $ns = $this->mTitle->getNamespace();
1389                 if( $wgNoFollowLinks && !in_array($ns, $wgNoFollowNsExceptions) ) {
1390                         $attribs['rel'] = 'nofollow';
1391
1392                         global $wgNoFollowDomainExceptions;
1393                         if ( $wgNoFollowDomainExceptions ) {
1394                                 $bits = wfParseUrl( $url );
1395                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1396                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1397                                                 if( substr( $bits['host'], -strlen( $domain ) )
1398                                                 == $domain ) {
1399                                                         unset( $attribs['rel'] );
1400                                                         break;
1401                                                 }
1402                                         }
1403                                 }
1404                         }
1405                 }
1406                 if ( $this->mOptions->getExternalLinkTarget() ) {
1407                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1408                 }
1409                 return $attribs;
1410         }
1411
1412
1413         /**
1414          * Replace unusual URL escape codes with their equivalent characters
1415          * @param string
1416          * @return string
1417          * @static
1418          * @todo  This can merge genuinely required bits in the path or query string,
1419          *        breaking legit URLs. A proper fix would treat the various parts of
1420          *        the URL differently; as a workaround, just use the output for
1421          *        statistical records, not for actual linking/output.
1422          */
1423         static function replaceUnusualEscapes( $url ) {
1424                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1425                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1426         }
1427
1428         /**
1429          * Callback function used in replaceUnusualEscapes().
1430          * Replaces unusual URL escape codes with their equivalent character
1431          * @static
1432          * @private
1433          */
1434         private static function replaceUnusualEscapesCallback( $matches ) {
1435                 $char = urldecode( $matches[0] );
1436                 $ord = ord( $char );
1437                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1438                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1439                         // No, shouldn't be escaped
1440                         return $char;
1441                 } else {
1442                         // Yes, leave it escaped
1443                         return $matches[0];
1444                 }
1445         }
1446
1447         /**
1448          * make an image if it's allowed, either through the global
1449          * option, through the exception, or through the on-wiki whitelist
1450          * @private
1451          */
1452         function maybeMakeExternalImage( $url ) {
1453                 $sk = $this->mOptions->getSkin();
1454                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1455                 $imagesexception = !empty($imagesfrom);
1456                 $text = false;
1457                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1458                 if( $imagesexception && is_array( $imagesfrom ) ) {
1459                         $imagematch = false;
1460                         foreach( $imagesfrom as $match ) {
1461                                 if( strpos( $url, $match ) === 0 ) {
1462                                         $imagematch = true;
1463                                         break;
1464                                 }
1465                         }
1466                 } elseif( $imagesexception ) {
1467                         $imagematch = (strpos( $url, $imagesfrom ) === 0);
1468                 } else {
1469                         $imagematch = false;
1470                 }
1471                 if ( $this->mOptions->getAllowExternalImages()
1472                      || ( $imagesexception && $imagematch ) ) {
1473                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1474                                 # Image found
1475                                 $text = $sk->makeExternalImage( $url );
1476                         }
1477                 }
1478                 if( !$text && $this->mOptions->getEnableImageWhitelist()
1479                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1480                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1481                         foreach( $whitelist as $entry ) {
1482                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1483                                 if( strpos( $entry, '#' ) === 0 || $entry === '' )
1484                                         continue;
1485                                 if( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1486                                         # Image matches a whitelist entry
1487                                         $text = $sk->makeExternalImage( $url );
1488                                         break;
1489                                 }
1490                         }
1491                 }
1492                 return $text;
1493         }
1494
1495         /**
1496          * Process [[ ]] wikilinks
1497          * @return processed text
1498          *
1499          * @private
1500          */
1501         function replaceInternalLinks( $s ) {
1502                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1503                 return $s;
1504         }
1505
1506         /**
1507          * Process [[ ]] wikilinks (RIL)
1508          * @return LinkHolderArray
1509          *
1510          * @private
1511          */
1512         function replaceInternalLinks2( &$s ) {
1513                 global $wgContLang;
1514
1515                 wfProfileIn( __METHOD__ );
1516
1517                 wfProfileIn( __METHOD__.'-setup' );
1518                 static $tc = FALSE, $e1, $e1_img;
1519                 # the % is needed to support urlencoded titles as well
1520                 if ( !$tc ) {
1521                         $tc = Title::legalChars() . '#%';
1522                         # Match a link having the form [[namespace:link|alternate]]trail
1523                         $e1 = "/^([{$tc}]*)(\\|.*?)?]](.*)\$/sD";
1524                         # Match cases where there is no "]]", which might still be images
1525                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1526                 }
1527
1528                 $sk = $this->mOptions->getSkin();
1529                 $holders = new LinkHolderArray( $this );
1530
1531                 #split the entire text string on occurences of [[
1532                 $a = StringUtils::explode( '[[', ' ' . $s );
1533                 #get the first element (all text up to first [[), and remove the space we added
1534                 $s = $a->current();
1535                 $a->next();
1536                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1537                 $s = substr( $s, 1 );
1538
1539                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1540                 $e2 = null;
1541                 if ( $useLinkPrefixExtension ) {
1542                         # Match the end of a line for a word that's not followed by whitespace,
1543                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1544                         $e2 = wfMsgForContent( 'linkprefix' );
1545                 }
1546
1547                 if( is_null( $this->mTitle ) ) {
1548                         wfProfileOut( __METHOD__.'-setup' );
1549                         wfProfileOut( __METHOD__ );
1550                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1551                 }
1552                 $nottalk = !$this->mTitle->isTalkPage();
1553
1554                 if ( $useLinkPrefixExtension ) {
1555                         $m = array();
1556                         if ( preg_match( $e2, $s, $m ) ) {
1557                                 $first_prefix = $m[2];
1558                         } else {
1559                                 $first_prefix = false;
1560                         }
1561                 } else {
1562                         $prefix = '';
1563                 }
1564
1565                 if($wgContLang->hasVariants()) {
1566                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1567                 } else {
1568                         $selflink = array($this->mTitle->getPrefixedText());
1569                 }
1570                 $useSubpages = $this->areSubpagesAllowed();
1571                 wfProfileOut( __METHOD__.'-setup' );
1572
1573                 # Loop for each link
1574                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1575                         # Check for excessive memory usage
1576                         if ( $holders->isBig() ) {
1577                                 # Too big
1578                                 # Do the existence check, replace the link holders and clear the array
1579                                 $holders->replace( $s );
1580                                 $holders->clear();
1581                         }
1582
1583                         if ( $useLinkPrefixExtension ) {
1584                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1585                                 if ( preg_match( $e2, $s, $m ) ) {
1586                                         $prefix = $m[2];
1587                                         $s = $m[1];
1588                                 } else {
1589                                         $prefix='';
1590                                 }
1591                                 # first link
1592                                 if($first_prefix) {
1593                                         $prefix = $first_prefix;
1594                                         $first_prefix = false;
1595                                 }
1596                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1597                         }
1598
1599                         $might_be_img = false;
1600
1601                         wfProfileIn( __METHOD__."-e1" );
1602                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1603
1604                                 if( $m[2] === '' ) {
1605                                         $text = '';
1606                                 } elseif( $m[2] === '|' ) {
1607                                         $text = $this->getPipeTrickText( $m[1] );
1608                                 } else {
1609                                         $text = substr( $m[2], 1 );
1610                                 }
1611
1612                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1613                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1614                                 # the real problem is with the $e1 regex
1615                                 # See bug 1300.
1616                                 #
1617                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1618                                 # and no image is in sight. See bug 2095.
1619                                 #
1620                                 if( $text !== '' &&
1621                                         substr( $m[3], 0, 1 ) === ']' &&
1622                                         strpos($text, '[') !== false
1623                                 )
1624                                 {
1625                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1626                                         $m[3] = substr( $m[3], 1 );
1627                                 }
1628
1629                                 # Handle pipe-trick for [[|<blah>]]
1630                                 $lnk = $m[1] === '' ? $this->getPipeTrickLink( $text ) : $m[1];
1631                                 # fix up urlencoded title texts
1632                                 if( strpos( $lnk, '%' ) !== false ) {
1633                                         # Should anchors '#' also be rejected?
1634                                         $lnk = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($lnk) );
1635                                 }
1636
1637                                 $trail = $m[3];
1638                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1639                                 $might_be_img = true;
1640                                 $text = $m[2];
1641                                 $lnk = strpos( $m[1], '%' ) === false ? $m[1] : urldecode( $m[1] );
1642                                 $trail = "";
1643                         } else { # Invalid form; output directly
1644                                 $s .= $prefix . '[[' . $line ;
1645                                 wfProfileOut( __METHOD__."-e1" );
1646                                 continue;
1647                         }
1648                         wfProfileOut( __METHOD__."-e1" );
1649                         wfProfileIn( __METHOD__."-misc" );
1650
1651                         # Don't allow internal links to pages containing
1652                         # PROTO: where PROTO is a valid URL protocol; these
1653                         # should be external links.
1654                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $lnk ) ) {
1655                                 $s .= $prefix . '[[' . $line ;
1656                                 wfProfileOut( __METHOD__."-misc" );
1657                                 continue;
1658                         }
1659
1660                         # Make subpage if necessary
1661                         if ( $useSubpages ) {
1662                                 $link = $this->maybeDoSubpageLink( $lnk, $text );
1663                         } else {
1664                                 $link = $lnk;
1665                         }
1666
1667                         $noforce = (substr( $lnk, 0, 1 ) !== ':');
1668                         if (!$noforce) {
1669                                 # Strip off leading ':'
1670                                 $link = substr( $link, 1 );
1671                         }
1672
1673                         wfProfileOut( __METHOD__."-misc" );
1674                         wfProfileIn( __METHOD__."-title" );
1675                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1676                         if ( $nt === null ) {
1677                                 $s .= $prefix . '[[' . $line;
1678                                 wfProfileOut( __METHOD__."-title" );
1679                                 continue;
1680                         }
1681
1682                         $ns = $nt->getNamespace();
1683                         $iw = $nt->getInterWiki();
1684                         wfProfileOut( __METHOD__."-title" );
1685
1686                         if ( $might_be_img ) { # if this is actually an invalid link
1687                                 wfProfileIn( __METHOD__."-might_be_img" );
1688                                 if ( $ns == NS_FILE && $noforce ) { #but might be an image
1689                                         $found = false;
1690                                         while ( true ) {
1691                                                 #look at the next 'line' to see if we can close it there
1692                                                 $a->next();
1693                                                 $next_line = $a->current();
1694                                                 if ( $next_line === false || $next_line === null ) {
1695                                                         break;
1696                                                 }
1697                                                 $m = explode( ']]', $next_line, 3 );
1698                                                 if ( count( $m ) == 3 ) {
1699                                                         # the first ]] closes the inner link, the second the image
1700                                                         $found = true;
1701                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1702                                                         $trail = $m[2];
1703                                                         break;
1704                                                 } elseif ( count( $m ) == 2 ) {
1705                                                         #if there's exactly one ]] that's fine, we'll keep looking
1706                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1707                                                 } else {
1708                                                         #if $next_line is invalid too, we need look no further
1709                                                         $text .= '[[' . $next_line;
1710                                                         break;
1711                                                 }
1712                                         }
1713                                         if ( !$found ) {
1714                                                 # we couldn't find the end of this imageLink, so output it raw
1715                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1716                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1717                                                 $s .= "{$prefix}[[$link|$text";
1718                                                 # note: no $trail, because without an end, there *is* no trail
1719                                                 wfProfileOut( __METHOD__."-might_be_img" );
1720                                                 continue;
1721                                         }
1722                                 } else { #it's not an image, so output it raw
1723                                         $s .= "{$prefix}[[$link|$text";
1724                                         # note: no $trail, because without an end, there *is* no trail
1725                                         wfProfileOut( __METHOD__."-might_be_img" );
1726                                         continue;
1727                                 }
1728                                 wfProfileOut( __METHOD__."-might_be_img" );
1729                         }
1730
1731                         $wasblank = ( $text  == '' );
1732                         if ( $wasblank ) $text = $link;
1733
1734                         # Link not escaped by : , create the various objects
1735                         if ( $noforce ) {
1736
1737                                 # Interwikis
1738                                 wfProfileIn( __METHOD__."-interwiki" );
1739                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1740                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1741                                         $s = rtrim($s . $prefix);
1742                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1743                                         wfProfileOut( __METHOD__."-interwiki" );
1744                                         continue;
1745                                 }
1746                                 wfProfileOut( __METHOD__."-interwiki" );
1747
1748                                 if ( $ns == NS_FILE ) {
1749                                         wfProfileIn( __METHOD__."-image" );
1750                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1751                                                 if ( $wasblank ) {
1752                                                         # if no parameters were passed, $text
1753                                                         # becomes something like "File:Foo.png",
1754                                                         # which we don't want to pass on to the
1755                                                         # image generator
1756                                                         $text = '';
1757                                                 } else {
1758                                                         # recursively parse links inside the image caption
1759                                                         # actually, this will parse them in any other parameters, too,
1760                                                         # but it might be hard to fix that, and it doesn't matter ATM
1761                                                         $text = $this->replaceExternalLinks($text);
1762                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1763                                                 }
1764                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1765                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1766                                         } else {
1767                                                 $s .= $prefix . $trail;
1768                                         }
1769                                         $this->mOutput->addImage( $nt->getDBkey() );
1770                                         wfProfileOut( __METHOD__."-image" );
1771                                         continue;
1772
1773                                 }
1774
1775                                 if ( $ns == NS_CATEGORY ) {
1776                                         wfProfileIn( __METHOD__."-category" );
1777                                         $s = rtrim($s . "\n"); # bug 87
1778
1779                                         if ( $wasblank ) {
1780                                                 $sortkey = $this->getDefaultSort();
1781                                         } else {
1782                                                 $sortkey = $text;
1783                                         }
1784                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1785                                         $sortkey = str_replace( "\n", '', $sortkey );
1786                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1787                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1788
1789                                         /**
1790                                          * Strip the whitespace Category links produce, see bug 87
1791                                          * @todo We might want to use trim($tmp, "\n") here.
1792                                          */
1793                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1794
1795                                         wfProfileOut( __METHOD__."-category" );
1796                                         continue;
1797                                 }
1798                         }
1799
1800                         # Self-link checking
1801                         if( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1802                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1803                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1804                                         continue;
1805                                 }
1806                         }
1807
1808                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1809                         # FIXME: Should do batch file existence checks, see comment below
1810                         if( $ns == NS_MEDIA ) {
1811                                 wfProfileIn( __METHOD__."-media" );
1812                                 # Give extensions a chance to select the file revision for us
1813                                 $skip = $time = false;
1814                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1815                                 if ( $skip ) {
1816                                         $link = $sk->link( $nt );
1817                                 } else {
1818                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1819                                 }
1820                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1821                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1822                                 $this->mOutput->addImage( $nt->getDBkey() );
1823                                 wfProfileOut( __METHOD__."-media" );
1824                                 continue;
1825                         }
1826
1827                         wfProfileIn( __METHOD__."-always_known" );
1828                         # Some titles, such as valid special pages or files in foreign repos, should
1829                         # be shown as bluelinks even though they're not included in the page table
1830                         #
1831                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1832                         # batch file existence checks for NS_FILE and NS_MEDIA
1833                         if( $iw == '' && $nt->isAlwaysKnown() ) {
1834                                 $this->mOutput->addLink( $nt );
1835                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1836                         } else {
1837                                 # Links will be added to the output link list after checking
1838                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1839                         }
1840                         wfProfileOut( __METHOD__."-always_known" );
1841                 }
1842                 wfProfileOut( __METHOD__ );
1843                 return $holders;
1844         }
1845
1846         /**
1847          * Make a link placeholder. The text returned can be later resolved to a real link with
1848          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1849          * parsing of interwiki links, and secondly to allow all existence checks and
1850          * article length checks (for stub links) to be bundled into a single query.
1851          *
1852          * @deprecated
1853          */
1854         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1855                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1856         }
1857
1858         /**
1859          * Render a forced-blue link inline; protect against double expansion of
1860          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1861          * Since this little disaster has to split off the trail text to avoid
1862          * breaking URLs in the following text without breaking trails on the
1863          * wiki links, it's been made into a horrible function.
1864          *
1865          * @param Title $nt
1866          * @param string $text
1867          * @param string $query
1868          * @param string $trail
1869          * @param string $prefix
1870          * @return string HTML-wikitext mix oh yuck
1871          */
1872         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1873                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1874                 $sk = $this->mOptions->getSkin();
1875                 // FIXME: use link() instead of deprecated makeKnownLinkObj()
1876                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1877                 return $this->armorLinks( $link ) . $trail;
1878         }
1879
1880         /**
1881          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1882          * going to go through further parsing steps before inline URL expansion.
1883          *
1884          * Not needed quite as much as it used to be since free links are a bit
1885          * more sensible these days. But bracketed links are still an issue.
1886          *
1887          * @param string more-or-less HTML
1888          * @return string less-or-more HTML with NOPARSE bits
1889          */
1890         function armorLinks( $text ) {
1891                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1892                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1893         }
1894
1895         /**
1896          * Return true if subpage links should be expanded on this page.
1897          * @return bool
1898          */
1899         function areSubpagesAllowed() {
1900                 # Some namespaces don't allow subpages
1901                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1902         }
1903
1904         /**
1905          * Handle link to subpage if necessary
1906          * @param string $target the source of the link
1907          * @param string &$text the link text, modified as necessary
1908          * @return string the full name of the link
1909          * @private
1910          */
1911         function maybeDoSubpageLink($target, &$text) {
1912                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
1913         }
1914
1915         /**
1916          * From the [[title|]] return link-text as though the used typed [[title|link-text]]
1917          * @param string $link from [[$link|]]
1918          * @return string $text for [[$link|$text]]
1919          */
1920         function getPipeTrickText( $link ) {
1921                 return Linker::getPipeTrickText( $link );
1922         }
1923
1924         /**
1925          * From the [[|link-text]] return the title as though the user typed [[title|link-text]]
1926          * @param string $text from [[|$text]]
1927          * @param Title $title to resolve the link against
1928          * @return string $link for [[$link|$text]]
1929          */
1930         function getPipeTrickLink( $text ) {
1931                 return Linker::getPipeTrickLink( $text, $this->mTitle );
1932         }
1933
1934         /**#@+
1935          * Used by doBlockLevels()
1936          * @private
1937          */
1938         /* private */ function closeParagraph() {
1939                 $result = '';
1940                 if ( $this->mLastSection != '' ) {
1941                         $result = '</' . $this->mLastSection  . ">\n";
1942                 }
1943                 $this->mInPre = false;
1944                 $this->mLastSection = '';
1945                 return $result;
1946         }
1947         # getCommon() returns the length of the longest common substring
1948         # of both arguments, starting at the beginning of both.
1949         #
1950         /* private */ function getCommon( $st1, $st2 ) {
1951                 $fl = strlen( $st1 );
1952                 $shorter = strlen( $st2 );
1953                 if ( $fl < $shorter ) { $shorter = $fl; }
1954
1955                 for ( $i = 0; $i < $shorter; ++$i ) {
1956                         if ( $st1{$i} != $st2{$i} ) { break; }
1957                 }
1958                 return $i;
1959         }
1960         # These next three functions open, continue, and close the list
1961         # element appropriate to the prefix character passed into them.
1962         #
1963         /* private */ function openList( $char ) {
1964                 $result = $this->closeParagraph();
1965
1966                 if ( '*' === $char ) { $result .= '<ul><li>'; }
1967                 elseif ( '#' === $char ) { $result .= '<ol><li>'; }
1968                 elseif ( ':' === $char ) { $result .= '<dl><dd>'; }
1969                 elseif ( ';' === $char ) {
1970                         $result .= '<dl><dt>';
1971                         $this->mDTopen = true;
1972                 }
1973                 else { $result = '<!-- ERR 1 -->'; }
1974
1975                 return $result;
1976         }
1977
1978         /* private */ function nextItem( $char ) {
1979                 if ( '*' === $char || '#' === $char ) { return '</li><li>'; }
1980                 elseif ( ':' === $char || ';' === $char ) {
1981                         $close = '</dd>';
1982                         if ( $this->mDTopen ) { $close = '</dt>'; }
1983                         if ( ';' === $char ) {
1984                                 $this->mDTopen = true;
1985                                 return $close . '<dt>';
1986                         } else {
1987                                 $this->mDTopen = false;
1988                                 return $close . '<dd>';
1989                         }
1990                 }
1991                 return '<!-- ERR 2 -->';
1992         }
1993
1994         /* private */ function closeList( $char ) {
1995                 if ( '*' === $char ) { $text = '</li></ul>'; }
1996                 elseif ( '#' === $char ) { $text = '</li></ol>'; }
1997                 elseif ( ':' === $char ) {
1998                         if ( $this->mDTopen ) {
1999                                 $this->mDTopen = false;
2000                                 $text = '</dt></dl>';
2001                         } else {
2002                                 $text = '</dd></dl>';
2003                         }
2004                 }
2005                 else {  return '<!-- ERR 3 -->'; }
2006                 return $text."\n";
2007         }
2008         /**#@-*/
2009
2010         /**
2011          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2012          *
2013          * @param $linestart bool whether or not this is at the start of a line.
2014          * @private
2015          * @return string the lists rendered as HTML
2016          */
2017         function doBlockLevels( $text, $linestart ) {
2018                 wfProfileIn( __METHOD__ );
2019
2020                 # Parsing through the text line by line.  The main thing
2021                 # happening here is handling of block-level elements p, pre,
2022                 # and making lists from lines starting with * # : etc.
2023                 #
2024                 $textLines = StringUtils::explode( "\n", $text );
2025
2026                 $lastPrefix = $output = '';
2027                 $this->mDTopen = $inBlockElem = false;
2028                 $prefixLength = 0;
2029                 $paragraphStack = false;
2030
2031                 foreach ( $textLines as $oLine ) {
2032                         # Fix up $linestart
2033                         if ( !$linestart ) {
2034                                 $output .= $oLine;
2035                                 $linestart = true;
2036                                 continue;
2037                         }
2038                         // * = ul
2039                         // # = ol
2040                         // ; = dt
2041                         // : = dd
2042
2043                         $lastPrefixLength = strlen( $lastPrefix );
2044                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2045                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2046                         // If not in a <pre> element, scan for and figure out what prefixes are there.
2047                         if ( !$this->mInPre ) {
2048                                 # Multiple prefixes may abut each other for nested lists.
2049                                 $prefixLength = strspn( $oLine, '*#:;' );
2050                                 $prefix = substr( $oLine, 0, $prefixLength );
2051
2052                                 # eh?
2053                                 // ; and : are both from definition-lists, so they're equivalent
2054                                 //  for the purposes of determining whether or not we need to open/close
2055                                 //  elements.
2056                                 $prefix2 = str_replace( ';', ':', $prefix );
2057                                 $t = substr( $oLine, $prefixLength );
2058                                 $this->mInPre = (bool)$preOpenMatch;
2059                         } else {
2060                                 # Don't interpret any other prefixes in preformatted text
2061                                 $prefixLength = 0;
2062                                 $prefix = $prefix2 = '';
2063                                 $t = $oLine;
2064                         }
2065
2066                         # List generation
2067                         if( $prefixLength && $lastPrefix === $prefix2 ) {
2068                                 # Same as the last item, so no need to deal with nesting or opening stuff
2069                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2070                                 $paragraphStack = false;
2071
2072                                 if ( substr( $prefix, -1 ) === ';') {
2073                                         # The one nasty exception: definition lists work like this:
2074                                         # ; title : definition text
2075                                         # So we check for : in the remainder text to split up the
2076                                         # title and definition, without b0rking links.
2077                                         $term = $t2 = '';
2078                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2079                                                 $t = $t2;
2080                                                 $output .= $term . $this->nextItem( ':' );
2081                                         }
2082                                 }
2083                         } elseif( $prefixLength || $lastPrefixLength ) {
2084                                 // We need to open or close prefixes, or both.
2085
2086                                 # Either open or close a level...
2087                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2088                                 $paragraphStack = false;
2089
2090                                 // Close all the prefixes which aren't shared.
2091                                 while( $commonPrefixLength < $lastPrefixLength ) {
2092                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2093                                         --$lastPrefixLength;
2094                                 }
2095
2096                                 // Continue the current prefix if appropriate.
2097                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2098                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2099                                 }
2100
2101                                 // Open prefixes where appropriate.
2102                                 while ( $prefixLength > $commonPrefixLength ) {
2103                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2104                                         $output .= $this->openList( $char );
2105
2106                                         if ( ';' === $char ) {
2107                                                 # FIXME: This is dupe of code above
2108                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2109                                                         $t = $t2;
2110                                                         $output .= $term . $this->nextItem( ':' );
2111                                                 }
2112                                         }
2113                                         ++$commonPrefixLength;
2114                                 }
2115                                 $lastPrefix = $prefix2;
2116                         }
2117
2118                         // If we have no prefixes, go to paragraph mode.
2119                         if( 0 == $prefixLength ) {
2120                                 wfProfileIn( __METHOD__."-paragraph" );
2121                                 # No prefix (not in list)--go to paragraph mode
2122                                 // XXX: use a stack for nestable elements like span, table and div
2123                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2124                                 $closematch = preg_match(
2125                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2126                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2127                                 if ( $openmatch or $closematch ) {
2128                                         $paragraphStack = false;
2129                                         # TODO bug 5718: paragraph closed
2130                                         $output .= $this->closeParagraph();
2131                                         if ( $preOpenMatch and !$preCloseMatch ) {
2132                                                 $this->mInPre = true;
2133                                         }
2134                                         if ( $closematch ) {
2135                                                 $inBlockElem = false;
2136                                         } else {
2137                                                 $inBlockElem = true;
2138                                         }
2139                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2140                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' or trim($t) != '' ) ) {
2141                                                 // pre
2142                                                 if ($this->mLastSection !== 'pre') {
2143                                                         $paragraphStack = false;
2144                                                         $output .= $this->closeParagraph().'<pre>';
2145                                                         $this->mLastSection = 'pre';
2146                                                 }
2147                                                 $t = substr( $t, 1 );
2148                                         } else {
2149                                                 // paragraph
2150                                                 if ( trim($t) == '' ) {
2151                                                         if ( $paragraphStack ) {
2152                                                                 $output .= $paragraphStack.'<br />';
2153                                                                 $paragraphStack = false;
2154                                                                 $this->mLastSection = 'p';
2155                                                         } else {
2156                                                                 if ($this->mLastSection !== 'p' ) {
2157                                                                         $output .= $this->closeParagraph();
2158                                                                         $this->mLastSection = '';
2159                                                                         $paragraphStack = '<p>';
2160                                                                 } else {
2161                                                                         $paragraphStack = '</p><p>';
2162                                                                 }
2163                                                         }
2164                                                 } else {
2165                                                         if ( $paragraphStack ) {
2166                                                                 $output .= $paragraphStack;
2167                                                                 $paragraphStack = false;
2168                                                                 $this->mLastSection = 'p';
2169                                                         } else if ($this->mLastSection !== 'p') {
2170                                                                 $output .= $this->closeParagraph().'<p>';
2171                                                                 $this->mLastSection = 'p';
2172                                                         }
2173                                                 }
2174                                         }
2175                                 }
2176                                 wfProfileOut( __METHOD__."-paragraph" );
2177                         }
2178                         // somewhere above we forget to get out of pre block (bug 785)
2179                         if($preCloseMatch && $this->mInPre) {
2180                                 $this->mInPre = false;
2181                         }
2182                         if ($paragraphStack === false) {
2183                                 $output .= $t."\n";
2184                         }
2185                 }
2186                 while ( $prefixLength ) {
2187                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2188                         --$prefixLength;
2189                 }
2190                 if ( $this->mLastSection != '' ) {
2191                         $output .= '</' . $this->mLastSection . '>';
2192                         $this->mLastSection = '';
2193                 }
2194
2195                 wfProfileOut( __METHOD__ );
2196                 return $output;
2197         }
2198
2199         /**
2200          * Split up a string on ':', ignoring any occurences inside tags
2201          * to prevent illegal overlapping.
2202          * @param string $str the string to split
2203          * @param string &$before set to everything before the ':'
2204          * @param string &$after set to everything after the ':'
2205          * return string the position of the ':', or false if none found
2206          */
2207         function findColonNoLinks($str, &$before, &$after) {
2208                 wfProfileIn( __METHOD__ );
2209
2210                 $pos = strpos( $str, ':' );
2211                 if( $pos === false ) {
2212                         // Nothing to find!
2213                         wfProfileOut( __METHOD__ );
2214                         return false;
2215                 }
2216
2217                 $lt = strpos( $str, '<' );
2218                 if( $lt === false || $lt > $pos ) {
2219                         // Easy; no tag nesting to worry about
2220                         $before = substr( $str, 0, $pos );
2221                         $after = substr( $str, $pos+1 );
2222                         wfProfileOut( __METHOD__ );
2223                         return $pos;
2224                 }
2225
2226                 // Ugly state machine to walk through avoiding tags.
2227                 $state = self::COLON_STATE_TEXT;
2228                 $stack = 0;
2229                 $len = strlen( $str );
2230                 for( $i = 0; $i < $len; $i++ ) {
2231                         $c = $str{$i};
2232
2233                         switch( $state ) {
2234                         // (Using the number is a performance hack for common cases)
2235                         case 0: // self::COLON_STATE_TEXT:
2236                                 switch( $c ) {
2237                                 case "<":
2238                                         // Could be either a <start> tag or an </end> tag
2239                                         $state = self::COLON_STATE_TAGSTART;
2240                                         break;
2241                                 case ":":
2242                                         if( $stack == 0 ) {
2243                                                 // We found it!
2244                                                 $before = substr( $str, 0, $i );
2245                                                 $after = substr( $str, $i + 1 );
2246                                                 wfProfileOut( __METHOD__ );
2247                                                 return $i;
2248                                         }
2249                                         // Embedded in a tag; don't break it.
2250                                         break;
2251                                 default:
2252                                         // Skip ahead looking for something interesting
2253                                         $colon = strpos( $str, ':', $i );
2254                                         if( $colon === false ) {
2255                                                 // Nothing else interesting
2256                                                 wfProfileOut( __METHOD__ );
2257                                                 return false;
2258                                         }
2259                                         $lt = strpos( $str, '<', $i );
2260                                         if( $stack === 0 ) {
2261                                                 if( $lt === false || $colon < $lt ) {
2262                                                         // We found it!
2263                                                         $before = substr( $str, 0, $colon );
2264                                                         $after = substr( $str, $colon + 1 );
2265                                                         wfProfileOut( __METHOD__ );
2266                                                         return $i;
2267                                                 }
2268                                         }
2269                                         if( $lt === false ) {
2270                                                 // Nothing else interesting to find; abort!
2271                                                 // We're nested, but there's no close tags left. Abort!
2272                                                 break 2;
2273                                         }
2274                                         // Skip ahead to next tag start
2275                                         $i = $lt;
2276                                         $state = self::COLON_STATE_TAGSTART;
2277                                 }
2278                                 break;
2279                         case 1: // self::COLON_STATE_TAG:
2280                                 // In a <tag>
2281                                 switch( $c ) {
2282                                 case ">":
2283                                         $stack++;
2284                                         $state = self::COLON_STATE_TEXT;
2285                                         break;
2286                                 case "/":
2287                                         // Slash may be followed by >?
2288                                         $state = self::COLON_STATE_TAGSLASH;
2289                                         break;
2290                                 default:
2291                                         // ignore
2292                                 }
2293                                 break;
2294                         case 2: // self::COLON_STATE_TAGSTART:
2295                                 switch( $c ) {
2296                                 case "/":
2297                                         $state = self::COLON_STATE_CLOSETAG;
2298                                         break;
2299                                 case "!":
2300                                         $state = self::COLON_STATE_COMMENT;
2301                                         break;
2302                                 case ">":
2303                                         // Illegal early close? This shouldn't happen D:
2304                                         $state = self::COLON_STATE_TEXT;
2305                                         break;
2306                                 default:
2307                                         $state = self::COLON_STATE_TAG;
2308                                 }
2309                                 break;
2310                         case 3: // self::COLON_STATE_CLOSETAG:
2311                                 // In a </tag>
2312                                 if( $c === ">" ) {
2313                                         $stack--;
2314                                         if( $stack < 0 ) {
2315                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2316                                                 wfProfileOut( __METHOD__ );
2317                                                 return false;
2318                                         }
2319                                         $state = self::COLON_STATE_TEXT;
2320                                 }
2321                                 break;
2322                         case self::COLON_STATE_TAGSLASH:
2323                                 if( $c === ">" ) {
2324                                         // Yes, a self-closed tag <blah/>
2325                                         $state = self::COLON_STATE_TEXT;
2326                                 } else {
2327                                         // Probably we're jumping the gun, and this is an attribute
2328                                         $state = self::COLON_STATE_TAG;
2329                                 }
2330                                 break;
2331                         case 5: // self::COLON_STATE_COMMENT:
2332                                 if( $c === "-" ) {
2333                                         $state = self::COLON_STATE_COMMENTDASH;
2334                                 }
2335                                 break;
2336                         case self::COLON_STATE_COMMENTDASH:
2337                                 if( $c === "-" ) {
2338                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2339                                 } else {
2340                                         $state = self::COLON_STATE_COMMENT;
2341                                 }
2342                                 break;
2343                         case self::COLON_STATE_COMMENTDASHDASH:
2344                                 if( $c === ">" ) {
2345                                         $state = self::COLON_STATE_TEXT;
2346                                 } else {
2347                                         $state = self::COLON_STATE_COMMENT;
2348                                 }
2349                                 break;
2350                         default:
2351                                 throw new MWException( "State machine error in " . __METHOD__ );
2352                         }
2353                 }
2354                 if( $stack > 0 ) {
2355                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2356                         return false;
2357                 }
2358                 wfProfileOut( __METHOD__ );
2359                 return false;
2360         }
2361
2362         /**
2363          * Return value of a magic variable (like PAGENAME)
2364          *
2365          * @private
2366          */
2367         function getVariableValue( $index, $frame=false ) {
2368                 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2369                 global $wgScriptPath, $wgStylePath;
2370
2371                 /**
2372                  * Some of these require message or data lookups and can be
2373                  * expensive to check many times.
2374                  */
2375                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2376                         if ( isset( $this->mVarCache[$index] ) ) {
2377                                 return $this->mVarCache[$index];
2378                         }
2379                 }
2380
2381                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2382                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2383
2384                 # Use the time zone
2385                 global $wgLocaltimezone;
2386                 if ( isset( $wgLocaltimezone ) ) {
2387                         $oldtz = date_default_timezone_get();
2388                         date_default_timezone_set( $wgLocaltimezone );
2389                 }
2390
2391                 $localTimestamp = date( 'YmdHis', $ts );
2392                 $localMonth = date( 'm', $ts );
2393                 $localMonth1 = date( 'n', $ts );
2394                 $localMonthName = date( 'n', $ts );
2395                 $localDay = date( 'j', $ts );
2396                 $localDay2 = date( 'd', $ts );
2397                 $localDayOfWeek = date( 'w', $ts );
2398                 $localWeek = date( 'W', $ts );
2399                 $localYear = date( 'Y', $ts );
2400                 $localHour = date( 'H', $ts );
2401                 if ( isset( $wgLocaltimezone ) ) {
2402                         date_default_timezone_set( $oldtz );
2403                 }
2404
2405                 switch ( $index ) {
2406                         case 'currentmonth':
2407                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2408                                 break;
2409                         case 'currentmonth1':
2410                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2411                                 break;
2412                         case 'currentmonthname':
2413                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2414                                 break;
2415                         case 'currentmonthnamegen':
2416                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2417                                 break;
2418                         case 'currentmonthabbrev':
2419                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2420                                 break;
2421                         case 'currentday':
2422                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2423                                 break;
2424                         case 'currentday2':
2425                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2426                                 break;
2427                         case 'localmonth':
2428                                 $value = $wgContLang->formatNum( $localMonth );
2429                                 break;
2430                         case 'localmonth1':
2431                                 $value = $wgContLang->formatNum( $localMonth1 );
2432                                 break;
2433                         case 'localmonthname':
2434                                 $value = $wgContLang->getMonthName( $localMonthName );
2435                                 break;
2436                         case 'localmonthnamegen':
2437                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2438                                 break;
2439                         case 'localmonthabbrev':
2440                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2441                                 break;
2442                         case 'localday':
2443                                 $value = $wgContLang->formatNum( $localDay );
2444                                 break;
2445                         case 'localday2':
2446                                 $value = $wgContLang->formatNum( $localDay2 );
2447                                 break;
2448                         case 'pagename':
2449                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2450                                 break;
2451                         case 'pagenamee':
2452                                 $value = $this->mTitle->getPartialURL();
2453                                 break;
2454                         case 'fullpagename':
2455                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2456                                 break;
2457                         case 'fullpagenamee':
2458                                 $value = $this->mTitle->getPrefixedURL();
2459                                 break;
2460                         case 'subpagename':
2461                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2462                                 break;
2463                         case 'subpagenamee':
2464                                 $value = $this->mTitle->getSubpageUrlForm();
2465                                 break;
2466                         case 'basepagename':
2467                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2468                                 break;
2469                         case 'basepagenamee':
2470                                 $value = wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2471                                 break;
2472                         case 'talkpagename':
2473                                 if( $this->mTitle->canTalk() ) {
2474                                         $talkPage = $this->mTitle->getTalkPage();
2475                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2476                                 } else {
2477                                         $value = '';
2478                                 }
2479                                 break;
2480                         case 'talkpagenamee':
2481                                 if( $this->mTitle->canTalk() ) {
2482                                         $talkPage = $this->mTitle->getTalkPage();
2483                                         $value = $talkPage->getPrefixedUrl();
2484                                 } else {
2485                                         $value = '';
2486                                 }
2487                                 break;
2488                         case 'subjectpagename':
2489                                 $subjPage = $this->mTitle->getSubjectPage();
2490                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2491                                 break;
2492                         case 'subjectpagenamee':
2493                                 $subjPage = $this->mTitle->getSubjectPage();
2494                                 $value = $subjPage->getPrefixedUrl();
2495                                 break;
2496                         case 'pipetrick':
2497                                 $text = $this->mTitle->getText();
2498                                 $value = $this->getPipeTrickText( $text );
2499                                 break;
2500                         case 'pipetricke':
2501                                 $text = $this->mTitle->getText();
2502                                 $value = wfUrlEncode( str_replace( ' ', '_', $this->getPipeTrickText( $text ) ) );
2503                                 break;
2504                         case 'revisionid':
2505                                 // Let the edit saving system know we should parse the page
2506                                 // *after* a revision ID has been assigned.
2507                                 $this->mOutput->setFlag( 'vary-revision' );
2508                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2509                                 $value = $this->mRevisionId;
2510                                 break;
2511                         case 'revisionday':
2512                                 // Let the edit saving system know we should parse the page
2513                                 // *after* a revision ID has been assigned. This is for null edits.
2514                                 $this->mOutput->setFlag( 'vary-revision' );
2515                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2516                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2517                                 break;
2518                         case 'revisionday2':
2519                                 // Let the edit saving system know we should parse the page
2520                                 // *after* a revision ID has been assigned. This is for null edits.
2521                                 $this->mOutput->setFlag( 'vary-revision' );
2522                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2523                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2524                                 break;
2525                         case 'revisionmonth':
2526                                 // Let the edit saving system know we should parse the page
2527                                 // *after* a revision ID has been assigned. This is for null edits.
2528                                 $this->mOutput->setFlag( 'vary-revision' );
2529                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2530                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2531                                 break;
2532                         case 'revisionyear':
2533                                 // Let the edit saving system know we should parse the page
2534                                 // *after* a revision ID has been assigned. This is for null edits.
2535                                 $this->mOutput->setFlag( 'vary-revision' );
2536                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2537                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2538                                 break;
2539                         case 'revisiontimestamp':
2540                                 // Let the edit saving system know we should parse the page
2541                                 // *after* a revision ID has been assigned. This is for null edits.
2542                                 $this->mOutput->setFlag( 'vary-revision' );
2543                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2544                                 $value = $this->getRevisionTimestamp();
2545                                 break;
2546                         case 'revisionuser':
2547                                 // Let the edit saving system know we should parse the page
2548                                 // *after* a revision ID has been assigned. This is for null edits.
2549                                 $this->mOutput->setFlag( 'vary-revision' );
2550                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2551                                 $value = $this->getRevisionUser();
2552                                 break;
2553                         case 'namespace':
2554                                 $value = str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2555                                 break;
2556                         case 'namespacee':
2557                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2558                                 break;
2559                         case 'talkspace':
2560                                 $value = $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2561                                 break;
2562                         case 'talkspacee':
2563                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2564                                 break;
2565                         case 'subjectspace':
2566                                 $value = $this->mTitle->getSubjectNsText();
2567                                 break;
2568                         case 'subjectspacee':
2569                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2570                                 break;
2571                         case 'currentdayname':
2572                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2573                                 break;
2574                         case 'currentyear':
2575                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2576                                 break;
2577                         case 'currenttime':
2578                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2579                                 break;
2580                         case 'currenthour':
2581                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2582                                 break;
2583                         case 'currentweek':
2584                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2585                                 // int to remove the padding
2586                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2587                                 break;
2588                         case 'currentdow':
2589                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2590                                 break;
2591                         case 'localdayname':
2592                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2593                                 break;
2594                         case 'localyear':
2595                                 $value = $wgContLang->formatNum( $localYear, true );
2596                                 break;
2597                         case 'localtime':
2598                                 $value = $wgContLang->time( $localTimestamp, false, false );
2599                                 break;
2600                         case 'localhour':
2601                                 $value = $wgContLang->formatNum( $localHour, true );
2602                                 break;
2603                         case 'localweek':
2604                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2605                                 // int to remove the padding
2606                                 $value = $wgContLang->formatNum( (int)$localWeek );
2607                                 break;
2608                         case 'localdow':
2609                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2610                                 break;
2611                         case 'numberofarticles':
2612                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2613                                 break;
2614                         case 'numberoffiles':
2615                                 $value = $wgContLang->formatNum( SiteStats::images() );
2616                                 break;
2617                         case 'numberofusers':
2618                                 $value = $wgContLang->formatNum( SiteStats::users() );
2619                                 break;
2620                         case 'numberofactiveusers':
2621                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2622                                 break;
2623                         case 'numberofpages':
2624                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2625                                 break;
2626                         case 'numberofadmins':
2627                                 $value = $wgContLang->formatNum( SiteStats::numberingroup('sysop') );
2628                                 break;
2629                         case 'numberofedits':
2630                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2631                                 break;
2632                         case 'numberofviews':
2633                                 $value = $wgContLang->formatNum( SiteStats::views() );
2634                                 break;
2635                         case 'currenttimestamp':
2636                                 $value = wfTimestamp( TS_MW, $ts );
2637                                 break;
2638                         case 'localtimestamp':
2639                                 $value = $localTimestamp;
2640                                 break;
2641                         case 'currentversion':
2642                                 $value = SpecialVersion::getVersion();
2643                                 break;
2644                         case 'sitename':
2645                                 return $wgSitename;
2646                         case 'server':
2647                                 return $wgServer;
2648                         case 'servername':
2649                                 return $wgServerName;
2650                         case 'scriptpath':
2651                                 return $wgScriptPath;
2652                         case 'stylepath':
2653                                 return $wgStylePath;
2654                         case 'directionmark':
2655                                 return $wgContLang->getDirMark();
2656                         case 'contentlanguage':
2657                                 global $wgContLanguageCode;
2658                                 return $wgContLanguageCode;
2659                         default:
2660                                 $ret = null;
2661                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) )
2662                                         return $ret;
2663                                 else
2664                                         return null;
2665                 }
2666
2667                 if ( $index )
2668                         $this->mVarCache[$index] = $value;
2669
2670                 return $value;
2671         }
2672
2673         /**
2674          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2675          *
2676          * @private
2677          */
2678         function initialiseVariables() {
2679                 wfProfileIn( __METHOD__ );
2680                 $variableIDs = MagicWord::getVariableIDs();
2681                 $substIDs = MagicWord::getSubstIDs();
2682
2683                 $this->mVariables = new MagicWordArray( $variableIDs );
2684                 $this->mSubsts = new MagicWordArray( $substIDs );
2685                 wfProfileOut( __METHOD__ );
2686         }
2687
2688         /**
2689          * Preprocess some wikitext and return the document tree.
2690          * This is the ghost of replace_variables().
2691          *
2692          * @param string $text The text to parse
2693          * @param integer flags Bitwise combination of:
2694          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2695          *                                     included. Default is to assume a direct page view.
2696          *
2697          * The generated DOM tree must depend only on the input text and the flags.
2698          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2699          *
2700          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2701          * change in the DOM tree for a given text, must be passed through the section identifier
2702          * in the section edit link and thus back to extractSections().
2703          *
2704          * The output of this function is currently only cached in process memory, but a persistent
2705          * cache may be implemented at a later date which takes further advantage of these strict
2706          * dependency requirements.
2707          *
2708          * @private
2709          */
2710         function preprocessToDom ( $text, $flags = 0 ) {
2711                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2712                 return $dom;
2713         }
2714
2715         /*
2716          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2717          */
2718         public static function splitWhitespace( $s ) {
2719                 $ltrimmed = ltrim( $s );
2720                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2721                 $trimmed = rtrim( $ltrimmed );
2722                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2723                 if ( $diff > 0 ) {
2724                         $w2 = substr( $ltrimmed, -$diff );
2725                 } else {
2726                         $w2 = '';
2727                 }
2728                 return array( $w1, $trimmed, $w2 );
2729         }
2730
2731         /**
2732          * Replace magic variables, templates, and template arguments
2733          * with the appropriate text. Templates are substituted recursively,
2734          * taking care to avoid infinite loops.
2735          *
2736          * Note that the substitution depends on value of $mOutputType:
2737          *  self::OT_WIKI: only {{subst:}} templates
2738          *  self::OT_PREPROCESS: templates but not extension tags
2739          *  self::OT_HTML: all templates and extension tags
2740          *
2741          * @param string $tex The text to transform
2742          * @param PPFrame $frame Object describing the arguments passed to the template.
2743          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2744          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2745          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2746          * @private
2747          */
2748         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2749                 # Is there any text? Also, Prevent too big inclusions!
2750                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2751                         return $text;
2752                 }
2753                 wfProfileIn( __METHOD__ );
2754
2755                 if ( $frame === false ) {
2756                         $frame = $this->getPreprocessor()->newFrame();
2757                 } elseif ( !( $frame instanceof PPFrame ) ) {
2758                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2759                         $frame = $this->getPreprocessor()->newCustomFrame($frame);
2760                 }
2761
2762                 $dom = $this->preprocessToDom( $text );
2763                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2764                 $text = $frame->expand( $dom, $flags );
2765
2766                 wfProfileOut( __METHOD__ );
2767                 return $text;
2768         }
2769
2770         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2771         static function createAssocArgs( $args ) {
2772                 $assocArgs = array();
2773                 $index = 1;
2774                 foreach( $args as $arg ) {
2775                         $eqpos = strpos( $arg, '=' );
2776                         if ( $eqpos === false ) {
2777                                 $assocArgs[$index++] = $arg;
2778                         } else {
2779                                 $name = trim( substr( $arg, 0, $eqpos ) );
2780                                 $value = trim( substr( $arg, $eqpos+1 ) );
2781                                 if ( $value === false ) {
2782                                         $value = '';
2783                                 }
2784                                 if ( $name !== false ) {
2785                                         $assocArgs[$name] = $value;
2786                                 }
2787                         }
2788                 }
2789
2790                 return $assocArgs;
2791         }
2792
2793         /**
2794          * Warn the user when a parser limitation is reached
2795          * Will warn at most once the user per limitation type
2796          *
2797          * @param string $limitationType, should be one of:
2798          *   'expensive-parserfunction' (corresponding messages: 'expensive-parserfunction-warning', 'expensive-parserfunction-category')
2799          *   'post-expand-template-argument' (corresponding messages: 'post-expand-template-argument-warning', 'post-expand-template-argument-category')
2800          *   'post-expand-template-inclusion' (corresponding messages: 'post-expand-template-inclusion-warning', 'post-expand-template-inclusion-category')
2801          * @params int $current, $max When an explicit limit has been
2802          *       exceeded, provide the values (optional)
2803          */
2804         function limitationWarn( $limitationType, $current=null, $max=null) {
2805                 //does no harm if $current and $max are present but are unnecessary for the message
2806                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
2807                 $this->mOutput->addWarning( $warning );
2808                 $this->addTrackingCategory( "$limitationType-category" );
2809         }
2810
2811         /**
2812          * Return the text of a template, after recursively
2813          * replacing any variables or templates within the template.
2814          *
2815          * @param array $piece The parts of the template
2816          *  $piece['title']: the title, i.e. the part before the |
2817          *  $piece['parts']: the parameter array
2818          *  $piece['lineStart']: whether the brace was at the start of a line
2819          * @param PPFrame The current frame, contains template arguments
2820          * @return string the text of the template
2821          * @private
2822          */
2823         function braceSubstitution( $piece, $frame ) {
2824                 global $wgContLang, $wgNonincludableNamespaces;
2825                 wfProfileIn( __METHOD__ );
2826                 wfProfileIn( __METHOD__.'-setup' );
2827
2828                 # Flags
2829                 $found = false;             # $text has been filled
2830                 $nowiki = false;            # wiki markup in $text should be escaped
2831                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2832                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2833                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2834                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2835
2836                 # Title object, where $text came from
2837                 $title = null;
2838
2839                 # $part1 is the bit before the first |, and must contain only title characters.
2840                 # Various prefixes will be stripped from it later.
2841                 $titleWithSpaces = $frame->expand( $piece['title'] );
2842                 $part1 = trim( $titleWithSpaces );
2843                 $titleText = false;
2844
2845                 # Original title text preserved for various purposes
2846                 $originalTitle = $part1;
2847
2848                 # $args is a list of argument nodes, starting from index 0, not including $part1
2849                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2850                 wfProfileOut( __METHOD__.'-setup' );
2851
2852                 # SUBST
2853                 wfProfileIn( __METHOD__.'-modifiers' );
2854                 if ( !$found ) {
2855
2856                         $substMatch = $this->mSubsts->matchStartAndRemove( $part1 );
2857
2858                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
2859                         # Whether to include depends also on whether we are in the pre-save-transform
2860                         #
2861                         # safesubst || (subst && PST) || (false && !PST) => transclude (skip the if)
2862                         # (false && PST) || (subst && !PST)  => return input (handled by if)
2863                         if ( $substMatch != 'safesubst' && ($substMatch == 'subst' xor $this->ot['wiki']) ) {
2864                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2865                                 $isLocalObj = true;
2866                                 $found = true;
2867                         }
2868                 }
2869
2870                 # Variables
2871                 if ( !$found && $args->getLength() == 0 ) {
2872                         $id = $this->mVariables->matchStartToEnd( $part1 );
2873                         if ( $id !== false ) {
2874                                 $text = $this->getVariableValue( $id, $frame );
2875                                 if (MagicWord::getCacheTTL($id)>-1)
2876                                         $this->mOutput->mContainsOldMagic = true;
2877                                 $found = true;
2878                         }
2879                 }
2880
2881                 # MSG, MSGNW and RAW
2882                 if ( !$found ) {
2883                         # Check for MSGNW:
2884                         $mwMsgnw = MagicWord::get( 'msgnw' );
2885                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2886                                 $nowiki = true;
2887                         } else {
2888                                 # Remove obsolete MSG:
2889                                 $mwMsg = MagicWord::get( 'msg' );
2890                                 $mwMsg->matchStartAndRemove( $part1 );
2891                         }
2892
2893                         # Check for RAW:
2894                         $mwRaw = MagicWord::get( 'raw' );
2895                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2896                                 $forceRawInterwiki = true;
2897                         }
2898                 }
2899                 wfProfileOut( __METHOD__.'-modifiers' );
2900
2901                 # Parser functions
2902                 if ( !$found ) {
2903                         wfProfileIn( __METHOD__ . '-pfunc' );
2904
2905                         $colonPos = strpos( $part1, ':' );
2906                         if ( $colonPos !== false ) {
2907                                 # Case sensitive functions
2908                                 $function = substr( $part1, 0, $colonPos );
2909                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2910                                         $function = $this->mFunctionSynonyms[1][$function];
2911                                 } else {
2912                                         # Case insensitive functions
2913                                         $function = $wgContLang->lc( $function );
2914                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2915                                                 $function = $this->mFunctionSynonyms[0][$function];
2916                                         } else {
2917                                                 $function = false;
2918                                         }
2919                                 }
2920                                 if ( $function ) {
2921                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2922                                         $initialArgs = array( &$this );
2923                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2924                                         if ( $flags & SFH_OBJECT_ARGS ) {
2925                                                 # Add a frame parameter, and pass the arguments as an array
2926                                                 $allArgs = $initialArgs;
2927                                                 $allArgs[] = $frame;
2928                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2929                                                         $funcArgs[] = $args->item( $i );
2930                                                 }
2931                                                 $allArgs[] = $funcArgs;
2932                                         } else {
2933                                                 # Convert arguments to plain text
2934                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2935                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2936                                                 }
2937                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2938                                         }
2939
2940                                         # Workaround for PHP bug 35229 and similar
2941                                         if ( !is_callable( $callback ) ) {
2942                                                 wfProfileOut( __METHOD__ . '-pfunc' );
2943                                                 wfProfileOut( __METHOD__ );
2944                                                 throw new MWException( "Tag hook for $function is not callable\n" );
2945                                         }
2946                                         $result = call_user_func_array( $callback, $allArgs );
2947                                         $found = true;
2948                                         $noparse = true;
2949                                         $preprocessFlags = 0;
2950
2951                                         if ( is_array( $result ) ) {
2952                                                 if ( isset( $result[0] ) ) {
2953                                                         $text = $result[0];
2954                                                         unset( $result[0] );
2955                                                 }
2956
2957                                                 // Extract flags into the local scope
2958                                                 // This allows callers to set flags such as nowiki, found, etc.
2959                                                 extract( $result );
2960                                         } else {
2961                                                 $text = $result;
2962                                         }
2963                                         if ( !$noparse ) {
2964                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
2965                                                 $isChildObj = true;
2966                                         }
2967                                 }
2968                         }
2969                         wfProfileOut( __METHOD__ . '-pfunc' );
2970                 }
2971
2972                 # Finish mangling title and then check for loops.
2973                 # Set $title to a Title object and $titleText to the PDBK
2974                 if ( !$found ) {
2975                         $ns = NS_TEMPLATE;
2976                         # Split the title into page and subpage
2977                         $subpage = '';
2978                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2979                         if ($subpage !== '') {
2980                                 $ns = $this->mTitle->getNamespace();
2981                         }
2982                         $title = Title::newFromText( $part1, $ns );
2983                         if ( $title ) {
2984                                 $titleText = $title->getPrefixedText();
2985                                 # Check for language variants if the template is not found
2986                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
2987                                         $wgContLang->findVariantLink( $part1, $title, true );
2988                                 }
2989                                 # Do recursion depth check
2990                                 $limit = $this->mOptions->getMaxTemplateDepth();
2991                                 if ( $frame->depth >= $limit ) {
2992                                         $found = true;
2993                                         $text = '<span class="error">' . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit ) . '</span>';
2994                                 }
2995                         }
2996                 }
2997
2998                 # Load from database
2999                 if ( !$found && $title ) {
3000                         wfProfileIn( __METHOD__ . '-loadtpl' );
3001                         if ( !$title->isExternal() ) {
3002                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
3003                                         $text = SpecialPage::capturePath( $title );
3004                                         if ( is_string( $text ) ) {
3005                                                 $found = true;
3006                                                 $isHTML = true;
3007                                                 $this->disableCache();
3008                                         }
3009                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3010                                         $found = false; //access denied
3011                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3012                                 } else {
3013                                         list( $text, $title ) = $this->getTemplateDom( $title );
3014                                         if ( $text !== false ) {
3015                                                 $found = true;
3016                                                 $isChildObj = true;
3017                                         }
3018                                 }
3019
3020                                 # If the title is valid but undisplayable, make a link to it
3021                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3022                                         $text = "[[:$titleText]]";
3023                                         $found = true;
3024                                 }
3025                         } elseif ( $title->isTrans() ) {
3026                                 // Interwiki transclusion
3027                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3028                                         $text = $this->interwikiTransclude( $title, 'render' );
3029                                         $isHTML = true;
3030                                 } else {
3031                                         $text = $this->interwikiTransclude( $title, 'raw' );
3032                                         // Preprocess it like a template
3033                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3034                                         $isChildObj = true;
3035                                 }
3036                                 $found = true;
3037                         }
3038
3039                         # Do infinite loop check
3040                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3041                         if ( !$frame->loopCheck( $title ) ) {
3042                                 $found = true;
3043                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3044                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3045                         }
3046                         wfProfileOut( __METHOD__ . '-loadtpl' );
3047                 }
3048
3049                 # If we haven't found text to substitute by now, we're done
3050                 # Recover the source wikitext and return it
3051                 if ( !$found ) {
3052                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3053                         wfProfileOut( __METHOD__ );
3054                         return array( 'object' => $text );
3055                 }
3056
3057                 # Expand DOM-style return values in a child frame
3058                 if ( $isChildObj ) {
3059                         # Clean up argument array
3060                         $newFrame = $frame->newChild( $args, $title );
3061
3062                         if ( $nowiki ) {
3063                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3064                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3065                                 # Expansion is eligible for the empty-frame cache
3066                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3067                                         $text = $this->mTplExpandCache[$titleText];
3068                                 } else {
3069                                         $text = $newFrame->expand( $text );
3070                                         $this->mTplExpandCache[$titleText] = $text;
3071                                 }
3072                         } else {
3073                                 # Uncached expansion
3074                                 $text = $newFrame->expand( $text );
3075                         }
3076                 }
3077                 if ( $isLocalObj && $nowiki ) {
3078                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3079                         $isLocalObj = false;
3080                 }
3081
3082                 # Replace raw HTML by a placeholder
3083                 # Add a blank line preceding, to prevent it from mucking up
3084                 # immediately preceding headings
3085                 if ( $isHTML ) {
3086                         $text = "\n\n" . $this->insertStripItem( $text );
3087                 }
3088                 # Escape nowiki-style return values
3089                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3090                         $text = wfEscapeWikiText( $text );
3091                 }
3092                 # Bug 529: if the template begins with a table or block-level
3093                 # element, it should be treated as beginning a new line.
3094                 # This behaviour is somewhat controversial.
3095                 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3096                         $text = "\n" . $text;
3097                 }
3098
3099                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3100                         # Error, oversize inclusion
3101                         $text = "[[$originalTitle]]" .
3102                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3103                         $this->limitationWarn( 'post-expand-template-inclusion' );
3104                 }
3105
3106                 if ( $isLocalObj ) {
3107                         $ret = array( 'object' => $text );
3108                 } else {
3109                         $ret = array( 'text' => $text );
3110                 }
3111
3112                 wfProfileOut( __METHOD__ );
3113                 return $ret;
3114         }
3115
3116         /**
3117          * Get the semi-parsed DOM representation of a template with a given title,
3118          * and its redirect destination title. Cached.
3119          */
3120         function getTemplateDom( $title ) {
3121                 $cacheTitle = $title;
3122                 $titleText = $title->getPrefixedDBkey();
3123
3124                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3125                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3126                         $title = Title::makeTitle( $ns, $dbk );
3127                         $titleText = $title->getPrefixedDBkey();
3128                 }
3129                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3130                         return array( $this->mTplDomCache[$titleText], $title );
3131                 }
3132
3133                 // Cache miss, go to the database
3134                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3135
3136                 if ( $text === false ) {
3137                         $this->mTplDomCache[$titleText] = false;
3138                         return array( false, $title );
3139                 }
3140
3141                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3142                 $this->mTplDomCache[ $titleText ] = $dom;
3143
3144                 if (! $title->equals($cacheTitle)) {
3145                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3146                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3147                 }
3148
3149                 return array( $dom, $title );
3150         }
3151
3152         /**
3153          * Fetch the unparsed text of a template and register a reference to it.
3154          */
3155         function fetchTemplateAndTitle( $title ) {
3156                 $templateCb = $this->mOptions->getTemplateCallback();
3157                 $stuff = call_user_func( $templateCb, $title, $this );
3158                 $text = $stuff['text'];
3159                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3160                 if ( isset( $stuff['deps'] ) ) {
3161                         foreach ( $stuff['deps'] as $dep ) {
3162                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3163                         }
3164                 }
3165                 return array($text,$finalTitle);
3166         }
3167
3168         function fetchTemplate( $title ) {
3169                 $rv = $this->fetchTemplateAndTitle($title);
3170                 return $rv[0];
3171         }
3172
3173         /**
3174          * Static function to get a template
3175          * Can be overridden via ParserOptions::setTemplateCallback().
3176          */
3177         static function statelessFetchTemplate( $title, $parser=false ) {
3178                 $text = $skip = false;
3179                 $finalTitle = $title;
3180                 $deps = array();
3181
3182                 // Loop to fetch the article, with up to 1 redirect
3183                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3184                         # Give extensions a chance to select the revision instead
3185                         $id = false; // Assume current
3186                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3187
3188                         if( $skip ) {
3189                                 $text = false;
3190                                 $deps[] = array(
3191                                         'title' => $title,
3192                                         'page_id' => $title->getArticleID(),
3193                                         'rev_id' => null );
3194                                 break;
3195                         }
3196                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3197                         $rev_id = $rev ? $rev->getId() : 0;
3198                         // If there is no current revision, there is no page
3199                         if( $id === false && !$rev ) {
3200                                 $linkCache = LinkCache::singleton();
3201                                 $linkCache->addBadLinkObj( $title );
3202                         }
3203
3204                         $deps[] = array(
3205                                 'title' => $title,
3206                                 'page_id' => $title->getArticleID(),
3207                                 'rev_id' => $rev_id );
3208
3209                         if( $rev ) {
3210                                 $text = $rev->getText();
3211                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3212                                 global $wgContLang;
3213                                 $message = $wgContLang->lcfirst( $title->getText() );
3214                                 $text = wfMsgForContentNoTrans( $message );
3215                                 if( wfEmptyMsg( $message, $text ) ) {
3216                                         $text = false;
3217                                         break;
3218                                 }
3219                         } else {
3220                                 break;
3221                         }
3222                         if ( $text === false ) {
3223                                 break;
3224                         }
3225                         // Redirect?
3226                         $finalTitle = $title;
3227                         $title = Title::newFromRedirect( $text );
3228                 }
3229                 return array(
3230                         'text' => $text,
3231                         'finalTitle' => $finalTitle,
3232                         'deps' => $deps );
3233         }
3234
3235         /**
3236          * Transclude an interwiki link.
3237          */
3238         function interwikiTransclude( $title, $action ) {
3239                 global $wgEnableScaryTranscluding;
3240
3241                 if (!$wgEnableScaryTranscluding)
3242                         return wfMsg('scarytranscludedisabled');
3243
3244                 $url = $title->getFullUrl( "action=$action" );
3245
3246                 if (strlen($url) > 255)
3247                         return wfMsg('scarytranscludetoolong');
3248                 return $this->fetchScaryTemplateMaybeFromCache($url);
3249         }
3250
3251         function fetchScaryTemplateMaybeFromCache($url) {
3252                 global $wgTranscludeCacheExpiry;
3253                 $dbr = wfGetDB(DB_SLAVE);
3254                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3255                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3256                                 array('tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3257                 if ($obj) {
3258                         return $obj->tc_contents;
3259                 }
3260
3261                 $text = Http::get($url);
3262                 if (!$text)
3263                         return wfMsg('scarytranscludefailed', $url);
3264
3265                 $dbw = wfGetDB(DB_MASTER);
3266                 $dbw->replace('transcache', array('tc_url'), array(
3267                         'tc_url' => $url,
3268                         'tc_time' => $dbw->timestamp( time() ),
3269                         'tc_contents' => $text));
3270                 return $text;
3271         }
3272
3273
3274         /**
3275          * Triple brace replacement -- used for template arguments
3276          * @private
3277          */
3278         function argSubstitution( $piece, $frame ) {
3279                 wfProfileIn( __METHOD__ );
3280
3281                 $error = false;
3282                 $parts = $piece['parts'];
3283                 $nameWithSpaces = $frame->expand( $piece['title'] );
3284                 $argName = trim( $nameWithSpaces );
3285                 $object = false;
3286                 $text = $frame->getArgument( $argName );
3287                 if (  $text === false && $parts->getLength() > 0
3288                   && (
3289                     $this->ot['html']
3290                     || $this->ot['pre']
3291                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3292                   )
3293                 ) {
3294                         # No match in frame, use the supplied default
3295                         $object = $parts->item( 0 )->getChildren();
3296                 }
3297                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3298                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3299                         $this->limitationWarn( 'post-expand-template-argument' );
3300                 }
3301
3302                 if ( $text === false && $object === false ) {
3303                         # No match anywhere
3304                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3305                 }
3306                 if ( $error !== false ) {
3307                         $text .= $error;
3308                 }
3309                 if ( $object !== false ) {
3310                         $ret = array( 'object' => $object );
3311                 } else {
3312                         $ret = array( 'text' => $text );
3313                 }
3314
3315                 wfProfileOut( __METHOD__ );
3316                 return $ret;
3317         }
3318
3319         /**
3320          * Return the text to be used for a given extension tag.
3321          * This is the ghost of strip().
3322          *
3323          * @param array $params Associative array of parameters:
3324          *     name       PPNode for the tag name
3325          *     attr       PPNode for unparsed text where tag attributes are thought to be
3326          *     attributes Optional associative array of parsed attributes
3327          *     inner      Contents of extension element
3328          *     noClose    Original text did not have a close tag
3329          * @param PPFrame $frame
3330          */
3331         function extensionSubstitution( $params, $frame ) {
3332                 global $wgRawHtml, $wgContLang;
3333
3334                 $name = $frame->expand( $params['name'] );
3335                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3336                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3337                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $this->mMarkerIndex++) . self::MARKER_SUFFIX;
3338
3339                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3340                         ( $this->ot['html'] || $this->ot['pre'] );
3341                 if ( $isFunctionTag ) {
3342                         $markerType = 'none';
3343                 } else {
3344                         $markerType = 'general';
3345                 }
3346                 if ( $this->ot['html'] || $isFunctionTag ) {
3347                         $name = strtolower( $name );
3348                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3349                         if ( isset( $params['attributes'] ) ) {
3350                                 $attributes = $attributes + $params['attributes'];
3351                         }
3352
3353                         if( isset( $this->mTagHooks[$name] ) ) {
3354                                 # Workaround for PHP bug 35229 and similar
3355                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3356                                         throw new MWException( "Tag hook for $name is not callable\n" );
3357                                 }
3358                                 $output = call_user_func_array( $this->mTagHooks[$name],
3359                                         array( $content, $attributes, $this, $frame ) );
3360                         } elseif( isset( $this->mFunctionTagHooks[$name] ) ) {
3361                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3362                                 if( !is_callable( $callback ) )
3363                                         throw new MWException( "Tag hook for $name is not callable\n" );
3364
3365                                 $output = call_user_func_array( $callback,
3366                                         array( &$this, $frame, $content, $attributes ) );
3367                         } else {
3368                                 $output = '<span class="error">Invalid tag extension name: ' .
3369                                         htmlspecialchars( $name ) . '</span>';
3370                         }
3371
3372                         if ( is_array( $output ) ) {
3373                                 // Extract flags to local scope (to override $markerType)
3374                                 $flags = $output;
3375                                 $output = $flags[0];
3376                                 unset( $flags[0] );
3377                                 extract( $flags );
3378                         }
3379                 } else {
3380                         if ( is_null( $attrText ) ) {
3381                                 $attrText = '';
3382                         }
3383                         if ( isset( $params['attributes'] ) ) {
3384                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3385                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3386                                                 htmlspecialchars( $attrValue ) . '"';
3387                                 }
3388                         }
3389                         if ( $content === null ) {
3390                                 $output = "<$name$attrText/>";
3391                         } else {
3392                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3393                                 $output = "<$name$attrText>$content$close";
3394                         }
3395                 }
3396
3397                 if( $markerType === 'none' ) {
3398                         return $output;
3399                 } elseif ( $markerType === 'nowiki' ) {
3400                         $this->mStripState->nowiki->setPair( $marker, $output );
3401                 } elseif ( $markerType === 'general' ) {
3402                         $this->mStripState->general->setPair( $marker, $output );
3403                 } else {
3404                         throw new MWException( __METHOD__.': invalid marker type' );
3405                 }
3406                 return $marker;
3407         }
3408
3409         /**
3410          * Increment an include size counter
3411          *
3412          * @param string $type The type of expansion
3413          * @param integer $size The size of the text
3414          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3415          */
3416         function incrementIncludeSize( $type, $size ) {
3417                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3418                         return false;
3419                 } else {
3420                         $this->mIncludeSizes[$type] += $size;
3421                         return true;
3422                 }
3423         }
3424
3425         /**
3426          * Increment the expensive function count
3427          *
3428          * @return boolean False if the limit has been exceeded
3429          */
3430         function incrementExpensiveFunctionCount() {
3431                 global $wgExpensiveParserFunctionLimit;
3432                 $this->mExpensiveFunctionCount++;
3433                 if($this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit) {
3434                         return true;
3435                 }
3436                 return false;
3437         }
3438
3439         /**
3440          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3441          * Fills $this->mDoubleUnderscores, returns the modified text
3442          */
3443         function doDoubleUnderscore( $text ) {
3444                 wfProfileIn( __METHOD__ );
3445
3446                 // The position of __TOC__ needs to be recorded
3447                 $mw = MagicWord::get( 'toc' );
3448                 if( $mw->match( $text ) ) {
3449                         $this->mShowToc = true;
3450                         $this->mForceTocPosition = true;
3451
3452                         // Set a placeholder. At the end we'll fill it in with the TOC.
3453                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3454
3455                         // Only keep the first one.
3456                         $text = $mw->replace( '', $text );
3457                 }
3458
3459                 // Now match and remove the rest of them
3460                 $mwa = MagicWord::getDoubleUnderscoreArray();
3461                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3462
3463                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3464                         $this->mOutput->mNoGallery = true;
3465                 }
3466                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3467                         $this->mShowToc = false;
3468                 }
3469                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3470                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3471                         $this->addTrackingCategory( 'hidden-category-category' );
3472                 }
3473                 # (bug 8068) Allow control over whether robots index a page.
3474                 #
3475                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3476                 # is not desirable, the last one on the page should win.
3477                 if( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3478                         $this->mOutput->setIndexPolicy( 'noindex' );
3479                         $this->addTrackingCategory( 'noindex-category' );
3480                 }
3481                 if( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ){
3482                         $this->mOutput->setIndexPolicy( 'index' );
3483                         $this->addTrackingCategory( 'index-category' );
3484                 }
3485
3486                 wfProfileOut( __METHOD__ );
3487                 return $text;
3488         }
3489
3490         /**
3491          * Add a tracking category, getting the title from a system message,
3492          * or print a debug message if the title is invalid.
3493          * @param $msg String message key
3494          * @return Bool whether the addition was successful
3495          */
3496         protected function addTrackingCategory( $msg ){
3497                 $cat = wfMsgForContent( $msg );
3498
3499                 # Allow tracking categories to be disabled by setting them to "-"
3500                 if( $cat === '-' ) return false;
3501
3502                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3503                 if ( $containerCategory ) {
3504                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3505                         return true;
3506                 } else {
3507                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3508                         return false;
3509                 }
3510         }
3511
3512         /**
3513          * This function accomplishes several tasks:
3514          * 1) Auto-number headings if that option is enabled
3515          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3516          * 3) Add a Table of contents on the top for users who have enabled the option
3517          * 4) Auto-anchor headings
3518          *
3519          * It loops through all headlines, collects the necessary data, then splits up the
3520          * string and re-inserts the newly formatted headlines.
3521          *
3522          * @param string $text
3523          * @param string $origText Original, untouched wikitext
3524          * @param boolean $isMain
3525          * @private
3526          */
3527         function formatHeadings( $text, $origText, $isMain=true ) {
3528                 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3529
3530                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3531                 $showEditLink = $this->mOptions->getEditSection();
3532
3533                 // Do not call quickUserCan unless necessary
3534                 if( $showEditLink && !$this->mTitle->quickUserCan( 'edit' ) ) {
3535                         $showEditLink = 0;
3536                 }
3537
3538                 # Inhibit editsection links if requested in the page
3539                 if ( isset( $this->mDoubleUnderscores['noeditsection'] )  || $this->mOptions->getIsPrintable() ) {
3540                         $showEditLink = 0;
3541                 }
3542
3543                 # Get all headlines for numbering them and adding funky stuff like [edit]
3544                 # links - this is for later, but we need the number of headlines right now
3545                 $matches = array();
3546                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3547
3548                 # if there are fewer than 4 headlines in the article, do not show TOC
3549                 # unless it's been explicitly enabled.
3550                 $enoughToc = $this->mShowToc &&
3551                         (($numMatches >= 4) || $this->mForceTocPosition);
3552
3553                 # Allow user to stipulate that a page should have a "new section"
3554                 # link added via __NEWSECTIONLINK__
3555                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3556                         $this->mOutput->setNewSection( true );
3557                 }
3558
3559                 # Allow user to remove the "new section"
3560                 # link via __NONEWSECTIONLINK__
3561                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3562                         $this->mOutput->hideNewSection( true );
3563                 }
3564
3565                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3566                 # override above conditions and always show TOC above first header
3567                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3568                         $this->mShowToc = true;
3569                         $enoughToc = true;
3570                 }
3571
3572                 # We need this to perform operations on the HTML
3573                 $sk = $this->mOptions->getSkin();
3574
3575                 # headline counter
3576                 $headlineCount = 0;
3577                 $numVisible = 0;
3578
3579                 # Ugh .. the TOC should have neat indentation levels which can be
3580                 # passed to the skin functions. These are determined here
3581                 $toc = '';
3582                 $full = '';
3583                 $head = array();
3584                 $sublevelCount = array();
3585                 $levelCount = array();
3586                 $toclevel = 0;
3587                 $level = 0;
3588                 $prevlevel = 0;
3589                 $toclevel = 0;
3590                 $prevtoclevel = 0;
3591                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3592                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3593                 $oldType = $this->mOutputType;
3594                 $this->setOutputType( self::OT_WIKI );
3595                 $frame = $this->getPreprocessor()->newFrame();
3596                 $root = $this->preprocessToDom( $origText );
3597                 $node = $root->getFirstChild();
3598                 $byteOffset = 0;
3599                 $tocraw = array();
3600
3601                 foreach( $matches[3] as $headline ) {
3602                         $isTemplate = false;
3603                         $titleText = false;
3604                         $sectionIndex = false;
3605                         $numbering = '';
3606                         $markerMatches = array();
3607                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3608                                 $serial = $markerMatches[1];
3609                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3610                                 $isTemplate = ($titleText != $baseTitleText);
3611                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3612                         }
3613
3614                         if( $toclevel ) {
3615                                 $prevlevel = $level;
3616                                 $prevtoclevel = $toclevel;
3617                         }
3618                         $level = $matches[1][$headlineCount];
3619
3620                         if ( $level > $prevlevel ) {
3621                                 # Increase TOC level
3622                                 $toclevel++;
3623                                 $sublevelCount[$toclevel] = 0;
3624                                 if( $toclevel<$wgMaxTocLevel ) {
3625                                         $prevtoclevel = $toclevel;
3626                                         $toc .= $sk->tocIndent();
3627                                         $numVisible++;
3628                                 }
3629                         }
3630                         elseif ( $level < $prevlevel && $toclevel > 1 ) {
3631                                 # Decrease TOC level, find level to jump to
3632
3633                                 for ($i = $toclevel; $i > 0; $i--) {
3634                                         if ( $levelCount[$i] == $level ) {
3635                                                 # Found last matching level
3636                                                 $toclevel = $i;
3637                                                 break;
3638                                         }
3639                                         elseif ( $levelCount[$i] < $level ) {
3640                                                 # Found first matching level below current level
3641                                                 $toclevel = $i + 1;
3642                                                 break;
3643                                         }
3644                                 }
3645                                 if( $i == 0 ) $toclevel = 1;
3646                                 if( $toclevel<$wgMaxTocLevel ) {
3647                                         if($prevtoclevel < $wgMaxTocLevel) {
3648                                                 # Unindent only if the previous toc level was shown :p
3649                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3650                                                 $prevtoclevel = $toclevel;
3651                                         } else {
3652                                                 $toc .= $sk->tocLineEnd();
3653                                         }
3654                                 }
3655                         }
3656                         else {
3657                                 # No change in level, end TOC line
3658                                 if( $toclevel<$wgMaxTocLevel ) {
3659                                         $toc .= $sk->tocLineEnd();
3660                                 }
3661                         }
3662
3663                         $levelCount[$toclevel] = $level;
3664
3665                         # count number of headlines for each level
3666                         @$sublevelCount[$toclevel]++;
3667                         $dot = 0;
3668                         for( $i = 1; $i <= $toclevel; $i++ ) {
3669                                 if( !empty( $sublevelCount[$i] ) ) {
3670                                         if( $dot ) {
3671                                                 $numbering .= '.';
3672                                         }
3673                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3674                                         $dot = 1;
3675                                 }
3676                         }
3677
3678                         # The safe header is a version of the header text safe to use for links
3679                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3680                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3681
3682                         # Remove link placeholders by the link text.
3683                         #     <!--LINK number-->
3684                         # turns into
3685                         #     link text with suffix
3686                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3687
3688                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3689                         $tocline = preg_replace(
3690                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3691                                 array( '',                          '<$1>'),
3692                                 $safeHeadline
3693                         );
3694                         $tocline = trim( $tocline );
3695
3696                         # For the anchor, strip out HTML-y stuff period
3697                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3698                         $safeHeadline = preg_replace( '/[ _]+/', ' ', $safeHeadline );
3699                         $safeHeadline = trim( $safeHeadline );
3700
3701                         # Save headline for section edit hint before it's escaped
3702                         $headlineHint = $safeHeadline;
3703
3704                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
3705                                 # For reverse compatibility, provide an id that's
3706                                 # HTML4-compatible, like we used to.
3707                                 #
3708                                 # It may be worth noting, academically, that it's possible for
3709                                 # the legacy anchor to conflict with a non-legacy headline
3710                                 # anchor on the page.  In this case likely the "correct" thing
3711                                 # would be to either drop the legacy anchors or make sure
3712                                 # they're numbered first.  However, this would require people
3713                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3714                                 # manually, so let's not bother worrying about it.
3715                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3716                                         array( 'noninitial', 'legacy' ) );
3717                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3718
3719                                 if ( $legacyHeadline == $safeHeadline ) {
3720                                         # No reason to have both (in fact, we can't)
3721                                         $legacyHeadline = false;
3722                                 }
3723                         } else {
3724                                 $legacyHeadline = false;
3725                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3726                                         'noninitial' );
3727                         }
3728
3729                         # HTML names must be case-insensitively unique (bug 10721).  FIXME:
3730                         # Does this apply to Unicode characters?  Because we aren't
3731                         # handling those here.
3732                         $arrayKey = strtolower( $safeHeadline );
3733                         if ( $legacyHeadline === false ) {
3734                                 $legacyArrayKey = false;
3735                         } else {
3736                                 $legacyArrayKey = strtolower( $legacyHeadline );
3737                         }
3738
3739                         # count how many in assoc. array so we can track dupes in anchors
3740                         if ( isset( $refers[$arrayKey] ) ) {
3741                                 $refers[$arrayKey]++;
3742                         } else {
3743                                 $refers[$arrayKey] = 1;
3744                         }
3745                         if ( isset( $refers[$legacyArrayKey] ) ) {
3746                                 $refers[$legacyArrayKey]++;
3747                         } else {
3748                                 $refers[$legacyArrayKey] = 1;
3749                         }
3750
3751                         # Don't number the heading if it is the only one (looks silly)
3752                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3753                                 # the two are different if the line contains a link
3754                                 $headline = $numbering . ' ' . $headline;
3755                         }
3756
3757                         # Create the anchor for linking from the TOC to the section
3758                         $anchor = $safeHeadline;
3759                         $legacyAnchor = $legacyHeadline;
3760                         if ( $refers[$arrayKey] > 1 ) {
3761                                 $anchor .= '_' . $refers[$arrayKey];
3762                         }
3763                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3764                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3765                         }
3766                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3767                                 $toc .= $sk->tocLine($anchor, $tocline,
3768                                         $numbering, $toclevel, ($isTemplate ? false : $sectionIndex));
3769                         }
3770
3771                         # Add the section to the section tree
3772                         # Find the DOM node for this header
3773                         while ( $node && !$isTemplate ) {
3774                                 if ( $node->getName() === 'h' ) {
3775                                         $bits = $node->splitHeading();
3776                                         if ( $bits['i'] == $sectionIndex )
3777                                                 break;
3778                                 }
3779                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
3780                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
3781                                 $node = $node->getNextSibling();
3782                         }
3783                         $tocraw[] = array(
3784                                 'toclevel' => $toclevel,
3785                                 'level' => $level,
3786                                 'line' => $tocline,
3787                                 'number' => $numbering,
3788                                 'index' => ($isTemplate ? 'T-' : '' ) . $sectionIndex,
3789                                 'fromtitle' => $titleText,
3790                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
3791                                 'anchor' => $anchor,
3792                         );
3793
3794                         # give headline the correct <h#> tag
3795                         if( $showEditLink && $sectionIndex !== false ) {
3796                                 if( $isTemplate ) {
3797                                         # Put a T flag in the section identifier, to indicate to extractSections()
3798                                         # that sections inside <includeonly> should be counted.
3799                                         $editlink = $sk->doEditSectionLink(Title::newFromText( $titleText ), "T-$sectionIndex");
3800                                 } else {
3801                                         $editlink = $sk->doEditSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3802                                 }
3803                         } else {
3804                                 $editlink = '';
3805                         }
3806                         $head[$headlineCount] = $sk->makeHeadline( $level,
3807                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3808                                 $editlink, $legacyAnchor );
3809
3810                         $headlineCount++;
3811                 }
3812
3813                 $this->setOutputType( $oldType );
3814
3815                 # Never ever show TOC if no headers
3816                 if( $numVisible < 1 ) {
3817                         $enoughToc = false;
3818                 }
3819
3820                 if( $enoughToc ) {
3821                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3822                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3823                         }
3824                         $toc = $sk->tocList( $toc );
3825                         $this->mOutput->setTOCHTML( $toc );
3826                 }
3827
3828                 if ( $isMain ) {
3829                         $this->mOutput->setSections( $tocraw );
3830                 }
3831
3832                 # split up and insert constructed headlines
3833
3834                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3835                 $i = 0;
3836
3837                 foreach( $blocks as $block ) {
3838                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3839                                 # This is the [edit] link that appears for the top block of text when
3840                                 # section editing is enabled
3841
3842                                 # Disabled because it broke block formatting
3843                                 # For example, a bullet point in the top line
3844                                 # $full .= $sk->editSectionLink(0);
3845                         }
3846                         $full .= $block;
3847                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3848                                 # Top anchor now in skin
3849                                 $full = $full.$toc;
3850                         }
3851
3852                         if( !empty( $head[$i] ) ) {
3853                                 $full .= $head[$i];
3854                         }
3855                         $i++;
3856                 }
3857                 if( $this->mForceTocPosition ) {
3858                         return str_replace( '<!--MWTOC-->', $toc, $full );
3859                 } else {
3860                         return $full;
3861                 }
3862         }
3863
3864         /**
3865          * Merge $tree2 into $tree1 by replacing the section with index
3866          * $section in $tree1 and its descendants with the sections in $tree2.
3867          * Note that in the returned section tree, only the 'index' and
3868          * 'byteoffset' fields are guaranteed to be correct.
3869          * @param $tree1 array Section tree from ParserOutput::getSectons()
3870         *  @param $tree2 array Section tree
3871          * @param $section int Section index
3872          * @param $title Title Title both section trees come from
3873          * @param $len2 int Length of the original wikitext for $tree2
3874          * @return array Merged section tree
3875          */
3876         public static function mergeSectionTrees( $tree1, $tree2, $section, $title, $len2 ) {
3877                 global $wgContLang;
3878                 $newTree = array();
3879                 $targetLevel = false;
3880                 $merged = false;
3881                 $lastLevel = 1;
3882                 $nextIndex = 1;
3883                 $numbering = array( 0 );
3884                 $titletext = $title->getPrefixedDBkey();
3885                 foreach ( $tree1 as $s ) {
3886                         if ( $targetLevel !== false ) {
3887                                 if ( $s['level'] <= $targetLevel )
3888                                         // We've skipped enough
3889                                         $targetLevel = false;
3890                                 else
3891                                         continue;
3892                         }
3893                         if ( $s['index'] != $section ||
3894                                         $s['fromtitle'] != $titletext ) {
3895                                 self::incrementNumbering( $numbering,
3896                                         $s['toclevel'], $lastLevel );
3897
3898                                 // Rewrite index, byteoffset and number
3899                                 if ( $s['fromtitle'] == $titletext ) {
3900                                         $s['index'] = $nextIndex++;
3901                                         if ( $merged )
3902                                                 $s['byteoffset'] += $len2;
3903                                 }
3904                                 $s['number']  = implode( '.', array_map(
3905                                         array( $wgContLang, 'formatnum' ),
3906                                         $numbering ) );
3907                                 $lastLevel = $s['toclevel'];
3908                                 $newTree[] = $s;
3909                         } else {
3910                                 // We're at $section
3911                                 // Insert sections from $tree2 here
3912                                 foreach ( $tree2 as $s2 ) {
3913                                         // Rewrite the fields in $s2
3914                                         // before inserting it
3915                                         $s2['toclevel'] += $s['toclevel'] - 1;
3916                                         $s2['level'] += $s['level'] - 1;
3917                                         $s2['index'] = $nextIndex++;
3918                                         $s2['byteoffset'] += $s['byteoffset'];
3919
3920                                         self::incrementNumbering( $numbering,
3921                                                 $s2['toclevel'], $lastLevel );
3922                                         $s2['number']  = implode( '.', array_map(
3923                                                 array( $wgContLang, 'formatnum' ),
3924                                                 $numbering ) );
3925                                         $lastLevel = $s2['toclevel'];
3926                                         $newTree[] = $s2;
3927                                 }
3928                                 // Skip all descendants of $section in $tree1
3929                                 $targetLevel = $s['level'];
3930                                 $merged = true;
3931                         }
3932                 }
3933                 return $newTree;
3934         }
3935
3936         /**
3937          * Increment a section number. Helper function for mergeSectionTrees()
3938          * @param $number array Array representing a section number
3939          * @param $level int Current TOC level (depth)
3940          * @param $lastLevel int Level of previous TOC entry
3941          */
3942         private static function incrementNumbering( &$number, $level, $lastLevel ) {
3943                 if ( $level > $lastLevel )
3944                         $number[$level - 1] = 1;
3945                 else if ( $level < $lastLevel ) {
3946                         foreach ( $number as $key => $unused )
3947                                 if ( $key >= $level )
3948                                         unset( $number[$key] );
3949                         $number[$level - 1]++;
3950                 } else
3951                         $number[$level - 1]++;
3952         }
3953
3954         /**
3955          * Transform wiki markup when saving a page by doing \r\n -> \n
3956          * conversion, substitting signatures, {{subst:}} templates, etc.
3957          *
3958          * @param string $text the text to transform
3959          * @param Title &$title the Title object for the current article
3960          * @param User $user the User object describing the current user
3961          * @param ParserOptions $options parsing options
3962          * @param bool $clearState whether to clear the parser state first
3963          * @return string the altered wiki markup
3964          * @public
3965          */
3966         function preSaveTransform( $text, Title $title, $user, $options, $clearState = true ) {
3967                 $this->mOptions = $options;
3968                 $this->setTitle( $title );
3969                 $this->setOutputType( self::OT_WIKI );
3970
3971                 if ( $clearState ) {
3972                         $this->clearState();
3973                 }
3974
3975                 $pairs = array(
3976                         "\r\n" => "\n",
3977                 );
3978                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3979                 $text = $this->pstPass2( $text, $user );
3980                 $text = $this->mStripState->unstripBoth( $text );
3981                 return $text;
3982         }
3983
3984         /**
3985          * Pre-save transform helper function
3986          * @private
3987          */
3988         function pstPass2( $text, $user ) {
3989                 global $wgContLang, $wgLocaltimezone;
3990
3991                 /* Note: This is the timestamp saved as hardcoded wikitext to
3992                  * the database, we use $wgContLang here in order to give
3993                  * everyone the same signature and use the default one rather
3994                  * than the one selected in each user's preferences.
3995                  *
3996                  * (see also bug 12815)
3997                  */
3998                 $ts = $this->mOptions->getTimestamp();
3999                 if ( isset( $wgLocaltimezone ) ) {
4000                         $tz = $wgLocaltimezone;
4001                 } else {
4002                         $tz = date_default_timezone_get();
4003                 }
4004
4005                 $unixts = wfTimestamp( TS_UNIX, $ts );
4006                 $oldtz = date_default_timezone_get();
4007                 date_default_timezone_set( $tz );
4008                 $ts = date( 'YmdHis', $unixts );
4009                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4010
4011                 /* Allow translation of timezones trough wiki. date() can return
4012                  * whatever crap the system uses, localised or not, so we cannot
4013                  * ship premade translations.
4014                  */
4015                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4016                 $value = wfMsgForContent( $key );
4017                 if ( !wfEmptyMsg( $key, $value ) ) $tzMsg = $value;
4018
4019                 date_default_timezone_set( $oldtz );
4020
4021                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4022
4023                 # Variable replacement
4024                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4025                 $text = $this->replaceVariables( $text );
4026
4027                 # Signatures
4028                 $sigText = $this->getUserSig( $user );
4029                 $text = strtr( $text, array(
4030                         '~~~~~' => $d,
4031                         '~~~~' => "$sigText $d",
4032                         '~~~' => $sigText
4033                 ) );
4034
4035                 # Links of the form [[|<blah>]] or [[<blah>|]] perform pipe tricks
4036                 # Note this only allows the # in the one position it works.
4037                 global $wgLegalTitleChars;
4038                 $pipeTrickRe = "/\[\[(?:(\\|)([$wgLegalTitleChars]+)|(#?[$wgLegalTitleChars]+)\\|)\]\]/";
4039                 $text = preg_replace_callback( $pipeTrickRe, array( $this, 'pstPipeTrickCallback' ), $text );
4040
4041                 # Trim trailing whitespace
4042                 $text = rtrim( $text );
4043
4044                 return $text;
4045         }
4046
4047         /**
4048          * Called from pstPass2 to perform the pipe trick on links.
4049          * Original was either [[|text]] or [[link|]]
4050          *
4051          * @param Array ("|" or "", text, link) $m
4052          */
4053         function pstPipeTrickCallback( $m )
4054         {
4055                 if( $m[1] ) { # [[|<blah>]]
4056                         $text = $m[2];
4057                         $link = $this->getPipeTrickLink( $text );
4058                 } else { # [[<blah>|]]
4059                         $link = $m[3];
4060                         $text = $this->getPipeTrickText( $link );
4061                 }
4062
4063                 return $link === $text ? "[[$link]]" : "[[$link|$text]]";
4064         }
4065
4066         /**
4067          * Fetch the user's signature text, if any, and normalize to
4068          * validated, ready-to-insert wikitext.
4069          * If you have pre-fetched the nickname or the fancySig option, you can
4070          * specify them here to save a database query.
4071          *
4072          * @param User $user
4073          * @return string
4074          */
4075         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4076                 global $wgMaxSigChars;
4077
4078                 $username = $user->getName();
4079
4080                 // If not given, retrieve from the user object.
4081                 if ( $nickname === false )
4082                         $nickname = $user->getOption( 'nickname' );
4083
4084                 if ( is_null( $fancySig ) )
4085                         $fancySig = $user->getBoolOption( 'fancysig' );
4086
4087                 $nickname = $nickname == null ? $username : $nickname;
4088
4089                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4090                         $nickname = $username;
4091                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4092                 } elseif( $fancySig !== false ) {
4093                         # Sig. might contain markup; validate this
4094                         if( $this->validateSig( $nickname ) !== false ) {
4095                                 # Validated; clean up (if needed) and return it
4096                                 return $this->cleanSig( $nickname, true );
4097                         } else {
4098                                 # Failed to validate; fall back to the default
4099                                 $nickname = $username;
4100                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4101                         }
4102                 }
4103
4104                 // Make sure nickname doesnt get a sig in a sig
4105                 $nickname = $this->cleanSigInSig( $nickname );
4106
4107                 # If we're still here, make it a link to the user page
4108                 $userText = wfEscapeWikiText( $username );
4109                 $nickText = wfEscapeWikiText( $nickname );
4110                 if ( $user->isAnon() )  {
4111                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4112                 } else {
4113                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4114                 }
4115         }
4116
4117         /**
4118          * Check that the user's signature contains no bad XML
4119          *
4120          * @param string $text
4121          * @return mixed An expanded string, or false if invalid.
4122          */
4123         function validateSig( $text ) {
4124                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4125         }
4126
4127         /**
4128          * Clean up signature text
4129          *
4130          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4131          * 2) Substitute all transclusions
4132          *
4133          * @param string $text
4134          * @param $parsing Whether we're cleaning (preferences save) or parsing
4135          * @return string Signature text
4136          */
4137         function cleanSig( $text, $parsing = false ) {
4138                 if ( !$parsing ) {
4139                         global $wgTitle;
4140                         $this->clearState();
4141                         $this->setTitle( $wgTitle );
4142                         $this->mOptions = new ParserOptions;
4143                         $this->setOutputType = self::OT_PREPROCESS;
4144                 }
4145
4146                 # Option to disable this feature
4147                 if ( !$this->mOptions->getCleanSignatures() ) {
4148                         return $text;
4149                 }
4150
4151                 # FIXME: regex doesn't respect extension tags or nowiki
4152                 #  => Move this logic to braceSubstitution()
4153                 $substWord = MagicWord::get( 'subst' );
4154                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4155                 $substText = '{{' . $substWord->getSynonym( 0 );
4156
4157                 $text = preg_replace( $substRegex, $substText, $text );
4158                 $text = $this->cleanSigInSig( $text );
4159                 $dom = $this->preprocessToDom( $text );
4160                 $frame = $this->getPreprocessor()->newFrame();
4161                 $text = $frame->expand( $dom );
4162
4163                 if ( !$parsing ) {
4164                         $text = $this->mStripState->unstripBoth( $text );
4165                 }
4166
4167                 return $text;
4168         }
4169
4170         /**
4171          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4172          * @param string $text
4173          * @return string Signature text with /~{3,5}/ removed
4174          */
4175         function cleanSigInSig( $text ) {
4176                 $text = preg_replace( '/~{3,5}/', '', $text );
4177                 return $text;
4178         }
4179
4180         /**
4181          * Set up some variables which are usually set up in parse()
4182          * so that an external function can call some class members with confidence
4183          * @public
4184          */
4185         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4186                 $this->setTitle( $title );
4187                 $this->mOptions = $options;
4188                 $this->setOutputType( $outputType );
4189                 if ( $clearState ) {
4190                         $this->clearState();
4191                 }
4192         }
4193
4194         /**
4195          * Wrapper for preprocess()
4196          *
4197          * @param string $text the text to preprocess
4198          * @param ParserOptions $options  options
4199          * @return string
4200          * @public
4201          */
4202         function transformMsg( $text, $options ) {
4203                 global $wgTitle;
4204                 static $executing = false;
4205
4206                 # Guard against infinite recursion
4207                 if ( $executing ) {
4208                         return $text;
4209                 }
4210                 $executing = true;
4211
4212                 wfProfileIn(__METHOD__);
4213                 $text = $this->preprocess( $text, $wgTitle, $options );
4214
4215                 $executing = false;
4216                 wfProfileOut(__METHOD__);
4217                 return $text;
4218         }
4219
4220         /**
4221          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4222          * The callback should have the following form:
4223          *    function myParserHook( $text, $params, &$parser ) { ... }
4224          *
4225          * Transform and return $text. Use $parser for any required context, e.g. use
4226          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4227          *
4228          * @public
4229          *
4230          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4231          * @param mixed $callback The callback function (and object) to use for the tag
4232          *
4233          * @return The old value of the mTagHooks array associated with the hook
4234          */
4235         function setHook( $tag, $callback ) {
4236                 $tag = strtolower( $tag );
4237                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4238                 $this->mTagHooks[$tag] = $callback;
4239                 if( !in_array( $tag, $this->mStripList ) ) {
4240                         $this->mStripList[] = $tag;
4241                 }
4242
4243                 return $oldVal;
4244         }
4245
4246         function setTransparentTagHook( $tag, $callback ) {
4247                 $tag = strtolower( $tag );
4248                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4249                 $this->mTransparentTagHooks[$tag] = $callback;
4250
4251                 return $oldVal;
4252         }
4253
4254         /**
4255          * Remove all tag hooks
4256          */
4257         function clearTagHooks() {
4258                 $this->mTagHooks = array();
4259                 $this->mStripList = $this->mDefaultStripList;
4260         }
4261
4262         /**
4263          * Create a function, e.g. {{sum:1|2|3}}
4264          * The callback function should have the form:
4265          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4266          *
4267          * Or with SFH_OBJECT_ARGS:
4268          *    function myParserFunction( $parser, $frame, $args ) { ... }
4269          *
4270          * The callback may either return the text result of the function, or an array with the text
4271          * in element 0, and a number of flags in the other elements. The names of the flags are
4272          * specified in the keys. Valid flags are:
4273          *   found                     The text returned is valid, stop processing the template. This
4274          *                             is on by default.
4275          *   nowiki                    Wiki markup in the return value should be escaped
4276          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4277          *
4278          * @public
4279          *
4280          * @param string $id The magic word ID
4281          * @param mixed $callback The callback function (and object) to use
4282          * @param integer $flags a combination of the following flags:
4283          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4284          *
4285          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4286          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4287          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4288          *     the arguments, and to control the way they are expanded.
4289          *
4290          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4291          *     arguments, for instance:
4292          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4293          *
4294          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4295          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4296          *     working if/when this is changed.
4297          *
4298          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4299          *     expansion.
4300          *
4301          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4302          *     about the methods available in PPFrame and PPNode.
4303          *
4304          * @return The old callback function for this name, if any
4305          */
4306         function setFunctionHook( $id, $callback, $flags = 0 ) {
4307                 global $wgContLang;
4308
4309                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4310                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4311
4312                 # Add to function cache
4313                 $mw = MagicWord::get( $id );
4314                 if( !$mw )
4315                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4316
4317                 $synonyms = $mw->getSynonyms();
4318                 $sensitive = intval( $mw->isCaseSensitive() );
4319
4320                 foreach ( $synonyms as $syn ) {
4321                         # Case
4322                         if ( !$sensitive ) {
4323                                 $syn = $wgContLang->lc( $syn );
4324                         }
4325                         # Add leading hash
4326                         if ( !( $flags & SFH_NO_HASH ) ) {
4327                                 $syn = '#' . $syn;
4328                         }
4329                         # Remove trailing colon
4330                         if ( substr( $syn, -1, 1 ) === ':' ) {
4331                                 $syn = substr( $syn, 0, -1 );
4332                         }
4333                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4334                 }
4335                 return $oldVal;
4336         }
4337
4338         /**
4339          * Get all registered function hook identifiers
4340          *
4341          * @return array
4342          */
4343         function getFunctionHooks() {
4344                 return array_keys( $this->mFunctionHooks );
4345         }
4346
4347         /**
4348          * Create a tag function, e.g. <test>some stuff</test>.
4349          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4350          * Unlike parser functions, their content is not preprocessed.
4351          */
4352         function setFunctionTagHook( $tag, $callback, $flags ) {
4353                 $tag = strtolower( $tag );
4354                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4355                         $this->mFunctionTagHooks[$tag] : null;
4356                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4357
4358                 if( !in_array( $tag, $this->mStripList ) ) {
4359                         $this->mStripList[] = $tag;
4360                 }
4361
4362                 return $old;
4363         }
4364
4365         /**
4366          * FIXME: update documentation. makeLinkObj() is deprecated.
4367          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4368          * Placeholders created in Skin::makeLinkObj()
4369          * Returns an array of link CSS classes, indexed by PDBK.
4370          */
4371         function replaceLinkHolders( &$text, $options = 0 ) {
4372                 return $this->mLinkHolders->replace( $text );
4373         }
4374
4375         /**
4376          * Replace <!--LINK--> link placeholders with plain text of links
4377          * (not HTML-formatted).
4378          * @param string $text
4379          * @return string
4380          */
4381         function replaceLinkHoldersText( $text ) {
4382                 return $this->mLinkHolders->replaceText( $text );
4383         }
4384
4385         /**
4386          * Renders an image gallery from a text with one line per image.
4387          * text labels may be given by using |-style alternative text. E.g.
4388          *   Image:one.jpg|The number "1"
4389          *   Image:tree.jpg|A tree
4390          * given as text will return the HTML of a gallery with two images,
4391          * labeled 'The number "1"' and
4392          * 'A tree'.
4393          */
4394         function renderImageGallery( $text, $params ) {
4395                 $ig = new ImageGallery();
4396                 $ig->setContextTitle( $this->mTitle );
4397                 $ig->setShowBytes( false );
4398                 $ig->setShowFilename( false );
4399                 $ig->setParser( $this );
4400                 $ig->setHideBadImages();
4401                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4402                 $ig->useSkin( $this->mOptions->getSkin() );
4403                 $ig->mRevisionId = $this->mRevisionId;
4404
4405                 if( isset( $params['caption'] ) ) {
4406                         $caption = $params['caption'];
4407                         $caption = htmlspecialchars( $caption );
4408                         $caption = $this->replaceInternalLinks( $caption );
4409                         $ig->setCaptionHtml( $caption );
4410                 }
4411                 if( isset( $params['perrow'] ) ) {
4412                         $ig->setPerRow( $params['perrow'] );
4413                 }
4414                 if( isset( $params['widths'] ) ) {
4415                         $ig->setWidths( $params['widths'] );
4416                 }
4417                 if( isset( $params['heights'] ) ) {
4418                         $ig->setHeights( $params['heights'] );
4419                 }
4420
4421                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4422
4423                 $lines = StringUtils::explode( "\n", $text );
4424                 foreach ( $lines as $line ) {
4425                         # match lines like these:
4426                         # Image:someimage.jpg|This is some image
4427                         $matches = array();
4428                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4429                         # Skip empty lines
4430                         if ( count( $matches ) == 0 ) {
4431                                 continue;
4432                         }
4433
4434                         if ( strpos( $matches[0], '%' ) !== false )
4435                                 $matches[1] = urldecode( $matches[1] );
4436                         $tp = Title::newFromText( $matches[1]/*, NS_FILE*/ );
4437                         $nt =& $tp;
4438                         if( is_null( $nt ) ) {
4439                                 # Bogus title. Ignore these so we don't bomb out later.
4440                                 continue;
4441                         }
4442                         if ( isset( $matches[3] ) ) {
4443                                 $label = $matches[3];
4444                         } else {
4445                                 $label = '';
4446                         }
4447
4448                         $html = $this->recursiveTagParse( trim( $label ) );
4449
4450                         $ig->add( $nt, $html );
4451
4452                         # Only add real images (bug #5586)
4453                         if ( $nt->getNamespace() == NS_FILE ) {
4454                                 $this->mOutput->addImage( $nt->getDBkey() );
4455                         }
4456                 }
4457                 return $ig->toHTML();
4458         }
4459
4460         function getImageParams( $handler ) {
4461                 if ( $handler ) {
4462                         $handlerClass = get_class( $handler );
4463                 } else {
4464                         $handlerClass = '';
4465                 }
4466                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4467                         // Initialise static lists
4468                         static $internalParamNames = array(
4469                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4470                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4471                                         'bottom', 'text-bottom' ),
4472                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4473                                         'upright', 'border', 'link', 'alt' ),
4474                         );
4475                         static $internalParamMap;
4476                         if ( !$internalParamMap ) {
4477                                 $internalParamMap = array();
4478                                 foreach ( $internalParamNames as $type => $names ) {
4479                                         foreach ( $names as $name ) {
4480                                                 $magicName = str_replace( '-', '_', "img_$name" );
4481                                                 $internalParamMap[$magicName] = array( $type, $name );
4482                                         }
4483                                 }
4484                         }
4485
4486                         // Add handler params
4487                         $paramMap = $internalParamMap;
4488                         if ( $handler ) {
4489                                 $handlerParamMap = $handler->getParamMap();
4490                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4491                                         $paramMap[$magic] = array( 'handler', $paramName );
4492                                 }
4493                         }
4494                         $this->mImageParams[$handlerClass] = $paramMap;
4495                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4496                 }
4497                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4498         }
4499
4500         /**
4501          * Parse image options text and use it to make an image
4502          * @param Title $title
4503          * @param string $options
4504          * @param LinkHolderArray $holders
4505          */
4506         function makeImage( $title, $options, $holders = false ) {
4507                 # Check if the options text is of the form "options|alt text"
4508                 # Options are:
4509                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4510                 #  * left       no resizing, just left align. label is used for alt= only
4511                 #  * right      same, but right aligned
4512                 #  * none       same, but not aligned
4513                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4514                 #  * center     center the image
4515                 #  * frame      Keep original image size, no magnify-button.
4516                 #  * framed     Same as "frame"
4517                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4518                 #  * upright    reduce width for upright images, rounded to full __0 px
4519                 #  * border     draw a 1px border around the image
4520                 #  * alt        Text for HTML alt attribute (defaults to empty)
4521                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4522                 # vertical-align values (no % or length right now):
4523                 #  * baseline
4524                 #  * sub
4525                 #  * super
4526                 #  * top
4527                 #  * text-top
4528                 #  * middle
4529                 #  * bottom
4530                 #  * text-bottom
4531
4532                 $parts = StringUtils::explode( "|", $options );
4533                 $sk = $this->mOptions->getSkin();
4534
4535                 # Give extensions a chance to select the file revision for us
4536                 $skip = $time = $descQuery = false;
4537                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4538
4539                 if ( $skip ) {
4540                         return $sk->link( $title );
4541                 }
4542
4543                 # Get the file
4544                 $imagename = $title->getDBkey();
4545                 $file = wfFindFile( $title, array( 'time' => $time ) );
4546                 # Get parameter map
4547                 $handler = $file ? $file->getHandler() : false;
4548
4549                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4550
4551                 # Process the input parameters
4552                 $caption = '';
4553                 $params = array( 'frame' => array(), 'handler' => array(),
4554                         'horizAlign' => array(), 'vertAlign' => array() );
4555                 foreach( $parts as $part ) {
4556                         $part = trim( $part );
4557                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4558                         $validated = false;
4559                         if( isset( $paramMap[$magicName] ) ) {
4560                                 list( $type, $paramName ) = $paramMap[$magicName];
4561
4562                                 // Special case; width and height come in one variable together
4563                                 if( $type === 'handler' && $paramName === 'width' ) {
4564                                         $m = array();
4565                                         # (bug 13500) In both cases (width/height and width only),
4566                                         # permit trailing "px" for backward compatibility.
4567                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4568                                                 $width = intval( $m[1] );
4569                                                 $height = intval( $m[2] );
4570                                                 if ( $handler->validateParam( 'width', $width ) ) {
4571                                                         $params[$type]['width'] = $width;
4572                                                         $validated = true;
4573                                                 }
4574                                                 if ( $handler->validateParam( 'height', $height ) ) {
4575                                                         $params[$type]['height'] = $height;
4576                                                         $validated = true;
4577                                                 }
4578                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4579                                                 $width = intval( $value );
4580                                                 if ( $handler->validateParam( 'width', $width ) ) {
4581                                                         $params[$type]['width'] = $width;
4582                                                         $validated = true;
4583                                                 }
4584                                         } // else no validation -- bug 13436
4585                                 } else {
4586                                         if ( $type === 'handler' ) {
4587                                                 # Validate handler parameter
4588                                                 $validated = $handler->validateParam( $paramName, $value );
4589                                         } else {
4590                                                 # Validate internal parameters
4591                                                 switch( $paramName ) {
4592                                                 case 'manualthumb':
4593                                                 case 'alt':
4594                                                         // @todo Fixme: possibly check validity here for
4595                                                         // manualthumb? downstream behavior seems odd with
4596                                                         // missing manual thumbs.
4597                                                         $validated = true;
4598                                                         $value = $this->stripAltText( $value, $holders );
4599                                                         break;
4600                                                 case 'link':
4601                                                         $chars = self::EXT_LINK_URL_CLASS;
4602                                                         $prots = $this->mUrlProtocols;
4603                                                         if ( $value === '' ) {
4604                                                                 $paramName = 'no-link';
4605                                                                 $value = true;
4606                                                                 $validated = true;
4607                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4608                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4609                                                                         $paramName = 'link-url';
4610                                                                         $this->mOutput->addExternalLink( $value );
4611                                                                         $validated = true;
4612                                                                 }
4613                                                         } else {
4614                                                                 $linkTitle = Title::newFromText( $value );
4615                                                                 if ( $linkTitle ) {
4616                                                                         $paramName = 'link-title';
4617                                                                         $value = $linkTitle;
4618                                                                         $this->mOutput->addLink( $linkTitle );
4619                                                                         $validated = true;
4620                                                                 }
4621                                                         }
4622                                                         break;
4623                                                 default:
4624                                                         // Most other things appear to be empty or numeric...
4625                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4626                                                 }
4627                                         }
4628
4629                                         if ( $validated ) {
4630                                                 $params[$type][$paramName] = $value;
4631                                         }
4632                                 }
4633                         }
4634                         if ( !$validated ) {
4635                                 $caption = $part;
4636                         }
4637                 }
4638
4639                 # Process alignment parameters
4640                 if ( $params['horizAlign'] ) {
4641                         $params['frame']['align'] = key( $params['horizAlign'] );
4642                 }
4643                 if ( $params['vertAlign'] ) {
4644                         $params['frame']['valign'] = key( $params['vertAlign'] );
4645                 }
4646
4647                 $params['frame']['caption'] = $caption;
4648
4649                 # Will the image be presented in a frame, with the caption below?
4650                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4651                                  isset( $params['frame']['framed'] ) ||
4652                                  isset( $params['frame']['thumbnail'] ) ||
4653                                  isset( $params['frame']['manualthumb'] );
4654
4655                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4656                 # came to also set the caption, ordinary text after the image -- which
4657                 # makes no sense, because that just repeats the text multiple times in
4658                 # screen readers.  It *also* came to set the title attribute.
4659                 #
4660                 # Now that we have an alt attribute, we should not set the alt text to
4661                 # equal the caption: that's worse than useless, it just repeats the
4662                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4663                 # use the unnamed parameter for alt text as well, just for the time be-
4664                 # ing, if the unnamed param is set and the alt param is not.
4665                 #
4666                 # For the future, we need to figure out if we want to tweak this more,
4667                 # e.g., introducing a title= parameter for the title; ignoring the un-
4668                 # named parameter entirely for images without a caption; adding an ex-
4669                 # plicit caption= parameter and preserving the old magic unnamed para-
4670                 # meter for BC; ...
4671                 if ( $imageIsFramed ) { # Framed image
4672                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4673                                 # No caption or alt text, add the filename as the alt text so
4674                                 # that screen readers at least get some description of the image
4675                                 $params['frame']['alt'] = $title->getText();
4676                         }
4677                         # Do not set $params['frame']['title'] because tooltips don't make sense
4678                         # for framed images
4679                 } else { # Inline image
4680                         if ( !isset( $params['frame']['alt'] ) ) {
4681                                 # No alt text, use the "caption" for the alt text
4682                                 if ( $caption !== '') {
4683                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4684                                 } else {
4685                                         # No caption, fall back to using the filename for the
4686                                         # alt text
4687                                         $params['frame']['alt'] = $title->getText();
4688                                 }
4689                         }
4690                         # Use the "caption" for the tooltip text
4691                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4692                 }
4693
4694                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4695
4696                 # Linker does the rest
4697                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4698
4699                 # Give the handler a chance to modify the parser object
4700                 if ( $handler ) {
4701                         $handler->parserTransformHook( $this, $file );
4702                 }
4703
4704                 return $ret;
4705         }
4706
4707         protected function stripAltText( $caption, $holders ) {
4708                 # Strip bad stuff out of the title (tooltip).  We can't just use
4709                 # replaceLinkHoldersText() here, because if this function is called
4710                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4711                 if ( $holders ) {
4712                         $tooltip = $holders->replaceText( $caption );
4713                 } else {
4714                         $tooltip = $this->replaceLinkHoldersText( $caption );
4715                 }
4716
4717                 # make sure there are no placeholders in thumbnail attributes
4718                 # that are later expanded to html- so expand them now and
4719                 # remove the tags
4720                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4721                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4722
4723                 return $tooltip;
4724         }
4725
4726         /**
4727          * Set a flag in the output object indicating that the content is dynamic and
4728          * shouldn't be cached.
4729          */
4730         function disableCache() {
4731                 wfDebug( "Parser output marked as uncacheable.\n" );
4732                 $this->mOutput->mCacheTime = -1;
4733         }
4734
4735         /**#@+
4736          * Callback from the Sanitizer for expanding items found in HTML attribute
4737          * values, so they can be safely tested and escaped.
4738          * @param string $text
4739          * @param PPFrame $frame
4740          * @return string
4741          * @private
4742          */
4743         function attributeStripCallback( &$text, $frame = false ) {
4744                 $text = $this->replaceVariables( $text, $frame );
4745                 $text = $this->mStripState->unstripBoth( $text );
4746                 return $text;
4747         }
4748
4749         /**#@-*/
4750
4751         /**#@+
4752          * Accessor/mutator
4753          */
4754         function Title( $x = null ) { return wfSetVar( $this->mTitle, $x ); }
4755         function Options( $x = null ) { return wfSetVar( $this->mOptions, $x ); }
4756         function OutputType( $x = null ) { return wfSetVar( $this->mOutputType, $x ); }
4757         /**#@-*/
4758
4759         /**#@+
4760          * Accessor
4761          */
4762         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4763         /**#@-*/
4764
4765
4766         /**
4767          * Break wikitext input into sections, and either pull or replace
4768          * some particular section's text.
4769          *
4770          * External callers should use the getSection and replaceSection methods.
4771          *
4772          * @param string $text Page wikitext
4773          * @param string $section A section identifier string of the form:
4774          *   <flag1> - <flag2> - ... - <section number>
4775          *
4776          * Currently the only recognised flag is "T", which means the target section number
4777          * was derived during a template inclusion parse, in other words this is a template
4778          * section edit link. If no flags are given, it was an ordinary section edit link.
4779          * This flag is required to avoid a section numbering mismatch when a section is
4780          * enclosed by <includeonly> (bug 6563).
4781          *
4782          * The section number 0 pulls the text before the first heading; other numbers will
4783          * pull the given section along with its lower-level subsections. If the section is
4784          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4785          *
4786          * @param string $mode One of "get" or "replace"
4787          * @param string $newText Replacement text for section data.
4788          * @return string for "get", the extracted section text.
4789          *                for "replace", the whole page with the section replaced.
4790          */
4791         private function extractSections( $text, $section, $mode, $newText='' ) {
4792                 global $wgTitle;
4793                 $this->clearState();
4794                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4795                 $this->mOptions = new ParserOptions;
4796                 $this->setOutputType( self::OT_WIKI );
4797                 $outText = '';
4798                 $frame = $this->getPreprocessor()->newFrame();
4799
4800                 // Process section extraction flags
4801                 $flags = 0;
4802                 $sectionParts = explode( '-', $section );
4803                 $sectionIndex = array_pop( $sectionParts );
4804                 foreach ( $sectionParts as $part ) {
4805                         if ( $part === 'T' ) {
4806                                 $flags |= self::PTD_FOR_INCLUSION;
4807                         }
4808                 }
4809                 // Preprocess the text
4810                 $root = $this->preprocessToDom( $text, $flags );
4811
4812                 // <h> nodes indicate section breaks
4813                 // They can only occur at the top level, so we can find them by iterating the root's children
4814                 $node = $root->getFirstChild();
4815
4816                 // Find the target section
4817                 if ( $sectionIndex == 0 ) {
4818                         // Section zero doesn't nest, level=big
4819                         $targetLevel = 1000;
4820                 } else {
4821             while ( $node ) {
4822                 if ( $node->getName() === 'h' ) {
4823                     $bits = $node->splitHeading();
4824                                         if ( $bits['i'] == $sectionIndex ) {
4825                                         $targetLevel = $bits['level'];
4826                                                 break;
4827                                         }
4828                                 }
4829                                 if ( $mode === 'replace' ) {
4830                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4831                                 }
4832                                 $node = $node->getNextSibling();
4833                         }
4834                 }
4835
4836                 if ( !$node ) {
4837                         // Not found
4838                         if ( $mode === 'get' ) {
4839                                 return $newText;
4840                         } else {
4841                                 return $text;
4842                         }
4843                 }
4844
4845                 // Find the end of the section, including nested sections
4846                 do {
4847                         if ( $node->getName() === 'h' ) {
4848                                 $bits = $node->splitHeading();
4849                                 $curLevel = $bits['level'];
4850                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4851                                         break;
4852                                 }
4853                         }
4854                         if ( $mode === 'get' ) {
4855                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4856                         }
4857                         $node = $node->getNextSibling();
4858                 } while ( $node );
4859
4860                 // Write out the remainder (in replace mode only)
4861                 if ( $mode === 'replace' ) {
4862                         // Output the replacement text
4863                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4864                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4865                         // Only add trailing whitespace if there is newText
4866                         if($newText != "") {
4867                                 $outText .= $newText . "\n\n";
4868                         }
4869
4870                         while ( $node ) {
4871                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4872                                 $node = $node->getNextSibling();
4873                         }
4874                 }
4875
4876                 if ( is_string( $outText ) ) {
4877                         // Re-insert stripped tags
4878                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4879                 }
4880
4881                 return $outText;
4882         }
4883
4884         /**
4885          * This function returns the text of a section, specified by a number ($section).
4886          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4887          * the first section before any such heading (section 0).
4888          *
4889          * If a section contains subsections, these are also returned.
4890          *
4891          * @param string $text text to look in
4892          * @param string $section section identifier
4893          * @param string $deftext default to return if section is not found
4894          * @return string text of the requested section
4895          */
4896         public function getSection( $text, $section, $deftext='' ) {
4897                 return $this->extractSections( $text, $section, "get", $deftext );
4898         }
4899
4900         public function replaceSection( $oldtext, $section, $text ) {
4901                 return $this->extractSections( $oldtext, $section, "replace", $text );
4902         }
4903
4904         /**
4905          * Get the timestamp associated with the current revision, adjusted for
4906          * the default server-local timestamp
4907          */
4908         function getRevisionTimestamp() {
4909                 if ( is_null( $this->mRevisionTimestamp ) ) {
4910                         wfProfileIn( __METHOD__ );
4911                         global $wgContLang;
4912                         $dbr = wfGetDB( DB_SLAVE );
4913                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4914                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4915
4916                         // Normalize timestamp to internal MW format for timezone processing.
4917                         // This has the added side-effect of replacing a null value with
4918                         // the current time, which gives us more sensible behavior for
4919                         // previews.
4920                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4921
4922                         // The cryptic '' timezone parameter tells to use the site-default
4923                         // timezone offset instead of the user settings.
4924                         //
4925                         // Since this value will be saved into the parser cache, served
4926                         // to other users, and potentially even used inside links and such,
4927                         // it needs to be consistent for all visitors.
4928                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4929
4930                         wfProfileOut( __METHOD__ );
4931                 }
4932                 return $this->mRevisionTimestamp;
4933         }
4934
4935         /**
4936          * Get the name of the user that edited the last revision
4937          */
4938         function getRevisionUser() {
4939                 // if this template is subst: the revision id will be blank,
4940                 // so just use the current user's name
4941                 if( $this->mRevisionId ) {
4942                         $revision = Revision::newFromId( $this->mRevisionId );
4943                         $revuser = $revision->getUserText();
4944                 } else {
4945                         global $wgUser;
4946                         $revuser = $wgUser->getName();
4947                 }
4948                 return $revuser;
4949         }
4950
4951         /**
4952          * Mutator for $mDefaultSort
4953          *
4954          * @param $sort New value
4955          */
4956         public function setDefaultSort( $sort ) {
4957                 $this->mDefaultSort = $sort;
4958         }
4959
4960         /**
4961          * Accessor for $mDefaultSort
4962          * Will use the title/prefixed title if none is set
4963          *
4964          * @return string
4965          */
4966         public function getDefaultSort() {
4967                 global $wgCategoryPrefixedDefaultSortkey;
4968                 if( $this->mDefaultSort !== false ) {
4969                         return $this->mDefaultSort;
4970                 } elseif ($this->mTitle->getNamespace() == NS_CATEGORY ||
4971                         !$wgCategoryPrefixedDefaultSortkey) {
4972                         return $this->mTitle->getText();
4973                 } else {
4974                         return $this->mTitle->getPrefixedText();
4975                 }
4976         }
4977
4978         /**
4979          * Accessor for $mDefaultSort
4980          * Unlike getDefaultSort(), will return false if none is set
4981          *
4982          * @return string or false
4983          */
4984         public function getCustomDefaultSort() {
4985                 return $this->mDefaultSort;
4986         }
4987
4988         /**
4989          * Try to guess the section anchor name based on a wikitext fragment
4990          * presumably extracted from a heading, for example "Header" from
4991          * "== Header ==".
4992          */
4993         public function guessSectionNameFromWikiText( $text ) {
4994                 # Strip out wikitext links(they break the anchor)
4995                 $text = $this->stripSectionName( $text );
4996                 $headline = Sanitizer::decodeCharReferences( $text );
4997                 # strip out HTML
4998                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
4999                 $headline = trim( $headline );
5000                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
5001                 $replacearray = array(
5002                         '%3A' => ':',
5003                         '%' => '.'
5004                 );
5005                 return str_replace(
5006                         array_keys( $replacearray ),
5007                         array_values( $replacearray ),
5008                         $sectionanchor );
5009         }
5010
5011         /**
5012          * Strips a text string of wikitext for use in a section anchor
5013          *
5014          * Accepts a text string and then removes all wikitext from the
5015          * string and leaves only the resultant text (i.e. the result of
5016          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5017          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5018          * to create valid section anchors by mimicing the output of the
5019          * parser when headings are parsed.
5020          *
5021          * @param $text string Text string to be stripped of wikitext
5022          * for use in a Section anchor
5023          * @return Filtered text string
5024          */
5025         public function stripSectionName( $text ) {
5026                 # Strip internal link markup
5027                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
5028                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
5029
5030                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5031                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5032                 # on how many empty links there are on the page - need to figure that out.
5033                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
5034
5035                 # Parse wikitext quotes (italics & bold)
5036                 $text = $this->doQuotes($text);
5037
5038                 # Strip HTML tags
5039                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5040                 return $text;
5041         }
5042
5043         function srvus( $text ) {
5044                 return $this->testSrvus( $text, $this->mOutputType );
5045         }
5046
5047         /**
5048          * strip/replaceVariables/unstrip for preprocessor regression testing
5049          */
5050         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
5051                 $this->clearState();
5052                 if ( ! ( $title instanceof Title ) ) {
5053                         $title = Title::newFromText( $title );
5054                 }
5055                 $this->mTitle = $title;
5056                 $this->mOptions = $options;
5057                 $this->setOutputType( $outputType );
5058                 $text = $this->replaceVariables( $text );
5059                 $text = $this->mStripState->unstripBoth( $text );
5060                 $text = Sanitizer::removeHTMLtags( $text );
5061                 return $text;
5062         }
5063
5064         function testPst( $text, $title, $options ) {
5065                 global $wgUser;
5066                 if ( ! ( $title instanceof Title ) ) {
5067                         $title = Title::newFromText( $title );
5068                 }
5069                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5070         }
5071
5072         function testPreprocess( $text, $title, $options ) {
5073                 if ( ! ( $title instanceof Title ) ) {
5074                         $title = Title::newFromText( $title );
5075                 }
5076                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5077         }
5078
5079         function markerSkipCallback( $s, $callback ) {
5080                 $i = 0;
5081                 $out = '';
5082                 while ( $i < strlen( $s ) ) {
5083                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5084                         if ( $markerStart === false ) {
5085                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5086                                 break;
5087                         } else {
5088                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5089                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5090                                 if ( $markerEnd === false ) {
5091                                         $out .= substr( $s, $markerStart );
5092                                         break;
5093                                 } else {
5094                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5095                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5096                                         $i = $markerEnd;
5097                                 }
5098                         }
5099                 }
5100                 return $out;
5101         }
5102
5103         function serialiseHalfParsedText( $text ) {
5104                 $data = array();
5105                 $data['text'] = $text;
5106
5107                 // First, find all strip markers, and store their
5108                 //  data in an array.
5109                 $stripState = new StripState;
5110                 $pos = 0;
5111                 while( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) ) && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) ) {
5112                         $end_pos += strlen( self::MARKER_SUFFIX );
5113                         $marker = substr( $text, $start_pos, $end_pos-$start_pos );
5114
5115                         if ( !empty( $this->mStripState->general->data[$marker] ) ) {
5116                                 $replaceArray = $stripState->general;
5117                                 $stripText = $this->mStripState->general->data[$marker];
5118                         } elseif ( !empty( $this->mStripState->nowiki->data[$marker] ) ) {
5119                                 $replaceArray = $stripState->nowiki;
5120                                 $stripText = $this->mStripState->nowiki->data[$marker];
5121                         } else {
5122                                 throw new MWException( "Hanging strip marker: '$marker'." );
5123                         }
5124
5125                         $replaceArray->setPair( $marker, $stripText );
5126                         $pos = $end_pos;
5127                 }
5128                 $data['stripstate'] = $stripState;
5129
5130                 // Now, find all of our links, and store THEIR
5131                 //  data in an array! :)
5132                 $links = array( 'internal' => array(), 'interwiki' => array() );
5133                 $pos = 0;
5134
5135                 // Internal links
5136                 while( ( $start_pos = strpos( $text, '<!--LINK ', $pos ) ) ) {
5137                         list( $ns, $trail ) = explode( ':', substr( $text, $start_pos + strlen( '<!--LINK ' ) ), 2 );
5138
5139                         $ns = trim($ns);
5140                         if (empty( $links['internal'][$ns] )) {
5141                                 $links['internal'][$ns] = array();
5142                         }
5143
5144                         $key = trim( substr( $trail, 0, strpos( $trail, '-->' ) ) );
5145                         $links['internal'][$ns][] = $this->mLinkHolders->internals[$ns][$key];
5146                         $pos = $start_pos + strlen( "<!--LINK $ns:$key-->" );
5147                 }
5148
5149                 $pos = 0;
5150
5151                 // Interwiki links
5152                 while( ( $start_pos = strpos( $text, '<!--IWLINK ', $pos ) ) ) {
5153                         $data = substr( $text, $start_pos );
5154                         $key = trim( substr( $data, 0, strpos( $data, '-->' ) ) );
5155                         $links['interwiki'][] = $this->mLinkHolders->interwiki[$key];
5156                         $pos = $start_pos + strlen( "<!--IWLINK $key-->" );
5157                 }
5158
5159                 $data['linkholder'] = $links;
5160
5161                 return $data;
5162         }
5163
5164         function unserialiseHalfParsedText( $data, $intPrefix = null /* Unique identifying prefix */ ) {
5165                 if (!$intPrefix)
5166                         $intPrefix = $this->getRandomString();
5167
5168                 // First, extract the strip state.
5169                 $stripState = $data['stripstate'];
5170                 $this->mStripState->general->merge( $stripState->general );
5171                 $this->mStripState->nowiki->merge( $stripState->nowiki );
5172
5173                 // Now, extract the text, and renumber links
5174                 $text = $data['text'];
5175                 $links = $data['linkholder'];
5176
5177                 // Internal...
5178                 foreach( $links['internal'] as $ns => $nsLinks ) {
5179                         foreach( $nsLinks as $key => $entry ) {
5180                                 $newKey = $intPrefix . '-' . $key;
5181                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
5182
5183                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
5184                         }
5185                 }
5186
5187                 // Interwiki...
5188                 foreach( $links['interwiki'] as $key => $entry ) {
5189                         $newKey = "$intPrefix-$key";
5190                         $this->mLinkHolders->interwikis[$newKey] = $entry;
5191
5192                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
5193                 }
5194
5195                 // Should be good to go.
5196                 return $text;
5197         }
5198 }
5199
5200 /**
5201  * @todo document, briefly.
5202  * @ingroup Parser
5203  */
5204 class StripState {
5205         var $general, $nowiki;
5206
5207         function __construct() {
5208                 $this->general = new ReplacementArray;
5209                 $this->nowiki = new ReplacementArray;
5210         }
5211
5212         function unstripGeneral( $text ) {
5213                 wfProfileIn( __METHOD__ );
5214                 do {
5215                         $oldText = $text;
5216                         $text = $this->general->replace( $text );
5217                 } while ( $text !== $oldText );
5218                 wfProfileOut( __METHOD__ );
5219                 return $text;
5220         }
5221
5222         function unstripNoWiki( $text ) {
5223                 wfProfileIn( __METHOD__ );
5224                 do {
5225                         $oldText = $text;
5226                         $text = $this->nowiki->replace( $text );
5227                 } while ( $text !== $oldText );
5228                 wfProfileOut( __METHOD__ );
5229                 return $text;
5230         }
5231
5232         function unstripBoth( $text ) {
5233                 wfProfileIn( __METHOD__ );
5234                 do {
5235                         $oldText = $text;
5236                         $text = $this->general->replace( $text );
5237                         $text = $this->nowiki->replace( $text );
5238                 } while ( $text !== $oldText );
5239                 wfProfileOut( __METHOD__ );
5240                 return $text;
5241         }
5242 }
5243
5244 /**
5245  * @todo document, briefly.
5246  * @ingroup Parser
5247  */
5248 class OnlyIncludeReplacer {
5249         var $output = '';
5250
5251         function replace( $matches ) {
5252                 if ( substr( $matches[1], -1 ) === "\n" ) {
5253                         $this->output .= substr( $matches[1], 0, -1 );
5254                 } else {
5255                         $this->output .= $matches[1];
5256                 }
5257         }
5258 }