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                         if ( $title->getNamespace() == NS_SPECIAL ) {
 371                                 $this->setTitle( $wgContLang->convert( $title ) );
 372                         } else {
 373                                 $this->mOutput->setTitleText( $wgContLang->convert( $title->getPrefixedText() ) );
 374                         }
 375                 }
 376
 377                 $text = $this->mStripState->unstripNoWiki( $text );
 378
 379                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 380
 381                 if ( $this->mTransparentTagHooks ) {
 382                         //!JF Move to its own function
 383                         $uniq_prefix = $this->mUniqPrefix;
 384                         $matches = array();
 385                         $elements = array_keys( $this->mTransparentTagHooks );
 386                         $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 387
 388                         foreach( $matches as $marker => $data ) {
 389                                 list( $element, $content, $params, $tag ) = $data;
 390                                 $tagName = strtolower( $element );
 391                                 if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 392                                         $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 393                                                 array( $content, $params, $this ) );
 394                                 } else {
 395                                         $output = $tag;
 396                                 }
 397                                 $this->mStripState->general->setPair( $marker, $output );
 398                         }
 399                 }
 400
 401                 # This was originally inserted for transparent tag hooks (now deprecated)
 402                 # but some extensions (notably <poem>) rely on the extra unstripGeneral()
 403                 # after unstripNoWiki() so they can modify the contents of <nowiki> tags.
 404                 $text = $this->mStripState->unstripGeneral( $text );
 405
 406                 $text = Sanitizer::normalizeCharReferences( $text );
 407
 408                 if ( ( $wgUseTidy && $this->mOptions->mTidy ) || $wgAlwaysUseTidy ) {
 409                         $text = MWTidy::tidy( $text );
 410                 } else {
 411                         # attempt to sanitize at least some nesting problems
 412                         # (bug #2702 and quite a few others)
 413                         $tidyregs = array(
 414                                 # ''Something [http://www.cool.com cool''] -->
 415                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 416                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 417                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 418                                 # fix up an anchor inside another anchor, only
 419                                 # at least for a single single nested link (bug 3695)
 420                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 421                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 422                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 423                                 # contains a div, so fix it up here; replace
 424                                 # div with escaped text
 425                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 426                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 427                                 # remove empty italic or bold tag pairs, some
 428                                 # introduced by rules above
 429                                 '/<([bi])><\/\\1>/' => '',
 430                         );
 431
 432                         $text = preg_replace(
 433                                 array_keys( $tidyregs ),
 434                                 array_values( $tidyregs ),
 435                                 $text );
 436                 }
 437                 global $wgExpensiveParserFunctionLimit;
 438                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 439                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 440                 }
 441
 442                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 443
 444                 # Information on include size limits, for the benefit of users who try to skirt them
 445                 if ( $this->mOptions->getEnableLimitReport() ) {
 446                         global $wgExpensiveParserFunctionLimit;
 447                         $max = $this->mOptions->getMaxIncludeSize();
 448                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 449                         $limitReport =
 450                                 "NewPP limit report\n" .
 451                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 452                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 453                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 454                                 $PFreport;
 455                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 456                         $text .= "\n<!-- \n$limitReport-->\n";
 457                 }
 458                 $this->mOutput->setText( $text );
 459
 460                 $this->mRevisionId = $oldRevisionId;
 461                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 462                 wfProfileOut( $fname );
 463                 wfProfileOut( __METHOD__ );
 464
 465                 return $this->mOutput;
 466         }
 467
 468         /**
 469          * Recursive parser entry point that can be called from an extension tag
 470          * hook.
 471          *
 472          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 473          *
 474          * @param $text String: text extension wants to have parsed
 475          * @param PPFrame $frame: The frame to use for expanding any template variables
 476          */
 477         function recursiveTagParse( $text, $frame=false ) {
 478                 wfProfileIn( __METHOD__ );
 479                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 480                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 481                 $text = $this->internalParse( $text, false, $frame );
 482                 wfProfileOut( __METHOD__ );
 483                 return $text;
 484         }
 485
 486         /**
 487          * Expand templates and variables in the text, producing valid, static wikitext.
 488          * Also removes comments.
 489          */
 490         function preprocess( $text, $title, $options, $revid = null ) {
 491                 wfProfileIn( __METHOD__ );
 492                 $this->clearState();
 493                 $this->setOutputType( self::OT_PREPROCESS );
 494                 $this->mOptions = $options;
 495                 $this->setTitle( $title );
 496                 if( $revid !== null ) {
 497                         $this->mRevisionId = $revid;
 498                 }
 499                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 500                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 501                 $text = $this->replaceVariables( $text );
 502                 $text = $this->mStripState->unstripBoth( $text );
 503                 wfProfileOut( __METHOD__ );
 504                 return $text;
 505         }
 506
 507         /**
 508          * Get the wikitext of a page as though it was transcluded.
 509          *
 510          * Specifically <includeonly> etc. are parsed, redirects are followed, comments
 511          * are removed, but templates arguments and parser functions are untouched.
 512          *
 513          * This is not called by the parser itself, see braceSubstitution for its transclusion.
 514          */
 515         public function getTransclusionText( $title, $options ) {
 516                 // Must initialize first
 517                 $this->clearState();
 518                 $this->setOutputType( self::OT_INCLUDES );
 519                 $this->mOptions = $options;
 520                 $this->setTitle( new FakeTitle );
 521
 522                 list( $text, $title ) = $this->getTemplateDom( $title );
 523                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 524                 return $this->getPreprocessor()->newFrame()->expand( $text, $flags );
 525         }
 526
 527         /**
 528          * Get a random string
 529          *
 530          * @private
 531          * @static
 532          */
 533         function getRandomString() {
 534                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 535         }
 536
 537         function &getTitle() { return $this->mTitle; }
 538         function getOptions() { return $this->mOptions; }
 539         function getRevisionId() { return $this->mRevisionId; }
 540         function getOutput() { return $this->mOutput; }
 541         function nextLinkID() { return $this->mLinkID++; }
 542
 543         function getFunctionLang() {
 544                 global $wgLang, $wgContLang;
 545
 546                 $target = $this->mOptions->getTargetLanguage();
 547                 if ( $target !== null ) {
 548                         return $target;
 549                 } else {
 550                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 551                 }
 552         }
 553
 554         /**
 555          * Get a preprocessor object
 556          */
 557         function getPreprocessor() {
 558                 if ( !isset( $this->mPreprocessor ) ) {
 559                         $class = $this->mPreprocessorClass;
 560                         $this->mPreprocessor = new $class( $this );
 561                 }
 562                 return $this->mPreprocessor;
 563         }
 564
 565         /**
 566          * Replaces all occurrences of HTML-style comments and the given tags
 567          * in the text with a random marker and returns the next text. The output
 568          * parameter $matches will be an associative array filled with data in
 569          * the form:
 570          *   'UNIQ-xxxxx' => array(
 571          *     'element',
 572          *     'tag content',
 573          *     array( 'param' => 'x' ),
 574          *     '<element param="x">tag content</element>' ) )
 575          *
 576          * @param $elements list of element names. Comments are always extracted.
 577          * @param $text Source text string.
 578          * @param $uniq_prefix
 579          *
 580          * @public
 581          * @static
 582          */
 583         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 584                 static $n = 1;
 585                 $stripped = '';
 586                 $matches = array();
 587
 588                 $taglist = implode( '|', $elements );
 589                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 590
 591                 while ( $text != '' ) {
 592                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 593                         $stripped .= $p[0];
 594                         if( count( $p ) < 5 ) {
 595                                 break;
 596                         }
 597                         if( count( $p ) > 5 ) {
 598                                 // comment
 599                                 $element    = $p[4];
 600                                 $attributes = '';
 601                                 $close      = '';
 602                                 $inside     = $p[5];
 603                         } else {
 604                                 // tag
 605                                 $element    = $p[1];
 606                                 $attributes = $p[2];
 607                                 $close      = $p[3];
 608                                 $inside     = $p[4];
 609                         }
 610
 611                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . self::MARKER_SUFFIX;
 612                         $stripped .= $marker;
 613
 614                         if ( $close === '/>' ) {
 615                                 // Empty element tag, <tag />
 616                                 $content = null;
 617                                 $text = $inside;
 618                                 $tail = null;
 619                         } else {
 620                                 if( $element === '!--' ) {
 621                                         $end = '/(-->)/';
 622                                 } else {
 623                                         $end = "/(<\\/$element\\s*>)/i";
 624                                 }
 625                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 626                                 $content = $q[0];
 627                                 if( count( $q ) < 3 ) {
 628                                         # No end tag -- let it run out to the end of the text.
 629                                         $tail = '';
 630                                         $text = '';
 631                                 } else {
 632                                         $tail = $q[1];
 633                                         $text = $q[2];
 634                                 }
 635                         }
 636
 637                         $matches[$marker] = array( $element,
 638                                 $content,
 639                                 Sanitizer::decodeTagAttributes( $attributes ),
 640                                 "<$element$attributes$close$content$tail" );
 641                 }
 642                 return $stripped;
 643         }
 644
 645         /**
 646          * Get a list of strippable XML-like elements
 647          */
 648         function getStripList() {
 649                 return $this->mStripList;
 650         }
 651
 652         /**
 653          * @deprecated use replaceVariables
 654          */
 655         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 656                 return $text;
 657         }
 658
 659         /**
 660          * Restores pre, math, and other extensions removed by strip()
 661          *
 662          * always call unstripNoWiki() after this one
 663          * @private
 664          * @deprecated use $this->mStripState->unstrip()
 665          */
 666         function unstrip( $text, $state ) {
 667                 return $state->unstripGeneral( $text );
 668         }
 669
 670         /**
 671          * Always call this after unstrip() to preserve the order
 672          *
 673          * @private
 674          * @deprecated use $this->mStripState->unstrip()
 675          */
 676         function unstripNoWiki( $text, $state ) {
 677                 return $state->unstripNoWiki( $text );
 678         }
 679
 680         /**
 681          * @deprecated use $this->mStripState->unstripBoth()
 682          */
 683         function unstripForHTML( $text ) {
 684                 return $this->mStripState->unstripBoth( $text );
 685         }
 686
 687         /**
 688          * Add an item to the strip state
 689          * Returns the unique tag which must be inserted into the stripped text
 690          * The tag will be replaced with the original text in unstrip()
 691          *
 692          * @private
 693          */
 694         function insertStripItem( $text ) {
 695                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 696                 $this->mMarkerIndex++;
 697                 $this->mStripState->general->setPair( $rnd, $text );
 698                 return $rnd;
 699         }
 700
 701         /**
 702          * Interface with html tidy
 703          * @deprecated Use MWTidy::tidy()
 704          */
 705         public static function tidy( $text ) {
 706                 wfDeprecated( __METHOD__ );
 707                 return MWTidy::tidy( $text );
 708         }
 709
 710         /**
 711          * parse the wiki syntax used to render tables
 712          *
 713          * @private
 714          */
 715         function doTableStuff ( $text ) {
 716                 wfProfileIn( __METHOD__ );
 717
 718                 $lines = StringUtils::explode( "\n", $text );
 719                 $out = '';
 720                 $td_history = array (); // Is currently a td tag open?
 721                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 722                 $tr_history = array (); // Is currently a tr tag open?
 723                 $tr_attributes = array (); // history of tr attributes
 724                 $has_opened_tr = array(); // Did this table open a <tr> element?
 725                 $indent_level = 0; // indent level of the table
 726
 727                 foreach ( $lines as $outLine ) {
 728                         $line = trim( $outLine );
 729
 730                         if( $line == '' ) { // empty line, go to next line
 731                                 $out .= $outLine."\n";
 732                                 continue;
 733                         }
 734                         $first_character = $line[0];
 735                         $matches = array();
 736
 737                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 738                                 // First check if we are starting a new table
 739                                 $indent_level = strlen( $matches[1] );
 740
 741                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 742                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 743
 744                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 745                                 array_push( $td_history , false );
 746                                 array_push( $last_tag_history , '' );
 747                                 array_push( $tr_history , false );
 748                                 array_push( $tr_attributes , '' );
 749                                 array_push( $has_opened_tr , false );
 750                         } else if ( count ( $td_history ) == 0 ) {
 751                                 // Don't do any of the following
 752                                 $out .= $outLine."\n";
 753                                 continue;
 754                         } else if ( substr ( $line , 0 , 2 ) === '|}' ) {
 755                                 // We are ending a table
 756                                 $line = '</table>' . substr ( $line , 2 );
 757                                 $last_tag = array_pop ( $last_tag_history );
 758
 759                                 if ( !array_pop ( $has_opened_tr ) ) {
 760                                         $line = "<tr><td></td></tr>{$line}";
 761                                 }
 762
 763                                 if ( array_pop ( $tr_history ) ) {
 764                                         $line = "</tr>{$line}";
 765                                 }
 766
 767                                 if ( array_pop ( $td_history ) ) {
 768                                         $line = "</{$last_tag}>{$line}";
 769                                 }
 770                                 array_pop ( $tr_attributes );
 771                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 772                         } else if ( substr ( $line , 0 , 2 ) === '|-' ) {
 773                                 // Now we have a table row
 774                                 $line = preg_replace( '#^\|-+#', '', $line );
 775
 776                                 // Whats after the tag is now only attributes
 777                                 $attributes = $this->mStripState->unstripBoth( $line );
 778                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 779                                 array_pop( $tr_attributes );
 780                                 array_push( $tr_attributes, $attributes );
 781
 782                                 $line = '';
 783                                 $last_tag = array_pop ( $last_tag_history );
 784                                 array_pop ( $has_opened_tr );
 785                                 array_push ( $has_opened_tr , true );
 786
 787                                 if ( array_pop ( $tr_history ) ) {
 788                                         $line = '</tr>';
 789                                 }
 790
 791                                 if ( array_pop ( $td_history ) ) {
 792                                         $line = "</{$last_tag}>{$line}";
 793                                 }
 794
 795                                 $outLine = $line;
 796                                 array_push ( $tr_history , false );
 797                                 array_push ( $td_history , false );
 798                                 array_push ( $last_tag_history , '' );
 799                         }
 800                         else if ( $first_character === '|' || $first_character === '!' || substr ( $line , 0 , 2 )  === '|+' ) {
 801                                 // This might be cell elements, td, th or captions
 802                                 if ( substr ( $line , 0 , 2 ) === '|+' ) {
 803                                         $first_character = '+';
 804                                         $line = substr ( $line , 1 );
 805                                 }
 806
 807                                 $line = substr ( $line , 1 );
 808
 809                                 if ( $first_character === '!' ) {
 810                                         $line = str_replace ( '!!' , '||' , $line );
 811                                 }
 812
 813                                 // Split up multiple cells on the same line.
 814                                 // FIXME : This can result in improper nesting of tags processed
 815                                 // by earlier parser steps, but should avoid splitting up eg
 816                                 // attribute values containing literal "||".
 817                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 818
 819                                 $outLine = '';
 820
 821                                 // Loop through each table cell
 822                                 foreach ( $cells as $cell )
 823                                 {
 824                                         $previous = '';
 825                                         if ( $first_character !== '+' )
 826                                         {
 827                                                 $tr_after = array_pop ( $tr_attributes );
 828                                                 if ( !array_pop ( $tr_history ) ) {
 829                                                         $previous = "<tr{$tr_after}>\n";
 830                                                 }
 831                                                 array_push ( $tr_history , true );
 832                                                 array_push ( $tr_attributes , '' );
 833                                                 array_pop ( $has_opened_tr );
 834                                                 array_push ( $has_opened_tr , true );
 835                                         }
 836
 837                                         $last_tag = array_pop ( $last_tag_history );
 838
 839                                         if ( array_pop ( $td_history ) ) {
 840                                                 $previous = "</{$last_tag}>{$previous}";
 841                                         }
 842
 843                                         if ( $first_character === '|' ) {
 844                                                 $last_tag = 'td';
 845                                         } else if ( $first_character === '!' ) {
 846                                                 $last_tag = 'th';
 847                                         } else if ( $first_character === '+' ) {
 848                                                 $last_tag = 'caption';
 849                                         } else {
 850                                                 $last_tag = '';
 851                                         }
 852
 853                                         array_push ( $last_tag_history , $last_tag );
 854
 855                                         // A cell could contain both parameters and data
 856                                         $cell_data = explode ( '|' , $cell , 2 );
 857
 858                                         // Bug 553: Note that a '|' inside an invalid link should not
 859                                         // be mistaken as delimiting cell parameters
 860                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 861                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 862                                         } else if ( count ( $cell_data ) == 1 )
 863                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 864                                         else {
 865                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 866                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 867                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 868                                         }
 869
 870                                         $outLine .= $cell;
 871                                         array_push ( $td_history , true );
 872                                 }
 873                         }
 874                         $out .= $outLine . "\n";
 875                 }
 876
 877                 // Closing open td, tr && table
 878                 while ( count ( $td_history ) > 0 )
 879                 {
 880                         if ( array_pop ( $td_history ) ) {
 881                                 $out .= "</td>\n";
 882                         }
 883                         if ( array_pop ( $tr_history ) ) {
 884                                 $out .= "</tr>\n";
 885                         }
 886                         if ( !array_pop ( $has_opened_tr ) ) {
 887                                 $out .= "<tr><td></td></tr>\n" ;
 888                         }
 889
 890                         $out .= "</table>\n";
 891                 }
 892
 893                 // Remove trailing line-ending (b/c)
 894                 if ( substr( $out, -1 ) === "\n" ) {
 895                         $out = substr( $out, 0, -1 );
 896                 }
 897
 898                 // special case: don't return empty table
 899                 if( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 900                         $out = '';
 901                 }
 902
 903                 wfProfileOut( __METHOD__ );
 904
 905                 return $out;
 906         }
 907
 908         /**
 909          * Helper function for parse() that transforms wiki markup into
 910          * HTML. Only called for $mOutputType == self::OT_HTML.
 911          *
 912          * @private
 913          */
 914         function internalParse( $text, $isMain = true, $frame=false ) {
 915                 wfProfileIn( __METHOD__ );
 916
 917                 $origText = $text;
 918
 919                 # Hook to suspend the parser in this state
 920                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 921                         wfProfileOut( __METHOD__ );
 922                         return $text ;
 923                 }
 924
 925                 // if $frame is provided, then use $frame for replacing any variables
 926                 if ($frame) {
 927                         // use frame depth to infer how include/noinclude tags should be handled
 928                         // depth=0 means this is the top-level document; otherwise it's an included document
 929                         if( !$frame->depth )
 930                                 $flag = 0;
 931                         else
 932                                 $flag = Parser::PTD_FOR_INCLUSION;
 933                         $dom = $this->preprocessToDom( $text, $flag );
 934                         $text = $frame->expand( $dom );
 935                 }
 936                 // if $frame is not provided, then use old-style replaceVariables
 937                 else {
 938                         $text = $this->replaceVariables( $text );
 939                 }
 940
 941                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 942                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 943
 944                 // Tables need to come after variable replacement for things to work
 945                 // properly; putting them before other transformations should keep
 946                 // exciting things like link expansions from showing up in surprising
 947                 // places.
 948                 $text = $this->doTableStuff( $text );
 949
 950                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 951
 952                 $text = $this->doDoubleUnderscore( $text );
 953
 954                 $text = $this->doHeadings( $text );
 955                 if( $this->mOptions->getUseDynamicDates() ) {
 956                         $df = DateFormatter::getInstance();
 957                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 958                 }
 959                 $text = $this->doAllQuotes( $text );
 960                 $text = $this->replaceInternalLinks( $text );
 961                 $text = $this->replaceExternalLinks( $text );
 962
 963                 # replaceInternalLinks may sometimes leave behind
 964                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 965                 $text = str_replace($this->mUniqPrefix.'NOPARSE', '', $text);
 966
 967                 $text = $this->doMagicLinks( $text );
 968                 $text = $this->formatHeadings( $text, $origText, $isMain );
 969
 970                 wfProfileOut( __METHOD__ );
 971                 return $text;
 972         }
 973
 974         /**
 975          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 976          * magic external links.
 977          *
 978          * DML
 979          * @private
 980          */
 981         function doMagicLinks( $text ) {
 982                 wfProfileIn( __METHOD__ );
 983                 $prots = $this->mUrlProtocols;
 984                 $urlChar = self::EXT_LINK_URL_CLASS;
 985                 $text = preg_replace_callback(
 986                         '!(?:                           # Start cases
 987                                 (<a.*?</a>) |               # m[1]: Skip link text
 988                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
 989                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
 990                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
 991                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
 992                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 993                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 994                                     [0-9Xx]                 # check digit
 995                                     \b)
 996                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 997                 wfProfileOut( __METHOD__ );
 998                 return $text;
 999         }
1000
1001         function magicLinkCallback( $m ) {
1002                 if ( isset( $m[1] ) && $m[1] !== '' ) {
1003                         # Skip anchor
1004                         return $m[0];
1005                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
1006                         # Skip HTML element
1007                         return $m[0];
1008                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1009                         # Free external link
1010                         return $this->makeFreeExternalLink( $m[0] );
1011                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1012                         # RFC or PMID
1013                         $CssClass = '';
1014                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1015                                 $keyword = 'RFC';
1016                                 $urlmsg = 'rfcurl';
1017                                 $CssClass = 'mw-magiclink-rfc';
1018                                 $id = $m[4];
1019                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1020                                 $keyword = 'PMID';
1021                                 $urlmsg = 'pubmedurl';
1022                                 $CssClass = 'mw-magiclink-pmid';
1023                                 $id = $m[4];
1024                         } else {
1025                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1026                                         substr($m[0], 0, 20 ) . '"' );
1027                         }
1028                         $url = wfMsg( $urlmsg, $id);
1029                         $sk = $this->mOptions->getSkin();
1030                         $la = $sk->getExternalLinkAttributes( "external $CssClass" );
1031                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1032                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1033                         # ISBN
1034                         $isbn = $m[5];
1035                         $num = strtr( $isbn, array(
1036                                 '-' => '',
1037                                 ' ' => '',
1038                                 'x' => 'X',
1039                         ));
1040                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1041                         return'<a href="' .
1042                                 $titleObj->escapeLocalUrl() .
1043                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1044                 } else {
1045                         return $m[0];
1046                 }
1047         }
1048
1049         /**
1050          * Make a free external link, given a user-supplied URL
1051          * @return HTML
1052          * @private
1053          */
1054         function makeFreeExternalLink( $url ) {
1055                 global $wgContLang;
1056                 wfProfileIn( __METHOD__ );
1057
1058                 $sk = $this->mOptions->getSkin();
1059                 $trail = '';
1060
1061                 # The characters '<' and '>' (which were escaped by
1062                 # removeHTMLtags()) should not be included in
1063                 # URLs, per RFC 2396.
1064                 $m2 = array();
1065                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1066                         $trail = substr($url, $m2[0][1]) . $trail;
1067                         $url = substr($url, 0, $m2[0][1]);
1068                 }
1069
1070                 # Move trailing punctuation to $trail
1071                 $sep = ',;\.:!?';
1072                 # If there is no left bracket, then consider right brackets fair game too
1073                 if ( strpos( $url, '(' ) === false ) {
1074                         $sep .= ')';
1075                 }
1076
1077                 $numSepChars = strspn( strrev( $url ), $sep );
1078                 if ( $numSepChars ) {
1079                         $trail = substr( $url, -$numSepChars ) . $trail;
1080                         $url = substr( $url, 0, -$numSepChars );
1081                 }
1082
1083                 $url = Sanitizer::cleanUrl( $url );
1084
1085                 # Is this an external image?
1086                 $text = $this->maybeMakeExternalImage( $url );
1087                 if ( $text === false ) {
1088                         # Not an image, make a link
1089                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1090                                 $this->getExternalLinkAttribs( $url ) );
1091                         # Register it in the output object...
1092                         # Replace unnecessary URL escape codes with their equivalent characters
1093                         $pasteurized = self::replaceUnusualEscapes( $url );
1094                         $this->mOutput->addExternalLink( $pasteurized );
1095                 }
1096                 wfProfileOut( __METHOD__ );
1097                 return $text . $trail;
1098         }
1099
1100
1101         /**
1102          * Parse headers and return html
1103          *
1104          * @private
1105          */
1106         function doHeadings( $text ) {
1107                 wfProfileIn( __METHOD__ );
1108                 for ( $i = 6; $i >= 1; --$i ) {
1109                         $h = str_repeat( '=', $i );
1110                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1111                           "<h$i>\\1</h$i>", $text );
1112                 }
1113                 wfProfileOut( __METHOD__ );
1114                 return $text;
1115         }
1116
1117         /**
1118          * Replace single quotes with HTML markup
1119          * @private
1120          * @return string the altered text
1121          */
1122         function doAllQuotes( $text ) {
1123                 wfProfileIn( __METHOD__ );
1124                 $outtext = '';
1125                 $lines = StringUtils::explode( "\n", $text );
1126                 foreach ( $lines as $line ) {
1127                         $outtext .= $this->doQuotes( $line ) . "\n";
1128                 }
1129                 $outtext = substr($outtext, 0,-1);
1130                 wfProfileOut( __METHOD__ );
1131                 return $outtext;
1132         }
1133
1134         /**
1135          * Helper function for doAllQuotes()
1136          */
1137         public function doQuotes( $text ) {
1138                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1139                 if ( count( $arr ) == 1 )
1140                         return $text;
1141                 else
1142                 {
1143                         # First, do some preliminary work. This may shift some apostrophes from
1144                         # being mark-up to being text. It also counts the number of occurrences
1145                         # of bold and italics mark-ups.
1146                         $i = 0;
1147                         $numbold = 0;
1148                         $numitalics = 0;
1149                         foreach ( $arr as $r )
1150                         {
1151                                 if ( ( $i % 2 ) == 1 )
1152                                 {
1153                                         # If there are ever four apostrophes, assume the first is supposed to
1154                                         # be text, and the remaining three constitute mark-up for bold text.
1155                                         if ( strlen( $arr[$i] ) == 4 )
1156                                         {
1157                                                 $arr[$i-1] .= "'";
1158                                                 $arr[$i] = "'''";
1159                                         }
1160                                         # If there are more than 5 apostrophes in a row, assume they're all
1161                                         # text except for the last 5.
1162                                         else if ( strlen( $arr[$i] ) > 5 )
1163                                         {
1164                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1165                                                 $arr[$i] = "'''''";
1166                                         }
1167                                         # Count the number of occurrences of bold and italics mark-ups.
1168                                         # We are not counting sequences of five apostrophes.
1169                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1170                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1171                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1172                                 }
1173                                 $i++;
1174                         }
1175
1176                         # If there is an odd number of both bold and italics, it is likely
1177                         # that one of the bold ones was meant to be an apostrophe followed
1178                         # by italics. Which one we cannot know for certain, but it is more
1179                         # likely to be one that has a single-letter word before it.
1180                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1181                         {
1182                                 $i = 0;
1183                                 $firstsingleletterword = -1;
1184                                 $firstmultiletterword = -1;
1185                                 $firstspace = -1;
1186                                 foreach ( $arr as $r )
1187                                 {
1188                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1189                                         {
1190                                                 $x1 = substr ($arr[$i-1], -1);
1191                                                 $x2 = substr ($arr[$i-1], -2, 1);
1192                                                 if ($x1 === ' ') {
1193                                                         if ($firstspace == -1) $firstspace = $i;
1194                                                 } else if ($x2 === ' ') {
1195                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1196                                                 } else {
1197                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1198                                                 }
1199                                         }
1200                                         $i++;
1201                                 }
1202
1203                                 # If there is a single-letter word, use it!
1204                                 if ($firstsingleletterword > -1)
1205                                 {
1206                                         $arr [ $firstsingleletterword ] = "''";
1207                                         $arr [ $firstsingleletterword-1 ] .= "'";
1208                                 }
1209                                 # If not, but there's a multi-letter word, use that one.
1210                                 else if ($firstmultiletterword > -1)
1211                                 {
1212                                         $arr [ $firstmultiletterword ] = "''";
1213                                         $arr [ $firstmultiletterword-1 ] .= "'";
1214                                 }
1215                                 # ... otherwise use the first one that has neither.
1216                                 # (notice that it is possible for all three to be -1 if, for example,
1217                                 # there is only one pentuple-apostrophe in the line)
1218                                 else if ($firstspace > -1)
1219                                 {
1220                                         $arr [ $firstspace ] = "''";
1221                                         $arr [ $firstspace-1 ] .= "'";
1222                                 }
1223                         }
1224
1225                         # Now let's actually convert our apostrophic mush to HTML!
1226                         $output = '';
1227                         $buffer = '';
1228                         $state = '';
1229                         $i = 0;
1230                         foreach ($arr as $r)
1231                         {
1232                                 if (($i % 2) == 0)
1233                                 {
1234                                         if ($state === 'both')
1235                                                 $buffer .= $r;
1236                                         else
1237                                                 $output .= $r;
1238                                 }
1239                                 else
1240                                 {
1241                                         if (strlen ($r) == 2)
1242                                         {
1243                                                 if ($state === 'i')
1244                                                 { $output .= '</i>'; $state = ''; }
1245                                                 else if ($state === 'bi')
1246                                                 { $output .= '</i>'; $state = 'b'; }
1247                                                 else if ($state === 'ib')
1248                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1249                                                 else if ($state === 'both')
1250                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1251                                                 else # $state can be 'b' or ''
1252                                                 { $output .= '<i>'; $state .= 'i'; }
1253                                         }
1254                                         else if (strlen ($r) == 3)
1255                                         {
1256                                                 if ($state === 'b')
1257                                                 { $output .= '</b>'; $state = ''; }
1258                                                 else if ($state === 'bi')
1259                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1260                                                 else if ($state === 'ib')
1261                                                 { $output .= '</b>'; $state = 'i'; }
1262                                                 else if ($state === 'both')
1263                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1264                                                 else # $state can be 'i' or ''
1265                                                 { $output .= '<b>'; $state .= 'b'; }
1266                                         }
1267                                         else if (strlen ($r) == 5)
1268                                         {
1269                                                 if ($state === 'b')
1270                                                 { $output .= '</b><i>'; $state = 'i'; }
1271                                                 else if ($state === 'i')
1272                                                 { $output .= '</i><b>'; $state = 'b'; }
1273                                                 else if ($state === 'bi')
1274                                                 { $output .= '</i></b>'; $state = ''; }
1275                                                 else if ($state === 'ib')
1276                                                 { $output .= '</b></i>'; $state = ''; }
1277                                                 else if ($state === 'both')
1278                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1279                                                 else # ($state == '')
1280                                                 { $buffer = ''; $state = 'both'; }
1281                                         }
1282                                 }
1283                                 $i++;
1284                         }
1285                         # Now close all remaining tags.  Notice that the order is important.
1286                         if ($state === 'b' || $state === 'ib')
1287                                 $output .= '</b>';
1288                         if ($state === 'i' || $state === 'bi' || $state === 'ib')
1289                                 $output .= '</i>';
1290                         if ($state === 'bi')
1291                                 $output .= '</b>';
1292                         # There might be lonely ''''', so make sure we have a buffer
1293                         if ($state === 'both' && $buffer)
1294                                 $output .= '<b><i>'.$buffer.'</i></b>';
1295                         return $output;
1296                 }
1297         }
1298
1299         /**
1300          * Replace external links (REL)
1301          *
1302          * Note: this is all very hackish and the order of execution matters a lot.
1303          * Make sure to run maintenance/parserTests.php if you change this code.
1304          *
1305          * @private
1306          */
1307         function replaceExternalLinks( $text ) {
1308                 global $wgContLang;
1309                 wfProfileIn( __METHOD__ );
1310
1311                 $sk = $this->mOptions->getSkin();
1312
1313                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1314                 $s = array_shift( $bits );
1315
1316                 $i = 0;
1317                 while ( $i<count( $bits ) ) {
1318                         $url = $bits[$i++];
1319                         $protocol = $bits[$i++];
1320                         $text = $bits[$i++];
1321                         $trail = $bits[$i++];
1322
1323                         # The characters '<' and '>' (which were escaped by
1324                         # removeHTMLtags()) should not be included in
1325                         # URLs, per RFC 2396.
1326                         $m2 = array();
1327                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1328                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1329                                 $url = substr($url, 0, $m2[0][1]);
1330                         }
1331
1332                         # If the link text is an image URL, replace it with an <img> tag
1333                         # This happened by accident in the original parser, but some people used it extensively
1334                         $img = $this->maybeMakeExternalImage( $text );
1335                         if ( $img !== false ) {
1336                                 $text = $img;
1337                         }
1338
1339                         $dtrail = '';
1340
1341                         # Set linktype for CSS - if URL==text, link is essentially free
1342                         $linktype = ($text === $url) ? 'free' : 'text';
1343
1344                         # No link text, e.g. [http://domain.tld/some.link]
1345                         if ( $text == '' ) {
1346                                 # Autonumber if allowed. See bug #5918
1347                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1348                                         $langObj = $this->getFunctionLang();
1349                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1350                                         $linktype = 'autonumber';
1351                                 } else {
1352                                         # Otherwise just use the URL
1353                                         $text = htmlspecialchars( $url );
1354                                         $linktype = 'free';
1355                                 }
1356                         } else {
1357                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1358                                 # Check for trail
1359                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1360                         }
1361
1362                         $text = $wgContLang->markNoConversion($text);
1363
1364                         $url = Sanitizer::cleanUrl( $url );
1365
1366                         # Use the encoded URL
1367                         # This means that users can paste URLs directly into the text
1368                         # Funny characters like &ouml; aren't valid in URLs anyway
1369                         # This was changed in August 2004
1370                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1371                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1372
1373                         # Register link in the output object.
1374                         # Replace unnecessary URL escape codes with the referenced character
1375                         # This prevents spammers from hiding links from the filters
1376                         $pasteurized = self::replaceUnusualEscapes( $url );
1377                         $this->mOutput->addExternalLink( $pasteurized );
1378                 }
1379
1380                 wfProfileOut( __METHOD__ );
1381                 return $s;
1382         }
1383
1384         /**
1385          * Get an associative array of additional HTML attributes appropriate for a
1386          * particular external link.  This currently may include rel => nofollow
1387          * (depending on configuration, namespace, and the URL's domain) and/or a
1388          * target attribute (depending on configuration).
1389          *
1390          * @param string $url Optional URL, to extract the domain from for rel =>
1391          *   nofollow if appropriate
1392          * @return array Associative array of HTML attributes
1393          */
1394         function getExternalLinkAttribs( $url = false ) {
1395                 $attribs = array();
1396                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1397                 $ns = $this->mTitle->getNamespace();
1398                 if( $wgNoFollowLinks && !in_array($ns, $wgNoFollowNsExceptions) ) {
1399                         $attribs['rel'] = 'nofollow';
1400
1401                         global $wgNoFollowDomainExceptions;
1402                         if ( $wgNoFollowDomainExceptions ) {
1403                                 $bits = wfParseUrl( $url );
1404                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1405                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1406                                                 if( substr( $bits['host'], -strlen( $domain ) )
1407                                                 == $domain ) {
1408                                                         unset( $attribs['rel'] );
1409                                                         break;
1410                                                 }
1411                                         }
1412                                 }
1413                         }
1414                 }
1415                 if ( $this->mOptions->getExternalLinkTarget() ) {
1416                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1417                 }
1418                 return $attribs;
1419         }
1420
1421
1422         /**
1423          * Replace unusual URL escape codes with their equivalent characters
1424          * @param string
1425          * @return string
1426          * @static
1427          * @todo  This can merge genuinely required bits in the path or query string,
1428          *        breaking legit URLs. A proper fix would treat the various parts of
1429          *        the URL differently; as a workaround, just use the output for
1430          *        statistical records, not for actual linking/output.
1431          */
1432         static function replaceUnusualEscapes( $url ) {
1433                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1434                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1435         }
1436
1437         /**
1438          * Callback function used in replaceUnusualEscapes().
1439          * Replaces unusual URL escape codes with their equivalent character
1440          * @static
1441          * @private
1442          */
1443         private static function replaceUnusualEscapesCallback( $matches ) {
1444                 $char = urldecode( $matches[0] );
1445                 $ord = ord( $char );
1446                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1447                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1448                         // No, shouldn't be escaped
1449                         return $char;
1450                 } else {
1451                         // Yes, leave it escaped
1452                         return $matches[0];
1453                 }
1454         }
1455
1456         /**
1457          * make an image if it's allowed, either through the global
1458          * option, through the exception, or through the on-wiki whitelist
1459          * @private
1460          */
1461         function maybeMakeExternalImage( $url ) {
1462                 $sk = $this->mOptions->getSkin();
1463                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1464                 $imagesexception = !empty($imagesfrom);
1465                 $text = false;
1466                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1467                 if( $imagesexception && is_array( $imagesfrom ) ) {
1468                         $imagematch = false;
1469                         foreach( $imagesfrom as $match ) {
1470                                 if( strpos( $url, $match ) === 0 ) {
1471                                         $imagematch = true;
1472                                         break;
1473                                 }
1474                         }
1475                 } elseif( $imagesexception ) {
1476                         $imagematch = (strpos( $url, $imagesfrom ) === 0);
1477                 } else {
1478                         $imagematch = false;
1479                 }
1480                 if ( $this->mOptions->getAllowExternalImages()
1481                      || ( $imagesexception && $imagematch ) ) {
1482                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1483                                 # Image found
1484                                 $text = $sk->makeExternalImage( $url );
1485                         }
1486                 }
1487                 if( !$text && $this->mOptions->getEnableImageWhitelist()
1488                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1489                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1490                         foreach( $whitelist as $entry ) {
1491                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1492                                 if( strpos( $entry, '#' ) === 0 || $entry === '' )
1493                                         continue;
1494                                 if( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1495                                         # Image matches a whitelist entry
1496                                         $text = $sk->makeExternalImage( $url );
1497                                         break;
1498                                 }
1499                         }
1500                 }
1501                 return $text;
1502         }
1503
1504         /**
1505          * Process [[ ]] wikilinks
1506          * @return processed text
1507          *
1508          * @private
1509          */
1510         function replaceInternalLinks( $s ) {
1511                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1512                 return $s;
1513         }
1514
1515         /**
1516          * Process [[ ]] wikilinks (RIL)
1517          * @return LinkHolderArray
1518          *
1519          * @private
1520          */
1521         function replaceInternalLinks2( &$s ) {
1522                 global $wgContLang;
1523
1524                 wfProfileIn( __METHOD__ );
1525
1526                 wfProfileIn( __METHOD__.'-setup' );
1527                 static $tc = FALSE, $e1, $e1_img;
1528                 # the % is needed to support urlencoded titles as well
1529                 if ( !$tc ) {
1530                         $tc = Title::legalChars() . '#%';
1531                         # Match a link having the form [[namespace:link|alternate]]trail
1532                         $e1 = "/^([{$tc}]*)(\\|.*?)?]](.*)\$/sD";
1533                         # Match cases where there is no "]]", which might still be images
1534                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1535                 }
1536
1537                 $sk = $this->mOptions->getSkin();
1538                 $holders = new LinkHolderArray( $this );
1539
1540                 #split the entire text string on occurences of [[
1541                 $a = StringUtils::explode( '[[', ' ' . $s );
1542                 #get the first element (all text up to first [[), and remove the space we added
1543                 $s = $a->current();
1544                 $a->next();
1545                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1546                 $s = substr( $s, 1 );
1547
1548                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1549                 $e2 = null;
1550                 if ( $useLinkPrefixExtension ) {
1551                         # Match the end of a line for a word that's not followed by whitespace,
1552                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1553                         $e2 = wfMsgForContent( 'linkprefix' );
1554                 }
1555
1556                 if( is_null( $this->mTitle ) ) {
1557                         wfProfileOut( __METHOD__.'-setup' );
1558                         wfProfileOut( __METHOD__ );
1559                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1560                 }
1561                 $nottalk = !$this->mTitle->isTalkPage();
1562
1563                 if ( $useLinkPrefixExtension ) {
1564                         $m = array();
1565                         if ( preg_match( $e2, $s, $m ) ) {
1566                                 $first_prefix = $m[2];
1567                         } else {
1568                                 $first_prefix = false;
1569                         }
1570                 } else {
1571                         $prefix = '';
1572                 }
1573
1574                 if($wgContLang->hasVariants()) {
1575                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1576                 } else {
1577                         $selflink = array($this->mTitle->getPrefixedText());
1578                 }
1579                 $useSubpages = $this->areSubpagesAllowed();
1580                 wfProfileOut( __METHOD__.'-setup' );
1581
1582                 # Loop for each link
1583                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1584                         # Check for excessive memory usage
1585                         if ( $holders->isBig() ) {
1586                                 # Too big
1587                                 # Do the existence check, replace the link holders and clear the array
1588                                 $holders->replace( $s );
1589                                 $holders->clear();
1590                         }
1591
1592                         if ( $useLinkPrefixExtension ) {
1593                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1594                                 if ( preg_match( $e2, $s, $m ) ) {
1595                                         $prefix = $m[2];
1596                                         $s = $m[1];
1597                                 } else {
1598                                         $prefix='';
1599                                 }
1600                                 # first link
1601                                 if($first_prefix) {
1602                                         $prefix = $first_prefix;
1603                                         $first_prefix = false;
1604                                 }
1605                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1606                         }
1607
1608                         $might_be_img = false;
1609
1610                         wfProfileIn( __METHOD__."-e1" );
1611                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1612
1613                                 if( $m[2] === '' ) {
1614                                         $text = '';
1615                                 } elseif( $m[2] === '|' ) {
1616                                         $text = $this->getPipeTrickText( $m[1] );
1617                                 } else {
1618                                         $text = substr( $m[2], 1 );
1619                                 }
1620
1621                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1622                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1623                                 # the real problem is with the $e1 regex
1624                                 # See bug 1300.
1625                                 #
1626                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1627                                 # and no image is in sight. See bug 2095.
1628                                 #
1629                                 if( $text !== '' &&
1630                                         substr( $m[3], 0, 1 ) === ']' &&
1631                                         strpos($text, '[') !== false
1632                                 )
1633                                 {
1634                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1635                                         $m[3] = substr( $m[3], 1 );
1636                                 }
1637
1638                                 # Handle pipe-trick for [[|<blah>]]
1639                                 $lnk = $m[1] === '' ? $this->getPipeTrickLink( $text ) : $m[1];
1640                                 # fix up urlencoded title texts
1641                                 if( strpos( $lnk, '%' ) !== false ) {
1642                                         # Should anchors '#' also be rejected?
1643                                         $lnk = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($lnk) );
1644                                 }
1645
1646                                 $trail = $m[3];
1647                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1648                                 $might_be_img = true;
1649                                 $text = $m[2];
1650                                 $lnk = strpos( $m[1], '%' ) === false ? $m[1] : urldecode( $m[1] );
1651                                 $trail = "";
1652                         } else { # Invalid form; output directly
1653                                 $s .= $prefix . '[[' . $line ;
1654                                 wfProfileOut( __METHOD__."-e1" );
1655                                 continue;
1656                         }
1657                         wfProfileOut( __METHOD__."-e1" );
1658                         wfProfileIn( __METHOD__."-misc" );
1659
1660                         # Don't allow internal links to pages containing
1661                         # PROTO: where PROTO is a valid URL protocol; these
1662                         # should be external links.
1663                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $lnk ) ) {
1664                                 $s .= $prefix . '[[' . $line ;
1665                                 wfProfileOut( __METHOD__."-misc" );
1666                                 continue;
1667                         }
1668
1669                         # Make subpage if necessary
1670                         if ( $useSubpages ) {
1671                                 $link = $this->maybeDoSubpageLink( $lnk, $text );
1672                         } else {
1673                                 $link = $lnk;
1674                         }
1675
1676                         $noforce = (substr( $lnk, 0, 1 ) !== ':');
1677                         if (!$noforce) {
1678                                 # Strip off leading ':'
1679                                 $link = substr( $link, 1 );
1680                         }
1681
1682                         wfProfileOut( __METHOD__."-misc" );
1683                         wfProfileIn( __METHOD__."-title" );
1684                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1685                         if ( $nt === null ) {
1686                                 $s .= $prefix . '[[' . $line;
1687                                 wfProfileOut( __METHOD__."-title" );
1688                                 continue;
1689                         }
1690
1691                         $ns = $nt->getNamespace();
1692                         $iw = $nt->getInterWiki();
1693                         wfProfileOut( __METHOD__."-title" );
1694
1695                         if ( $might_be_img ) { # if this is actually an invalid link
1696                                 wfProfileIn( __METHOD__."-might_be_img" );
1697                                 if ( $ns == NS_FILE && $noforce ) { #but might be an image
1698                                         $found = false;
1699                                         while ( true ) {
1700                                                 #look at the next 'line' to see if we can close it there
1701                                                 $a->next();
1702                                                 $next_line = $a->current();
1703                                                 if ( $next_line === false || $next_line === null ) {
1704                                                         break;
1705                                                 }
1706                                                 $m = explode( ']]', $next_line, 3 );
1707                                                 if ( count( $m ) == 3 ) {
1708                                                         # the first ]] closes the inner link, the second the image
1709                                                         $found = true;
1710                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1711                                                         $trail = $m[2];
1712                                                         break;
1713                                                 } elseif ( count( $m ) == 2 ) {
1714                                                         #if there's exactly one ]] that's fine, we'll keep looking
1715                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1716                                                 } else {
1717                                                         #if $next_line is invalid too, we need look no further
1718                                                         $text .= '[[' . $next_line;
1719                                                         break;
1720                                                 }
1721                                         }
1722                                         if ( !$found ) {
1723                                                 # we couldn't find the end of this imageLink, so output it raw
1724                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1725                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1726                                                 $s .= "{$prefix}[[$link|$text";
1727                                                 # note: no $trail, because without an end, there *is* no trail
1728                                                 wfProfileOut( __METHOD__."-might_be_img" );
1729                                                 continue;
1730                                         }
1731                                 } else { #it's not an image, so output it raw
1732                                         $s .= "{$prefix}[[$link|$text";
1733                                         # note: no $trail, because without an end, there *is* no trail
1734                                         wfProfileOut( __METHOD__."-might_be_img" );
1735                                         continue;
1736                                 }
1737                                 wfProfileOut( __METHOD__."-might_be_img" );
1738                         }
1739
1740                         $wasblank = ( $text  == '' );
1741                         if ( $wasblank ) $text = $link;
1742
1743                         # Link not escaped by : , create the various objects
1744                         if ( $noforce ) {
1745
1746                                 # Interwikis
1747                                 wfProfileIn( __METHOD__."-interwiki" );
1748                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1749                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1750                                         $s = rtrim($s . $prefix);
1751                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1752                                         wfProfileOut( __METHOD__."-interwiki" );
1753                                         continue;
1754                                 }
1755                                 wfProfileOut( __METHOD__."-interwiki" );
1756
1757                                 if ( $ns == NS_FILE ) {
1758                                         wfProfileIn( __METHOD__."-image" );
1759                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1760                                                 if ( $wasblank ) {
1761                                                         # if no parameters were passed, $text
1762                                                         # becomes something like "File:Foo.png",
1763                                                         # which we don't want to pass on to the
1764                                                         # image generator
1765                                                         $text = '';
1766                                                 } else {
1767                                                         # recursively parse links inside the image caption
1768                                                         # actually, this will parse them in any other parameters, too,
1769                                                         # but it might be hard to fix that, and it doesn't matter ATM
1770                                                         $text = $this->replaceExternalLinks($text);
1771                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1772                                                 }
1773                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1774                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1775                                         } else {
1776                                                 $s .= $prefix . $trail;
1777                                         }
1778                                         $this->mOutput->addImage( $nt->getDBkey() );
1779                                         wfProfileOut( __METHOD__."-image" );
1780                                         continue;
1781
1782                                 }
1783
1784                                 if ( $ns == NS_CATEGORY ) {
1785                                         wfProfileIn( __METHOD__."-category" );
1786                                         $s = rtrim($s . "\n"); # bug 87
1787
1788                                         if ( $wasblank ) {
1789                                                 $sortkey = $this->getDefaultSort();
1790                                         } else {
1791                                                 $sortkey = $text;
1792                                         }
1793                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1794                                         $sortkey = str_replace( "\n", '', $sortkey );
1795                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1796                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1797
1798                                         /**
1799                                          * Strip the whitespace Category links produce, see bug 87
1800                                          * @todo We might want to use trim($tmp, "\n") here.
1801                                          */
1802                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1803
1804                                         wfProfileOut( __METHOD__."-category" );
1805                                         continue;
1806                                 }
1807                         }
1808
1809                         # Self-link checking
1810                         if( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1811                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1812                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1813                                         continue;
1814                                 }
1815                         }
1816
1817                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1818                         # FIXME: Should do batch file existence checks, see comment below
1819                         if( $ns == NS_MEDIA ) {
1820                                 wfProfileIn( __METHOD__."-media" );
1821                                 # Give extensions a chance to select the file revision for us
1822                                 $skip = $time = false;
1823                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1824                                 if ( $skip ) {
1825                                         $link = $sk->link( $nt );
1826                                 } else {
1827                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1828                                 }
1829                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1830                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1831                                 $this->mOutput->addImage( $nt->getDBkey() );
1832                                 wfProfileOut( __METHOD__."-media" );
1833                                 continue;
1834                         }
1835
1836                         wfProfileIn( __METHOD__."-always_known" );
1837                         # Some titles, such as valid special pages or files in foreign repos, should
1838                         # be shown as bluelinks even though they're not included in the page table
1839                         #
1840                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1841                         # batch file existence checks for NS_FILE and NS_MEDIA
1842                         if( $iw == '' && $nt->isAlwaysKnown() ) {
1843                                 $this->mOutput->addLink( $nt );
1844                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1845                         } else {
1846                                 # Links will be added to the output link list after checking
1847                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1848                         }
1849                         wfProfileOut( __METHOD__."-always_known" );
1850                 }
1851                 wfProfileOut( __METHOD__ );
1852                 return $holders;
1853         }
1854
1855         /**
1856          * Make a link placeholder. The text returned can be later resolved to a real link with
1857          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1858          * parsing of interwiki links, and secondly to allow all existence checks and
1859          * article length checks (for stub links) to be bundled into a single query.
1860          *
1861          * @deprecated
1862          */
1863         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1864                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1865         }
1866
1867         /**
1868          * Render a forced-blue link inline; protect against double expansion of
1869          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1870          * Since this little disaster has to split off the trail text to avoid
1871          * breaking URLs in the following text without breaking trails on the
1872          * wiki links, it's been made into a horrible function.
1873          *
1874          * @param Title $nt
1875          * @param string $text
1876          * @param string $query
1877          * @param string $trail
1878          * @param string $prefix
1879          * @return string HTML-wikitext mix oh yuck
1880          */
1881         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1882                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1883                 $sk = $this->mOptions->getSkin();
1884                 // FIXME: use link() instead of deprecated makeKnownLinkObj()
1885                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1886                 return $this->armorLinks( $link ) . $trail;
1887         }
1888
1889         /**
1890          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1891          * going to go through further parsing steps before inline URL expansion.
1892          *
1893          * Not needed quite as much as it used to be since free links are a bit
1894          * more sensible these days. But bracketed links are still an issue.
1895          *
1896          * @param string more-or-less HTML
1897          * @return string less-or-more HTML with NOPARSE bits
1898          */
1899         function armorLinks( $text ) {
1900                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1901                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1902         }
1903
1904         /**
1905          * Return true if subpage links should be expanded on this page.
1906          * @return bool
1907          */
1908         function areSubpagesAllowed() {
1909                 # Some namespaces don't allow subpages
1910                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1911         }
1912
1913         /**
1914          * Handle link to subpage if necessary
1915          * @param string $target the source of the link
1916          * @param string &$text the link text, modified as necessary
1917          * @return string the full name of the link
1918          * @private
1919          */
1920         function maybeDoSubpageLink($target, &$text) {
1921                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
1922         }
1923
1924         /**
1925          * From the [[title|]] return link-text as though the used typed [[title|link-text]]
1926          * @param string $link from [[$link|]]
1927          * @return string $text for [[$link|$text]]
1928          */
1929         function getPipeTrickText( $link ) {
1930                 return Linker::getPipeTrickText( $link );
1931         }
1932
1933         /**
1934          * From the [[|link-text]] return the title as though the user typed [[title|link-text]]
1935          * @param string $text from [[|$text]]
1936          * @param Title $title to resolve the link against
1937          * @return string $link for [[$link|$text]]
1938          */
1939         function getPipeTrickLink( $text ) {
1940                 return Linker::getPipeTrickLink( $text, $this->mTitle );
1941         }
1942
1943         /**#@+
1944          * Used by doBlockLevels()
1945          * @private
1946          */
1947         /* private */ function closeParagraph() {
1948                 $result = '';
1949                 if ( $this->mLastSection != '' ) {
1950                         $result = '</' . $this->mLastSection  . ">\n";
1951                 }
1952                 $this->mInPre = false;
1953                 $this->mLastSection = '';
1954                 return $result;
1955         }
1956         # getCommon() returns the length of the longest common substring
1957         # of both arguments, starting at the beginning of both.
1958         #
1959         /* private */ function getCommon( $st1, $st2 ) {
1960                 $fl = strlen( $st1 );
1961                 $shorter = strlen( $st2 );
1962                 if ( $fl < $shorter ) { $shorter = $fl; }
1963
1964                 for ( $i = 0; $i < $shorter; ++$i ) {
1965                         if ( $st1{$i} != $st2{$i} ) { break; }
1966                 }
1967                 return $i;
1968         }
1969         # These next three functions open, continue, and close the list
1970         # element appropriate to the prefix character passed into them.
1971         #
1972         /* private */ function openList( $char ) {
1973                 $result = $this->closeParagraph();
1974
1975                 if ( '*' === $char ) { $result .= '<ul><li>'; }
1976                 elseif ( '#' === $char ) { $result .= '<ol><li>'; }
1977                 elseif ( ':' === $char ) { $result .= '<dl><dd>'; }
1978                 elseif ( ';' === $char ) {
1979                         $result .= '<dl><dt>';
1980                         $this->mDTopen = true;
1981                 }
1982                 else { $result = '<!-- ERR 1 -->'; }
1983
1984                 return $result;
1985         }
1986
1987         /* private */ function nextItem( $char ) {
1988                 if ( '*' === $char || '#' === $char ) { return '</li><li>'; }
1989                 elseif ( ':' === $char || ';' === $char ) {
1990                         $close = '</dd>';
1991                         if ( $this->mDTopen ) { $close = '</dt>'; }
1992                         if ( ';' === $char ) {
1993                                 $this->mDTopen = true;
1994                                 return $close . '<dt>';
1995                         } else {
1996                                 $this->mDTopen = false;
1997                                 return $close . '<dd>';
1998                         }
1999                 }
2000                 return '<!-- ERR 2 -->';
2001         }
2002
2003         /* private */ function closeList( $char ) {
2004                 if ( '*' === $char ) { $text = '</li></ul>'; }
2005                 elseif ( '#' === $char ) { $text = '</li></ol>'; }
2006                 elseif ( ':' === $char ) {
2007                         if ( $this->mDTopen ) {
2008                                 $this->mDTopen = false;
2009                                 $text = '</dt></dl>';
2010                         } else {
2011                                 $text = '</dd></dl>';
2012                         }
2013                 }
2014                 else {  return '<!-- ERR 3 -->'; }
2015                 return $text."\n";
2016         }
2017         /**#@-*/
2018
2019         /**
2020          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2021          *
2022          * @param $linestart bool whether or not this is at the start of a line.
2023          * @private
2024          * @return string the lists rendered as HTML
2025          */
2026         function doBlockLevels( $text, $linestart ) {
2027                 wfProfileIn( __METHOD__ );
2028
2029                 # Parsing through the text line by line.  The main thing
2030                 # happening here is handling of block-level elements p, pre,
2031                 # and making lists from lines starting with * # : etc.
2032                 #
2033                 $textLines = StringUtils::explode( "\n", $text );
2034
2035                 $lastPrefix = $output = '';
2036                 $this->mDTopen = $inBlockElem = false;
2037                 $prefixLength = 0;
2038                 $paragraphStack = false;
2039
2040                 foreach ( $textLines as $oLine ) {
2041                         # Fix up $linestart
2042                         if ( !$linestart ) {
2043                                 $output .= $oLine;
2044                                 $linestart = true;
2045                                 continue;
2046                         }
2047                         // * = ul
2048                         // # = ol
2049                         // ; = dt
2050                         // : = dd
2051
2052                         $lastPrefixLength = strlen( $lastPrefix );
2053                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2054                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2055                         // If not in a <pre> element, scan for and figure out what prefixes are there.
2056                         if ( !$this->mInPre ) {
2057                                 # Multiple prefixes may abut each other for nested lists.
2058                                 $prefixLength = strspn( $oLine, '*#:;' );
2059                                 $prefix = substr( $oLine, 0, $prefixLength );
2060
2061                                 # eh?
2062                                 // ; and : are both from definition-lists, so they're equivalent
2063                                 //  for the purposes of determining whether or not we need to open/close
2064                                 //  elements.
2065                                 $prefix2 = str_replace( ';', ':', $prefix );
2066                                 $t = substr( $oLine, $prefixLength );
2067                                 $this->mInPre = (bool)$preOpenMatch;
2068                         } else {
2069                                 # Don't interpret any other prefixes in preformatted text
2070                                 $prefixLength = 0;
2071                                 $prefix = $prefix2 = '';
2072                                 $t = $oLine;
2073                         }
2074
2075                         # List generation
2076                         if( $prefixLength && $lastPrefix === $prefix2 ) {
2077                                 # Same as the last item, so no need to deal with nesting or opening stuff
2078                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2079                                 $paragraphStack = false;
2080
2081                                 if ( substr( $prefix, -1 ) === ';') {
2082                                         # The one nasty exception: definition lists work like this:
2083                                         # ; title : definition text
2084                                         # So we check for : in the remainder text to split up the
2085                                         # title and definition, without b0rking links.
2086                                         $term = $t2 = '';
2087                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2088                                                 $t = $t2;
2089                                                 $output .= $term . $this->nextItem( ':' );
2090                                         }
2091                                 }
2092                         } elseif( $prefixLength || $lastPrefixLength ) {
2093                                 // We need to open or close prefixes, or both.
2094
2095                                 # Either open or close a level...
2096                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2097                                 $paragraphStack = false;
2098
2099                                 // Close all the prefixes which aren't shared.
2100                                 while( $commonPrefixLength < $lastPrefixLength ) {
2101                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2102                                         --$lastPrefixLength;
2103                                 }
2104
2105                                 // Continue the current prefix if appropriate.
2106                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2107                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2108                                 }
2109
2110                                 // Open prefixes where appropriate.
2111                                 while ( $prefixLength > $commonPrefixLength ) {
2112                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2113                                         $output .= $this->openList( $char );
2114
2115                                         if ( ';' === $char ) {
2116                                                 # FIXME: This is dupe of code above
2117                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2118                                                         $t = $t2;
2119                                                         $output .= $term . $this->nextItem( ':' );
2120                                                 }
2121                                         }
2122                                         ++$commonPrefixLength;
2123                                 }
2124                                 $lastPrefix = $prefix2;
2125                         }
2126
2127                         // If we have no prefixes, go to paragraph mode.
2128                         if( 0 == $prefixLength ) {
2129                                 wfProfileIn( __METHOD__."-paragraph" );
2130                                 # No prefix (not in list)--go to paragraph mode
2131                                 // XXX: use a stack for nestable elements like span, table and div
2132                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2133                                 $closematch = preg_match(
2134                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2135                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2136                                 if ( $openmatch or $closematch ) {
2137                                         $paragraphStack = false;
2138                                         # TODO bug 5718: paragraph closed
2139                                         $output .= $this->closeParagraph();
2140                                         if ( $preOpenMatch and !$preCloseMatch ) {
2141                                                 $this->mInPre = true;
2142                                         }
2143                                         if ( $closematch ) {
2144                                                 $inBlockElem = false;
2145                                         } else {
2146                                                 $inBlockElem = true;
2147                                         }
2148                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2149                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' or trim($t) != '' ) ) {
2150                                                 // pre
2151                                                 if ($this->mLastSection !== 'pre') {
2152                                                         $paragraphStack = false;
2153                                                         $output .= $this->closeParagraph().'<pre>';
2154                                                         $this->mLastSection = 'pre';
2155                                                 }
2156                                                 $t = substr( $t, 1 );
2157                                         } else {
2158                                                 // paragraph
2159                                                 if ( trim($t) == '' ) {
2160                                                         if ( $paragraphStack ) {
2161                                                                 $output .= $paragraphStack.'<br />';
2162                                                                 $paragraphStack = false;
2163                                                                 $this->mLastSection = 'p';
2164                                                         } else {
2165                                                                 if ($this->mLastSection !== 'p' ) {
2166                                                                         $output .= $this->closeParagraph();
2167                                                                         $this->mLastSection = '';
2168                                                                         $paragraphStack = '<p>';
2169                                                                 } else {
2170                                                                         $paragraphStack = '</p><p>';
2171                                                                 }
2172                                                         }
2173                                                 } else {
2174                                                         if ( $paragraphStack ) {
2175                                                                 $output .= $paragraphStack;
2176                                                                 $paragraphStack = false;
2177                                                                 $this->mLastSection = 'p';
2178                                                         } else if ($this->mLastSection !== 'p') {
2179                                                                 $output .= $this->closeParagraph().'<p>';
2180                                                                 $this->mLastSection = 'p';
2181                                                         }
2182                                                 }
2183                                         }
2184                                 }
2185                                 wfProfileOut( __METHOD__."-paragraph" );
2186                         }
2187                         // somewhere above we forget to get out of pre block (bug 785)
2188                         if($preCloseMatch && $this->mInPre) {
2189                                 $this->mInPre = false;
2190                         }
2191                         if ($paragraphStack === false) {
2192                                 $output .= $t."\n";
2193                         }
2194                 }
2195                 while ( $prefixLength ) {
2196                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2197                         --$prefixLength;
2198                 }
2199                 if ( $this->mLastSection != '' ) {
2200                         $output .= '</' . $this->mLastSection . '>';
2201                         $this->mLastSection = '';
2202                 }
2203
2204                 wfProfileOut( __METHOD__ );
2205                 return $output;
2206         }
2207
2208         /**
2209          * Split up a string on ':', ignoring any occurences inside tags
2210          * to prevent illegal overlapping.
2211          * @param string $str the string to split
2212          * @param string &$before set to everything before the ':'
2213          * @param string &$after set to everything after the ':'
2214          * return string the position of the ':', or false if none found
2215          */
2216         function findColonNoLinks($str, &$before, &$after) {
2217                 wfProfileIn( __METHOD__ );
2218
2219                 $pos = strpos( $str, ':' );
2220                 if( $pos === false ) {
2221                         // Nothing to find!
2222                         wfProfileOut( __METHOD__ );
2223                         return false;
2224                 }
2225
2226                 $lt = strpos( $str, '<' );
2227                 if( $lt === false || $lt > $pos ) {
2228                         // Easy; no tag nesting to worry about
2229                         $before = substr( $str, 0, $pos );
2230                         $after = substr( $str, $pos+1 );
2231                         wfProfileOut( __METHOD__ );
2232                         return $pos;
2233                 }
2234
2235                 // Ugly state machine to walk through avoiding tags.
2236                 $state = self::COLON_STATE_TEXT;
2237                 $stack = 0;
2238                 $len = strlen( $str );
2239                 for( $i = 0; $i < $len; $i++ ) {
2240                         $c = $str{$i};
2241
2242                         switch( $state ) {
2243                         // (Using the number is a performance hack for common cases)
2244                         case 0: // self::COLON_STATE_TEXT:
2245                                 switch( $c ) {
2246                                 case "<":
2247                                         // Could be either a <start> tag or an </end> tag
2248                                         $state = self::COLON_STATE_TAGSTART;
2249                                         break;
2250                                 case ":":
2251                                         if( $stack == 0 ) {
2252                                                 // We found it!
2253                                                 $before = substr( $str, 0, $i );
2254                                                 $after = substr( $str, $i + 1 );
2255                                                 wfProfileOut( __METHOD__ );
2256                                                 return $i;
2257                                         }
2258                                         // Embedded in a tag; don't break it.
2259                                         break;
2260                                 default:
2261                                         // Skip ahead looking for something interesting
2262                                         $colon = strpos( $str, ':', $i );
2263                                         if( $colon === false ) {
2264                                                 // Nothing else interesting
2265                                                 wfProfileOut( __METHOD__ );
2266                                                 return false;
2267                                         }
2268                                         $lt = strpos( $str, '<', $i );
2269                                         if( $stack === 0 ) {
2270                                                 if( $lt === false || $colon < $lt ) {
2271                                                         // We found it!
2272                                                         $before = substr( $str, 0, $colon );
2273                                                         $after = substr( $str, $colon + 1 );
2274                                                         wfProfileOut( __METHOD__ );
2275                                                         return $i;
2276                                                 }
2277                                         }
2278                                         if( $lt === false ) {
2279                                                 // Nothing else interesting to find; abort!
2280                                                 // We're nested, but there's no close tags left. Abort!
2281                                                 break 2;
2282                                         }
2283                                         // Skip ahead to next tag start
2284                                         $i = $lt;
2285                                         $state = self::COLON_STATE_TAGSTART;
2286                                 }
2287                                 break;
2288                         case 1: // self::COLON_STATE_TAG:
2289                                 // In a <tag>
2290                                 switch( $c ) {
2291                                 case ">":
2292                                         $stack++;
2293                                         $state = self::COLON_STATE_TEXT;
2294                                         break;
2295                                 case "/":
2296                                         // Slash may be followed by >?
2297                                         $state = self::COLON_STATE_TAGSLASH;
2298                                         break;
2299                                 default:
2300                                         // ignore
2301                                 }
2302                                 break;
2303                         case 2: // self::COLON_STATE_TAGSTART:
2304                                 switch( $c ) {
2305                                 case "/":
2306                                         $state = self::COLON_STATE_CLOSETAG;
2307                                         break;
2308                                 case "!":
2309                                         $state = self::COLON_STATE_COMMENT;
2310                                         break;
2311                                 case ">":
2312                                         // Illegal early close? This shouldn't happen D:
2313                                         $state = self::COLON_STATE_TEXT;
2314                                         break;
2315                                 default:
2316                                         $state = self::COLON_STATE_TAG;
2317                                 }
2318                                 break;
2319                         case 3: // self::COLON_STATE_CLOSETAG:
2320                                 // In a </tag>
2321                                 if( $c === ">" ) {
2322                                         $stack--;
2323                                         if( $stack < 0 ) {
2324                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2325                                                 wfProfileOut( __METHOD__ );
2326                                                 return false;
2327                                         }
2328                                         $state = self::COLON_STATE_TEXT;
2329                                 }
2330                                 break;
2331                         case self::COLON_STATE_TAGSLASH:
2332                                 if( $c === ">" ) {
2333                                         // Yes, a self-closed tag <blah/>
2334                                         $state = self::COLON_STATE_TEXT;
2335                                 } else {
2336                                         // Probably we're jumping the gun, and this is an attribute
2337                                         $state = self::COLON_STATE_TAG;
2338                                 }
2339                                 break;
2340                         case 5: // self::COLON_STATE_COMMENT:
2341                                 if( $c === "-" ) {
2342                                         $state = self::COLON_STATE_COMMENTDASH;
2343                                 }
2344                                 break;
2345                         case self::COLON_STATE_COMMENTDASH:
2346                                 if( $c === "-" ) {
2347                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2348                                 } else {
2349                                         $state = self::COLON_STATE_COMMENT;
2350                                 }
2351                                 break;
2352                         case self::COLON_STATE_COMMENTDASHDASH:
2353                                 if( $c === ">" ) {
2354                                         $state = self::COLON_STATE_TEXT;
2355                                 } else {
2356                                         $state = self::COLON_STATE_COMMENT;
2357                                 }
2358                                 break;
2359                         default:
2360                                 throw new MWException( "State machine error in " . __METHOD__ );
2361                         }
2362                 }
2363                 if( $stack > 0 ) {
2364                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2365                         return false;
2366                 }
2367                 wfProfileOut( __METHOD__ );
2368                 return false;
2369         }
2370
2371         /**
2372          * Return value of a magic variable (like PAGENAME)
2373          *
2374          * @private
2375          */
2376         function getVariableValue( $index, $frame=false ) {
2377                 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2378                 global $wgScriptPath, $wgStylePath;
2379
2380                 /**
2381                  * Some of these require message or data lookups and can be
2382                  * expensive to check many times.
2383                  */
2384                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2385                         if ( isset( $this->mVarCache[$index] ) ) {
2386                                 return $this->mVarCache[$index];
2387                         }
2388                 }
2389
2390                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2391                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2392
2393                 # Use the time zone
2394                 global $wgLocaltimezone;
2395                 if ( isset( $wgLocaltimezone ) ) {
2396                         $oldtz = date_default_timezone_get();
2397                         date_default_timezone_set( $wgLocaltimezone );
2398                 }
2399
2400                 $localTimestamp = date( 'YmdHis', $ts );
2401                 $localMonth = date( 'm', $ts );
2402                 $localMonth1 = date( 'n', $ts );
2403                 $localMonthName = date( 'n', $ts );
2404                 $localDay = date( 'j', $ts );
2405                 $localDay2 = date( 'd', $ts );
2406                 $localDayOfWeek = date( 'w', $ts );
2407                 $localWeek = date( 'W', $ts );
2408                 $localYear = date( 'Y', $ts );
2409                 $localHour = date( 'H', $ts );
2410                 if ( isset( $wgLocaltimezone ) ) {
2411                         date_default_timezone_set( $oldtz );
2412                 }
2413
2414                 switch ( $index ) {
2415                         case 'currentmonth':
2416                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2417                                 break;
2418                         case 'currentmonth1':
2419                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2420                                 break;
2421                         case 'currentmonthname':
2422                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2423                                 break;
2424                         case 'currentmonthnamegen':
2425                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2426                                 break;
2427                         case 'currentmonthabbrev':
2428                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2429                                 break;
2430                         case 'currentday':
2431                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2432                                 break;
2433                         case 'currentday2':
2434                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2435                                 break;
2436                         case 'localmonth':
2437                                 $value = $wgContLang->formatNum( $localMonth );
2438                                 break;
2439                         case 'localmonth1':
2440                                 $value = $wgContLang->formatNum( $localMonth1 );
2441                                 break;
2442                         case 'localmonthname':
2443                                 $value = $wgContLang->getMonthName( $localMonthName );
2444                                 break;
2445                         case 'localmonthnamegen':
2446                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2447                                 break;
2448                         case 'localmonthabbrev':
2449                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2450                                 break;
2451                         case 'localday':
2452                                 $value = $wgContLang->formatNum( $localDay );
2453                                 break;
2454                         case 'localday2':
2455                                 $value = $wgContLang->formatNum( $localDay2 );
2456                                 break;
2457                         case 'pagename':
2458                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2459                                 break;
2460                         case 'pagenamee':
2461                                 $value = $this->mTitle->getPartialURL();
2462                                 break;
2463                         case 'fullpagename':
2464                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2465                                 break;
2466                         case 'fullpagenamee':
2467                                 $value = $this->mTitle->getPrefixedURL();
2468                                 break;
2469                         case 'subpagename':
2470                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2471                                 break;
2472                         case 'subpagenamee':
2473                                 $value = $this->mTitle->getSubpageUrlForm();
2474                                 break;
2475                         case 'basepagename':
2476                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2477                                 break;
2478                         case 'basepagenamee':
2479                                 $value = wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2480                                 break;
2481                         case 'talkpagename':
2482                                 if( $this->mTitle->canTalk() ) {
2483                                         $talkPage = $this->mTitle->getTalkPage();
2484                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2485                                 } else {
2486                                         $value = '';
2487                                 }
2488                                 break;
2489                         case 'talkpagenamee':
2490                                 if( $this->mTitle->canTalk() ) {
2491                                         $talkPage = $this->mTitle->getTalkPage();
2492                                         $value = $talkPage->getPrefixedUrl();
2493                                 } else {
2494                                         $value = '';
2495                                 }
2496                                 break;
2497                         case 'subjectpagename':
2498                                 $subjPage = $this->mTitle->getSubjectPage();
2499                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2500                                 break;
2501                         case 'subjectpagenamee':
2502                                 $subjPage = $this->mTitle->getSubjectPage();
2503                                 $value = $subjPage->getPrefixedUrl();
2504                                 break;
2505                         case 'pipetrick':
2506                                 $text = $this->mTitle->getText();
2507                                 $value = $this->getPipeTrickText( $text );
2508                                 break;
2509                         case 'pipetricke':
2510                                 $text = $this->mTitle->getText();
2511                                 $value = wfUrlEncode( str_replace( ' ', '_', $this->getPipeTrickText( $text ) ) );
2512                                 break;
2513                         case 'revisionid':
2514                                 // Let the edit saving system know we should parse the page
2515                                 // *after* a revision ID has been assigned.
2516                                 $this->mOutput->setFlag( 'vary-revision' );
2517                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2518                                 $value = $this->mRevisionId;
2519                                 break;
2520                         case 'revisionday':
2521                                 // Let the edit saving system know we should parse the page
2522                                 // *after* a revision ID has been assigned. This is for null edits.
2523                                 $this->mOutput->setFlag( 'vary-revision' );
2524                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2525                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2526                                 break;
2527                         case 'revisionday2':
2528                                 // Let the edit saving system know we should parse the page
2529                                 // *after* a revision ID has been assigned. This is for null edits.
2530                                 $this->mOutput->setFlag( 'vary-revision' );
2531                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2532                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2533                                 break;
2534                         case 'revisionmonth':
2535                                 // Let the edit saving system know we should parse the page
2536                                 // *after* a revision ID has been assigned. This is for null edits.
2537                                 $this->mOutput->setFlag( 'vary-revision' );
2538                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2539                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2540                                 break;
2541                         case 'revisionyear':
2542                                 // Let the edit saving system know we should parse the page
2543                                 // *after* a revision ID has been assigned. This is for null edits.
2544                                 $this->mOutput->setFlag( 'vary-revision' );
2545                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2546                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2547                                 break;
2548                         case 'revisiontimestamp':
2549                                 // Let the edit saving system know we should parse the page
2550                                 // *after* a revision ID has been assigned. This is for null edits.
2551                                 $this->mOutput->setFlag( 'vary-revision' );
2552                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2553                                 $value = $this->getRevisionTimestamp();
2554                                 break;
2555                         case 'revisionuser':
2556                                 // Let the edit saving system know we should parse the page
2557                                 // *after* a revision ID has been assigned. This is for null edits.
2558                                 $this->mOutput->setFlag( 'vary-revision' );
2559                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2560                                 $value = $this->getRevisionUser();
2561                                 break;
2562                         case 'namespace':
2563                                 $value = str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2564                                 break;
2565                         case 'namespacee':
2566                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2567                                 break;
2568                         case 'talkspace':
2569                                 $value = $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2570                                 break;
2571                         case 'talkspacee':
2572                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2573                                 break;
2574                         case 'subjectspace':
2575                                 $value = $this->mTitle->getSubjectNsText();
2576                                 break;
2577                         case 'subjectspacee':
2578                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2579                                 break;
2580                         case 'currentdayname':
2581                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2582                                 break;
2583                         case 'currentyear':
2584                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2585                                 break;
2586                         case 'currenttime':
2587                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2588                                 break;
2589                         case 'currenthour':
2590                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2591                                 break;
2592                         case 'currentweek':
2593                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2594                                 // int to remove the padding
2595                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2596                                 break;
2597                         case 'currentdow':
2598                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2599                                 break;
2600                         case 'localdayname':
2601                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2602                                 break;
2603                         case 'localyear':
2604                                 $value = $wgContLang->formatNum( $localYear, true );
2605                                 break;
2606                         case 'localtime':
2607                                 $value = $wgContLang->time( $localTimestamp, false, false );
2608                                 break;
2609                         case 'localhour':
2610                                 $value = $wgContLang->formatNum( $localHour, true );
2611                                 break;
2612                         case 'localweek':
2613                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2614                                 // int to remove the padding
2615                                 $value = $wgContLang->formatNum( (int)$localWeek );
2616                                 break;
2617                         case 'localdow':
2618                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2619                                 break;
2620                         case 'numberofarticles':
2621                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2622                                 break;
2623                         case 'numberoffiles':
2624                                 $value = $wgContLang->formatNum( SiteStats::images() );
2625                                 break;
2626                         case 'numberofusers':
2627                                 $value = $wgContLang->formatNum( SiteStats::users() );
2628                                 break;
2629                         case 'numberofactiveusers':
2630                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2631                                 break;
2632                         case 'numberofpages':
2633                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2634                                 break;
2635                         case 'numberofadmins':
2636                                 $value = $wgContLang->formatNum( SiteStats::numberingroup('sysop') );
2637                                 break;
2638                         case 'numberofedits':
2639                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2640                                 break;
2641                         case 'numberofviews':
2642                                 $value = $wgContLang->formatNum( SiteStats::views() );
2643                                 break;
2644                         case 'currenttimestamp':
2645                                 $value = wfTimestamp( TS_MW, $ts );
2646                                 break;
2647                         case 'localtimestamp':
2648                                 $value = $localTimestamp;
2649                                 break;
2650                         case 'currentversion':
2651                                 $value = SpecialVersion::getVersion();
2652                                 break;
2653                         case 'sitename':
2654                                 return $wgSitename;
2655                         case 'server':
2656                                 return $wgServer;
2657                         case 'servername':
2658                                 return $wgServerName;
2659                         case 'scriptpath':
2660                                 return $wgScriptPath;
2661                         case 'stylepath':
2662                                 return $wgStylePath;
2663                         case 'directionmark':
2664                                 return $wgContLang->getDirMark();
2665                         case 'contentlanguage':
2666                                 global $wgContLanguageCode;
2667                                 return $wgContLanguageCode;
2668                         default:
2669                                 $ret = null;
2670                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) )
2671                                         return $ret;
2672                                 else
2673                                         return null;
2674                 }
2675
2676                 if ( $index )
2677                         $this->mVarCache[$index] = $value;
2678
2679                 return $value;
2680         }
2681
2682         /**
2683          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2684          *
2685          * @private
2686          */
2687         function initialiseVariables() {
2688                 wfProfileIn( __METHOD__ );
2689                 $variableIDs = MagicWord::getVariableIDs();
2690                 $substIDs = MagicWord::getSubstIDs();
2691
2692                 $this->mVariables = new MagicWordArray( $variableIDs );
2693                 $this->mSubsts = new MagicWordArray( $substIDs );
2694                 wfProfileOut( __METHOD__ );
2695         }
2696
2697         /**
2698          * Preprocess some wikitext and return the document tree.
2699          * This is the ghost of replace_variables().
2700          *
2701          * @param string $text The text to parse
2702          * @param integer flags Bitwise combination of:
2703          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2704          *                                     included. Default is to assume a direct page view.
2705          *
2706          * The generated DOM tree must depend only on the input text and the flags.
2707          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2708          *
2709          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2710          * change in the DOM tree for a given text, must be passed through the section identifier
2711          * in the section edit link and thus back to extractSections().
2712          *
2713          * The output of this function is currently only cached in process memory, but a persistent
2714          * cache may be implemented at a later date which takes further advantage of these strict
2715          * dependency requirements.
2716          *
2717          * @private
2718          */
2719         function preprocessToDom ( $text, $flags = 0 ) {
2720                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2721                 return $dom;
2722         }
2723
2724         /*
2725          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2726          */
2727         public static function splitWhitespace( $s ) {
2728                 $ltrimmed = ltrim( $s );
2729                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2730                 $trimmed = rtrim( $ltrimmed );
2731                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2732                 if ( $diff > 0 ) {
2733                         $w2 = substr( $ltrimmed, -$diff );
2734                 } else {
2735                         $w2 = '';
2736                 }
2737                 return array( $w1, $trimmed, $w2 );
2738         }
2739
2740         /**
2741          * Replace magic variables, templates, and template arguments
2742          * with the appropriate text. Templates are substituted recursively,
2743          * taking care to avoid infinite loops.
2744          *
2745          * Note that the substitution depends on value of $mOutputType:
2746          *  self::OT_WIKI: only {{subst:}} templates
2747          *  self::OT_PREPROCESS: templates but not extension tags
2748          *  self::OT_HTML: all templates and extension tags
2749          *
2750          * @param string $tex The text to transform
2751          * @param PPFrame $frame Object describing the arguments passed to the template.
2752          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2753          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2754          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2755          * @private
2756          */
2757         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2758                 # Is there any text? Also, Prevent too big inclusions!
2759                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2760                         return $text;
2761                 }
2762                 wfProfileIn( __METHOD__ );
2763
2764                 if ( $frame === false ) {
2765                         $frame = $this->getPreprocessor()->newFrame();
2766                 } elseif ( !( $frame instanceof PPFrame ) ) {
2767                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2768                         $frame = $this->getPreprocessor()->newCustomFrame($frame);
2769                 }
2770
2771                 $dom = $this->preprocessToDom( $text );
2772                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2773                 $text = $frame->expand( $dom, $flags );
2774
2775                 wfProfileOut( __METHOD__ );
2776                 return $text;
2777         }
2778
2779         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2780         static function createAssocArgs( $args ) {
2781                 $assocArgs = array();
2782                 $index = 1;
2783                 foreach( $args as $arg ) {
2784                         $eqpos = strpos( $arg, '=' );
2785                         if ( $eqpos === false ) {
2786                                 $assocArgs[$index++] = $arg;
2787                         } else {
2788                                 $name = trim( substr( $arg, 0, $eqpos ) );
2789                                 $value = trim( substr( $arg, $eqpos+1 ) );
2790                                 if ( $value === false ) {
2791                                         $value = '';
2792                                 }
2793                                 if ( $name !== false ) {
2794                                         $assocArgs[$name] = $value;
2795                                 }
2796                         }
2797                 }
2798
2799                 return $assocArgs;
2800         }
2801
2802         /**
2803          * Warn the user when a parser limitation is reached
2804          * Will warn at most once the user per limitation type
2805          *
2806          * @param string $limitationType, should be one of:
2807          *   'expensive-parserfunction' (corresponding messages: 'expensive-parserfunction-warning', 'expensive-parserfunction-category')
2808          *   'post-expand-template-argument' (corresponding messages: 'post-expand-template-argument-warning', 'post-expand-template-argument-category')
2809          *   'post-expand-template-inclusion' (corresponding messages: 'post-expand-template-inclusion-warning', 'post-expand-template-inclusion-category')
2810          * @params int $current, $max When an explicit limit has been
2811          *       exceeded, provide the values (optional)
2812          */
2813         function limitationWarn( $limitationType, $current=null, $max=null) {
2814                 //does no harm if $current and $max are present but are unnecessary for the message
2815                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
2816                 $this->mOutput->addWarning( $warning );
2817                 $this->addTrackingCategory( "$limitationType-category" );
2818         }
2819
2820         /**
2821          * Return the text of a template, after recursively
2822          * replacing any variables or templates within the template.
2823          *
2824          * @param array $piece The parts of the template
2825          *  $piece['title']: the title, i.e. the part before the |
2826          *  $piece['parts']: the parameter array
2827          *  $piece['lineStart']: whether the brace was at the start of a line
2828          * @param PPFrame The current frame, contains template arguments
2829          * @return string the text of the template
2830          * @private
2831          */
2832         function braceSubstitution( $piece, $frame ) {
2833                 global $wgContLang, $wgNonincludableNamespaces;
2834                 wfProfileIn( __METHOD__ );
2835                 wfProfileIn( __METHOD__.'-setup' );
2836
2837                 # Flags
2838                 $found = false;             # $text has been filled
2839                 $nowiki = false;            # wiki markup in $text should be escaped
2840                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2841                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2842                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2843                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2844
2845                 # Title object, where $text came from
2846                 $title = null;
2847
2848                 # $part1 is the bit before the first |, and must contain only title characters.
2849                 # Various prefixes will be stripped from it later.
2850                 $titleWithSpaces = $frame->expand( $piece['title'] );
2851                 $part1 = trim( $titleWithSpaces );
2852                 $titleText = false;
2853
2854                 # Original title text preserved for various purposes
2855                 $originalTitle = $part1;
2856
2857                 # $args is a list of argument nodes, starting from index 0, not including $part1
2858                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2859                 wfProfileOut( __METHOD__.'-setup' );
2860
2861                 # SUBST
2862                 wfProfileIn( __METHOD__.'-modifiers' );
2863                 if ( !$found ) {
2864
2865                         $substMatch = $this->mSubsts->matchStartAndRemove( $part1 );
2866
2867                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
2868                         # Whether to include depends also on whether we are in the pre-save-transform
2869                         #
2870                         # safesubst || (subst && PST) || (false && !PST) => transclude (skip the if)
2871                         # (false && PST) || (subst && !PST)  => return input (handled by if)
2872                         if ( $substMatch != 'safesubst' && ($substMatch == 'subst' xor $this->ot['wiki']) ) {
2873                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2874                                 $isLocalObj = true;
2875                                 $found = true;
2876                         }
2877                 }
2878
2879                 # Variables
2880                 if ( !$found && $args->getLength() == 0 ) {
2881                         $id = $this->mVariables->matchStartToEnd( $part1 );
2882                         if ( $id !== false ) {
2883                                 $text = $this->getVariableValue( $id, $frame );
2884                                 if (MagicWord::getCacheTTL($id)>-1)
2885                                         $this->mOutput->mContainsOldMagic = true;
2886                                 $found = true;
2887                         }
2888                 }
2889
2890                 # MSG, MSGNW and RAW
2891                 if ( !$found ) {
2892                         # Check for MSGNW:
2893                         $mwMsgnw = MagicWord::get( 'msgnw' );
2894                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2895                                 $nowiki = true;
2896                         } else {
2897                                 # Remove obsolete MSG:
2898                                 $mwMsg = MagicWord::get( 'msg' );
2899                                 $mwMsg->matchStartAndRemove( $part1 );
2900                         }
2901
2902                         # Check for RAW:
2903                         $mwRaw = MagicWord::get( 'raw' );
2904                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2905                                 $forceRawInterwiki = true;
2906                         }
2907                 }
2908                 wfProfileOut( __METHOD__.'-modifiers' );
2909
2910                 # Parser functions
2911                 if ( !$found ) {
2912                         wfProfileIn( __METHOD__ . '-pfunc' );
2913
2914                         $colonPos = strpos( $part1, ':' );
2915                         if ( $colonPos !== false ) {
2916                                 # Case sensitive functions
2917                                 $function = substr( $part1, 0, $colonPos );
2918                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2919                                         $function = $this->mFunctionSynonyms[1][$function];
2920                                 } else {
2921                                         # Case insensitive functions
2922                                         $function = $wgContLang->lc( $function );
2923                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2924                                                 $function = $this->mFunctionSynonyms[0][$function];
2925                                         } else {
2926                                                 $function = false;
2927                                         }
2928                                 }
2929                                 if ( $function ) {
2930                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2931                                         $initialArgs = array( &$this );
2932                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2933                                         if ( $flags & SFH_OBJECT_ARGS ) {
2934                                                 # Add a frame parameter, and pass the arguments as an array
2935                                                 $allArgs = $initialArgs;
2936                                                 $allArgs[] = $frame;
2937                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2938                                                         $funcArgs[] = $args->item( $i );
2939                                                 }
2940                                                 $allArgs[] = $funcArgs;
2941                                         } else {
2942                                                 # Convert arguments to plain text
2943                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2944                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2945                                                 }
2946                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2947                                         }
2948
2949                                         # Workaround for PHP bug 35229 and similar
2950                                         if ( !is_callable( $callback ) ) {
2951                                                 wfProfileOut( __METHOD__ . '-pfunc' );
2952                                                 wfProfileOut( __METHOD__ );
2953                                                 throw new MWException( "Tag hook for $function is not callable\n" );
2954                                         }
2955                                         $result = call_user_func_array( $callback, $allArgs );
2956                                         $found = true;
2957                                         $noparse = true;
2958                                         $preprocessFlags = 0;
2959
2960                                         if ( is_array( $result ) ) {
2961                                                 if ( isset( $result[0] ) ) {
2962                                                         $text = $result[0];
2963                                                         unset( $result[0] );
2964                                                 }
2965
2966                                                 // Extract flags into the local scope
2967                                                 // This allows callers to set flags such as nowiki, found, etc.
2968                                                 extract( $result );
2969                                         } else {
2970                                                 $text = $result;
2971                                         }
2972                                         if ( !$noparse ) {
2973                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
2974                                                 $isChildObj = true;
2975                                         }
2976                                 }
2977                         }
2978                         wfProfileOut( __METHOD__ . '-pfunc' );
2979                 }
2980
2981                 # Finish mangling title and then check for loops.
2982                 # Set $title to a Title object and $titleText to the PDBK
2983                 if ( !$found ) {
2984                         $ns = NS_TEMPLATE;
2985                         # Split the title into page and subpage
2986                         $subpage = '';
2987                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2988                         if ($subpage !== '') {
2989                                 $ns = $this->mTitle->getNamespace();
2990                         }
2991                         $title = Title::newFromText( $part1, $ns );
2992                         if ( $title ) {
2993                                 $titleText = $title->getPrefixedText();
2994                                 # Check for language variants if the template is not found
2995                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
2996                                         $wgContLang->findVariantLink( $part1, $title, true );
2997                                 }
2998                                 # Do recursion depth check
2999                                 $limit = $this->mOptions->getMaxTemplateDepth();
3000                                 if ( $frame->depth >= $limit ) {
3001                                         $found = true;
3002                                         $text = '<span class="error">' . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit ) . '</span>';
3003                                 }
3004                         }
3005                 }
3006
3007                 # Load from database
3008                 if ( !$found && $title ) {
3009                         wfProfileIn( __METHOD__ . '-loadtpl' );
3010                         if ( !$title->isExternal() ) {
3011                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
3012                                         $text = SpecialPage::capturePath( $title );
3013                                         if ( is_string( $text ) ) {
3014                                                 $found = true;
3015                                                 $isHTML = true;
3016                                                 $this->disableCache();
3017                                         }
3018                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3019                                         $found = false; //access denied
3020                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3021                                 } else {
3022                                         list( $text, $title ) = $this->getTemplateDom( $title );
3023                                         if ( $text !== false ) {
3024                                                 $found = true;
3025                                                 $isChildObj = true;
3026                                         }
3027                                 }
3028
3029                                 # If the title is valid but undisplayable, make a link to it
3030                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3031                                         $text = "[[:$titleText]]";
3032                                         $found = true;
3033                                 }
3034                         } elseif ( $title->isTrans() ) {
3035                                 // Interwiki transclusion
3036                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3037                                         $text = $this->interwikiTransclude( $title, 'render' );
3038                                         $isHTML = true;
3039                                 } else {
3040                                         $text = $this->interwikiTransclude( $title, 'raw' );
3041                                         // Preprocess it like a template
3042                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3043                                         $isChildObj = true;
3044                                 }
3045                                 $found = true;
3046                         }
3047
3048                         # Do infinite loop check
3049                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3050                         if ( !$frame->loopCheck( $title ) ) {
3051                                 $found = true;
3052                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3053                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3054                         }
3055                         wfProfileOut( __METHOD__ . '-loadtpl' );
3056                 }
3057
3058                 # If we haven't found text to substitute by now, we're done
3059                 # Recover the source wikitext and return it
3060                 if ( !$found ) {
3061                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3062                         wfProfileOut( __METHOD__ );
3063                         return array( 'object' => $text );
3064                 }
3065
3066                 # Expand DOM-style return values in a child frame
3067                 if ( $isChildObj ) {
3068                         # Clean up argument array
3069                         $newFrame = $frame->newChild( $args, $title );
3070
3071                         if ( $nowiki ) {
3072                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3073                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3074                                 # Expansion is eligible for the empty-frame cache
3075                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3076                                         $text = $this->mTplExpandCache[$titleText];
3077                                 } else {
3078                                         $text = $newFrame->expand( $text );
3079                                         $this->mTplExpandCache[$titleText] = $text;
3080                                 }
3081                         } else {
3082                                 # Uncached expansion
3083                                 $text = $newFrame->expand( $text );
3084                         }
3085                 }
3086                 if ( $isLocalObj && $nowiki ) {
3087                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3088                         $isLocalObj = false;
3089                 }
3090
3091                 # Replace raw HTML by a placeholder
3092                 # Add a blank line preceding, to prevent it from mucking up
3093                 # immediately preceding headings
3094                 if ( $isHTML ) {
3095                         $text = "\n\n" . $this->insertStripItem( $text );
3096                 }
3097                 # Escape nowiki-style return values
3098                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3099                         $text = wfEscapeWikiText( $text );
3100                 }
3101                 # Bug 529: if the template begins with a table or block-level
3102                 # element, it should be treated as beginning a new line.
3103                 # This behaviour is somewhat controversial.
3104                 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3105                         $text = "\n" . $text;
3106                 }
3107
3108                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3109                         # Error, oversize inclusion
3110                         $text = "[[$originalTitle]]" .
3111                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3112                         $this->limitationWarn( 'post-expand-template-inclusion' );
3113                 }
3114
3115                 if ( $isLocalObj ) {
3116                         $ret = array( 'object' => $text );
3117                 } else {
3118                         $ret = array( 'text' => $text );
3119                 }
3120
3121                 wfProfileOut( __METHOD__ );
3122                 return $ret;
3123         }
3124
3125         /**
3126          * Get the semi-parsed DOM representation of a template with a given title,
3127          * and its redirect destination title. Cached.
3128          */
3129         function getTemplateDom( $title ) {
3130                 $cacheTitle = $title;
3131                 $titleText = $title->getPrefixedDBkey();
3132
3133                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3134                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3135                         $title = Title::makeTitle( $ns, $dbk );
3136                         $titleText = $title->getPrefixedDBkey();
3137                 }
3138                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3139                         return array( $this->mTplDomCache[$titleText], $title );
3140                 }
3141
3142                 // Cache miss, go to the database
3143                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3144
3145                 if ( $text === false ) {
3146                         $this->mTplDomCache[$titleText] = false;
3147                         return array( false, $title );
3148                 }
3149
3150                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3151                 $this->mTplDomCache[ $titleText ] = $dom;
3152
3153                 if (! $title->equals($cacheTitle)) {
3154                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3155                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3156                 }
3157
3158                 return array( $dom, $title );
3159         }
3160
3161         /**
3162          * Fetch the unparsed text of a template and register a reference to it.
3163          */
3164         function fetchTemplateAndTitle( $title ) {
3165                 $templateCb = $this->mOptions->getTemplateCallback();
3166                 $stuff = call_user_func( $templateCb, $title, $this );
3167                 $text = $stuff['text'];
3168                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3169                 if ( isset( $stuff['deps'] ) ) {
3170                         foreach ( $stuff['deps'] as $dep ) {
3171                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3172                         }
3173                 }
3174                 return array($text,$finalTitle);
3175         }
3176
3177         function fetchTemplate( $title ) {
3178                 $rv = $this->fetchTemplateAndTitle($title);
3179                 return $rv[0];
3180         }
3181
3182         /**
3183          * Static function to get a template
3184          * Can be overridden via ParserOptions::setTemplateCallback().
3185          */
3186         static function statelessFetchTemplate( $title, $parser=false ) {
3187                 $text = $skip = false;
3188                 $finalTitle = $title;
3189                 $deps = array();
3190
3191                 // Loop to fetch the article, with up to 1 redirect
3192                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3193                         # Give extensions a chance to select the revision instead
3194                         $id = false; // Assume current
3195                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3196
3197                         if( $skip ) {
3198                                 $text = false;
3199                                 $deps[] = array(
3200                                         'title' => $title,
3201                                         'page_id' => $title->getArticleID(),
3202                                         'rev_id' => null );
3203                                 break;
3204                         }
3205                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3206                         $rev_id = $rev ? $rev->getId() : 0;
3207                         // If there is no current revision, there is no page
3208                         if( $id === false && !$rev ) {
3209                                 $linkCache = LinkCache::singleton();
3210                                 $linkCache->addBadLinkObj( $title );
3211                         }
3212
3213                         $deps[] = array(
3214                                 'title' => $title,
3215                                 'page_id' => $title->getArticleID(),
3216                                 'rev_id' => $rev_id );
3217
3218                         if( $rev ) {
3219                                 $text = $rev->getText();
3220                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3221                                 global $wgContLang;
3222                                 $message = $wgContLang->lcfirst( $title->getText() );
3223                                 $text = wfMsgForContentNoTrans( $message );
3224                                 if( wfEmptyMsg( $message, $text ) ) {
3225                                         $text = false;
3226                                         break;
3227                                 }
3228                         } else {
3229                                 break;
3230                         }
3231                         if ( $text === false ) {
3232                                 break;
3233                         }
3234                         // Redirect?
3235                         $finalTitle = $title;
3236                         $title = Title::newFromRedirect( $text );
3237                 }
3238                 return array(
3239                         'text' => $text,
3240                         'finalTitle' => $finalTitle,
3241                         'deps' => $deps );
3242         }
3243
3244         /**
3245          * Transclude an interwiki link.
3246          */
3247         function interwikiTransclude( $title, $action ) {
3248                 global $wgEnableScaryTranscluding;
3249
3250                 if (!$wgEnableScaryTranscluding)
3251                         return wfMsg('scarytranscludedisabled');
3252
3253                 $url = $title->getFullUrl( "action=$action" );
3254
3255                 if (strlen($url) > 255)
3256                         return wfMsg('scarytranscludetoolong');
3257                 return $this->fetchScaryTemplateMaybeFromCache($url);
3258         }
3259
3260         function fetchScaryTemplateMaybeFromCache($url) {
3261                 global $wgTranscludeCacheExpiry;
3262                 $dbr = wfGetDB(DB_SLAVE);
3263                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3264                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3265                                 array('tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3266                 if ($obj) {
3267                         return $obj->tc_contents;
3268                 }
3269
3270                 $text = Http::get($url);
3271                 if (!$text)
3272                         return wfMsg('scarytranscludefailed', $url);
3273
3274                 $dbw = wfGetDB(DB_MASTER);
3275                 $dbw->replace('transcache', array('tc_url'), array(
3276                         'tc_url' => $url,
3277                         'tc_time' => $dbw->timestamp( time() ),
3278                         'tc_contents' => $text));
3279                 return $text;
3280         }
3281
3282
3283         /**
3284          * Triple brace replacement -- used for template arguments
3285          * @private
3286          */
3287         function argSubstitution( $piece, $frame ) {
3288                 wfProfileIn( __METHOD__ );
3289
3290                 $error = false;
3291                 $parts = $piece['parts'];
3292                 $nameWithSpaces = $frame->expand( $piece['title'] );
3293                 $argName = trim( $nameWithSpaces );
3294                 $object = false;
3295                 $text = $frame->getArgument( $argName );
3296                 if (  $text === false && $parts->getLength() > 0
3297                   && (
3298                     $this->ot['html']
3299                     || $this->ot['pre']
3300                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3301                   )
3302                 ) {
3303                         # No match in frame, use the supplied default
3304                         $object = $parts->item( 0 )->getChildren();
3305                 }
3306                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3307                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3308                         $this->limitationWarn( 'post-expand-template-argument' );
3309                 }
3310
3311                 if ( $text === false && $object === false ) {
3312                         # No match anywhere
3313                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3314                 }
3315                 if ( $error !== false ) {
3316                         $text .= $error;
3317                 }
3318                 if ( $object !== false ) {
3319                         $ret = array( 'object' => $object );
3320                 } else {
3321                         $ret = array( 'text' => $text );
3322                 }
3323
3324                 wfProfileOut( __METHOD__ );
3325                 return $ret;
3326         }
3327
3328         /**
3329          * Return the text to be used for a given extension tag.
3330          * This is the ghost of strip().
3331          *
3332          * @param array $params Associative array of parameters:
3333          *     name       PPNode for the tag name
3334          *     attr       PPNode for unparsed text where tag attributes are thought to be
3335          *     attributes Optional associative array of parsed attributes
3336          *     inner      Contents of extension element
3337          *     noClose    Original text did not have a close tag
3338          * @param PPFrame $frame
3339          */
3340         function extensionSubstitution( $params, $frame ) {
3341                 global $wgRawHtml, $wgContLang;
3342
3343                 $name = $frame->expand( $params['name'] );
3344                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3345                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3346                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $this->mMarkerIndex++) . self::MARKER_SUFFIX;
3347
3348                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3349                         ( $this->ot['html'] || $this->ot['pre'] );
3350                 if ( $isFunctionTag ) {
3351                         $markerType = 'none';
3352                 } else {
3353                         $markerType = 'general';
3354                 }
3355                 if ( $this->ot['html'] || $isFunctionTag ) {
3356                         $name = strtolower( $name );
3357                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3358                         if ( isset( $params['attributes'] ) ) {
3359                                 $attributes = $attributes + $params['attributes'];
3360                         }
3361
3362                         if( isset( $this->mTagHooks[$name] ) ) {
3363                                 # Workaround for PHP bug 35229 and similar
3364                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3365                                         throw new MWException( "Tag hook for $name is not callable\n" );
3366                                 }
3367                                 $output = call_user_func_array( $this->mTagHooks[$name],
3368                                         array( $content, $attributes, $this, $frame ) );
3369                         } elseif( isset( $this->mFunctionTagHooks[$name] ) ) {
3370                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3371                                 if( !is_callable( $callback ) )
3372                                         throw new MWException( "Tag hook for $name is not callable\n" );
3373
3374                                 $output = call_user_func_array( $callback,
3375                                         array( &$this, $frame, $content, $attributes ) );
3376                         } else {
3377                                 $output = '<span class="error">Invalid tag extension name: ' .
3378                                         htmlspecialchars( $name ) . '</span>';
3379                         }
3380
3381                         if ( is_array( $output ) ) {
3382                                 // Extract flags to local scope (to override $markerType)
3383                                 $flags = $output;
3384                                 $output = $flags[0];
3385                                 unset( $flags[0] );
3386                                 extract( $flags );
3387                         }
3388                 } else {
3389                         if ( is_null( $attrText ) ) {
3390                                 $attrText = '';
3391                         }
3392                         if ( isset( $params['attributes'] ) ) {
3393                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3394                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3395                                                 htmlspecialchars( $attrValue ) . '"';
3396                                 }
3397                         }
3398                         if ( $content === null ) {
3399                                 $output = "<$name$attrText/>";
3400                         } else {
3401                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3402                                 $output = "<$name$attrText>$content$close";
3403                         }
3404                 }
3405
3406                 if( $markerType === 'none' ) {
3407                         return $output;
3408                 } elseif ( $markerType === 'nowiki' ) {
3409                         $this->mStripState->nowiki->setPair( $marker, $output );
3410                 } elseif ( $markerType === 'general' ) {
3411                         $this->mStripState->general->setPair( $marker, $output );
3412                 } else {
3413                         throw new MWException( __METHOD__.': invalid marker type' );
3414                 }
3415                 return $marker;
3416         }
3417
3418         /**
3419          * Increment an include size counter
3420          *
3421          * @param string $type The type of expansion
3422          * @param integer $size The size of the text
3423          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3424          */
3425         function incrementIncludeSize( $type, $size ) {
3426                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3427                         return false;
3428                 } else {
3429                         $this->mIncludeSizes[$type] += $size;
3430                         return true;
3431                 }
3432         }
3433
3434         /**
3435          * Increment the expensive function count
3436          *
3437          * @return boolean False if the limit has been exceeded
3438          */
3439         function incrementExpensiveFunctionCount() {
3440                 global $wgExpensiveParserFunctionLimit;
3441                 $this->mExpensiveFunctionCount++;
3442                 if($this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit) {
3443                         return true;
3444                 }
3445                 return false;
3446         }
3447
3448         /**
3449          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3450          * Fills $this->mDoubleUnderscores, returns the modified text
3451          */
3452         function doDoubleUnderscore( $text ) {
3453                 wfProfileIn( __METHOD__ );
3454
3455                 // The position of __TOC__ needs to be recorded
3456                 $mw = MagicWord::get( 'toc' );
3457                 if( $mw->match( $text ) ) {
3458                         $this->mShowToc = true;
3459                         $this->mForceTocPosition = true;
3460
3461                         // Set a placeholder. At the end we'll fill it in with the TOC.
3462                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3463
3464                         // Only keep the first one.
3465                         $text = $mw->replace( '', $text );
3466                 }
3467
3468                 // Now match and remove the rest of them
3469                 $mwa = MagicWord::getDoubleUnderscoreArray();
3470                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3471
3472                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3473                         $this->mOutput->mNoGallery = true;
3474                 }
3475                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3476                         $this->mShowToc = false;
3477                 }
3478                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3479                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3480                         $this->addTrackingCategory( 'hidden-category-category' );
3481                 }
3482                 # (bug 8068) Allow control over whether robots index a page.
3483                 #
3484                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3485                 # is not desirable, the last one on the page should win.
3486                 if( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3487                         $this->mOutput->setIndexPolicy( 'noindex' );
3488                         $this->addTrackingCategory( 'noindex-category' );
3489                 }
3490                 if( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ){
3491                         $this->mOutput->setIndexPolicy( 'index' );
3492                         $this->addTrackingCategory( 'index-category' );
3493                 }
3494
3495                 wfProfileOut( __METHOD__ );
3496                 return $text;
3497         }
3498
3499         /**
3500          * Add a tracking category, getting the title from a system message,
3501          * or print a debug message if the title is invalid.
3502          * @param $msg String message key
3503          * @return Bool whether the addition was successful
3504          */
3505         protected function addTrackingCategory( $msg ){
3506                 $cat = wfMsgForContent( $msg );
3507
3508                 # Allow tracking categories to be disabled by setting them to "-"
3509                 if( $cat === '-' ) return false;
3510
3511                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3512                 if ( $containerCategory ) {
3513                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3514                         return true;
3515                 } else {
3516                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3517                         return false;
3518                 }
3519         }
3520
3521         /**
3522          * This function accomplishes several tasks:
3523          * 1) Auto-number headings if that option is enabled
3524          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3525          * 3) Add a Table of contents on the top for users who have enabled the option
3526          * 4) Auto-anchor headings
3527          *
3528          * It loops through all headlines, collects the necessary data, then splits up the
3529          * string and re-inserts the newly formatted headlines.
3530          *
3531          * @param string $text
3532          * @param string $origText Original, untouched wikitext
3533          * @param boolean $isMain
3534          * @private
3535          */
3536         function formatHeadings( $text, $origText, $isMain=true ) {
3537                 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3538
3539                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3540                 $showEditLink = $this->mOptions->getEditSection();
3541
3542                 // Do not call quickUserCan unless necessary
3543                 if( $showEditLink && !$this->mTitle->quickUserCan( 'edit' ) ) {
3544                         $showEditLink = 0;
3545                 }
3546
3547                 # Inhibit editsection links if requested in the page
3548                 if ( isset( $this->mDoubleUnderscores['noeditsection'] )  || $this->mOptions->getIsPrintable() ) {
3549                         $showEditLink = 0;
3550                 }
3551
3552                 # Get all headlines for numbering them and adding funky stuff like [edit]
3553                 # links - this is for later, but we need the number of headlines right now
3554                 $matches = array();
3555                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3556
3557                 # if there are fewer than 4 headlines in the article, do not show TOC
3558                 # unless it's been explicitly enabled.
3559                 $enoughToc = $this->mShowToc &&
3560                         (($numMatches >= 4) || $this->mForceTocPosition);
3561
3562                 # Allow user to stipulate that a page should have a "new section"
3563                 # link added via __NEWSECTIONLINK__
3564                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3565                         $this->mOutput->setNewSection( true );
3566                 }
3567
3568                 # Allow user to remove the "new section"
3569                 # link via __NONEWSECTIONLINK__
3570                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3571                         $this->mOutput->hideNewSection( true );
3572                 }
3573
3574                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3575                 # override above conditions and always show TOC above first header
3576                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3577                         $this->mShowToc = true;
3578                         $enoughToc = true;
3579                 }
3580
3581                 # We need this to perform operations on the HTML
3582                 $sk = $this->mOptions->getSkin();
3583
3584                 # headline counter
3585                 $headlineCount = 0;
3586                 $numVisible = 0;
3587
3588                 # Ugh .. the TOC should have neat indentation levels which can be
3589                 # passed to the skin functions. These are determined here
3590                 $toc = '';
3591                 $full = '';
3592                 $head = array();
3593                 $sublevelCount = array();
3594                 $levelCount = array();
3595                 $toclevel = 0;
3596                 $level = 0;
3597                 $prevlevel = 0;
3598                 $toclevel = 0;
3599                 $prevtoclevel = 0;
3600                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3601                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3602                 $oldType = $this->mOutputType;
3603                 $this->setOutputType( self::OT_WIKI );
3604                 $frame = $this->getPreprocessor()->newFrame();
3605                 $root = $this->preprocessToDom( $origText );
3606                 $node = $root->getFirstChild();
3607                 $byteOffset = 0;
3608                 $tocraw = array();
3609
3610                 foreach( $matches[3] as $headline ) {
3611                         $isTemplate = false;
3612                         $titleText = false;
3613                         $sectionIndex = false;
3614                         $numbering = '';
3615                         $markerMatches = array();
3616                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3617                                 $serial = $markerMatches[1];
3618                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3619                                 $isTemplate = ($titleText != $baseTitleText);
3620                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3621                         }
3622
3623                         if( $toclevel ) {
3624                                 $prevlevel = $level;
3625                                 $prevtoclevel = $toclevel;
3626                         }
3627                         $level = $matches[1][$headlineCount];
3628
3629                         if ( $level > $prevlevel ) {
3630                                 # Increase TOC level
3631                                 $toclevel++;
3632                                 $sublevelCount[$toclevel] = 0;
3633                                 if( $toclevel<$wgMaxTocLevel ) {
3634                                         $prevtoclevel = $toclevel;
3635                                         $toc .= $sk->tocIndent();
3636                                         $numVisible++;
3637                                 }
3638                         }
3639                         elseif ( $level < $prevlevel && $toclevel > 1 ) {
3640                                 # Decrease TOC level, find level to jump to
3641
3642                                 for ($i = $toclevel; $i > 0; $i--) {
3643                                         if ( $levelCount[$i] == $level ) {
3644                                                 # Found last matching level
3645                                                 $toclevel = $i;
3646                                                 break;
3647                                         }
3648                                         elseif ( $levelCount[$i] < $level ) {
3649                                                 # Found first matching level below current level
3650                                                 $toclevel = $i + 1;
3651                                                 break;
3652                                         }
3653                                 }
3654                                 if( $i == 0 ) $toclevel = 1;
3655                                 if( $toclevel<$wgMaxTocLevel ) {
3656                                         if($prevtoclevel < $wgMaxTocLevel) {
3657                                                 # Unindent only if the previous toc level was shown :p
3658                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3659                                                 $prevtoclevel = $toclevel;
3660                                         } else {
3661                                                 $toc .= $sk->tocLineEnd();
3662                                         }
3663                                 }
3664                         }
3665                         else {
3666                                 # No change in level, end TOC line
3667                                 if( $toclevel<$wgMaxTocLevel ) {
3668                                         $toc .= $sk->tocLineEnd();
3669                                 }
3670                         }
3671
3672                         $levelCount[$toclevel] = $level;
3673
3674                         # count number of headlines for each level
3675                         @$sublevelCount[$toclevel]++;
3676                         $dot = 0;
3677                         for( $i = 1; $i <= $toclevel; $i++ ) {
3678                                 if( !empty( $sublevelCount[$i] ) ) {
3679                                         if( $dot ) {
3680                                                 $numbering .= '.';
3681                                         }
3682                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3683                                         $dot = 1;
3684                                 }
3685                         }
3686
3687                         # The safe header is a version of the header text safe to use for links
3688                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3689                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3690
3691                         # Remove link placeholders by the link text.
3692                         #     <!--LINK number-->
3693                         # turns into
3694                         #     link text with suffix
3695                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3696
3697                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3698                         $tocline = preg_replace(
3699                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3700                                 array( '',                          '<$1>'),
3701                                 $safeHeadline
3702                         );
3703                         $tocline = trim( $tocline );
3704
3705                         # For the anchor, strip out HTML-y stuff period
3706                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3707                         $safeHeadline = preg_replace( '/[ _]+/', ' ', $safeHeadline );
3708                         $safeHeadline = trim( $safeHeadline );
3709
3710                         # Save headline for section edit hint before it's escaped
3711                         $headlineHint = $safeHeadline;
3712
3713                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
3714                                 # For reverse compatibility, provide an id that's
3715                                 # HTML4-compatible, like we used to.
3716                                 #
3717                                 # It may be worth noting, academically, that it's possible for
3718                                 # the legacy anchor to conflict with a non-legacy headline
3719                                 # anchor on the page.  In this case likely the "correct" thing
3720                                 # would be to either drop the legacy anchors or make sure
3721                                 # they're numbered first.  However, this would require people
3722                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3723                                 # manually, so let's not bother worrying about it.
3724                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3725                                         array( 'noninitial', 'legacy' ) );
3726                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3727
3728                                 if ( $legacyHeadline == $safeHeadline ) {
3729                                         # No reason to have both (in fact, we can't)
3730                                         $legacyHeadline = false;
3731                                 }
3732                         } else {
3733                                 $legacyHeadline = false;
3734                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3735                                         'noninitial' );
3736                         }
3737
3738                         # HTML names must be case-insensitively unique (bug 10721).  FIXME:
3739                         # Does this apply to Unicode characters?  Because we aren't
3740                         # handling those here.
3741                         $arrayKey = strtolower( $safeHeadline );
3742                         if ( $legacyHeadline === false ) {
3743                                 $legacyArrayKey = false;
3744                         } else {
3745                                 $legacyArrayKey = strtolower( $legacyHeadline );
3746                         }
3747
3748                         # count how many in assoc. array so we can track dupes in anchors
3749                         if ( isset( $refers[$arrayKey] ) ) {
3750                                 $refers[$arrayKey]++;
3751                         } else {
3752                                 $refers[$arrayKey] = 1;
3753                         }
3754                         if ( isset( $refers[$legacyArrayKey] ) ) {
3755                                 $refers[$legacyArrayKey]++;
3756                         } else {
3757                                 $refers[$legacyArrayKey] = 1;
3758                         }
3759
3760                         # Don't number the heading if it is the only one (looks silly)
3761                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3762                                 # the two are different if the line contains a link
3763                                 $headline = $numbering . ' ' . $headline;
3764                         }
3765
3766                         # Create the anchor for linking from the TOC to the section
3767                         $anchor = $safeHeadline;
3768                         $legacyAnchor = $legacyHeadline;
3769                         if ( $refers[$arrayKey] > 1 ) {
3770                                 $anchor .= '_' . $refers[$arrayKey];
3771                         }
3772                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3773                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3774                         }
3775                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3776                                 $toc .= $sk->tocLine($anchor, $tocline,
3777                                         $numbering, $toclevel, ($isTemplate ? false : $sectionIndex));
3778                         }
3779
3780                         # Add the section to the section tree
3781                         # Find the DOM node for this header
3782                         while ( $node && !$isTemplate ) {
3783                                 if ( $node->getName() === 'h' ) {
3784                                         $bits = $node->splitHeading();
3785                                         if ( $bits['i'] == $sectionIndex )
3786                                                 break;
3787                                 }
3788                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
3789                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
3790                                 $node = $node->getNextSibling();
3791                         }
3792                         $tocraw[] = array(
3793                                 'toclevel' => $toclevel,
3794                                 'level' => $level,
3795                                 'line' => $tocline,
3796                                 'number' => $numbering,
3797                                 'index' => ($isTemplate ? 'T-' : '' ) . $sectionIndex,
3798                                 'fromtitle' => $titleText,
3799                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
3800                                 'anchor' => $anchor,
3801                         );
3802
3803                         # give headline the correct <h#> tag
3804                         if( $showEditLink && $sectionIndex !== false ) {
3805                                 if( $isTemplate ) {
3806                                         # Put a T flag in the section identifier, to indicate to extractSections()
3807                                         # that sections inside <includeonly> should be counted.
3808                                         $editlink = $sk->doEditSectionLink(Title::newFromText( $titleText ), "T-$sectionIndex");
3809                                 } else {
3810                                         $editlink = $sk->doEditSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3811                                 }
3812                         } else {
3813                                 $editlink = '';
3814                         }
3815                         $head[$headlineCount] = $sk->makeHeadline( $level,
3816                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3817                                 $editlink, $legacyAnchor );
3818
3819                         $headlineCount++;
3820                 }
3821
3822                 $this->setOutputType( $oldType );
3823
3824                 # Never ever show TOC if no headers
3825                 if( $numVisible < 1 ) {
3826                         $enoughToc = false;
3827                 }
3828
3829                 if( $enoughToc ) {
3830                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3831                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3832                         }
3833                         $toc = $sk->tocList( $toc );
3834                         $this->mOutput->setTOCHTML( $toc );
3835                 }
3836
3837                 if ( $isMain ) {
3838                         $this->mOutput->setSections( $tocraw );
3839                 }
3840
3841                 # split up and insert constructed headlines
3842
3843                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3844                 $i = 0;
3845
3846                 foreach( $blocks as $block ) {
3847                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3848                                 # This is the [edit] link that appears for the top block of text when
3849                                 # section editing is enabled
3850
3851                                 # Disabled because it broke block formatting
3852                                 # For example, a bullet point in the top line
3853                                 # $full .= $sk->editSectionLink(0);
3854                         }
3855                         $full .= $block;
3856                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3857                                 # Top anchor now in skin
3858                                 $full = $full.$toc;
3859                         }
3860
3861                         if( !empty( $head[$i] ) ) {
3862                                 $full .= $head[$i];
3863                         }
3864                         $i++;
3865                 }
3866                 if( $this->mForceTocPosition ) {
3867                         return str_replace( '<!--MWTOC-->', $toc, $full );
3868                 } else {
3869                         return $full;
3870                 }
3871         }
3872
3873         /**
3874          * Merge $tree2 into $tree1 by replacing the section with index
3875          * $section in $tree1 and its descendants with the sections in $tree2.
3876          * Note that in the returned section tree, only the 'index' and
3877          * 'byteoffset' fields are guaranteed to be correct.
3878          * @param $tree1 array Section tree from ParserOutput::getSectons()
3879         *  @param $tree2 array Section tree
3880          * @param $section int Section index
3881          * @param $title Title Title both section trees come from
3882          * @param $len2 int Length of the original wikitext for $tree2
3883          * @return array Merged section tree
3884          */
3885         public static function mergeSectionTrees( $tree1, $tree2, $section, $title, $len2 ) {
3886                 global $wgContLang;
3887                 $newTree = array();
3888                 $targetLevel = false;
3889                 $merged = false;
3890                 $lastLevel = 1;
3891                 $nextIndex = 1;
3892                 $numbering = array( 0 );
3893                 $titletext = $title->getPrefixedDBkey();
3894                 foreach ( $tree1 as $s ) {
3895                         if ( $targetLevel !== false ) {
3896                                 if ( $s['level'] <= $targetLevel )
3897                                         // We've skipped enough
3898                                         $targetLevel = false;
3899                                 else
3900                                         continue;
3901                         }
3902                         if ( $s['index'] != $section ||
3903                                         $s['fromtitle'] != $titletext ) {
3904                                 self::incrementNumbering( $numbering,
3905                                         $s['toclevel'], $lastLevel );
3906
3907                                 // Rewrite index, byteoffset and number
3908                                 if ( $s['fromtitle'] == $titletext ) {
3909                                         $s['index'] = $nextIndex++;
3910                                         if ( $merged )
3911                                                 $s['byteoffset'] += $len2;
3912                                 }
3913                                 $s['number']  = implode( '.', array_map(
3914                                         array( $wgContLang, 'formatnum' ),
3915                                         $numbering ) );
3916                                 $lastLevel = $s['toclevel'];
3917                                 $newTree[] = $s;
3918                         } else {
3919                                 // We're at $section
3920                                 // Insert sections from $tree2 here
3921                                 foreach ( $tree2 as $s2 ) {
3922                                         // Rewrite the fields in $s2
3923                                         // before inserting it
3924                                         $s2['toclevel'] += $s['toclevel'] - 1;
3925                                         $s2['level'] += $s['level'] - 1;
3926                                         $s2['index'] = $nextIndex++;
3927                                         $s2['byteoffset'] += $s['byteoffset'];
3928
3929                                         self::incrementNumbering( $numbering,
3930                                                 $s2['toclevel'], $lastLevel );
3931                                         $s2['number']  = implode( '.', array_map(
3932                                                 array( $wgContLang, 'formatnum' ),
3933                                                 $numbering ) );
3934                                         $lastLevel = $s2['toclevel'];
3935                                         $newTree[] = $s2;
3936                                 }
3937                                 // Skip all descendants of $section in $tree1
3938                                 $targetLevel = $s['level'];
3939                                 $merged = true;
3940                         }
3941                 }
3942                 return $newTree;
3943         }
3944
3945         /**
3946          * Increment a section number. Helper function for mergeSectionTrees()
3947          * @param $number array Array representing a section number
3948          * @param $level int Current TOC level (depth)
3949          * @param $lastLevel int Level of previous TOC entry
3950          */
3951         private static function incrementNumbering( &$number, $level, $lastLevel ) {
3952                 if ( $level > $lastLevel )
3953                         $number[$level - 1] = 1;
3954                 else if ( $level < $lastLevel ) {
3955                         foreach ( $number as $key => $unused )
3956                                 if ( $key >= $level )
3957                                         unset( $number[$key] );
3958                         $number[$level - 1]++;
3959                 } else
3960                         $number[$level - 1]++;
3961         }
3962
3963         /**
3964          * Transform wiki markup when saving a page by doing \r\n -> \n
3965          * conversion, substitting signatures, {{subst:}} templates, etc.
3966          *
3967          * @param string $text the text to transform
3968          * @param Title &$title the Title object for the current article
3969          * @param User $user the User object describing the current user
3970          * @param ParserOptions $options parsing options
3971          * @param bool $clearState whether to clear the parser state first
3972          * @return string the altered wiki markup
3973          * @public
3974          */
3975         function preSaveTransform( $text, Title $title, $user, $options, $clearState = true ) {
3976                 $this->mOptions = $options;
3977                 $this->setTitle( $title );
3978                 $this->setOutputType( self::OT_WIKI );
3979
3980                 if ( $clearState ) {
3981                         $this->clearState();
3982                 }
3983
3984                 $pairs = array(
3985                         "\r\n" => "\n",
3986                 );
3987                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3988                 $text = $this->pstPass2( $text, $user );
3989                 $text = $this->mStripState->unstripBoth( $text );
3990                 return $text;
3991         }
3992
3993         /**
3994          * Pre-save transform helper function
3995          * @private
3996          */
3997         function pstPass2( $text, $user ) {
3998                 global $wgContLang, $wgLocaltimezone;
3999
4000                 /* Note: This is the timestamp saved as hardcoded wikitext to
4001                  * the database, we use $wgContLang here in order to give
4002                  * everyone the same signature and use the default one rather
4003                  * than the one selected in each user's preferences.
4004                  *
4005                  * (see also bug 12815)
4006                  */
4007                 $ts = $this->mOptions->getTimestamp();
4008                 if ( isset( $wgLocaltimezone ) ) {
4009                         $tz = $wgLocaltimezone;
4010                 } else {
4011                         $tz = date_default_timezone_get();
4012                 }
4013
4014                 $unixts = wfTimestamp( TS_UNIX, $ts );
4015                 $oldtz = date_default_timezone_get();
4016                 date_default_timezone_set( $tz );
4017                 $ts = date( 'YmdHis', $unixts );
4018                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4019
4020                 /* Allow translation of timezones trough wiki. date() can return
4021                  * whatever crap the system uses, localised or not, so we cannot
4022                  * ship premade translations.
4023                  */
4024                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4025                 $value = wfMsgForContent( $key );
4026                 if ( !wfEmptyMsg( $key, $value ) ) $tzMsg = $value;
4027
4028                 date_default_timezone_set( $oldtz );
4029
4030                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4031
4032                 # Variable replacement
4033                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4034                 $text = $this->replaceVariables( $text );
4035
4036                 # Signatures
4037                 $sigText = $this->getUserSig( $user );
4038                 $text = strtr( $text, array(
4039                         '~~~~~' => $d,
4040                         '~~~~' => "$sigText $d",
4041                         '~~~' => $sigText
4042                 ) );
4043
4044                 # Links of the form [[|<blah>]] or [[<blah>|]] perform pipe tricks
4045                 # Note this only allows the # in the position it works.
4046                 global $wgLegalTitleChars;
4047                 $pipeTrickRe = "/\[\[(?:(\\|)([$wgLegalTitleChars]+)|([#$wgLegalTitleChars]+)\\|)\]\]/";
4048                 $text = preg_replace_callback( $pipeTrickRe, array( $this, 'pstPipeTrickCallback' ), $text );
4049
4050                 # Trim trailing whitespace
4051                 $text = rtrim( $text );
4052
4053                 return $text;
4054         }
4055
4056         /**
4057          * Called from pstPass2 to perform the pipe trick on links.
4058          * Original was either [[|text]] or [[link|]]
4059          *
4060          * @param Array ("|" or "", text, link) $m
4061          */
4062         function pstPipeTrickCallback( $m )
4063         {
4064                 if( $m[1] ) { # [[|<blah>]]
4065                         $text = $m[2];
4066                         $link = $this->getPipeTrickLink( $text );
4067                 } else { # [[<blah>|]]
4068                         $link = $m[3];
4069                         $text = $this->getPipeTrickText( $link );
4070                 }
4071
4072                 return $link === $text ? "[[$link]]" : "[[$link|$text]]";
4073         }
4074
4075         /**
4076          * Fetch the user's signature text, if any, and normalize to
4077          * validated, ready-to-insert wikitext.
4078          * If you have pre-fetched the nickname or the fancySig option, you can
4079          * specify them here to save a database query.
4080          *
4081          * @param User $user
4082          * @return string
4083          */
4084         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4085                 global $wgMaxSigChars;
4086
4087                 $username = $user->getName();
4088
4089                 // If not given, retrieve from the user object.
4090                 if ( $nickname === false )
4091                         $nickname = $user->getOption( 'nickname' );
4092
4093                 if ( is_null( $fancySig ) )
4094                         $fancySig = $user->getBoolOption( 'fancysig' );
4095
4096                 $nickname = $nickname == null ? $username : $nickname;
4097
4098                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4099                         $nickname = $username;
4100                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4101                 } elseif( $fancySig !== false ) {
4102                         # Sig. might contain markup; validate this
4103                         if( $this->validateSig( $nickname ) !== false ) {
4104                                 # Validated; clean up (if needed) and return it
4105                                 return $this->cleanSig( $nickname, true );
4106                         } else {
4107                                 # Failed to validate; fall back to the default
4108                                 $nickname = $username;
4109                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4110                         }
4111                 }
4112
4113                 // Make sure nickname doesnt get a sig in a sig
4114                 $nickname = $this->cleanSigInSig( $nickname );
4115
4116                 # If we're still here, make it a link to the user page
4117                 $userText = wfEscapeWikiText( $username );
4118                 $nickText = wfEscapeWikiText( $nickname );
4119                 if ( $user->isAnon() )  {
4120                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4121                 } else {
4122                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4123                 }
4124         }
4125
4126         /**
4127          * Check that the user's signature contains no bad XML
4128          *
4129          * @param string $text
4130          * @return mixed An expanded string, or false if invalid.
4131          */
4132         function validateSig( $text ) {
4133                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4134         }
4135
4136         /**
4137          * Clean up signature text
4138          *
4139          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4140          * 2) Substitute all transclusions
4141          *
4142          * @param string $text
4143          * @param $parsing Whether we're cleaning (preferences save) or parsing
4144          * @return string Signature text
4145          */
4146         function cleanSig( $text, $parsing = false ) {
4147                 if ( !$parsing ) {
4148                         global $wgTitle;
4149                         $this->clearState();
4150                         $this->setTitle( $wgTitle );
4151                         $this->mOptions = new ParserOptions;
4152                         $this->setOutputType = self::OT_PREPROCESS;
4153                 }
4154
4155                 # Option to disable this feature
4156                 if ( !$this->mOptions->getCleanSignatures() ) {
4157                         return $text;
4158                 }
4159
4160                 # FIXME: regex doesn't respect extension tags or nowiki
4161                 #  => Move this logic to braceSubstitution()
4162                 $substWord = MagicWord::get( 'subst' );
4163                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4164                 $substText = '{{' . $substWord->getSynonym( 0 );
4165
4166                 $text = preg_replace( $substRegex, $substText, $text );
4167                 $text = $this->cleanSigInSig( $text );
4168                 $dom = $this->preprocessToDom( $text );
4169                 $frame = $this->getPreprocessor()->newFrame();
4170                 $text = $frame->expand( $dom );
4171
4172                 if ( !$parsing ) {
4173                         $text = $this->mStripState->unstripBoth( $text );
4174                 }
4175
4176                 return $text;
4177         }
4178
4179         /**
4180          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4181          * @param string $text
4182          * @return string Signature text with /~{3,5}/ removed
4183          */
4184         function cleanSigInSig( $text ) {
4185                 $text = preg_replace( '/~{3,5}/', '', $text );
4186                 return $text;
4187         }
4188
4189         /**
4190          * Set up some variables which are usually set up in parse()
4191          * so that an external function can call some class members with confidence
4192          * @public
4193          */
4194         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4195                 $this->setTitle( $title );
4196                 $this->mOptions = $options;
4197                 $this->setOutputType( $outputType );
4198                 if ( $clearState ) {
4199                         $this->clearState();
4200                 }
4201         }
4202
4203         /**
4204          * Wrapper for preprocess()
4205          *
4206          * @param string $text the text to preprocess
4207          * @param ParserOptions $options  options
4208          * @return string
4209          * @public
4210          */
4211         function transformMsg( $text, $options ) {
4212                 global $wgTitle;
4213                 static $executing = false;
4214
4215                 # Guard against infinite recursion
4216                 if ( $executing ) {
4217                         return $text;
4218                 }
4219                 $executing = true;
4220
4221                 wfProfileIn(__METHOD__);
4222                 $text = $this->preprocess( $text, $wgTitle, $options );
4223
4224                 $executing = false;
4225                 wfProfileOut(__METHOD__);
4226                 return $text;
4227         }
4228
4229         /**
4230          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4231          * The callback should have the following form:
4232          *    function myParserHook( $text, $params, &$parser ) { ... }
4233          *
4234          * Transform and return $text. Use $parser for any required context, e.g. use
4235          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4236          *
4237          * @public
4238          *
4239          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4240          * @param mixed $callback The callback function (and object) to use for the tag
4241          *
4242          * @return The old value of the mTagHooks array associated with the hook
4243          */
4244         function setHook( $tag, $callback ) {
4245                 $tag = strtolower( $tag );
4246                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4247                 $this->mTagHooks[$tag] = $callback;
4248                 if( !in_array( $tag, $this->mStripList ) ) {
4249                         $this->mStripList[] = $tag;
4250                 }
4251
4252                 return $oldVal;
4253         }
4254
4255         /* An old work-around for bug 2257 - deprecated 2010-02-13 */
4256         function setTransparentTagHook( $tag, $callback ) {
4257                 wfDeprecated( __METHOD__ );
4258                 $tag = strtolower( $tag );
4259                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4260                 $this->mTransparentTagHooks[$tag] = $callback;
4261
4262                 return $oldVal;
4263         }
4264
4265         /**
4266          * Remove all tag hooks
4267          */
4268         function clearTagHooks() {
4269                 $this->mTagHooks = array();
4270                 $this->mStripList = $this->mDefaultStripList;
4271         }
4272
4273         /**
4274          * Create a function, e.g. {{sum:1|2|3}}
4275          * The callback function should have the form:
4276          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4277          *
4278          * Or with SFH_OBJECT_ARGS:
4279          *    function myParserFunction( $parser, $frame, $args ) { ... }
4280          *
4281          * The callback may either return the text result of the function, or an array with the text
4282          * in element 0, and a number of flags in the other elements. The names of the flags are
4283          * specified in the keys. Valid flags are:
4284          *   found                     The text returned is valid, stop processing the template. This
4285          *                             is on by default.
4286          *   nowiki                    Wiki markup in the return value should be escaped
4287          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4288          *
4289          * @public
4290          *
4291          * @param string $id The magic word ID
4292          * @param mixed $callback The callback function (and object) to use
4293          * @param integer $flags a combination of the following flags:
4294          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4295          *
4296          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4297          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4298          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4299          *     the arguments, and to control the way they are expanded.
4300          *
4301          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4302          *     arguments, for instance:
4303          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4304          *
4305          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4306          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4307          *     working if/when this is changed.
4308          *
4309          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4310          *     expansion.
4311          *
4312          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4313          *     about the methods available in PPFrame and PPNode.
4314          *
4315          * @return The old callback function for this name, if any
4316          */
4317         function setFunctionHook( $id, $callback, $flags = 0 ) {
4318                 global $wgContLang;
4319
4320                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4321                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4322
4323                 # Add to function cache
4324                 $mw = MagicWord::get( $id );
4325                 if( !$mw )
4326                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4327
4328                 $synonyms = $mw->getSynonyms();
4329                 $sensitive = intval( $mw->isCaseSensitive() );
4330
4331                 foreach ( $synonyms as $syn ) {
4332                         # Case
4333                         if ( !$sensitive ) {
4334                                 $syn = $wgContLang->lc( $syn );
4335                         }
4336                         # Add leading hash
4337                         if ( !( $flags & SFH_NO_HASH ) ) {
4338                                 $syn = '#' . $syn;
4339                         }
4340                         # Remove trailing colon
4341                         if ( substr( $syn, -1, 1 ) === ':' ) {
4342                                 $syn = substr( $syn, 0, -1 );
4343                         }
4344                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4345                 }
4346                 return $oldVal;
4347         }
4348
4349         /**
4350          * Get all registered function hook identifiers
4351          *
4352          * @return array
4353          */
4354         function getFunctionHooks() {
4355                 return array_keys( $this->mFunctionHooks );
4356         }
4357
4358         /**
4359          * Create a tag function, e.g. <test>some stuff</test>.
4360          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4361          * Unlike parser functions, their content is not preprocessed.
4362          */
4363         function setFunctionTagHook( $tag, $callback, $flags ) {
4364                 $tag = strtolower( $tag );
4365                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4366                         $this->mFunctionTagHooks[$tag] : null;
4367                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4368
4369                 if( !in_array( $tag, $this->mStripList ) ) {
4370                         $this->mStripList[] = $tag;
4371                 }
4372
4373                 return $old;
4374         }
4375
4376         /**
4377          * FIXME: update documentation. makeLinkObj() is deprecated.
4378          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4379          * Placeholders created in Skin::makeLinkObj()
4380          * Returns an array of link CSS classes, indexed by PDBK.
4381          */
4382         function replaceLinkHolders( &$text, $options = 0 ) {
4383                 return $this->mLinkHolders->replace( $text );
4384         }
4385
4386         /**
4387          * Replace <!--LINK--> link placeholders with plain text of links
4388          * (not HTML-formatted).
4389          * @param string $text
4390          * @return string
4391          */
4392         function replaceLinkHoldersText( $text ) {
4393                 return $this->mLinkHolders->replaceText( $text );
4394         }
4395
4396         /**
4397          * Renders an image gallery from a text with one line per image.
4398          * text labels may be given by using |-style alternative text. E.g.
4399          *   Image:one.jpg|The number "1"
4400          *   Image:tree.jpg|A tree
4401          * given as text will return the HTML of a gallery with two images,
4402          * labeled 'The number "1"' and
4403          * 'A tree'.
4404          */
4405         function renderImageGallery( $text, $params ) {
4406                 $ig = new ImageGallery();
4407                 $ig->setContextTitle( $this->mTitle );
4408                 $ig->setShowBytes( false );
4409                 $ig->setShowFilename( false );
4410                 $ig->setParser( $this );
4411                 $ig->setHideBadImages();
4412                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4413                 $ig->useSkin( $this->mOptions->getSkin() );
4414                 $ig->mRevisionId = $this->mRevisionId;
4415
4416                 if( isset( $params['caption'] ) ) {
4417                         $caption = $params['caption'];
4418                         $caption = htmlspecialchars( $caption );
4419                         $caption = $this->replaceInternalLinks( $caption );
4420                         $ig->setCaptionHtml( $caption );
4421                 }
4422                 if( isset( $params['perrow'] ) ) {
4423                         $ig->setPerRow( $params['perrow'] );
4424                 }
4425                 if( isset( $params['widths'] ) ) {
4426                         $ig->setWidths( $params['widths'] );
4427                 }
4428                 if( isset( $params['heights'] ) ) {
4429                         $ig->setHeights( $params['heights'] );
4430                 }
4431
4432                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4433
4434                 $lines = StringUtils::explode( "\n", $text );
4435                 foreach ( $lines as $line ) {
4436                         # match lines like these:
4437                         # Image:someimage.jpg|This is some image
4438                         $matches = array();
4439                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4440                         # Skip empty lines
4441                         if ( count( $matches ) == 0 ) {
4442                                 continue;
4443                         }
4444
4445                         if ( strpos( $matches[0], '%' ) !== false )
4446                                 $matches[1] = urldecode( $matches[1] );
4447                         $tp = Title::newFromText( $matches[1]/*, NS_FILE*/ );
4448                         $nt =& $tp;
4449                         if( is_null( $nt ) ) {
4450                                 # Bogus title. Ignore these so we don't bomb out later.
4451                                 continue;
4452                         }
4453                         if ( isset( $matches[3] ) ) {
4454                                 $label = $matches[3];
4455                         } else {
4456                                 $label = '';
4457                         }
4458
4459                         $html = $this->recursiveTagParse( trim( $label ) );
4460
4461                         $ig->add( $nt, $html );
4462
4463                         # Only add real images (bug #5586)
4464                         if ( $nt->getNamespace() == NS_FILE ) {
4465                                 $this->mOutput->addImage( $nt->getDBkey() );
4466                         }
4467                 }
4468                 return $ig->toHTML();
4469         }
4470
4471         function getImageParams( $handler ) {
4472                 if ( $handler ) {
4473                         $handlerClass = get_class( $handler );
4474                 } else {
4475                         $handlerClass = '';
4476                 }
4477                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4478                         // Initialise static lists
4479                         static $internalParamNames = array(
4480                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4481                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4482                                         'bottom', 'text-bottom' ),
4483                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4484                                         'upright', 'border', 'link', 'alt' ),
4485                         );
4486                         static $internalParamMap;
4487                         if ( !$internalParamMap ) {
4488                                 $internalParamMap = array();
4489                                 foreach ( $internalParamNames as $type => $names ) {
4490                                         foreach ( $names as $name ) {
4491                                                 $magicName = str_replace( '-', '_', "img_$name" );
4492                                                 $internalParamMap[$magicName] = array( $type, $name );
4493                                         }
4494                                 }
4495                         }
4496
4497                         // Add handler params
4498                         $paramMap = $internalParamMap;
4499                         if ( $handler ) {
4500                                 $handlerParamMap = $handler->getParamMap();
4501                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4502                                         $paramMap[$magic] = array( 'handler', $paramName );
4503                                 }
4504                         }
4505                         $this->mImageParams[$handlerClass] = $paramMap;
4506                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4507                 }
4508                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4509         }
4510
4511         /**
4512          * Parse image options text and use it to make an image
4513          * @param Title $title
4514          * @param string $options
4515          * @param LinkHolderArray $holders
4516          */
4517         function makeImage( $title, $options, $holders = false ) {
4518                 # Check if the options text is of the form "options|alt text"
4519                 # Options are:
4520                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4521                 #  * left       no resizing, just left align. label is used for alt= only
4522                 #  * right      same, but right aligned
4523                 #  * none       same, but not aligned
4524                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4525                 #  * center     center the image
4526                 #  * frame      Keep original image size, no magnify-button.
4527                 #  * framed     Same as "frame"
4528                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4529                 #  * upright    reduce width for upright images, rounded to full __0 px
4530                 #  * border     draw a 1px border around the image
4531                 #  * alt        Text for HTML alt attribute (defaults to empty)
4532                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4533                 # vertical-align values (no % or length right now):
4534                 #  * baseline
4535                 #  * sub
4536                 #  * super
4537                 #  * top
4538                 #  * text-top
4539                 #  * middle
4540                 #  * bottom
4541                 #  * text-bottom
4542
4543                 $parts = StringUtils::explode( "|", $options );
4544                 $sk = $this->mOptions->getSkin();
4545
4546                 # Give extensions a chance to select the file revision for us
4547                 $skip = $time = $descQuery = false;
4548                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4549
4550                 if ( $skip ) {
4551                         return $sk->link( $title );
4552                 }
4553
4554                 # Get the file
4555                 $imagename = $title->getDBkey();
4556                 $file = wfFindFile( $title, array( 'time' => $time ) );
4557                 # Get parameter map
4558                 $handler = $file ? $file->getHandler() : false;
4559
4560                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4561
4562                 # Process the input parameters
4563                 $caption = '';
4564                 $params = array( 'frame' => array(), 'handler' => array(),
4565                         'horizAlign' => array(), 'vertAlign' => array() );
4566                 foreach( $parts as $part ) {
4567                         $part = trim( $part );
4568                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4569                         $validated = false;
4570                         if( isset( $paramMap[$magicName] ) ) {
4571                                 list( $type, $paramName ) = $paramMap[$magicName];
4572
4573                                 // Special case; width and height come in one variable together
4574                                 if( $type === 'handler' && $paramName === 'width' ) {
4575                                         $m = array();
4576                                         # (bug 13500) In both cases (width/height and width only),
4577                                         # permit trailing "px" for backward compatibility.
4578                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4579                                                 $width = intval( $m[1] );
4580                                                 $height = intval( $m[2] );
4581                                                 if ( $handler->validateParam( 'width', $width ) ) {
4582                                                         $params[$type]['width'] = $width;
4583                                                         $validated = true;
4584                                                 }
4585                                                 if ( $handler->validateParam( 'height', $height ) ) {
4586                                                         $params[$type]['height'] = $height;
4587                                                         $validated = true;
4588                                                 }
4589                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4590                                                 $width = intval( $value );
4591                                                 if ( $handler->validateParam( 'width', $width ) ) {
4592                                                         $params[$type]['width'] = $width;
4593                                                         $validated = true;
4594                                                 }
4595                                         } // else no validation -- bug 13436
4596                                 } else {
4597                                         if ( $type === 'handler' ) {
4598                                                 # Validate handler parameter
4599                                                 $validated = $handler->validateParam( $paramName, $value );
4600                                         } else {
4601                                                 # Validate internal parameters
4602                                                 switch( $paramName ) {
4603                                                 case 'manualthumb':
4604                                                 case 'alt':
4605                                                         // @todo Fixme: possibly check validity here for
4606                                                         // manualthumb? downstream behavior seems odd with
4607                                                         // missing manual thumbs.
4608                                                         $validated = true;
4609                                                         $value = $this->stripAltText( $value, $holders );
4610                                                         break;
4611                                                 case 'link':
4612                                                         $chars = self::EXT_LINK_URL_CLASS;
4613                                                         $prots = $this->mUrlProtocols;
4614                                                         if ( $value === '' ) {
4615                                                                 $paramName = 'no-link';
4616                                                                 $value = true;
4617                                                                 $validated = true;
4618                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4619                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4620                                                                         $paramName = 'link-url';
4621                                                                         $this->mOutput->addExternalLink( $value );
4622                                                                         $validated = true;
4623                                                                 }
4624                                                         } else {
4625                                                                 $linkTitle = Title::newFromText( $value );
4626                                                                 if ( $linkTitle ) {
4627                                                                         $paramName = 'link-title';
4628                                                                         $value = $linkTitle;
4629                                                                         $this->mOutput->addLink( $linkTitle );
4630                                                                         $validated = true;
4631                                                                 }
4632                                                         }
4633                                                         break;
4634                                                 default:
4635                                                         // Most other things appear to be empty or numeric...
4636                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4637                                                 }
4638                                         }
4639
4640                                         if ( $validated ) {
4641                                                 $params[$type][$paramName] = $value;
4642                                         }
4643                                 }
4644                         }
4645                         if ( !$validated ) {
4646                                 $caption = $part;
4647                         }
4648                 }
4649
4650                 # Process alignment parameters
4651                 if ( $params['horizAlign'] ) {
4652                         $params['frame']['align'] = key( $params['horizAlign'] );
4653                 }
4654                 if ( $params['vertAlign'] ) {
4655                         $params['frame']['valign'] = key( $params['vertAlign'] );
4656                 }
4657
4658                 $params['frame']['caption'] = $caption;
4659
4660                 # Will the image be presented in a frame, with the caption below?
4661                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4662                                  isset( $params['frame']['framed'] ) ||
4663                                  isset( $params['frame']['thumbnail'] ) ||
4664                                  isset( $params['frame']['manualthumb'] );
4665
4666                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4667                 # came to also set the caption, ordinary text after the image -- which
4668                 # makes no sense, because that just repeats the text multiple times in
4669                 # screen readers.  It *also* came to set the title attribute.
4670                 #
4671                 # Now that we have an alt attribute, we should not set the alt text to
4672                 # equal the caption: that's worse than useless, it just repeats the
4673                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4674                 # use the unnamed parameter for alt text as well, just for the time be-
4675                 # ing, if the unnamed param is set and the alt param is not.
4676                 #
4677                 # For the future, we need to figure out if we want to tweak this more,
4678                 # e.g., introducing a title= parameter for the title; ignoring the un-
4679                 # named parameter entirely for images without a caption; adding an ex-
4680                 # plicit caption= parameter and preserving the old magic unnamed para-
4681                 # meter for BC; ...
4682                 if ( $imageIsFramed ) { # Framed image
4683                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4684                                 # No caption or alt text, add the filename as the alt text so
4685                                 # that screen readers at least get some description of the image
4686                                 $params['frame']['alt'] = $title->getText();
4687                         }
4688                         # Do not set $params['frame']['title'] because tooltips don't make sense
4689                         # for framed images
4690                 } else { # Inline image
4691                         if ( !isset( $params['frame']['alt'] ) ) {
4692                                 # No alt text, use the "caption" for the alt text
4693                                 if ( $caption !== '') {
4694                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4695                                 } else {
4696                                         # No caption, fall back to using the filename for the
4697                                         # alt text
4698                                         $params['frame']['alt'] = $title->getText();
4699                                 }
4700                         }
4701                         # Use the "caption" for the tooltip text
4702                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4703                 }
4704
4705                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4706
4707                 # Linker does the rest
4708                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4709
4710                 # Give the handler a chance to modify the parser object
4711                 if ( $handler ) {
4712                         $handler->parserTransformHook( $this, $file );
4713                 }
4714
4715                 return $ret;
4716         }
4717
4718         protected function stripAltText( $caption, $holders ) {
4719                 # Strip bad stuff out of the title (tooltip).  We can't just use
4720                 # replaceLinkHoldersText() here, because if this function is called
4721                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4722                 if ( $holders ) {
4723                         $tooltip = $holders->replaceText( $caption );
4724                 } else {
4725                         $tooltip = $this->replaceLinkHoldersText( $caption );
4726                 }
4727
4728                 # make sure there are no placeholders in thumbnail attributes
4729                 # that are later expanded to html- so expand them now and
4730                 # remove the tags
4731                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4732                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4733
4734                 return $tooltip;
4735         }
4736
4737         /**
4738          * Set a flag in the output object indicating that the content is dynamic and
4739          * shouldn't be cached.
4740          */
4741         function disableCache() {
4742                 wfDebug( "Parser output marked as uncacheable.\n" );
4743                 $this->mOutput->mCacheTime = -1;
4744         }
4745
4746         /**#@+
4747          * Callback from the Sanitizer for expanding items found in HTML attribute
4748          * values, so they can be safely tested and escaped.
4749          * @param string $text
4750          * @param PPFrame $frame
4751          * @return string
4752          * @private
4753          */
4754         function attributeStripCallback( &$text, $frame = false ) {
4755                 $text = $this->replaceVariables( $text, $frame );
4756                 $text = $this->mStripState->unstripBoth( $text );
4757                 return $text;
4758         }
4759
4760         /**#@-*/
4761
4762         /**#@+
4763          * Accessor/mutator
4764          */
4765         function Title( $x = null ) { return wfSetVar( $this->mTitle, $x ); }
4766         function Options( $x = null ) { return wfSetVar( $this->mOptions, $x ); }
4767         function OutputType( $x = null ) { return wfSetVar( $this->mOutputType, $x ); }
4768         /**#@-*/
4769
4770         /**#@+
4771          * Accessor
4772          */
4773         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4774         /**#@-*/
4775
4776
4777         /**
4778          * Break wikitext input into sections, and either pull or replace
4779          * some particular section's text.
4780          *
4781          * External callers should use the getSection and replaceSection methods.
4782          *
4783          * @param string $text Page wikitext
4784          * @param string $section A section identifier string of the form:
4785          *   <flag1> - <flag2> - ... - <section number>
4786          *
4787          * Currently the only recognised flag is "T", which means the target section number
4788          * was derived during a template inclusion parse, in other words this is a template
4789          * section edit link. If no flags are given, it was an ordinary section edit link.
4790          * This flag is required to avoid a section numbering mismatch when a section is
4791          * enclosed by <includeonly> (bug 6563).
4792          *
4793          * The section number 0 pulls the text before the first heading; other numbers will
4794          * pull the given section along with its lower-level subsections. If the section is
4795          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4796          *
4797          * @param string $mode One of "get" or "replace"
4798          * @param string $newText Replacement text for section data.
4799          * @return string for "get", the extracted section text.
4800          *                for "replace", the whole page with the section replaced.
4801          */
4802         private function extractSections( $text, $section, $mode, $newText='' ) {
4803                 global $wgTitle;
4804                 $this->clearState();
4805                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4806                 $this->mOptions = new ParserOptions;
4807                 $this->setOutputType( self::OT_WIKI );
4808                 $outText = '';
4809                 $frame = $this->getPreprocessor()->newFrame();
4810
4811                 // Process section extraction flags
4812                 $flags = 0;
4813                 $sectionParts = explode( '-', $section );
4814                 $sectionIndex = array_pop( $sectionParts );
4815                 foreach ( $sectionParts as $part ) {
4816                         if ( $part === 'T' ) {
4817                                 $flags |= self::PTD_FOR_INCLUSION;
4818                         }
4819                 }
4820                 // Preprocess the text
4821                 $root = $this->preprocessToDom( $text, $flags );
4822
4823                 // <h> nodes indicate section breaks
4824                 // They can only occur at the top level, so we can find them by iterating the root's children
4825                 $node = $root->getFirstChild();
4826
4827                 // Find the target section
4828                 if ( $sectionIndex == 0 ) {
4829                         // Section zero doesn't nest, level=big
4830                         $targetLevel = 1000;
4831                 } else {
4832             while ( $node ) {
4833                 if ( $node->getName() === 'h' ) {
4834                     $bits = $node->splitHeading();
4835                                         if ( $bits['i'] == $sectionIndex ) {
4836                                         $targetLevel = $bits['level'];
4837                                                 break;
4838                                         }
4839                                 }
4840                                 if ( $mode === 'replace' ) {
4841                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4842                                 }
4843                                 $node = $node->getNextSibling();
4844                         }
4845                 }
4846
4847                 if ( !$node ) {
4848                         // Not found
4849                         if ( $mode === 'get' ) {
4850                                 return $newText;
4851                         } else {
4852                                 return $text;
4853                         }
4854                 }
4855
4856                 // Find the end of the section, including nested sections
4857                 do {
4858                         if ( $node->getName() === 'h' ) {
4859                                 $bits = $node->splitHeading();
4860                                 $curLevel = $bits['level'];
4861                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4862                                         break;
4863                                 }
4864                         }
4865                         if ( $mode === 'get' ) {
4866                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4867                         }
4868                         $node = $node->getNextSibling();
4869                 } while ( $node );
4870
4871                 // Write out the remainder (in replace mode only)
4872                 if ( $mode === 'replace' ) {
4873                         // Output the replacement text
4874                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4875                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4876                         // Only add trailing whitespace if there is newText
4877                         if($newText != "") {
4878                                 $outText .= $newText . "\n\n";
4879                         }
4880
4881                         while ( $node ) {
4882                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4883                                 $node = $node->getNextSibling();
4884                         }
4885                 }
4886
4887                 if ( is_string( $outText ) ) {
4888                         // Re-insert stripped tags
4889                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4890                 }
4891
4892                 return $outText;
4893         }
4894
4895         /**
4896          * This function returns the text of a section, specified by a number ($section).
4897          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4898          * the first section before any such heading (section 0).
4899          *
4900          * If a section contains subsections, these are also returned.
4901          *
4902          * @param string $text text to look in
4903          * @param string $section section identifier
4904          * @param string $deftext default to return if section is not found
4905          * @return string text of the requested section
4906          */
4907         public function getSection( $text, $section, $deftext='' ) {
4908                 return $this->extractSections( $text, $section, "get", $deftext );
4909         }
4910
4911         public function replaceSection( $oldtext, $section, $text ) {
4912                 return $this->extractSections( $oldtext, $section, "replace", $text );
4913         }
4914
4915         /**
4916          * Get the timestamp associated with the current revision, adjusted for
4917          * the default server-local timestamp
4918          */
4919         function getRevisionTimestamp() {
4920                 if ( is_null( $this->mRevisionTimestamp ) ) {
4921                         wfProfileIn( __METHOD__ );
4922                         global $wgContLang;
4923                         $dbr = wfGetDB( DB_SLAVE );
4924                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4925                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4926
4927                         // Normalize timestamp to internal MW format for timezone processing.
4928                         // This has the added side-effect of replacing a null value with
4929                         // the current time, which gives us more sensible behavior for
4930                         // previews.
4931                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4932
4933                         // The cryptic '' timezone parameter tells to use the site-default
4934                         // timezone offset instead of the user settings.
4935                         //
4936                         // Since this value will be saved into the parser cache, served
4937                         // to other users, and potentially even used inside links and such,
4938                         // it needs to be consistent for all visitors.
4939                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4940
4941                         wfProfileOut( __METHOD__ );
4942                 }
4943                 return $this->mRevisionTimestamp;
4944         }
4945
4946         /**
4947          * Get the name of the user that edited the last revision
4948          */
4949         function getRevisionUser() {
4950                 // if this template is subst: the revision id will be blank,
4951                 // so just use the current user's name
4952                 if( $this->mRevisionId ) {
4953                         $revision = Revision::newFromId( $this->mRevisionId );
4954                         $revuser = $revision->getUserText();
4955                 } else {
4956                         global $wgUser;
4957                         $revuser = $wgUser->getName();
4958                 }
4959                 return $revuser;
4960         }
4961
4962         /**
4963          * Mutator for $mDefaultSort
4964          *
4965          * @param $sort New value
4966          */
4967         public function setDefaultSort( $sort ) {
4968                 $this->mDefaultSort = $sort;
4969         }
4970
4971         /**
4972          * Accessor for $mDefaultSort
4973          * Will use the title/prefixed title if none is set
4974          *
4975          * @return string
4976          */
4977         public function getDefaultSort() {
4978                 global $wgCategoryPrefixedDefaultSortkey;
4979                 if( $this->mDefaultSort !== false ) {
4980                         return $this->mDefaultSort;
4981                 } elseif ($this->mTitle->getNamespace() == NS_CATEGORY ||
4982                         !$wgCategoryPrefixedDefaultSortkey) {
4983                         return $this->mTitle->getText();
4984                 } else {
4985                         return $this->mTitle->getPrefixedText();
4986                 }
4987         }
4988
4989         /**
4990          * Accessor for $mDefaultSort
4991          * Unlike getDefaultSort(), will return false if none is set
4992          *
4993          * @return string or false
4994          */
4995         public function getCustomDefaultSort() {
4996                 return $this->mDefaultSort;
4997         }
4998
4999         /**
5000          * Try to guess the section anchor name based on a wikitext fragment
5001          * presumably extracted from a heading, for example "Header" from
5002          * "== Header ==".
5003          */
5004         public function guessSectionNameFromWikiText( $text ) {
5005                 # Strip out wikitext links(they break the anchor)
5006                 $text = $this->stripSectionName( $text );
5007                 $headline = Sanitizer::decodeCharReferences( $text );
5008                 # strip out HTML
5009                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
5010                 $headline = trim( $headline );
5011                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
5012                 $replacearray = array(
5013                         '%3A' => ':',
5014                         '%' => '.'
5015                 );
5016                 return str_replace(
5017                         array_keys( $replacearray ),
5018                         array_values( $replacearray ),
5019                         $sectionanchor );
5020         }
5021
5022         /**
5023          * Strips a text string of wikitext for use in a section anchor
5024          *
5025          * Accepts a text string and then removes all wikitext from the
5026          * string and leaves only the resultant text (i.e. the result of
5027          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5028          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5029          * to create valid section anchors by mimicing the output of the
5030          * parser when headings are parsed.
5031          *
5032          * @param $text string Text string to be stripped of wikitext
5033          * for use in a Section anchor
5034          * @return Filtered text string
5035          */
5036         public function stripSectionName( $text ) {
5037                 # Strip internal link markup
5038                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
5039                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
5040
5041                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5042                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5043                 # on how many empty links there are on the page - need to figure that out.
5044                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
5045
5046                 # Parse wikitext quotes (italics & bold)
5047                 $text = $this->doQuotes($text);
5048
5049                 # Strip HTML tags
5050                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5051                 return $text;
5052         }
5053
5054         function srvus( $text ) {
5055                 return $this->testSrvus( $text, $this->mOutputType );
5056         }
5057
5058         /**
5059          * strip/replaceVariables/unstrip for preprocessor regression testing
5060          */
5061         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
5062                 $this->clearState();
5063                 if ( ! ( $title instanceof Title ) ) {
5064                         $title = Title::newFromText( $title );
5065                 }
5066                 $this->mTitle = $title;
5067                 $this->mOptions = $options;
5068                 $this->setOutputType( $outputType );
5069                 $text = $this->replaceVariables( $text );
5070                 $text = $this->mStripState->unstripBoth( $text );
5071                 $text = Sanitizer::removeHTMLtags( $text );
5072                 return $text;
5073         }
5074
5075         function testPst( $text, $title, $options ) {
5076                 global $wgUser;
5077                 if ( ! ( $title instanceof Title ) ) {
5078                         $title = Title::newFromText( $title );
5079                 }
5080                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5081         }
5082
5083         function testPreprocess( $text, $title, $options ) {
5084                 if ( ! ( $title instanceof Title ) ) {
5085                         $title = Title::newFromText( $title );
5086                 }
5087                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5088         }
5089
5090         function markerSkipCallback( $s, $callback ) {
5091                 $i = 0;
5092                 $out = '';
5093                 while ( $i < strlen( $s ) ) {
5094                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5095                         if ( $markerStart === false ) {
5096                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5097                                 break;
5098                         } else {
5099                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5100                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5101                                 if ( $markerEnd === false ) {
5102                                         $out .= substr( $s, $markerStart );
5103                                         break;
5104                                 } else {
5105                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5106                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5107                                         $i = $markerEnd;
5108                                 }
5109                         }
5110                 }
5111                 return $out;
5112         }
5113
5114         function serialiseHalfParsedText( $text ) {
5115                 $data = array();
5116                 $data['text'] = $text;
5117
5118                 // First, find all strip markers, and store their
5119                 //  data in an array.
5120                 $stripState = new StripState;
5121                 $pos = 0;
5122                 while( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) ) && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) ) {
5123                         $end_pos += strlen( self::MARKER_SUFFIX );
5124                         $marker = substr( $text, $start_pos, $end_pos-$start_pos );
5125
5126                         if ( !empty( $this->mStripState->general->data[$marker] ) ) {
5127                                 $replaceArray = $stripState->general;
5128                                 $stripText = $this->mStripState->general->data[$marker];
5129                         } elseif ( !empty( $this->mStripState->nowiki->data[$marker] ) ) {
5130                                 $replaceArray = $stripState->nowiki;
5131                                 $stripText = $this->mStripState->nowiki->data[$marker];
5132                         } else {
5133                                 throw new MWException( "Hanging strip marker: '$marker'." );
5134                         }
5135
5136                         $replaceArray->setPair( $marker, $stripText );
5137                         $pos = $end_pos;
5138                 }
5139                 $data['stripstate'] = $stripState;
5140
5141                 // Now, find all of our links, and store THEIR
5142                 //  data in an array! :)
5143                 $links = array( 'internal' => array(), 'interwiki' => array() );
5144                 $pos = 0;
5145
5146                 // Internal links
5147                 while( ( $start_pos = strpos( $text, '<!--LINK ', $pos ) ) ) {
5148                         list( $ns, $trail ) = explode( ':', substr( $text, $start_pos + strlen( '<!--LINK ' ) ), 2 );
5149
5150                         $ns = trim($ns);
5151                         if (empty( $links['internal'][$ns] )) {
5152                                 $links['internal'][$ns] = array();
5153                         }
5154
5155                         $key = trim( substr( $trail, 0, strpos( $trail, '-->' ) ) );
5156                         $links['internal'][$ns][] = $this->mLinkHolders->internals[$ns][$key];
5157                         $pos = $start_pos + strlen( "<!--LINK $ns:$key-->" );
5158                 }
5159
5160                 $pos = 0;
5161
5162                 // Interwiki links
5163                 while( ( $start_pos = strpos( $text, '<!--IWLINK ', $pos ) ) ) {
5164                         $data = substr( $text, $start_pos );
5165                         $key = trim( substr( $data, 0, strpos( $data, '-->' ) ) );
5166                         $links['interwiki'][] = $this->mLinkHolders->interwiki[$key];
5167                         $pos = $start_pos + strlen( "<!--IWLINK $key-->" );
5168                 }
5169
5170                 $data['linkholder'] = $links;
5171
5172                 return $data;
5173         }
5174
5175         function unserialiseHalfParsedText( $data, $intPrefix = null /* Unique identifying prefix */ ) {
5176                 if (!$intPrefix)
5177                         $intPrefix = $this->getRandomString();
5178
5179                 // First, extract the strip state.
5180                 $stripState = $data['stripstate'];
5181                 $this->mStripState->general->merge( $stripState->general );
5182                 $this->mStripState->nowiki->merge( $stripState->nowiki );
5183
5184                 // Now, extract the text, and renumber links
5185                 $text = $data['text'];
5186                 $links = $data['linkholder'];
5187
5188                 // Internal...
5189                 foreach( $links['internal'] as $ns => $nsLinks ) {
5190                         foreach( $nsLinks as $key => $entry ) {
5191                                 $newKey = $intPrefix . '-' . $key;
5192                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
5193
5194                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
5195                         }
5196                 }
5197
5198                 // Interwiki...
5199                 foreach( $links['interwiki'] as $key => $entry ) {
5200                         $newKey = "$intPrefix-$key";
5201                         $this->mLinkHolders->interwikis[$newKey] = $entry;
5202
5203                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
5204                 }
5205
5206                 // Should be good to go.
5207                 return $text;
5208         }
5209 }
5210
5211 /**
5212  * @todo document, briefly.
5213  * @ingroup Parser
5214  */
5215 class StripState {
5216         var $general, $nowiki;
5217
5218         function __construct() {
5219                 $this->general = new ReplacementArray;
5220                 $this->nowiki = new ReplacementArray;
5221         }
5222
5223         function unstripGeneral( $text ) {
5224                 wfProfileIn( __METHOD__ );
5225                 do {
5226                         $oldText = $text;
5227                         $text = $this->general->replace( $text );
5228                 } while ( $text !== $oldText );
5229                 wfProfileOut( __METHOD__ );
5230                 return $text;
5231         }
5232
5233         function unstripNoWiki( $text ) {
5234                 wfProfileIn( __METHOD__ );
5235                 do {
5236                         $oldText = $text;
5237                         $text = $this->nowiki->replace( $text );
5238                 } while ( $text !== $oldText );
5239                 wfProfileOut( __METHOD__ );
5240                 return $text;
5241         }
5242
5243         function unstripBoth( $text ) {
5244                 wfProfileIn( __METHOD__ );
5245                 do {
5246                         $oldText = $text;
5247                         $text = $this->general->replace( $text );
5248                         $text = $this->nowiki->replace( $text );
5249                 } while ( $text !== $oldText );
5250                 wfProfileOut( __METHOD__ );
5251                 return $text;
5252         }
5253 }
5254
5255 /**
5256  * @todo document, briefly.
5257  * @ingroup Parser
5258  */
5259 class OnlyIncludeReplacer {
5260         var $output = '';
5261
5262         function replace( $matches ) {
5263                 if ( substr( $matches[1], -1 ) === "\n" ) {
5264                         $this->output .= substr( $matches[1], 0, -1 );
5265                 } else {
5266                         $this->output .= $matches[1];
5267                 }
5268         }
5269 }