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                 $mSubstWords, $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                 }
 369
 370                 $text = $this->mStripState->unstripNoWiki( $text );
 371
 372                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 373
 374                 if ( $this->mTransparentTagHooks ) {
 375                         //!JF Move to its own function
 376                         $uniq_prefix = $this->mUniqPrefix;
 377                         $matches = array();
 378                         $elements = array_keys( $this->mTransparentTagHooks );
 379                         $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 380
 381                         foreach( $matches as $marker => $data ) {
 382                                 list( $element, $content, $params, $tag ) = $data;
 383                                 $tagName = strtolower( $element );
 384                                 if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 385                                         $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 386                                                 array( $content, $params, $this ) );
 387                                 } else {
 388                                         $output = $tag;
 389                                 }
 390                                 $this->mStripState->general->setPair( $marker, $output );
 391                         }
 392                 }
 393
 394                 # This was originally inserted for transparent tag hooks (now deprecated)
 395                 # but some extensions (notably <poem>) rely on the extra unstripGeneral()
 396                 # after unstripNoWiki() so they can modify the contents of <nowiki> tags.
 397                 $text = $this->mStripState->unstripGeneral( $text );
 398
 399                 $text = Sanitizer::normalizeCharReferences( $text );
 400
 401                 if ( ( $wgUseTidy && $this->mOptions->mTidy ) || $wgAlwaysUseTidy ) {
 402                         $text = MWTidy::tidy( $text );
 403                 } else {
 404                         # attempt to sanitize at least some nesting problems
 405                         # (bug #2702 and quite a few others)
 406                         $tidyregs = array(
 407                                 # ''Something [http://www.cool.com cool''] -->
 408                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 409                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 410                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 411                                 # fix up an anchor inside another anchor, only
 412                                 # at least for a single single nested link (bug 3695)
 413                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 414                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 415                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 416                                 # contains a div, so fix it up here; replace
 417                                 # div with escaped text
 418                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 419                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 420                                 # remove empty italic or bold tag pairs, some
 421                                 # introduced by rules above
 422                                 '/<([bi])><\/\\1>/' => '',
 423                         );
 424
 425                         $text = preg_replace(
 426                                 array_keys( $tidyregs ),
 427                                 array_values( $tidyregs ),
 428                                 $text );
 429                 }
 430                 global $wgExpensiveParserFunctionLimit;
 431                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 432                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 433                 }
 434
 435                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 436
 437                 # Information on include size limits, for the benefit of users who try to skirt them
 438                 if ( $this->mOptions->getEnableLimitReport() ) {
 439                         global $wgExpensiveParserFunctionLimit;
 440                         $max = $this->mOptions->getMaxIncludeSize();
 441                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 442                         $limitReport =
 443                                 "NewPP limit report\n" .
 444                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 445                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 446                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 447                                 $PFreport;
 448                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 449                         $text .= "\n<!-- \n$limitReport-->\n";
 450                 }
 451                 $this->mOutput->setText( $text );
 452
 453                 $this->mRevisionId = $oldRevisionId;
 454                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 455                 wfProfileOut( $fname );
 456                 wfProfileOut( __METHOD__ );
 457
 458                 return $this->mOutput;
 459         }
 460
 461         /**
 462          * Recursive parser entry point that can be called from an extension tag
 463          * hook.
 464          *
 465          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 466          *
 467          * @param $text String: text extension wants to have parsed
 468          * @param PPFrame $frame: The frame to use for expanding any template variables
 469          */
 470         function recursiveTagParse( $text, $frame=false ) {
 471                 wfProfileIn( __METHOD__ );
 472                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 473                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 474                 $text = $this->internalParse( $text, false, $frame );
 475                 wfProfileOut( __METHOD__ );
 476                 return $text;
 477         }
 478
 479         /**
 480          * Expand templates and variables in the text, producing valid, static wikitext.
 481          * Also removes comments.
 482          */
 483         function preprocess( $text, $title, $options, $revid = null ) {
 484                 wfProfileIn( __METHOD__ );
 485                 $this->clearState();
 486                 $this->setOutputType( self::OT_PREPROCESS );
 487                 $this->mOptions = $options;
 488                 $this->setTitle( $title );
 489                 if( $revid !== null ) {
 490                         $this->mRevisionId = $revid;
 491                 }
 492                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 493                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 494                 $text = $this->replaceVariables( $text );
 495                 $text = $this->mStripState->unstripBoth( $text );
 496                 wfProfileOut( __METHOD__ );
 497                 return $text;
 498         }
 499
 500         /**
 501          * Get the wikitext of a page as though it was transcluded.
 502          *
 503          * Specifically <includeonly> etc. are parsed, redirects are followed, comments
 504          * are removed, but templates arguments and parser functions are untouched.
 505          *
 506          * This is not called by the parser itself, see braceSubstitution for its transclusion.
 507          */
 508         public function getTransclusionText( $title, $options ) {
 509                 // Must initialize first
 510                 $this->clearState();
 511                 $this->setOutputType( self::OT_INCLUDES );
 512                 $this->mOptions = $options;
 513                 $this->setTitle( new FakeTitle );
 514
 515                 list( $text, $title ) = $this->getTemplateDom( $title );
 516                 $flags = PPFrame::NO_ARGS | PPFrame::NO_TEMPLATES;
 517                 return $this->getPreprocessor()->newFrame()->expand( $text, $flags );
 518         }
 519
 520         /**
 521          * Get a random string
 522          *
 523          * @private
 524          * @static
 525          */
 526         function getRandomString() {
 527                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 528         }
 529
 530         function &getTitle() { return $this->mTitle; }
 531         function getOptions() { return $this->mOptions; }
 532         function getRevisionId() { return $this->mRevisionId; }
 533         function getOutput() { return $this->mOutput; }
 534         function nextLinkID() { return $this->mLinkID++; }
 535
 536         function getFunctionLang() {
 537                 global $wgLang, $wgContLang;
 538
 539                 $target = $this->mOptions->getTargetLanguage();
 540                 if ( $target !== null ) {
 541                         return $target;
 542                 } else {
 543                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 544                 }
 545         }
 546
 547         /**
 548          * Get a preprocessor object
 549          */
 550         function getPreprocessor() {
 551                 if ( !isset( $this->mPreprocessor ) ) {
 552                         $class = $this->mPreprocessorClass;
 553                         $this->mPreprocessor = new $class( $this );
 554                 }
 555                 return $this->mPreprocessor;
 556         }
 557
 558         /**
 559          * Replaces all occurrences of HTML-style comments and the given tags
 560          * in the text with a random marker and returns the next text. The output
 561          * parameter $matches will be an associative array filled with data in
 562          * the form:
 563          *   'UNIQ-xxxxx' => array(
 564          *     'element',
 565          *     'tag content',
 566          *     array( 'param' => 'x' ),
 567          *     '<element param="x">tag content</element>' ) )
 568          *
 569          * @param $elements list of element names. Comments are always extracted.
 570          * @param $text Source text string.
 571          * @param $uniq_prefix
 572          *
 573          * @public
 574          * @static
 575          */
 576         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 577                 static $n = 1;
 578                 $stripped = '';
 579                 $matches = array();
 580
 581                 $taglist = implode( '|', $elements );
 582                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 583
 584                 while ( $text != '' ) {
 585                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 586                         $stripped .= $p[0];
 587                         if( count( $p ) < 5 ) {
 588                                 break;
 589                         }
 590                         if( count( $p ) > 5 ) {
 591                                 // comment
 592                                 $element    = $p[4];
 593                                 $attributes = '';
 594                                 $close      = '';
 595                                 $inside     = $p[5];
 596                         } else {
 597                                 // tag
 598                                 $element    = $p[1];
 599                                 $attributes = $p[2];
 600                                 $close      = $p[3];
 601                                 $inside     = $p[4];
 602                         }
 603
 604                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . self::MARKER_SUFFIX;
 605                         $stripped .= $marker;
 606
 607                         if ( $close === '/>' ) {
 608                                 // Empty element tag, <tag />
 609                                 $content = null;
 610                                 $text = $inside;
 611                                 $tail = null;
 612                         } else {
 613                                 if( $element === '!--' ) {
 614                                         $end = '/(-->)/';
 615                                 } else {
 616                                         $end = "/(<\\/$element\\s*>)/i";
 617                                 }
 618                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 619                                 $content = $q[0];
 620                                 if( count( $q ) < 3 ) {
 621                                         # No end tag -- let it run out to the end of the text.
 622                                         $tail = '';
 623                                         $text = '';
 624                                 } else {
 625                                         $tail = $q[1];
 626                                         $text = $q[2];
 627                                 }
 628                         }
 629
 630                         $matches[$marker] = array( $element,
 631                                 $content,
 632                                 Sanitizer::decodeTagAttributes( $attributes ),
 633                                 "<$element$attributes$close$content$tail" );
 634                 }
 635                 return $stripped;
 636         }
 637
 638         /**
 639          * Get a list of strippable XML-like elements
 640          */
 641         function getStripList() {
 642                 return $this->mStripList;
 643         }
 644
 645         /**
 646          * @deprecated use replaceVariables
 647          */
 648         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 649                 return $text;
 650         }
 651
 652         /**
 653          * Restores pre, math, and other extensions removed by strip()
 654          *
 655          * always call unstripNoWiki() after this one
 656          * @private
 657          * @deprecated use $this->mStripState->unstrip()
 658          */
 659         function unstrip( $text, $state ) {
 660                 return $state->unstripGeneral( $text );
 661         }
 662
 663         /**
 664          * Always call this after unstrip() to preserve the order
 665          *
 666          * @private
 667          * @deprecated use $this->mStripState->unstrip()
 668          */
 669         function unstripNoWiki( $text, $state ) {
 670                 return $state->unstripNoWiki( $text );
 671         }
 672
 673         /**
 674          * @deprecated use $this->mStripState->unstripBoth()
 675          */
 676         function unstripForHTML( $text ) {
 677                 return $this->mStripState->unstripBoth( $text );
 678         }
 679
 680         /**
 681          * Add an item to the strip state
 682          * Returns the unique tag which must be inserted into the stripped text
 683          * The tag will be replaced with the original text in unstrip()
 684          *
 685          * @private
 686          */
 687         function insertStripItem( $text ) {
 688                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 689                 $this->mMarkerIndex++;
 690                 $this->mStripState->general->setPair( $rnd, $text );
 691                 return $rnd;
 692         }
 693
 694         /**
 695          * Interface with html tidy
 696          * @deprecated Use MWTidy::tidy()
 697          */
 698         public static function tidy( $text ) {
 699                 wfDeprecated( __METHOD__ );
 700                 return MWTidy::tidy( $text );
 701         }
 702
 703         /**
 704          * parse the wiki syntax used to render tables
 705          *
 706          * @private
 707          */
 708         function doTableStuff ( $text ) {
 709                 wfProfileIn( __METHOD__ );
 710
 711                 $lines = StringUtils::explode( "\n", $text );
 712                 $out = '';
 713                 $td_history = array (); // Is currently a td tag open?
 714                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 715                 $tr_history = array (); // Is currently a tr tag open?
 716                 $tr_attributes = array (); // history of tr attributes
 717                 $has_opened_tr = array(); // Did this table open a <tr> element?
 718                 $indent_level = 0; // indent level of the table
 719
 720                 foreach ( $lines as $outLine ) {
 721                         $line = trim( $outLine );
 722
 723                         if( $line == '' ) { // empty line, go to next line
 724                                 $out .= $outLine."\n";
 725                                 continue;
 726                         }
 727                         $first_character = $line[0];
 728                         $matches = array();
 729
 730                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 731                                 // First check if we are starting a new table
 732                                 $indent_level = strlen( $matches[1] );
 733
 734                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 735                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 736
 737                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 738                                 array_push( $td_history , false );
 739                                 array_push( $last_tag_history , '' );
 740                                 array_push( $tr_history , false );
 741                                 array_push( $tr_attributes , '' );
 742                                 array_push( $has_opened_tr , false );
 743                         } else if ( count ( $td_history ) == 0 ) {
 744                                 // Don't do any of the following
 745                                 $out .= $outLine."\n";
 746                                 continue;
 747                         } else if ( substr ( $line , 0 , 2 ) === '|}' ) {
 748                                 // We are ending a table
 749                                 $line = '</table>' . substr ( $line , 2 );
 750                                 $last_tag = array_pop ( $last_tag_history );
 751
 752                                 if ( !array_pop ( $has_opened_tr ) ) {
 753                                         $line = "<tr><td></td></tr>{$line}";
 754                                 }
 755
 756                                 if ( array_pop ( $tr_history ) ) {
 757                                         $line = "</tr>{$line}";
 758                                 }
 759
 760                                 if ( array_pop ( $td_history ) ) {
 761                                         $line = "</{$last_tag}>{$line}";
 762                                 }
 763                                 array_pop ( $tr_attributes );
 764                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 765                         } else if ( substr ( $line , 0 , 2 ) === '|-' ) {
 766                                 // Now we have a table row
 767                                 $line = preg_replace( '#^\|-+#', '', $line );
 768
 769                                 // Whats after the tag is now only attributes
 770                                 $attributes = $this->mStripState->unstripBoth( $line );
 771                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 772                                 array_pop( $tr_attributes );
 773                                 array_push( $tr_attributes, $attributes );
 774
 775                                 $line = '';
 776                                 $last_tag = array_pop ( $last_tag_history );
 777                                 array_pop ( $has_opened_tr );
 778                                 array_push ( $has_opened_tr , true );
 779
 780                                 if ( array_pop ( $tr_history ) ) {
 781                                         $line = '</tr>';
 782                                 }
 783
 784                                 if ( array_pop ( $td_history ) ) {
 785                                         $line = "</{$last_tag}>{$line}";
 786                                 }
 787
 788                                 $outLine = $line;
 789                                 array_push ( $tr_history , false );
 790                                 array_push ( $td_history , false );
 791                                 array_push ( $last_tag_history , '' );
 792                         }
 793                         else if ( $first_character === '|' || $first_character === '!' || substr ( $line , 0 , 2 )  === '|+' ) {
 794                                 // This might be cell elements, td, th or captions
 795                                 if ( substr ( $line , 0 , 2 ) === '|+' ) {
 796                                         $first_character = '+';
 797                                         $line = substr ( $line , 1 );
 798                                 }
 799
 800                                 $line = substr ( $line , 1 );
 801
 802                                 if ( $first_character === '!' ) {
 803                                         $line = str_replace ( '!!' , '||' , $line );
 804                                 }
 805
 806                                 // Split up multiple cells on the same line.
 807                                 // FIXME : This can result in improper nesting of tags processed
 808                                 // by earlier parser steps, but should avoid splitting up eg
 809                                 // attribute values containing literal "||".
 810                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 811
 812                                 $outLine = '';
 813
 814                                 // Loop through each table cell
 815                                 foreach ( $cells as $cell )
 816                                 {
 817                                         $previous = '';
 818                                         if ( $first_character !== '+' )
 819                                         {
 820                                                 $tr_after = array_pop ( $tr_attributes );
 821                                                 if ( !array_pop ( $tr_history ) ) {
 822                                                         $previous = "<tr{$tr_after}>\n";
 823                                                 }
 824                                                 array_push ( $tr_history , true );
 825                                                 array_push ( $tr_attributes , '' );
 826                                                 array_pop ( $has_opened_tr );
 827                                                 array_push ( $has_opened_tr , true );
 828                                         }
 829
 830                                         $last_tag = array_pop ( $last_tag_history );
 831
 832                                         if ( array_pop ( $td_history ) ) {
 833                                                 $previous = "</{$last_tag}>{$previous}";
 834                                         }
 835
 836                                         if ( $first_character === '|' ) {
 837                                                 $last_tag = 'td';
 838                                         } else if ( $first_character === '!' ) {
 839                                                 $last_tag = 'th';
 840                                         } else if ( $first_character === '+' ) {
 841                                                 $last_tag = 'caption';
 842                                         } else {
 843                                                 $last_tag = '';
 844                                         }
 845
 846                                         array_push ( $last_tag_history , $last_tag );
 847
 848                                         // A cell could contain both parameters and data
 849                                         $cell_data = explode ( '|' , $cell , 2 );
 850
 851                                         // Bug 553: Note that a '|' inside an invalid link should not
 852                                         // be mistaken as delimiting cell parameters
 853                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 854                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 855                                         } else if ( count ( $cell_data ) == 1 )
 856                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 857                                         else {
 858                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 859                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 860                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 861                                         }
 862
 863                                         $outLine .= $cell;
 864                                         array_push ( $td_history , true );
 865                                 }
 866                         }
 867                         $out .= $outLine . "\n";
 868                 }
 869
 870                 // Closing open td, tr && table
 871                 while ( count ( $td_history ) > 0 )
 872                 {
 873                         if ( array_pop ( $td_history ) ) {
 874                                 $out .= "</td>\n";
 875                         }
 876                         if ( array_pop ( $tr_history ) ) {
 877                                 $out .= "</tr>\n";
 878                         }
 879                         if ( !array_pop ( $has_opened_tr ) ) {
 880                                 $out .= "<tr><td></td></tr>\n" ;
 881                         }
 882
 883                         $out .= "</table>\n";
 884                 }
 885
 886                 // Remove trailing line-ending (b/c)
 887                 if ( substr( $out, -1 ) === "\n" ) {
 888                         $out = substr( $out, 0, -1 );
 889                 }
 890
 891                 // special case: don't return empty table
 892                 if( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 893                         $out = '';
 894                 }
 895
 896                 wfProfileOut( __METHOD__ );
 897
 898                 return $out;
 899         }
 900
 901         /**
 902          * Helper function for parse() that transforms wiki markup into
 903          * HTML. Only called for $mOutputType == self::OT_HTML.
 904          *
 905          * @private
 906          */
 907         function internalParse( $text, $isMain = true, $frame=false ) {
 908                 wfProfileIn( __METHOD__ );
 909
 910                 $origText = $text;
 911
 912                 # Hook to suspend the parser in this state
 913                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 914                         wfProfileOut( __METHOD__ );
 915                         return $text ;
 916                 }
 917
 918                 // if $frame is provided, then use $frame for replacing any variables
 919                 if ($frame) {
 920                         // use frame depth to infer how include/noinclude tags should be handled
 921                         // depth=0 means this is the top-level document; otherwise it's an included document
 922                         if( !$frame->depth )
 923                                 $flag = 0;
 924                         else
 925                                 $flag = Parser::PTD_FOR_INCLUSION;
 926                         $dom = $this->preprocessToDom( $text, $flag );
 927                         $text = $frame->expand( $dom );
 928                 }
 929                 // if $frame is not provided, then use old-style replaceVariables
 930                 else {
 931                         $text = $this->replaceVariables( $text );
 932                 }
 933
 934                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 935                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 936
 937                 // Tables need to come after variable replacement for things to work
 938                 // properly; putting them before other transformations should keep
 939                 // exciting things like link expansions from showing up in surprising
 940                 // places.
 941                 $text = $this->doTableStuff( $text );
 942
 943                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 944
 945                 $text = $this->doDoubleUnderscore( $text );
 946
 947                 $text = $this->doHeadings( $text );
 948                 if( $this->mOptions->getUseDynamicDates() ) {
 949                         $df = DateFormatter::getInstance();
 950                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 951                 }
 952                 $text = $this->doAllQuotes( $text );
 953                 $text = $this->replaceInternalLinks( $text );
 954                 $text = $this->replaceExternalLinks( $text );
 955
 956                 # replaceInternalLinks may sometimes leave behind
 957                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 958                 $text = str_replace($this->mUniqPrefix.'NOPARSE', '', $text);
 959
 960                 $text = $this->doMagicLinks( $text );
 961                 $text = $this->formatHeadings( $text, $origText, $isMain );
 962
 963                 wfProfileOut( __METHOD__ );
 964                 return $text;
 965         }
 966
 967         /**
 968          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 969          * magic external links.
 970          *
 971          * DML
 972          * @private
 973          */
 974         function doMagicLinks( $text ) {
 975                 wfProfileIn( __METHOD__ );
 976                 $prots = $this->mUrlProtocols;
 977                 $urlChar = self::EXT_LINK_URL_CLASS;
 978                 $text = preg_replace_callback(
 979                         '!(?:                           # Start cases
 980                                 (<a.*?</a>) |               # m[1]: Skip link text
 981                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
 982                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
 983                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
 984                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
 985                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 986                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 987                                     [0-9Xx]                 # check digit
 988                                     \b)
 989                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 990                 wfProfileOut( __METHOD__ );
 991                 return $text;
 992         }
 993
 994         function magicLinkCallback( $m ) {
 995                 if ( isset( $m[1] ) && $m[1] !== '' ) {
 996                         # Skip anchor
 997                         return $m[0];
 998                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
 999                         # Skip HTML element
1000                         return $m[0];
1001                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
1002                         # Free external link
1003                         return $this->makeFreeExternalLink( $m[0] );
1004                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
1005                         # RFC or PMID
1006                         $CssClass = '';
1007                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1008                                 $keyword = 'RFC';
1009                                 $urlmsg = 'rfcurl';
1010                                 $CssClass = 'mw-magiclink-rfc';
1011                                 $id = $m[4];
1012                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1013                                 $keyword = 'PMID';
1014                                 $urlmsg = 'pubmedurl';
1015                                 $CssClass = 'mw-magiclink-pmid';
1016                                 $id = $m[4];
1017                         } else {
1018                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1019                                         substr($m[0], 0, 20 ) . '"' );
1020                         }
1021                         $url = wfMsg( $urlmsg, $id);
1022                         $sk = $this->mOptions->getSkin();
1023                         $la = $sk->getExternalLinkAttributes( "external $CssClass" );
1024                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1025                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1026                         # ISBN
1027                         $isbn = $m[5];
1028                         $num = strtr( $isbn, array(
1029                                 '-' => '',
1030                                 ' ' => '',
1031                                 'x' => 'X',
1032                         ));
1033                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1034                         return'<a href="' .
1035                                 $titleObj->escapeLocalUrl() .
1036                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1037                 } else {
1038                         return $m[0];
1039                 }
1040         }
1041
1042         /**
1043          * Make a free external link, given a user-supplied URL
1044          * @return HTML
1045          * @private
1046          */
1047         function makeFreeExternalLink( $url ) {
1048                 global $wgContLang;
1049                 wfProfileIn( __METHOD__ );
1050
1051                 $sk = $this->mOptions->getSkin();
1052                 $trail = '';
1053
1054                 # The characters '<' and '>' (which were escaped by
1055                 # removeHTMLtags()) should not be included in
1056                 # URLs, per RFC 2396.
1057                 $m2 = array();
1058                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1059                         $trail = substr($url, $m2[0][1]) . $trail;
1060                         $url = substr($url, 0, $m2[0][1]);
1061                 }
1062
1063                 # Move trailing punctuation to $trail
1064                 $sep = ',;\.:!?';
1065                 # If there is no left bracket, then consider right brackets fair game too
1066                 if ( strpos( $url, '(' ) === false ) {
1067                         $sep .= ')';
1068                 }
1069
1070                 $numSepChars = strspn( strrev( $url ), $sep );
1071                 if ( $numSepChars ) {
1072                         $trail = substr( $url, -$numSepChars ) . $trail;
1073                         $url = substr( $url, 0, -$numSepChars );
1074                 }
1075
1076                 $url = Sanitizer::cleanUrl( $url );
1077
1078                 # Is this an external image?
1079                 $text = $this->maybeMakeExternalImage( $url );
1080                 if ( $text === false ) {
1081                         # Not an image, make a link
1082                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1083                                 $this->getExternalLinkAttribs( $url ) );
1084                         # Register it in the output object...
1085                         # Replace unnecessary URL escape codes with their equivalent characters
1086                         $pasteurized = self::replaceUnusualEscapes( $url );
1087                         $this->mOutput->addExternalLink( $pasteurized );
1088                 }
1089                 wfProfileOut( __METHOD__ );
1090                 return $text . $trail;
1091         }
1092
1093
1094         /**
1095          * Parse headers and return html
1096          *
1097          * @private
1098          */
1099         function doHeadings( $text ) {
1100                 wfProfileIn( __METHOD__ );
1101                 for ( $i = 6; $i >= 1; --$i ) {
1102                         $h = str_repeat( '=', $i );
1103                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1104                           "<h$i>\\1</h$i>", $text );
1105                 }
1106                 wfProfileOut( __METHOD__ );
1107                 return $text;
1108         }
1109
1110         /**
1111          * Replace single quotes with HTML markup
1112          * @private
1113          * @return string the altered text
1114          */
1115         function doAllQuotes( $text ) {
1116                 wfProfileIn( __METHOD__ );
1117                 $outtext = '';
1118                 $lines = StringUtils::explode( "\n", $text );
1119                 foreach ( $lines as $line ) {
1120                         $outtext .= $this->doQuotes( $line ) . "\n";
1121                 }
1122                 $outtext = substr($outtext, 0,-1);
1123                 wfProfileOut( __METHOD__ );
1124                 return $outtext;
1125         }
1126
1127         /**
1128          * Helper function for doAllQuotes()
1129          */
1130         public function doQuotes( $text ) {
1131                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1132                 if ( count( $arr ) == 1 )
1133                         return $text;
1134                 else
1135                 {
1136                         # First, do some preliminary work. This may shift some apostrophes from
1137                         # being mark-up to being text. It also counts the number of occurrences
1138                         # of bold and italics mark-ups.
1139                         $i = 0;
1140                         $numbold = 0;
1141                         $numitalics = 0;
1142                         foreach ( $arr as $r )
1143                         {
1144                                 if ( ( $i % 2 ) == 1 )
1145                                 {
1146                                         # If there are ever four apostrophes, assume the first is supposed to
1147                                         # be text, and the remaining three constitute mark-up for bold text.
1148                                         if ( strlen( $arr[$i] ) == 4 )
1149                                         {
1150                                                 $arr[$i-1] .= "'";
1151                                                 $arr[$i] = "'''";
1152                                         }
1153                                         # If there are more than 5 apostrophes in a row, assume they're all
1154                                         # text except for the last 5.
1155                                         else if ( strlen( $arr[$i] ) > 5 )
1156                                         {
1157                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1158                                                 $arr[$i] = "'''''";
1159                                         }
1160                                         # Count the number of occurrences of bold and italics mark-ups.
1161                                         # We are not counting sequences of five apostrophes.
1162                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1163                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1164                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1165                                 }
1166                                 $i++;
1167                         }
1168
1169                         # If there is an odd number of both bold and italics, it is likely
1170                         # that one of the bold ones was meant to be an apostrophe followed
1171                         # by italics. Which one we cannot know for certain, but it is more
1172                         # likely to be one that has a single-letter word before it.
1173                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1174                         {
1175                                 $i = 0;
1176                                 $firstsingleletterword = -1;
1177                                 $firstmultiletterword = -1;
1178                                 $firstspace = -1;
1179                                 foreach ( $arr as $r )
1180                                 {
1181                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1182                                         {
1183                                                 $x1 = substr ($arr[$i-1], -1);
1184                                                 $x2 = substr ($arr[$i-1], -2, 1);
1185                                                 if ($x1 === ' ') {
1186                                                         if ($firstspace == -1) $firstspace = $i;
1187                                                 } else if ($x2 === ' ') {
1188                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1189                                                 } else {
1190                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1191                                                 }
1192                                         }
1193                                         $i++;
1194                                 }
1195
1196                                 # If there is a single-letter word, use it!
1197                                 if ($firstsingleletterword > -1)
1198                                 {
1199                                         $arr [ $firstsingleletterword ] = "''";
1200                                         $arr [ $firstsingleletterword-1 ] .= "'";
1201                                 }
1202                                 # If not, but there's a multi-letter word, use that one.
1203                                 else if ($firstmultiletterword > -1)
1204                                 {
1205                                         $arr [ $firstmultiletterword ] = "''";
1206                                         $arr [ $firstmultiletterword-1 ] .= "'";
1207                                 }
1208                                 # ... otherwise use the first one that has neither.
1209                                 # (notice that it is possible for all three to be -1 if, for example,
1210                                 # there is only one pentuple-apostrophe in the line)
1211                                 else if ($firstspace > -1)
1212                                 {
1213                                         $arr [ $firstspace ] = "''";
1214                                         $arr [ $firstspace-1 ] .= "'";
1215                                 }
1216                         }
1217
1218                         # Now let's actually convert our apostrophic mush to HTML!
1219                         $output = '';
1220                         $buffer = '';
1221                         $state = '';
1222                         $i = 0;
1223                         foreach ($arr as $r)
1224                         {
1225                                 if (($i % 2) == 0)
1226                                 {
1227                                         if ($state === 'both')
1228                                                 $buffer .= $r;
1229                                         else
1230                                                 $output .= $r;
1231                                 }
1232                                 else
1233                                 {
1234                                         if (strlen ($r) == 2)
1235                                         {
1236                                                 if ($state === 'i')
1237                                                 { $output .= '</i>'; $state = ''; }
1238                                                 else if ($state === 'bi')
1239                                                 { $output .= '</i>'; $state = 'b'; }
1240                                                 else if ($state === 'ib')
1241                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1242                                                 else if ($state === 'both')
1243                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1244                                                 else # $state can be 'b' or ''
1245                                                 { $output .= '<i>'; $state .= 'i'; }
1246                                         }
1247                                         else if (strlen ($r) == 3)
1248                                         {
1249                                                 if ($state === 'b')
1250                                                 { $output .= '</b>'; $state = ''; }
1251                                                 else if ($state === 'bi')
1252                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1253                                                 else if ($state === 'ib')
1254                                                 { $output .= '</b>'; $state = 'i'; }
1255                                                 else if ($state === 'both')
1256                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1257                                                 else # $state can be 'i' or ''
1258                                                 { $output .= '<b>'; $state .= 'b'; }
1259                                         }
1260                                         else if (strlen ($r) == 5)
1261                                         {
1262                                                 if ($state === 'b')
1263                                                 { $output .= '</b><i>'; $state = 'i'; }
1264                                                 else if ($state === 'i')
1265                                                 { $output .= '</i><b>'; $state = 'b'; }
1266                                                 else if ($state === 'bi')
1267                                                 { $output .= '</i></b>'; $state = ''; }
1268                                                 else if ($state === 'ib')
1269                                                 { $output .= '</b></i>'; $state = ''; }
1270                                                 else if ($state === 'both')
1271                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1272                                                 else # ($state == '')
1273                                                 { $buffer = ''; $state = 'both'; }
1274                                         }
1275                                 }
1276                                 $i++;
1277                         }
1278                         # Now close all remaining tags.  Notice that the order is important.
1279                         if ($state === 'b' || $state === 'ib')
1280                                 $output .= '</b>';
1281                         if ($state === 'i' || $state === 'bi' || $state === 'ib')
1282                                 $output .= '</i>';
1283                         if ($state === 'bi')
1284                                 $output .= '</b>';
1285                         # There might be lonely ''''', so make sure we have a buffer
1286                         if ($state === 'both' && $buffer)
1287                                 $output .= '<b><i>'.$buffer.'</i></b>';
1288                         return $output;
1289                 }
1290         }
1291
1292         /**
1293          * Replace external links (REL)
1294          *
1295          * Note: this is all very hackish and the order of execution matters a lot.
1296          * Make sure to run maintenance/parserTests.php if you change this code.
1297          *
1298          * @private
1299          */
1300         function replaceExternalLinks( $text ) {
1301                 global $wgContLang;
1302                 wfProfileIn( __METHOD__ );
1303
1304                 $sk = $this->mOptions->getSkin();
1305
1306                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1307                 $s = array_shift( $bits );
1308
1309                 $i = 0;
1310                 while ( $i<count( $bits ) ) {
1311                         $url = $bits[$i++];
1312                         $protocol = $bits[$i++];
1313                         $text = $bits[$i++];
1314                         $trail = $bits[$i++];
1315
1316                         # The characters '<' and '>' (which were escaped by
1317                         # removeHTMLtags()) should not be included in
1318                         # URLs, per RFC 2396.
1319                         $m2 = array();
1320                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1321                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1322                                 $url = substr($url, 0, $m2[0][1]);
1323                         }
1324
1325                         # If the link text is an image URL, replace it with an <img> tag
1326                         # This happened by accident in the original parser, but some people used it extensively
1327                         $img = $this->maybeMakeExternalImage( $text );
1328                         if ( $img !== false ) {
1329                                 $text = $img;
1330                         }
1331
1332                         $dtrail = '';
1333
1334                         # Set linktype for CSS - if URL==text, link is essentially free
1335                         $linktype = ($text === $url) ? 'free' : 'text';
1336
1337                         # No link text, e.g. [http://domain.tld/some.link]
1338                         if ( $text == '' ) {
1339                                 # Autonumber if allowed. See bug #5918
1340                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1341                                         $langObj = $this->getFunctionLang();
1342                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1343                                         $linktype = 'autonumber';
1344                                 } else {
1345                                         # Otherwise just use the URL
1346                                         $text = htmlspecialchars( $url );
1347                                         $linktype = 'free';
1348                                 }
1349                         } else {
1350                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1351                                 # Check for trail
1352                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1353                         }
1354
1355                         $text = $wgContLang->markNoConversion($text);
1356
1357                         $url = Sanitizer::cleanUrl( $url );
1358
1359                         # Use the encoded URL
1360                         # This means that users can paste URLs directly into the text
1361                         # Funny characters like &ouml; aren't valid in URLs anyway
1362                         # This was changed in August 2004
1363                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1364                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1365
1366                         # Register link in the output object.
1367                         # Replace unnecessary URL escape codes with the referenced character
1368                         # This prevents spammers from hiding links from the filters
1369                         $pasteurized = self::replaceUnusualEscapes( $url );
1370                         $this->mOutput->addExternalLink( $pasteurized );
1371                 }
1372
1373                 wfProfileOut( __METHOD__ );
1374                 return $s;
1375         }
1376
1377         /**
1378          * Get an associative array of additional HTML attributes appropriate for a
1379          * particular external link.  This currently may include rel => nofollow
1380          * (depending on configuration, namespace, and the URL's domain) and/or a
1381          * target attribute (depending on configuration).
1382          *
1383          * @param string $url Optional URL, to extract the domain from for rel =>
1384          *   nofollow if appropriate
1385          * @return array Associative array of HTML attributes
1386          */
1387         function getExternalLinkAttribs( $url = false ) {
1388                 $attribs = array();
1389                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1390                 $ns = $this->mTitle->getNamespace();
1391                 if( $wgNoFollowLinks && !in_array($ns, $wgNoFollowNsExceptions) ) {
1392                         $attribs['rel'] = 'nofollow';
1393
1394                         global $wgNoFollowDomainExceptions;
1395                         if ( $wgNoFollowDomainExceptions ) {
1396                                 $bits = wfParseUrl( $url );
1397                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1398                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1399                                                 if( substr( $bits['host'], -strlen( $domain ) )
1400                                                 == $domain ) {
1401                                                         unset( $attribs['rel'] );
1402                                                         break;
1403                                                 }
1404                                         }
1405                                 }
1406                         }
1407                 }
1408                 if ( $this->mOptions->getExternalLinkTarget() ) {
1409                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1410                 }
1411                 return $attribs;
1412         }
1413
1414
1415         /**
1416          * Replace unusual URL escape codes with their equivalent characters
1417          * @param string
1418          * @return string
1419          * @static
1420          * @todo  This can merge genuinely required bits in the path or query string,
1421          *        breaking legit URLs. A proper fix would treat the various parts of
1422          *        the URL differently; as a workaround, just use the output for
1423          *        statistical records, not for actual linking/output.
1424          */
1425         static function replaceUnusualEscapes( $url ) {
1426                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1427                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1428         }
1429
1430         /**
1431          * Callback function used in replaceUnusualEscapes().
1432          * Replaces unusual URL escape codes with their equivalent character
1433          * @static
1434          * @private
1435          */
1436         private static function replaceUnusualEscapesCallback( $matches ) {
1437                 $char = urldecode( $matches[0] );
1438                 $ord = ord( $char );
1439                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1440                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1441                         // No, shouldn't be escaped
1442                         return $char;
1443                 } else {
1444                         // Yes, leave it escaped
1445                         return $matches[0];
1446                 }
1447         }
1448
1449         /**
1450          * make an image if it's allowed, either through the global
1451          * option, through the exception, or through the on-wiki whitelist
1452          * @private
1453          */
1454         function maybeMakeExternalImage( $url ) {
1455                 $sk = $this->mOptions->getSkin();
1456                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1457                 $imagesexception = !empty($imagesfrom);
1458                 $text = false;
1459                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1460                 if( $imagesexception && is_array( $imagesfrom ) ) {
1461                         $imagematch = false;
1462                         foreach( $imagesfrom as $match ) {
1463                                 if( strpos( $url, $match ) === 0 ) {
1464                                         $imagematch = true;
1465                                         break;
1466                                 }
1467                         }
1468                 } elseif( $imagesexception ) {
1469                         $imagematch = (strpos( $url, $imagesfrom ) === 0);
1470                 } else {
1471                         $imagematch = false;
1472                 }
1473                 if ( $this->mOptions->getAllowExternalImages()
1474                      || ( $imagesexception && $imagematch ) ) {
1475                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1476                                 # Image found
1477                                 $text = $sk->makeExternalImage( $url );
1478                         }
1479                 }
1480                 if( !$text && $this->mOptions->getEnableImageWhitelist()
1481                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1482                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1483                         foreach( $whitelist as $entry ) {
1484                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1485                                 if( strpos( $entry, '#' ) === 0 || $entry === '' )
1486                                         continue;
1487                                 if( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1488                                         # Image matches a whitelist entry
1489                                         $text = $sk->makeExternalImage( $url );
1490                                         break;
1491                                 }
1492                         }
1493                 }
1494                 return $text;
1495         }
1496
1497         /**
1498          * Process [[ ]] wikilinks
1499          * @return processed text
1500          *
1501          * @private
1502          */
1503         function replaceInternalLinks( $s ) {
1504                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1505                 return $s;
1506         }
1507
1508         /**
1509          * Process [[ ]] wikilinks (RIL)
1510          * @return LinkHolderArray
1511          *
1512          * @private
1513          */
1514         function replaceInternalLinks2( &$s ) {
1515                 global $wgContLang;
1516
1517                 wfProfileIn( __METHOD__ );
1518
1519                 wfProfileIn( __METHOD__.'-setup' );
1520                 static $tc = FALSE, $e1, $e1_img;
1521                 # the % is needed to support urlencoded titles as well
1522                 if ( !$tc ) {
1523                         $tc = Title::legalChars() . '#%';
1524                         # Match a link having the form [[namespace:link|alternate]]trail
1525                         $e1 = "/^([{$tc}]*)(\\|.*?)?]](.*)\$/sD";
1526                         # Match cases where there is no "]]", which might still be images
1527                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1528                 }
1529
1530                 $sk = $this->mOptions->getSkin();
1531                 $holders = new LinkHolderArray( $this );
1532
1533                 #split the entire text string on occurences of [[
1534                 $a = StringUtils::explode( '[[', ' ' . $s );
1535                 #get the first element (all text up to first [[), and remove the space we added
1536                 $s = $a->current();
1537                 $a->next();
1538                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1539                 $s = substr( $s, 1 );
1540
1541                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1542                 $e2 = null;
1543                 if ( $useLinkPrefixExtension ) {
1544                         # Match the end of a line for a word that's not followed by whitespace,
1545                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1546                         $e2 = wfMsgForContent( 'linkprefix' );
1547                 }
1548
1549                 if( is_null( $this->mTitle ) ) {
1550                         wfProfileOut( __METHOD__.'-setup' );
1551                         wfProfileOut( __METHOD__ );
1552                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1553                 }
1554                 $nottalk = !$this->mTitle->isTalkPage();
1555
1556                 if ( $useLinkPrefixExtension ) {
1557                         $m = array();
1558                         if ( preg_match( $e2, $s, $m ) ) {
1559                                 $first_prefix = $m[2];
1560                         } else {
1561                                 $first_prefix = false;
1562                         }
1563                 } else {
1564                         $prefix = '';
1565                 }
1566
1567                 if($wgContLang->hasVariants()) {
1568                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1569                 } else {
1570                         $selflink = array($this->mTitle->getPrefixedText());
1571                 }
1572                 $useSubpages = $this->areSubpagesAllowed();
1573                 wfProfileOut( __METHOD__.'-setup' );
1574
1575                 # Loop for each link
1576                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1577                         # Check for excessive memory usage
1578                         if ( $holders->isBig() ) {
1579                                 # Too big
1580                                 # Do the existence check, replace the link holders and clear the array
1581                                 $holders->replace( $s );
1582                                 $holders->clear();
1583                         }
1584
1585                         if ( $useLinkPrefixExtension ) {
1586                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1587                                 if ( preg_match( $e2, $s, $m ) ) {
1588                                         $prefix = $m[2];
1589                                         $s = $m[1];
1590                                 } else {
1591                                         $prefix='';
1592                                 }
1593                                 # first link
1594                                 if($first_prefix) {
1595                                         $prefix = $first_prefix;
1596                                         $first_prefix = false;
1597                                 }
1598                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1599                         }
1600
1601                         $might_be_img = false;
1602
1603                         wfProfileIn( __METHOD__."-e1" );
1604                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1605
1606                                 if( $m[2] === '' ) {
1607                                         $text = '';
1608                                 } elseif( $m[2] === '|' ) {
1609                                         $text = $this->getPipeTrickText( $m[1] );
1610                                 } else {
1611                                         $text = substr( $m[2], 1 );
1612                                 }
1613
1614                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1615                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1616                                 # the real problem is with the $e1 regex
1617                                 # See bug 1300.
1618                                 #
1619                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1620                                 # and no image is in sight. See bug 2095.
1621                                 #
1622                                 if( $text !== '' &&
1623                                         substr( $m[3], 0, 1 ) === ']' &&
1624                                         strpos($text, '[') !== false
1625                                 )
1626                                 {
1627                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1628                                         $m[3] = substr( $m[3], 1 );
1629                                 }
1630
1631                                 # Handle pipe-trick for [[|<blah>]]
1632                                 $lnk = $m[1] === '' ? $this->getPipeTrickLink( $text ) : $m[1];
1633                                 # fix up urlencoded title texts
1634                                 if( strpos( $lnk, '%' ) !== false ) {
1635                                         # Should anchors '#' also be rejected?
1636                                         $lnk = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($lnk) );
1637                                 }
1638
1639                                 $trail = $m[3];
1640                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1641                                 $might_be_img = true;
1642                                 $text = $m[2];
1643                                 $lnk = strpos( $m[1], '%' ) === false ? $m[1] : urldecode( $m[1] );
1644                                 $trail = "";
1645                         } else { # Invalid form; output directly
1646                                 $s .= $prefix . '[[' . $line ;
1647                                 wfProfileOut( __METHOD__."-e1" );
1648                                 continue;
1649                         }
1650                         wfProfileOut( __METHOD__."-e1" );
1651                         wfProfileIn( __METHOD__."-misc" );
1652
1653                         # Don't allow internal links to pages containing
1654                         # PROTO: where PROTO is a valid URL protocol; these
1655                         # should be external links.
1656                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $lnk ) ) {
1657                                 $s .= $prefix . '[[' . $line ;
1658                                 wfProfileOut( __METHOD__."-misc" );
1659                                 continue;
1660                         }
1661
1662                         # Make subpage if necessary
1663                         if ( $useSubpages ) {
1664                                 $link = $this->maybeDoSubpageLink( $lnk, $text );
1665                         } else {
1666                                 $link = $lnk;
1667                         }
1668
1669                         $noforce = (substr( $lnk, 0, 1 ) !== ':');
1670                         if (!$noforce) {
1671                                 # Strip off leading ':'
1672                                 $link = substr( $link, 1 );
1673                         }
1674
1675                         wfProfileOut( __METHOD__."-misc" );
1676                         wfProfileIn( __METHOD__."-title" );
1677                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1678                         if ( $nt === null ) {
1679                                 $s .= $prefix . '[[' . $line;
1680                                 wfProfileOut( __METHOD__."-title" );
1681                                 continue;
1682                         }
1683
1684                         $ns = $nt->getNamespace();
1685                         $iw = $nt->getInterWiki();
1686                         wfProfileOut( __METHOD__."-title" );
1687
1688                         if ( $might_be_img ) { # if this is actually an invalid link
1689                                 wfProfileIn( __METHOD__."-might_be_img" );
1690                                 if ( $ns == NS_FILE && $noforce ) { #but might be an image
1691                                         $found = false;
1692                                         while ( true ) {
1693                                                 #look at the next 'line' to see if we can close it there
1694                                                 $a->next();
1695                                                 $next_line = $a->current();
1696                                                 if ( $next_line === false || $next_line === null ) {
1697                                                         break;
1698                                                 }
1699                                                 $m = explode( ']]', $next_line, 3 );
1700                                                 if ( count( $m ) == 3 ) {
1701                                                         # the first ]] closes the inner link, the second the image
1702                                                         $found = true;
1703                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1704                                                         $trail = $m[2];
1705                                                         break;
1706                                                 } elseif ( count( $m ) == 2 ) {
1707                                                         #if there's exactly one ]] that's fine, we'll keep looking
1708                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1709                                                 } else {
1710                                                         #if $next_line is invalid too, we need look no further
1711                                                         $text .= '[[' . $next_line;
1712                                                         break;
1713                                                 }
1714                                         }
1715                                         if ( !$found ) {
1716                                                 # we couldn't find the end of this imageLink, so output it raw
1717                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1718                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1719                                                 $s .= "{$prefix}[[$link|$text";
1720                                                 # note: no $trail, because without an end, there *is* no trail
1721                                                 wfProfileOut( __METHOD__."-might_be_img" );
1722                                                 continue;
1723                                         }
1724                                 } else { #it's not an image, so output it raw
1725                                         $s .= "{$prefix}[[$link|$text";
1726                                         # note: no $trail, because without an end, there *is* no trail
1727                                         wfProfileOut( __METHOD__."-might_be_img" );
1728                                         continue;
1729                                 }
1730                                 wfProfileOut( __METHOD__."-might_be_img" );
1731                         }
1732
1733                         $wasblank = ( $text  == '' );
1734                         if ( $wasblank ) $text = $link;
1735
1736                         # Link not escaped by : , create the various objects
1737                         if ( $noforce ) {
1738
1739                                 # Interwikis
1740                                 wfProfileIn( __METHOD__."-interwiki" );
1741                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1742                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1743                                         $s = rtrim($s . $prefix);
1744                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1745                                         wfProfileOut( __METHOD__."-interwiki" );
1746                                         continue;
1747                                 }
1748                                 wfProfileOut( __METHOD__."-interwiki" );
1749
1750                                 if ( $ns == NS_FILE ) {
1751                                         wfProfileIn( __METHOD__."-image" );
1752                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1753                                                 if ( $wasblank ) {
1754                                                         # if no parameters were passed, $text
1755                                                         # becomes something like "File:Foo.png",
1756                                                         # which we don't want to pass on to the
1757                                                         # image generator
1758                                                         $text = '';
1759                                                 } else {
1760                                                         # recursively parse links inside the image caption
1761                                                         # actually, this will parse them in any other parameters, too,
1762                                                         # but it might be hard to fix that, and it doesn't matter ATM
1763                                                         $text = $this->replaceExternalLinks($text);
1764                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1765                                                 }
1766                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1767                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1768                                         } else {
1769                                                 $s .= $prefix . $trail;
1770                                         }
1771                                         $this->mOutput->addImage( $nt->getDBkey() );
1772                                         wfProfileOut( __METHOD__."-image" );
1773                                         continue;
1774
1775                                 }
1776
1777                                 if ( $ns == NS_CATEGORY ) {
1778                                         wfProfileIn( __METHOD__."-category" );
1779                                         $s = rtrim($s . "\n"); # bug 87
1780
1781                                         if ( $wasblank ) {
1782                                                 $sortkey = $this->getDefaultSort();
1783                                         } else {
1784                                                 $sortkey = $text;
1785                                         }
1786                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1787                                         $sortkey = str_replace( "\n", '', $sortkey );
1788                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1789                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1790
1791                                         /**
1792                                          * Strip the whitespace Category links produce, see bug 87
1793                                          * @todo We might want to use trim($tmp, "\n") here.
1794                                          */
1795                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1796
1797                                         wfProfileOut( __METHOD__."-category" );
1798                                         continue;
1799                                 }
1800                         }
1801
1802                         # Self-link checking
1803                         if( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1804                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1805                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1806                                         continue;
1807                                 }
1808                         }
1809
1810                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1811                         # FIXME: Should do batch file existence checks, see comment below
1812                         if( $ns == NS_MEDIA ) {
1813                                 wfProfileIn( __METHOD__."-media" );
1814                                 # Give extensions a chance to select the file revision for us
1815                                 $skip = $time = false;
1816                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1817                                 if ( $skip ) {
1818                                         $link = $sk->link( $nt );
1819                                 } else {
1820                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1821                                 }
1822                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1823                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1824                                 $this->mOutput->addImage( $nt->getDBkey() );
1825                                 wfProfileOut( __METHOD__."-media" );
1826                                 continue;
1827                         }
1828
1829                         wfProfileIn( __METHOD__."-always_known" );
1830                         # Some titles, such as valid special pages or files in foreign repos, should
1831                         # be shown as bluelinks even though they're not included in the page table
1832                         #
1833                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1834                         # batch file existence checks for NS_FILE and NS_MEDIA
1835                         if( $iw == '' && $nt->isAlwaysKnown() ) {
1836                                 $this->mOutput->addLink( $nt );
1837                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1838                         } else {
1839                                 # Links will be added to the output link list after checking
1840                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1841                         }
1842                         wfProfileOut( __METHOD__."-always_known" );
1843                 }
1844                 wfProfileOut( __METHOD__ );
1845                 return $holders;
1846         }
1847
1848         /**
1849          * Make a link placeholder. The text returned can be later resolved to a real link with
1850          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1851          * parsing of interwiki links, and secondly to allow all existence checks and
1852          * article length checks (for stub links) to be bundled into a single query.
1853          *
1854          * @deprecated
1855          */
1856         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1857                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1858         }
1859
1860         /**
1861          * Render a forced-blue link inline; protect against double expansion of
1862          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1863          * Since this little disaster has to split off the trail text to avoid
1864          * breaking URLs in the following text without breaking trails on the
1865          * wiki links, it's been made into a horrible function.
1866          *
1867          * @param Title $nt
1868          * @param string $text
1869          * @param string $query
1870          * @param string $trail
1871          * @param string $prefix
1872          * @return string HTML-wikitext mix oh yuck
1873          */
1874         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1875                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1876                 $sk = $this->mOptions->getSkin();
1877                 // FIXME: use link() instead of deprecated makeKnownLinkObj()
1878                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1879                 return $this->armorLinks( $link ) . $trail;
1880         }
1881
1882         /**
1883          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1884          * going to go through further parsing steps before inline URL expansion.
1885          *
1886          * Not needed quite as much as it used to be since free links are a bit
1887          * more sensible these days. But bracketed links are still an issue.
1888          *
1889          * @param string more-or-less HTML
1890          * @return string less-or-more HTML with NOPARSE bits
1891          */
1892         function armorLinks( $text ) {
1893                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1894                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1895         }
1896
1897         /**
1898          * Return true if subpage links should be expanded on this page.
1899          * @return bool
1900          */
1901         function areSubpagesAllowed() {
1902                 # Some namespaces don't allow subpages
1903                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1904         }
1905
1906         /**
1907          * Handle link to subpage if necessary
1908          * @param string $target the source of the link
1909          * @param string &$text the link text, modified as necessary
1910          * @return string the full name of the link
1911          * @private
1912          */
1913         function maybeDoSubpageLink($target, &$text) {
1914                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
1915         }
1916
1917         /**
1918          * From the [[title|]] return link-text as though the used typed [[title|link-text]]
1919          * @param string $link from [[$link|]]
1920          * @return string $text for [[$link|$text]]
1921          */
1922         function getPipeTrickText( $link ) {
1923                 return Linker::getPipeTrickText( $link );
1924         }
1925
1926         /**
1927          * From the [[|link-text]] return the title as though the user typed [[title|link-text]]
1928          * @param string $text from [[|$text]]
1929          * @param Title $title to resolve the link against
1930          * @return string $link for [[$link|$text]]
1931          */
1932         function getPipeTrickLink( $text ) {
1933                 return Linker::getPipeTrickLink( $text, $this->mTitle );
1934         }
1935
1936         /**#@+
1937          * Used by doBlockLevels()
1938          * @private
1939          */
1940         /* private */ function closeParagraph() {
1941                 $result = '';
1942                 if ( $this->mLastSection != '' ) {
1943                         $result = '</' . $this->mLastSection  . ">\n";
1944                 }
1945                 $this->mInPre = false;
1946                 $this->mLastSection = '';
1947                 return $result;
1948         }
1949         # getCommon() returns the length of the longest common substring
1950         # of both arguments, starting at the beginning of both.
1951         #
1952         /* private */ function getCommon( $st1, $st2 ) {
1953                 $fl = strlen( $st1 );
1954                 $shorter = strlen( $st2 );
1955                 if ( $fl < $shorter ) { $shorter = $fl; }
1956
1957                 for ( $i = 0; $i < $shorter; ++$i ) {
1958                         if ( $st1{$i} != $st2{$i} ) { break; }
1959                 }
1960                 return $i;
1961         }
1962         # These next three functions open, continue, and close the list
1963         # element appropriate to the prefix character passed into them.
1964         #
1965         /* private */ function openList( $char ) {
1966                 $result = $this->closeParagraph();
1967
1968                 if ( '*' === $char ) { $result .= '<ul><li>'; }
1969                 elseif ( '#' === $char ) { $result .= '<ol><li>'; }
1970                 elseif ( ':' === $char ) { $result .= '<dl><dd>'; }
1971                 elseif ( ';' === $char ) {
1972                         $result .= '<dl><dt>';
1973                         $this->mDTopen = true;
1974                 }
1975                 else { $result = '<!-- ERR 1 -->'; }
1976
1977                 return $result;
1978         }
1979
1980         /* private */ function nextItem( $char ) {
1981                 if ( '*' === $char || '#' === $char ) { return '</li><li>'; }
1982                 elseif ( ':' === $char || ';' === $char ) {
1983                         $close = '</dd>';
1984                         if ( $this->mDTopen ) { $close = '</dt>'; }
1985                         if ( ';' === $char ) {
1986                                 $this->mDTopen = true;
1987                                 return $close . '<dt>';
1988                         } else {
1989                                 $this->mDTopen = false;
1990                                 return $close . '<dd>';
1991                         }
1992                 }
1993                 return '<!-- ERR 2 -->';
1994         }
1995
1996         /* private */ function closeList( $char ) {
1997                 if ( '*' === $char ) { $text = '</li></ul>'; }
1998                 elseif ( '#' === $char ) { $text = '</li></ol>'; }
1999                 elseif ( ':' === $char ) {
2000                         if ( $this->mDTopen ) {
2001                                 $this->mDTopen = false;
2002                                 $text = '</dt></dl>';
2003                         } else {
2004                                 $text = '</dd></dl>';
2005                         }
2006                 }
2007                 else {  return '<!-- ERR 3 -->'; }
2008                 return $text."\n";
2009         }
2010         /**#@-*/
2011
2012         /**
2013          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2014          *
2015          * @param $linestart bool whether or not this is at the start of a line.
2016          * @private
2017          * @return string the lists rendered as HTML
2018          */
2019         function doBlockLevels( $text, $linestart ) {
2020                 wfProfileIn( __METHOD__ );
2021
2022                 # Parsing through the text line by line.  The main thing
2023                 # happening here is handling of block-level elements p, pre,
2024                 # and making lists from lines starting with * # : etc.
2025                 #
2026                 $textLines = StringUtils::explode( "\n", $text );
2027
2028                 $lastPrefix = $output = '';
2029                 $this->mDTopen = $inBlockElem = false;
2030                 $prefixLength = 0;
2031                 $paragraphStack = false;
2032
2033                 foreach ( $textLines as $oLine ) {
2034                         # Fix up $linestart
2035                         if ( !$linestart ) {
2036                                 $output .= $oLine;
2037                                 $linestart = true;
2038                                 continue;
2039                         }
2040                         // * = ul
2041                         // # = ol
2042                         // ; = dt
2043                         // : = dd
2044
2045                         $lastPrefixLength = strlen( $lastPrefix );
2046                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2047                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2048                         // If not in a <pre> element, scan for and figure out what prefixes are there.
2049                         if ( !$this->mInPre ) {
2050                                 # Multiple prefixes may abut each other for nested lists.
2051                                 $prefixLength = strspn( $oLine, '*#:;' );
2052                                 $prefix = substr( $oLine, 0, $prefixLength );
2053
2054                                 # eh?
2055                                 // ; and : are both from definition-lists, so they're equivalent
2056                                 //  for the purposes of determining whether or not we need to open/close
2057                                 //  elements.
2058                                 $prefix2 = str_replace( ';', ':', $prefix );
2059                                 $t = substr( $oLine, $prefixLength );
2060                                 $this->mInPre = (bool)$preOpenMatch;
2061                         } else {
2062                                 # Don't interpret any other prefixes in preformatted text
2063                                 $prefixLength = 0;
2064                                 $prefix = $prefix2 = '';
2065                                 $t = $oLine;
2066                         }
2067
2068                         # List generation
2069                         if( $prefixLength && $lastPrefix === $prefix2 ) {
2070                                 # Same as the last item, so no need to deal with nesting or opening stuff
2071                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2072                                 $paragraphStack = false;
2073
2074                                 if ( substr( $prefix, -1 ) === ';') {
2075                                         # The one nasty exception: definition lists work like this:
2076                                         # ; title : definition text
2077                                         # So we check for : in the remainder text to split up the
2078                                         # title and definition, without b0rking links.
2079                                         $term = $t2 = '';
2080                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2081                                                 $t = $t2;
2082                                                 $output .= $term . $this->nextItem( ':' );
2083                                         }
2084                                 }
2085                         } elseif( $prefixLength || $lastPrefixLength ) {
2086                                 // We need to open or close prefixes, or both.
2087
2088                                 # Either open or close a level...
2089                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2090                                 $paragraphStack = false;
2091
2092                                 // Close all the prefixes which aren't shared.
2093                                 while( $commonPrefixLength < $lastPrefixLength ) {
2094                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2095                                         --$lastPrefixLength;
2096                                 }
2097
2098                                 // Continue the current prefix if appropriate.
2099                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2100                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2101                                 }
2102
2103                                 // Open prefixes where appropriate.
2104                                 while ( $prefixLength > $commonPrefixLength ) {
2105                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2106                                         $output .= $this->openList( $char );
2107
2108                                         if ( ';' === $char ) {
2109                                                 # FIXME: This is dupe of code above
2110                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2111                                                         $t = $t2;
2112                                                         $output .= $term . $this->nextItem( ':' );
2113                                                 }
2114                                         }
2115                                         ++$commonPrefixLength;
2116                                 }
2117                                 $lastPrefix = $prefix2;
2118                         }
2119
2120                         // If we have no prefixes, go to paragraph mode.
2121                         if( 0 == $prefixLength ) {
2122                                 wfProfileIn( __METHOD__."-paragraph" );
2123                                 # No prefix (not in list)--go to paragraph mode
2124                                 // XXX: use a stack for nestable elements like span, table and div
2125                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2126                                 $closematch = preg_match(
2127                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2128                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2129                                 if ( $openmatch or $closematch ) {
2130                                         $paragraphStack = false;
2131                                         # TODO bug 5718: paragraph closed
2132                                         $output .= $this->closeParagraph();
2133                                         if ( $preOpenMatch and !$preCloseMatch ) {
2134                                                 $this->mInPre = true;
2135                                         }
2136                                         if ( $closematch ) {
2137                                                 $inBlockElem = false;
2138                                         } else {
2139                                                 $inBlockElem = true;
2140                                         }
2141                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2142                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' or trim($t) != '' ) ) {
2143                                                 // pre
2144                                                 if ($this->mLastSection !== 'pre') {
2145                                                         $paragraphStack = false;
2146                                                         $output .= $this->closeParagraph().'<pre>';
2147                                                         $this->mLastSection = 'pre';
2148                                                 }
2149                                                 $t = substr( $t, 1 );
2150                                         } else {
2151                                                 // paragraph
2152                                                 if ( trim($t) == '' ) {
2153                                                         if ( $paragraphStack ) {
2154                                                                 $output .= $paragraphStack.'<br />';
2155                                                                 $paragraphStack = false;
2156                                                                 $this->mLastSection = 'p';
2157                                                         } else {
2158                                                                 if ($this->mLastSection !== 'p' ) {
2159                                                                         $output .= $this->closeParagraph();
2160                                                                         $this->mLastSection = '';
2161                                                                         $paragraphStack = '<p>';
2162                                                                 } else {
2163                                                                         $paragraphStack = '</p><p>';
2164                                                                 }
2165                                                         }
2166                                                 } else {
2167                                                         if ( $paragraphStack ) {
2168                                                                 $output .= $paragraphStack;
2169                                                                 $paragraphStack = false;
2170                                                                 $this->mLastSection = 'p';
2171                                                         } else if ($this->mLastSection !== 'p') {
2172                                                                 $output .= $this->closeParagraph().'<p>';
2173                                                                 $this->mLastSection = 'p';
2174                                                         }
2175                                                 }
2176                                         }
2177                                 }
2178                                 wfProfileOut( __METHOD__."-paragraph" );
2179                         }
2180                         // somewhere above we forget to get out of pre block (bug 785)
2181                         if($preCloseMatch && $this->mInPre) {
2182                                 $this->mInPre = false;
2183                         }
2184                         if ($paragraphStack === false) {
2185                                 $output .= $t."\n";
2186                         }
2187                 }
2188                 while ( $prefixLength ) {
2189                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2190                         --$prefixLength;
2191                 }
2192                 if ( $this->mLastSection != '' ) {
2193                         $output .= '</' . $this->mLastSection . '>';
2194                         $this->mLastSection = '';
2195                 }
2196
2197                 wfProfileOut( __METHOD__ );
2198                 return $output;
2199         }
2200
2201         /**
2202          * Split up a string on ':', ignoring any occurences inside tags
2203          * to prevent illegal overlapping.
2204          * @param string $str the string to split
2205          * @param string &$before set to everything before the ':'
2206          * @param string &$after set to everything after the ':'
2207          * return string the position of the ':', or false if none found
2208          */
2209         function findColonNoLinks($str, &$before, &$after) {
2210                 wfProfileIn( __METHOD__ );
2211
2212                 $pos = strpos( $str, ':' );
2213                 if( $pos === false ) {
2214                         // Nothing to find!
2215                         wfProfileOut( __METHOD__ );
2216                         return false;
2217                 }
2218
2219                 $lt = strpos( $str, '<' );
2220                 if( $lt === false || $lt > $pos ) {
2221                         // Easy; no tag nesting to worry about
2222                         $before = substr( $str, 0, $pos );
2223                         $after = substr( $str, $pos+1 );
2224                         wfProfileOut( __METHOD__ );
2225                         return $pos;
2226                 }
2227
2228                 // Ugly state machine to walk through avoiding tags.
2229                 $state = self::COLON_STATE_TEXT;
2230                 $stack = 0;
2231                 $len = strlen( $str );
2232                 for( $i = 0; $i < $len; $i++ ) {
2233                         $c = $str{$i};
2234
2235                         switch( $state ) {
2236                         // (Using the number is a performance hack for common cases)
2237                         case 0: // self::COLON_STATE_TEXT:
2238                                 switch( $c ) {
2239                                 case "<":
2240                                         // Could be either a <start> tag or an </end> tag
2241                                         $state = self::COLON_STATE_TAGSTART;
2242                                         break;
2243                                 case ":":
2244                                         if( $stack == 0 ) {
2245                                                 // We found it!
2246                                                 $before = substr( $str, 0, $i );
2247                                                 $after = substr( $str, $i + 1 );
2248                                                 wfProfileOut( __METHOD__ );
2249                                                 return $i;
2250                                         }
2251                                         // Embedded in a tag; don't break it.
2252                                         break;
2253                                 default:
2254                                         // Skip ahead looking for something interesting
2255                                         $colon = strpos( $str, ':', $i );
2256                                         if( $colon === false ) {
2257                                                 // Nothing else interesting
2258                                                 wfProfileOut( __METHOD__ );
2259                                                 return false;
2260                                         }
2261                                         $lt = strpos( $str, '<', $i );
2262                                         if( $stack === 0 ) {
2263                                                 if( $lt === false || $colon < $lt ) {
2264                                                         // We found it!
2265                                                         $before = substr( $str, 0, $colon );
2266                                                         $after = substr( $str, $colon + 1 );
2267                                                         wfProfileOut( __METHOD__ );
2268                                                         return $i;
2269                                                 }
2270                                         }
2271                                         if( $lt === false ) {
2272                                                 // Nothing else interesting to find; abort!
2273                                                 // We're nested, but there's no close tags left. Abort!
2274                                                 break 2;
2275                                         }
2276                                         // Skip ahead to next tag start
2277                                         $i = $lt;
2278                                         $state = self::COLON_STATE_TAGSTART;
2279                                 }
2280                                 break;
2281                         case 1: // self::COLON_STATE_TAG:
2282                                 // In a <tag>
2283                                 switch( $c ) {
2284                                 case ">":
2285                                         $stack++;
2286                                         $state = self::COLON_STATE_TEXT;
2287                                         break;
2288                                 case "/":
2289                                         // Slash may be followed by >?
2290                                         $state = self::COLON_STATE_TAGSLASH;
2291                                         break;
2292                                 default:
2293                                         // ignore
2294                                 }
2295                                 break;
2296                         case 2: // self::COLON_STATE_TAGSTART:
2297                                 switch( $c ) {
2298                                 case "/":
2299                                         $state = self::COLON_STATE_CLOSETAG;
2300                                         break;
2301                                 case "!":
2302                                         $state = self::COLON_STATE_COMMENT;
2303                                         break;
2304                                 case ">":
2305                                         // Illegal early close? This shouldn't happen D:
2306                                         $state = self::COLON_STATE_TEXT;
2307                                         break;
2308                                 default:
2309                                         $state = self::COLON_STATE_TAG;
2310                                 }
2311                                 break;
2312                         case 3: // self::COLON_STATE_CLOSETAG:
2313                                 // In a </tag>
2314                                 if( $c === ">" ) {
2315                                         $stack--;
2316                                         if( $stack < 0 ) {
2317                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2318                                                 wfProfileOut( __METHOD__ );
2319                                                 return false;
2320                                         }
2321                                         $state = self::COLON_STATE_TEXT;
2322                                 }
2323                                 break;
2324                         case self::COLON_STATE_TAGSLASH:
2325                                 if( $c === ">" ) {
2326                                         // Yes, a self-closed tag <blah/>
2327                                         $state = self::COLON_STATE_TEXT;
2328                                 } else {
2329                                         // Probably we're jumping the gun, and this is an attribute
2330                                         $state = self::COLON_STATE_TAG;
2331                                 }
2332                                 break;
2333                         case 5: // self::COLON_STATE_COMMENT:
2334                                 if( $c === "-" ) {
2335                                         $state = self::COLON_STATE_COMMENTDASH;
2336                                 }
2337                                 break;
2338                         case self::COLON_STATE_COMMENTDASH:
2339                                 if( $c === "-" ) {
2340                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2341                                 } else {
2342                                         $state = self::COLON_STATE_COMMENT;
2343                                 }
2344                                 break;
2345                         case self::COLON_STATE_COMMENTDASHDASH:
2346                                 if( $c === ">" ) {
2347                                         $state = self::COLON_STATE_TEXT;
2348                                 } else {
2349                                         $state = self::COLON_STATE_COMMENT;
2350                                 }
2351                                 break;
2352                         default:
2353                                 throw new MWException( "State machine error in " . __METHOD__ );
2354                         }
2355                 }
2356                 if( $stack > 0 ) {
2357                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2358                         return false;
2359                 }
2360                 wfProfileOut( __METHOD__ );
2361                 return false;
2362         }
2363
2364         /**
2365          * Return value of a magic variable (like PAGENAME)
2366          *
2367          * @private
2368          */
2369         function getVariableValue( $index, $frame=false ) {
2370                 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2371                 global $wgScriptPath, $wgStylePath;
2372
2373                 /**
2374                  * Some of these require message or data lookups and can be
2375                  * expensive to check many times.
2376                  */
2377                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2378                         if ( isset( $this->mVarCache[$index] ) ) {
2379                                 return $this->mVarCache[$index];
2380                         }
2381                 }
2382
2383                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2384                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2385
2386                 # Use the time zone
2387                 global $wgLocaltimezone;
2388                 if ( isset( $wgLocaltimezone ) ) {
2389                         $oldtz = date_default_timezone_get();
2390                         date_default_timezone_set( $wgLocaltimezone );
2391                 }
2392
2393                 $localTimestamp = date( 'YmdHis', $ts );
2394                 $localMonth = date( 'm', $ts );
2395                 $localMonth1 = date( 'n', $ts );
2396                 $localMonthName = date( 'n', $ts );
2397                 $localDay = date( 'j', $ts );
2398                 $localDay2 = date( 'd', $ts );
2399                 $localDayOfWeek = date( 'w', $ts );
2400                 $localWeek = date( 'W', $ts );
2401                 $localYear = date( 'Y', $ts );
2402                 $localHour = date( 'H', $ts );
2403                 if ( isset( $wgLocaltimezone ) ) {
2404                         date_default_timezone_set( $oldtz );
2405                 }
2406
2407                 switch ( $index ) {
2408                         case 'currentmonth':
2409                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2410                                 break;
2411                         case 'currentmonth1':
2412                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2413                                 break;
2414                         case 'currentmonthname':
2415                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2416                                 break;
2417                         case 'currentmonthnamegen':
2418                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2419                                 break;
2420                         case 'currentmonthabbrev':
2421                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2422                                 break;
2423                         case 'currentday':
2424                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2425                                 break;
2426                         case 'currentday2':
2427                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2428                                 break;
2429                         case 'localmonth':
2430                                 $value = $wgContLang->formatNum( $localMonth );
2431                                 break;
2432                         case 'localmonth1':
2433                                 $value = $wgContLang->formatNum( $localMonth1 );
2434                                 break;
2435                         case 'localmonthname':
2436                                 $value = $wgContLang->getMonthName( $localMonthName );
2437                                 break;
2438                         case 'localmonthnamegen':
2439                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2440                                 break;
2441                         case 'localmonthabbrev':
2442                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2443                                 break;
2444                         case 'localday':
2445                                 $value = $wgContLang->formatNum( $localDay );
2446                                 break;
2447                         case 'localday2':
2448                                 $value = $wgContLang->formatNum( $localDay2 );
2449                                 break;
2450                         case 'pagename':
2451                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2452                                 break;
2453                         case 'pagenamee':
2454                                 $value = $this->mTitle->getPartialURL();
2455                                 break;
2456                         case 'fullpagename':
2457                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2458                                 break;
2459                         case 'fullpagenamee':
2460                                 $value = $this->mTitle->getPrefixedURL();
2461                                 break;
2462                         case 'subpagename':
2463                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2464                                 break;
2465                         case 'subpagenamee':
2466                                 $value = $this->mTitle->getSubpageUrlForm();
2467                                 break;
2468                         case 'basepagename':
2469                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2470                                 break;
2471                         case 'basepagenamee':
2472                                 $value = wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2473                                 break;
2474                         case 'talkpagename':
2475                                 if( $this->mTitle->canTalk() ) {
2476                                         $talkPage = $this->mTitle->getTalkPage();
2477                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2478                                 } else {
2479                                         $value = '';
2480                                 }
2481                                 break;
2482                         case 'talkpagenamee':
2483                                 if( $this->mTitle->canTalk() ) {
2484                                         $talkPage = $this->mTitle->getTalkPage();
2485                                         $value = $talkPage->getPrefixedUrl();
2486                                 } else {
2487                                         $value = '';
2488                                 }
2489                                 break;
2490                         case 'subjectpagename':
2491                                 $subjPage = $this->mTitle->getSubjectPage();
2492                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2493                                 break;
2494                         case 'subjectpagenamee':
2495                                 $subjPage = $this->mTitle->getSubjectPage();
2496                                 $value = $subjPage->getPrefixedUrl();
2497                                 break;
2498                         case 'pipetrick':
2499                                 $text = $this->mTitle->getText();
2500                                 $value = $this->getPipeTrickText( $text );
2501                                 break;
2502                         case 'pipetricke':
2503                                 $text = $this->mTitle->getText();
2504                                 $value = wfUrlEncode( str_replace( ' ', '_', $this->getPipeTrickText( $text ) ) );
2505                                 break;
2506                         case 'revisionid':
2507                                 // Let the edit saving system know we should parse the page
2508                                 // *after* a revision ID has been assigned.
2509                                 $this->mOutput->setFlag( 'vary-revision' );
2510                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2511                                 $value = $this->mRevisionId;
2512                                 break;
2513                         case 'revisionday':
2514                                 // Let the edit saving system know we should parse the page
2515                                 // *after* a revision ID has been assigned. This is for null edits.
2516                                 $this->mOutput->setFlag( 'vary-revision' );
2517                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2518                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2519                                 break;
2520                         case 'revisionday2':
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__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2525                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2526                                 break;
2527                         case 'revisionmonth':
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__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2532                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2533                                 break;
2534                         case 'revisionyear':
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__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2539                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2540                                 break;
2541                         case 'revisiontimestamp':
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__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2546                                 $value = $this->getRevisionTimestamp();
2547                                 break;
2548                         case 'revisionuser':
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__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2553                                 $value = $this->getRevisionUser();
2554                                 break;
2555                         case 'namespace':
2556                                 $value = str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2557                                 break;
2558                         case 'namespacee':
2559                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2560                                 break;
2561                         case 'talkspace':
2562                                 $value = $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2563                                 break;
2564                         case 'talkspacee':
2565                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2566                                 break;
2567                         case 'subjectspace':
2568                                 $value = $this->mTitle->getSubjectNsText();
2569                                 break;
2570                         case 'subjectspacee':
2571                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2572                                 break;
2573                         case 'currentdayname':
2574                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2575                                 break;
2576                         case 'currentyear':
2577                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2578                                 break;
2579                         case 'currenttime':
2580                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2581                                 break;
2582                         case 'currenthour':
2583                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2584                                 break;
2585                         case 'currentweek':
2586                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2587                                 // int to remove the padding
2588                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2589                                 break;
2590                         case 'currentdow':
2591                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2592                                 break;
2593                         case 'localdayname':
2594                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2595                                 break;
2596                         case 'localyear':
2597                                 $value = $wgContLang->formatNum( $localYear, true );
2598                                 break;
2599                         case 'localtime':
2600                                 $value = $wgContLang->time( $localTimestamp, false, false );
2601                                 break;
2602                         case 'localhour':
2603                                 $value = $wgContLang->formatNum( $localHour, true );
2604                                 break;
2605                         case 'localweek':
2606                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2607                                 // int to remove the padding
2608                                 $value = $wgContLang->formatNum( (int)$localWeek );
2609                                 break;
2610                         case 'localdow':
2611                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2612                                 break;
2613                         case 'numberofarticles':
2614                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2615                                 break;
2616                         case 'numberoffiles':
2617                                 $value = $wgContLang->formatNum( SiteStats::images() );
2618                                 break;
2619                         case 'numberofusers':
2620                                 $value = $wgContLang->formatNum( SiteStats::users() );
2621                                 break;
2622                         case 'numberofactiveusers':
2623                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2624                                 break;
2625                         case 'numberofpages':
2626                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2627                                 break;
2628                         case 'numberofadmins':
2629                                 $value = $wgContLang->formatNum( SiteStats::numberingroup('sysop') );
2630                                 break;
2631                         case 'numberofedits':
2632                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2633                                 break;
2634                         case 'numberofviews':
2635                                 $value = $wgContLang->formatNum( SiteStats::views() );
2636                                 break;
2637                         case 'currenttimestamp':
2638                                 $value = wfTimestamp( TS_MW, $ts );
2639                                 break;
2640                         case 'localtimestamp':
2641                                 $value = $localTimestamp;
2642                                 break;
2643                         case 'currentversion':
2644                                 $value = SpecialVersion::getVersion();
2645                                 break;
2646                         case 'sitename':
2647                                 return $wgSitename;
2648                         case 'server':
2649                                 return $wgServer;
2650                         case 'servername':
2651                                 return $wgServerName;
2652                         case 'scriptpath':
2653                                 return $wgScriptPath;
2654                         case 'stylepath':
2655                                 return $wgStylePath;
2656                         case 'directionmark':
2657                                 return $wgContLang->getDirMark();
2658                         case 'contentlanguage':
2659                                 global $wgContLanguageCode;
2660                                 return $wgContLanguageCode;
2661                         default:
2662                                 $ret = null;
2663                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) )
2664                                         return $ret;
2665                                 else
2666                                         return null;
2667                 }
2668
2669                 if ( $index )
2670                         $this->mVarCache[$index] = $value;
2671
2672                 return $value;
2673         }
2674
2675         /**
2676          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2677          *
2678          * @private
2679          */
2680         function initialiseVariables() {
2681                 wfProfileIn( __METHOD__ );
2682                 $variableIDs = MagicWord::getVariableIDs();
2683                 $substIDs = MagicWord::getSubstIDs();
2684
2685                 $this->mVariables = new MagicWordArray( $variableIDs );
2686                 $this->mSubstWords = new MagicWordArray( $substIDs );
2687                 wfProfileOut( __METHOD__ );
2688         }
2689
2690         /**
2691          * Preprocess some wikitext and return the document tree.
2692          * This is the ghost of replace_variables().
2693          *
2694          * @param string $text The text to parse
2695          * @param integer flags Bitwise combination of:
2696          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2697          *                                     included. Default is to assume a direct page view.
2698          *
2699          * The generated DOM tree must depend only on the input text and the flags.
2700          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2701          *
2702          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2703          * change in the DOM tree for a given text, must be passed through the section identifier
2704          * in the section edit link and thus back to extractSections().
2705          *
2706          * The output of this function is currently only cached in process memory, but a persistent
2707          * cache may be implemented at a later date which takes further advantage of these strict
2708          * dependency requirements.
2709          *
2710          * @private
2711          */
2712         function preprocessToDom ( $text, $flags = 0 ) {
2713                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2714                 return $dom;
2715         }
2716
2717         /*
2718          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2719          */
2720         public static function splitWhitespace( $s ) {
2721                 $ltrimmed = ltrim( $s );
2722                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2723                 $trimmed = rtrim( $ltrimmed );
2724                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2725                 if ( $diff > 0 ) {
2726                         $w2 = substr( $ltrimmed, -$diff );
2727                 } else {
2728                         $w2 = '';
2729                 }
2730                 return array( $w1, $trimmed, $w2 );
2731         }
2732
2733         /**
2734          * Replace magic variables, templates, and template arguments
2735          * with the appropriate text. Templates are substituted recursively,
2736          * taking care to avoid infinite loops.
2737          *
2738          * Note that the substitution depends on value of $mOutputType:
2739          *  self::OT_WIKI: only {{subst:}} templates
2740          *  self::OT_PREPROCESS: templates but not extension tags
2741          *  self::OT_HTML: all templates and extension tags
2742          *
2743          * @param string $tex The text to transform
2744          * @param PPFrame $frame Object describing the arguments passed to the template.
2745          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2746          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2747          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2748          * @private
2749          */
2750         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2751                 # Is there any text? Also, Prevent too big inclusions!
2752                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2753                         return $text;
2754                 }
2755                 wfProfileIn( __METHOD__ );
2756
2757                 if ( $frame === false ) {
2758                         $frame = $this->getPreprocessor()->newFrame();
2759                 } elseif ( !( $frame instanceof PPFrame ) ) {
2760                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2761                         $frame = $this->getPreprocessor()->newCustomFrame($frame);
2762                 }
2763
2764                 $dom = $this->preprocessToDom( $text );
2765                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2766                 $text = $frame->expand( $dom, $flags );
2767
2768                 wfProfileOut( __METHOD__ );
2769                 return $text;
2770         }
2771
2772         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2773         static function createAssocArgs( $args ) {
2774                 $assocArgs = array();
2775                 $index = 1;
2776                 foreach( $args as $arg ) {
2777                         $eqpos = strpos( $arg, '=' );
2778                         if ( $eqpos === false ) {
2779                                 $assocArgs[$index++] = $arg;
2780                         } else {
2781                                 $name = trim( substr( $arg, 0, $eqpos ) );
2782                                 $value = trim( substr( $arg, $eqpos+1 ) );
2783                                 if ( $value === false ) {
2784                                         $value = '';
2785                                 }
2786                                 if ( $name !== false ) {
2787                                         $assocArgs[$name] = $value;
2788                                 }
2789                         }
2790                 }
2791
2792                 return $assocArgs;
2793         }
2794
2795         /**
2796          * Warn the user when a parser limitation is reached
2797          * Will warn at most once the user per limitation type
2798          *
2799          * @param string $limitationType, should be one of:
2800          *   'expensive-parserfunction' (corresponding messages: 'expensive-parserfunction-warning', 'expensive-parserfunction-category')
2801          *   'post-expand-template-argument' (corresponding messages: 'post-expand-template-argument-warning', 'post-expand-template-argument-category')
2802          *   'post-expand-template-inclusion' (corresponding messages: 'post-expand-template-inclusion-warning', 'post-expand-template-inclusion-category')
2803          * @params int $current, $max When an explicit limit has been
2804          *       exceeded, provide the values (optional)
2805          */
2806         function limitationWarn( $limitationType, $current=null, $max=null) {
2807                 //does no harm if $current and $max are present but are unnecessary for the message
2808                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
2809                 $this->mOutput->addWarning( $warning );
2810                 $this->addTrackingCategory( "$limitationType-category" );
2811         }
2812
2813         /**
2814          * Return the text of a template, after recursively
2815          * replacing any variables or templates within the template.
2816          *
2817          * @param array $piece The parts of the template
2818          *  $piece['title']: the title, i.e. the part before the |
2819          *  $piece['parts']: the parameter array
2820          *  $piece['lineStart']: whether the brace was at the start of a line
2821          * @param PPFrame The current frame, contains template arguments
2822          * @return string the text of the template
2823          * @private
2824          */
2825         function braceSubstitution( $piece, $frame ) {
2826                 global $wgContLang, $wgNonincludableNamespaces;
2827                 wfProfileIn( __METHOD__ );
2828                 wfProfileIn( __METHOD__.'-setup' );
2829
2830                 # Flags
2831                 $found = false;             # $text has been filled
2832                 $nowiki = false;            # wiki markup in $text should be escaped
2833                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2834                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2835                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2836                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2837
2838                 # Title object, where $text came from
2839                 $title = null;
2840
2841                 # $part1 is the bit before the first |, and must contain only title characters.
2842                 # Various prefixes will be stripped from it later.
2843                 $titleWithSpaces = $frame->expand( $piece['title'] );
2844                 $part1 = trim( $titleWithSpaces );
2845                 $titleText = false;
2846
2847                 # Original title text preserved for various purposes
2848                 $originalTitle = $part1;
2849
2850                 # $args is a list of argument nodes, starting from index 0, not including $part1
2851                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2852                 wfProfileOut( __METHOD__.'-setup' );
2853
2854                 # SUBST
2855                 wfProfileIn( __METHOD__.'-modifiers' );
2856                 if ( !$found ) {
2857
2858                         $substMatch = $this->mSubstWords->matchStartAndRemove( $part1 );
2859
2860                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
2861                         # Decide whether to expand template or keep wikitext as-is.
2862                         if ( $this->ot['wiki'] )
2863                         {
2864                                 if ( $substMatch === false ) {
2865                                         $literal = true;  # literal when in PST with no prefix
2866                                 } else {
2867                                         $literal = false; # expand when in PST with subst: or safesubst:
2868                                 }
2869                         } else {
2870                                 if ( $substMatch == 'subst' ) {
2871                                         $literal = true;  # literal when not in PST with plain subst:
2872                                 } else {
2873                                         $literal = false; # expand when not in PST with safesubst: or no prefix
2874                                 }
2875                         }
2876                         if ( $literal ) {
2877                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2878                                 $isLocalObj = true;
2879                                 $found = true;
2880                         }
2881                 }
2882
2883                 # Variables
2884                 if ( !$found && $args->getLength() == 0 ) {
2885                         $id = $this->mVariables->matchStartToEnd( $part1 );
2886                         if ( $id !== false ) {
2887                                 $text = $this->getVariableValue( $id, $frame );
2888                                 if (MagicWord::getCacheTTL($id)>-1)
2889                                         $this->mOutput->mContainsOldMagic = true;
2890                                 $found = true;
2891                         }
2892                 }
2893
2894                 # MSG, MSGNW and RAW
2895                 if ( !$found ) {
2896                         # Check for MSGNW:
2897                         $mwMsgnw = MagicWord::get( 'msgnw' );
2898                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2899                                 $nowiki = true;
2900                         } else {
2901                                 # Remove obsolete MSG:
2902                                 $mwMsg = MagicWord::get( 'msg' );
2903                                 $mwMsg->matchStartAndRemove( $part1 );
2904                         }
2905
2906                         # Check for RAW:
2907                         $mwRaw = MagicWord::get( 'raw' );
2908                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2909                                 $forceRawInterwiki = true;
2910                         }
2911                 }
2912                 wfProfileOut( __METHOD__.'-modifiers' );
2913
2914                 # Parser functions
2915                 if ( !$found ) {
2916                         wfProfileIn( __METHOD__ . '-pfunc' );
2917
2918                         $colonPos = strpos( $part1, ':' );
2919                         if ( $colonPos !== false ) {
2920                                 # Case sensitive functions
2921                                 $function = substr( $part1, 0, $colonPos );
2922                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2923                                         $function = $this->mFunctionSynonyms[1][$function];
2924                                 } else {
2925                                         # Case insensitive functions
2926                                         $function = $wgContLang->lc( $function );
2927                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2928                                                 $function = $this->mFunctionSynonyms[0][$function];
2929                                         } else {
2930                                                 $function = false;
2931                                         }
2932                                 }
2933                                 if ( $function ) {
2934                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2935                                         $initialArgs = array( &$this );
2936                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2937                                         if ( $flags & SFH_OBJECT_ARGS ) {
2938                                                 # Add a frame parameter, and pass the arguments as an array
2939                                                 $allArgs = $initialArgs;
2940                                                 $allArgs[] = $frame;
2941                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2942                                                         $funcArgs[] = $args->item( $i );
2943                                                 }
2944                                                 $allArgs[] = $funcArgs;
2945                                         } else {
2946                                                 # Convert arguments to plain text
2947                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2948                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2949                                                 }
2950                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2951                                         }
2952
2953                                         # Workaround for PHP bug 35229 and similar
2954                                         if ( !is_callable( $callback ) ) {
2955                                                 wfProfileOut( __METHOD__ . '-pfunc' );
2956                                                 wfProfileOut( __METHOD__ );
2957                                                 throw new MWException( "Tag hook for $function is not callable\n" );
2958                                         }
2959                                         $result = call_user_func_array( $callback, $allArgs );
2960                                         $found = true;
2961                                         $noparse = true;
2962                                         $preprocessFlags = 0;
2963
2964                                         if ( is_array( $result ) ) {
2965                                                 if ( isset( $result[0] ) ) {
2966                                                         $text = $result[0];
2967                                                         unset( $result[0] );
2968                                                 }
2969
2970                                                 // Extract flags into the local scope
2971                                                 // This allows callers to set flags such as nowiki, found, etc.
2972                                                 extract( $result );
2973                                         } else {
2974                                                 $text = $result;
2975                                         }
2976                                         if ( !$noparse ) {
2977                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
2978                                                 $isChildObj = true;
2979                                         }
2980                                 }
2981                         }
2982                         wfProfileOut( __METHOD__ . '-pfunc' );
2983                 }
2984
2985                 # Finish mangling title and then check for loops.
2986                 # Set $title to a Title object and $titleText to the PDBK
2987                 if ( !$found ) {
2988                         $ns = NS_TEMPLATE;
2989                         # Split the title into page and subpage
2990                         $subpage = '';
2991                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2992                         if ($subpage !== '') {
2993                                 $ns = $this->mTitle->getNamespace();
2994                         }
2995                         $title = Title::newFromText( $part1, $ns );
2996                         if ( $title ) {
2997                                 $titleText = $title->getPrefixedText();
2998                                 # Check for language variants if the template is not found
2999                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
3000                                         $wgContLang->findVariantLink( $part1, $title, true );
3001                                 }
3002                                 # Do recursion depth check
3003                                 $limit = $this->mOptions->getMaxTemplateDepth();
3004                                 if ( $frame->depth >= $limit ) {
3005                                         $found = true;
3006                                         $text = '<span class="error">' . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit ) . '</span>';
3007                                 }
3008                         }
3009                 }
3010
3011                 # Load from database
3012                 if ( !$found && $title ) {
3013                         wfProfileIn( __METHOD__ . '-loadtpl' );
3014                         if ( !$title->isExternal() ) {
3015                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
3016                                         $text = SpecialPage::capturePath( $title );
3017                                         if ( is_string( $text ) ) {
3018                                                 $found = true;
3019                                                 $isHTML = true;
3020                                                 $this->disableCache();
3021                                         }
3022                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3023                                         $found = false; //access denied
3024                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
3025                                 } else {
3026                                         list( $text, $title ) = $this->getTemplateDom( $title );
3027                                         if ( $text !== false ) {
3028                                                 $found = true;
3029                                                 $isChildObj = true;
3030                                         }
3031                                 }
3032
3033                                 # If the title is valid but undisplayable, make a link to it
3034                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3035                                         $text = "[[:$titleText]]";
3036                                         $found = true;
3037                                 }
3038                         } elseif ( $title->isTrans() ) {
3039                                 // Interwiki transclusion
3040                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
3041                                         $text = $this->interwikiTransclude( $title, 'render' );
3042                                         $isHTML = true;
3043                                 } else {
3044                                         $text = $this->interwikiTransclude( $title, 'raw' );
3045                                         // Preprocess it like a template
3046                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3047                                         $isChildObj = true;
3048                                 }
3049                                 $found = true;
3050                         }
3051
3052                         # Do infinite loop check
3053                         # This has to be done after redirect resolution to avoid infinite loops via redirects
3054                         if ( !$frame->loopCheck( $title ) ) {
3055                                 $found = true;
3056                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
3057                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
3058                         }
3059                         wfProfileOut( __METHOD__ . '-loadtpl' );
3060                 }
3061
3062                 # If we haven't found text to substitute by now, we're done
3063                 # Recover the source wikitext and return it
3064                 if ( !$found ) {
3065                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3066                         wfProfileOut( __METHOD__ );
3067                         return array( 'object' => $text );
3068                 }
3069
3070                 # Expand DOM-style return values in a child frame
3071                 if ( $isChildObj ) {
3072                         # Clean up argument array
3073                         $newFrame = $frame->newChild( $args, $title );
3074
3075                         if ( $nowiki ) {
3076                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3077                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3078                                 # Expansion is eligible for the empty-frame cache
3079                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3080                                         $text = $this->mTplExpandCache[$titleText];
3081                                 } else {
3082                                         $text = $newFrame->expand( $text );
3083                                         $this->mTplExpandCache[$titleText] = $text;
3084                                 }
3085                         } else {
3086                                 # Uncached expansion
3087                                 $text = $newFrame->expand( $text );
3088                         }
3089                 }
3090                 if ( $isLocalObj && $nowiki ) {
3091                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3092                         $isLocalObj = false;
3093                 }
3094
3095                 # Replace raw HTML by a placeholder
3096                 # Add a blank line preceding, to prevent it from mucking up
3097                 # immediately preceding headings
3098                 if ( $isHTML ) {
3099                         $text = "\n\n" . $this->insertStripItem( $text );
3100                 }
3101                 # Escape nowiki-style return values
3102                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3103                         $text = wfEscapeWikiText( $text );
3104                 }
3105                 # Bug 529: if the template begins with a table or block-level
3106                 # element, it should be treated as beginning a new line.
3107                 # This behaviour is somewhat controversial.
3108                 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3109                         $text = "\n" . $text;
3110                 }
3111
3112                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3113                         # Error, oversize inclusion
3114                         $text = "[[$originalTitle]]" .
3115                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3116                         $this->limitationWarn( 'post-expand-template-inclusion' );
3117                 }
3118
3119                 if ( $isLocalObj ) {
3120                         $ret = array( 'object' => $text );
3121                 } else {
3122                         $ret = array( 'text' => $text );
3123                 }
3124
3125                 wfProfileOut( __METHOD__ );
3126                 return $ret;
3127         }
3128
3129         /**
3130          * Get the semi-parsed DOM representation of a template with a given title,
3131          * and its redirect destination title. Cached.
3132          */
3133         function getTemplateDom( $title ) {
3134                 $cacheTitle = $title;
3135                 $titleText = $title->getPrefixedDBkey();
3136
3137                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3138                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3139                         $title = Title::makeTitle( $ns, $dbk );
3140                         $titleText = $title->getPrefixedDBkey();
3141                 }
3142                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3143                         return array( $this->mTplDomCache[$titleText], $title );
3144                 }
3145
3146                 // Cache miss, go to the database
3147                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3148
3149                 if ( $text === false ) {
3150                         $this->mTplDomCache[$titleText] = false;
3151                         return array( false, $title );
3152                 }
3153
3154                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3155                 $this->mTplDomCache[ $titleText ] = $dom;
3156
3157                 if (! $title->equals($cacheTitle)) {
3158                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3159                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3160                 }
3161
3162                 return array( $dom, $title );
3163         }
3164
3165         /**
3166          * Fetch the unparsed text of a template and register a reference to it.
3167          */
3168         function fetchTemplateAndTitle( $title ) {
3169                 $templateCb = $this->mOptions->getTemplateCallback();
3170                 $stuff = call_user_func( $templateCb, $title, $this );
3171                 $text = $stuff['text'];
3172                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3173                 if ( isset( $stuff['deps'] ) ) {
3174                         foreach ( $stuff['deps'] as $dep ) {
3175                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3176                         }
3177                 }
3178                 return array($text,$finalTitle);
3179         }
3180
3181         function fetchTemplate( $title ) {
3182                 $rv = $this->fetchTemplateAndTitle($title);
3183                 return $rv[0];
3184         }
3185
3186         /**
3187          * Static function to get a template
3188          * Can be overridden via ParserOptions::setTemplateCallback().
3189          */
3190         static function statelessFetchTemplate( $title, $parser=false ) {
3191                 $text = $skip = false;
3192                 $finalTitle = $title;
3193                 $deps = array();
3194
3195                 // Loop to fetch the article, with up to 1 redirect
3196                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3197                         # Give extensions a chance to select the revision instead
3198                         $id = false; // Assume current
3199                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3200
3201                         if( $skip ) {
3202                                 $text = false;
3203                                 $deps[] = array(
3204                                         'title' => $title,
3205                                         'page_id' => $title->getArticleID(),
3206                                         'rev_id' => null );
3207                                 break;
3208                         }
3209                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3210                         $rev_id = $rev ? $rev->getId() : 0;
3211                         // If there is no current revision, there is no page
3212                         if( $id === false && !$rev ) {
3213                                 $linkCache = LinkCache::singleton();
3214                                 $linkCache->addBadLinkObj( $title );
3215                         }
3216
3217                         $deps[] = array(
3218                                 'title' => $title,
3219                                 'page_id' => $title->getArticleID(),
3220                                 'rev_id' => $rev_id );
3221
3222                         if( $rev ) {
3223                                 $text = $rev->getText();
3224                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3225                                 global $wgContLang;
3226                                 $message = $wgContLang->lcfirst( $title->getText() );
3227                                 $text = wfMsgForContentNoTrans( $message );
3228                                 if( wfEmptyMsg( $message, $text ) ) {
3229                                         $text = false;
3230                                         break;
3231                                 }
3232                         } else {
3233                                 break;
3234                         }
3235                         if ( $text === false ) {
3236                                 break;
3237                         }
3238                         // Redirect?
3239                         $finalTitle = $title;
3240                         $title = Title::newFromRedirect( $text );
3241                 }
3242                 return array(
3243                         'text' => $text,
3244                         'finalTitle' => $finalTitle,
3245                         'deps' => $deps );
3246         }
3247
3248         /**
3249          * Transclude an interwiki link.
3250          */
3251         function interwikiTransclude( $title, $action ) {
3252                 global $wgEnableScaryTranscluding;
3253
3254                 if (!$wgEnableScaryTranscluding)
3255                         return wfMsg('scarytranscludedisabled');
3256
3257                 $url = $title->getFullUrl( "action=$action" );
3258
3259                 if (strlen($url) > 255)
3260                         return wfMsg('scarytranscludetoolong');
3261                 return $this->fetchScaryTemplateMaybeFromCache($url);
3262         }
3263
3264         function fetchScaryTemplateMaybeFromCache($url) {
3265                 global $wgTranscludeCacheExpiry;
3266                 $dbr = wfGetDB(DB_SLAVE);
3267                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3268                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3269                                 array('tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3270                 if ($obj) {
3271                         return $obj->tc_contents;
3272                 }
3273
3274                 $text = Http::get($url);
3275                 if (!$text)
3276                         return wfMsg('scarytranscludefailed', $url);
3277
3278                 $dbw = wfGetDB(DB_MASTER);
3279                 $dbw->replace('transcache', array('tc_url'), array(
3280                         'tc_url' => $url,
3281                         'tc_time' => $dbw->timestamp( time() ),
3282                         'tc_contents' => $text));
3283                 return $text;
3284         }
3285
3286
3287         /**
3288          * Triple brace replacement -- used for template arguments
3289          * @private
3290          */
3291         function argSubstitution( $piece, $frame ) {
3292                 wfProfileIn( __METHOD__ );
3293
3294                 $error = false;
3295                 $parts = $piece['parts'];
3296                 $nameWithSpaces = $frame->expand( $piece['title'] );
3297                 $argName = trim( $nameWithSpaces );
3298                 $object = false;
3299                 $text = $frame->getArgument( $argName );
3300                 if (  $text === false && $parts->getLength() > 0
3301                   && (
3302                     $this->ot['html']
3303                     || $this->ot['pre']
3304                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3305                   )
3306                 ) {
3307                         # No match in frame, use the supplied default
3308                         $object = $parts->item( 0 )->getChildren();
3309                 }
3310                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3311                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3312                         $this->limitationWarn( 'post-expand-template-argument' );
3313                 }
3314
3315                 if ( $text === false && $object === false ) {
3316                         # No match anywhere
3317                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3318                 }
3319                 if ( $error !== false ) {
3320                         $text .= $error;
3321                 }
3322                 if ( $object !== false ) {
3323                         $ret = array( 'object' => $object );
3324                 } else {
3325                         $ret = array( 'text' => $text );
3326                 }
3327
3328                 wfProfileOut( __METHOD__ );
3329                 return $ret;
3330         }
3331
3332         /**
3333          * Return the text to be used for a given extension tag.
3334          * This is the ghost of strip().
3335          *
3336          * @param array $params Associative array of parameters:
3337          *     name       PPNode for the tag name
3338          *     attr       PPNode for unparsed text where tag attributes are thought to be
3339          *     attributes Optional associative array of parsed attributes
3340          *     inner      Contents of extension element
3341          *     noClose    Original text did not have a close tag
3342          * @param PPFrame $frame
3343          */
3344         function extensionSubstitution( $params, $frame ) {
3345                 global $wgRawHtml, $wgContLang;
3346
3347                 $name = $frame->expand( $params['name'] );
3348                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3349                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3350                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $this->mMarkerIndex++) . self::MARKER_SUFFIX;
3351
3352                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3353                         ( $this->ot['html'] || $this->ot['pre'] );
3354                 if ( $isFunctionTag ) {
3355                         $markerType = 'none';
3356                 } else {
3357                         $markerType = 'general';
3358                 }
3359                 if ( $this->ot['html'] || $isFunctionTag ) {
3360                         $name = strtolower( $name );
3361                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3362                         if ( isset( $params['attributes'] ) ) {
3363                                 $attributes = $attributes + $params['attributes'];
3364                         }
3365
3366                         if( isset( $this->mTagHooks[$name] ) ) {
3367                                 # Workaround for PHP bug 35229 and similar
3368                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3369                                         throw new MWException( "Tag hook for $name is not callable\n" );
3370                                 }
3371                                 $output = call_user_func_array( $this->mTagHooks[$name],
3372                                         array( $content, $attributes, $this, $frame ) );
3373                         } elseif( isset( $this->mFunctionTagHooks[$name] ) ) {
3374                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3375                                 if( !is_callable( $callback ) )
3376                                         throw new MWException( "Tag hook for $name is not callable\n" );
3377
3378                                 $output = call_user_func_array( $callback,
3379                                         array( &$this, $frame, $content, $attributes ) );
3380                         } else {
3381                                 $output = '<span class="error">Invalid tag extension name: ' .
3382                                         htmlspecialchars( $name ) . '</span>';
3383                         }
3384
3385                         if ( is_array( $output ) ) {
3386                                 // Extract flags to local scope (to override $markerType)
3387                                 $flags = $output;
3388                                 $output = $flags[0];
3389                                 unset( $flags[0] );
3390                                 extract( $flags );
3391                         }
3392                 } else {
3393                         if ( is_null( $attrText ) ) {
3394                                 $attrText = '';
3395                         }
3396                         if ( isset( $params['attributes'] ) ) {
3397                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3398                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3399                                                 htmlspecialchars( $attrValue ) . '"';
3400                                 }
3401                         }
3402                         if ( $content === null ) {
3403                                 $output = "<$name$attrText/>";
3404                         } else {
3405                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3406                                 $output = "<$name$attrText>$content$close";
3407                         }
3408                 }
3409
3410                 if( $markerType === 'none' ) {
3411                         return $output;
3412                 } elseif ( $markerType === 'nowiki' ) {
3413                         $this->mStripState->nowiki->setPair( $marker, $output );
3414                 } elseif ( $markerType === 'general' ) {
3415                         $this->mStripState->general->setPair( $marker, $output );
3416                 } else {
3417                         throw new MWException( __METHOD__.': invalid marker type' );
3418                 }
3419                 return $marker;
3420         }
3421
3422         /**
3423          * Increment an include size counter
3424          *
3425          * @param string $type The type of expansion
3426          * @param integer $size The size of the text
3427          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3428          */
3429         function incrementIncludeSize( $type, $size ) {
3430                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3431                         return false;
3432                 } else {
3433                         $this->mIncludeSizes[$type] += $size;
3434                         return true;
3435                 }
3436         }
3437
3438         /**
3439          * Increment the expensive function count
3440          *
3441          * @return boolean False if the limit has been exceeded
3442          */
3443         function incrementExpensiveFunctionCount() {
3444                 global $wgExpensiveParserFunctionLimit;
3445                 $this->mExpensiveFunctionCount++;
3446                 if($this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit) {
3447                         return true;
3448                 }
3449                 return false;
3450         }
3451
3452         /**
3453          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3454          * Fills $this->mDoubleUnderscores, returns the modified text
3455          */
3456         function doDoubleUnderscore( $text ) {
3457                 wfProfileIn( __METHOD__ );
3458
3459                 // The position of __TOC__ needs to be recorded
3460                 $mw = MagicWord::get( 'toc' );
3461                 if( $mw->match( $text ) ) {
3462                         $this->mShowToc = true;
3463                         $this->mForceTocPosition = true;
3464
3465                         // Set a placeholder. At the end we'll fill it in with the TOC.
3466                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3467
3468                         // Only keep the first one.
3469                         $text = $mw->replace( '', $text );
3470                 }
3471
3472                 // Now match and remove the rest of them
3473                 $mwa = MagicWord::getDoubleUnderscoreArray();
3474                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3475
3476                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3477                         $this->mOutput->mNoGallery = true;
3478                 }
3479                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3480                         $this->mShowToc = false;
3481                 }
3482                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3483                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3484                         $this->addTrackingCategory( 'hidden-category-category' );
3485                 }
3486                 # (bug 8068) Allow control over whether robots index a page.
3487                 #
3488                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3489                 # is not desirable, the last one on the page should win.
3490                 if( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3491                         $this->mOutput->setIndexPolicy( 'noindex' );
3492                         $this->addTrackingCategory( 'noindex-category' );
3493                 }
3494                 if( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ){
3495                         $this->mOutput->setIndexPolicy( 'index' );
3496                         $this->addTrackingCategory( 'index-category' );
3497                 }
3498
3499                 wfProfileOut( __METHOD__ );
3500                 return $text;
3501         }
3502
3503         /**
3504          * Add a tracking category, getting the title from a system message,
3505          * or print a debug message if the title is invalid.
3506          * @param $msg String message key
3507          * @return Bool whether the addition was successful
3508          */
3509         protected function addTrackingCategory( $msg ){
3510                 $cat = wfMsgForContent( $msg );
3511
3512                 # Allow tracking categories to be disabled by setting them to "-"
3513                 if( $cat === '-' ) return false;
3514
3515                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3516                 if ( $containerCategory ) {
3517                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3518                         return true;
3519                 } else {
3520                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3521                         return false;
3522                 }
3523         }
3524
3525         /**
3526          * This function accomplishes several tasks:
3527          * 1) Auto-number headings if that option is enabled
3528          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3529          * 3) Add a Table of contents on the top for users who have enabled the option
3530          * 4) Auto-anchor headings
3531          *
3532          * It loops through all headlines, collects the necessary data, then splits up the
3533          * string and re-inserts the newly formatted headlines.
3534          *
3535          * @param string $text
3536          * @param string $origText Original, untouched wikitext
3537          * @param boolean $isMain
3538          * @private
3539          */
3540         function formatHeadings( $text, $origText, $isMain=true ) {
3541                 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3542
3543                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3544                 $showEditLink = $this->mOptions->getEditSection();
3545
3546                 // Do not call quickUserCan unless necessary
3547                 if( $showEditLink && !$this->mTitle->quickUserCan( 'edit' ) ) {
3548                         $showEditLink = 0;
3549                 }
3550
3551                 # Inhibit editsection links if requested in the page
3552                 if ( isset( $this->mDoubleUnderscores['noeditsection'] )  || $this->mOptions->getIsPrintable() ) {
3553                         $showEditLink = 0;
3554                 }
3555
3556                 # Get all headlines for numbering them and adding funky stuff like [edit]
3557                 # links - this is for later, but we need the number of headlines right now
3558                 $matches = array();
3559                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3560
3561                 # if there are fewer than 4 headlines in the article, do not show TOC
3562                 # unless it's been explicitly enabled.
3563                 $enoughToc = $this->mShowToc &&
3564                         (($numMatches >= 4) || $this->mForceTocPosition);
3565
3566                 # Allow user to stipulate that a page should have a "new section"
3567                 # link added via __NEWSECTIONLINK__
3568                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3569                         $this->mOutput->setNewSection( true );
3570                 }
3571
3572                 # Allow user to remove the "new section"
3573                 # link via __NONEWSECTIONLINK__
3574                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3575                         $this->mOutput->hideNewSection( true );
3576                 }
3577
3578                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3579                 # override above conditions and always show TOC above first header
3580                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3581                         $this->mShowToc = true;
3582                         $enoughToc = true;
3583                 }
3584
3585                 # We need this to perform operations on the HTML
3586                 $sk = $this->mOptions->getSkin();
3587
3588                 # headline counter
3589                 $headlineCount = 0;
3590                 $numVisible = 0;
3591
3592                 # Ugh .. the TOC should have neat indentation levels which can be
3593                 # passed to the skin functions. These are determined here
3594                 $toc = '';
3595                 $full = '';
3596                 $head = array();
3597                 $sublevelCount = array();
3598                 $levelCount = array();
3599                 $toclevel = 0;
3600                 $level = 0;
3601                 $prevlevel = 0;
3602                 $toclevel = 0;
3603                 $prevtoclevel = 0;
3604                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3605                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3606                 $oldType = $this->mOutputType;
3607                 $this->setOutputType( self::OT_WIKI );
3608                 $frame = $this->getPreprocessor()->newFrame();
3609                 $root = $this->preprocessToDom( $origText );
3610                 $node = $root->getFirstChild();
3611                 $byteOffset = 0;
3612                 $tocraw = array();
3613
3614                 foreach( $matches[3] as $headline ) {
3615                         $isTemplate = false;
3616                         $titleText = false;
3617                         $sectionIndex = false;
3618                         $numbering = '';
3619                         $markerMatches = array();
3620                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3621                                 $serial = $markerMatches[1];
3622                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3623                                 $isTemplate = ($titleText != $baseTitleText);
3624                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3625                         }
3626
3627                         if( $toclevel ) {
3628                                 $prevlevel = $level;
3629                                 $prevtoclevel = $toclevel;
3630                         }
3631                         $level = $matches[1][$headlineCount];
3632
3633                         if ( $level > $prevlevel ) {
3634                                 # Increase TOC level
3635                                 $toclevel++;
3636                                 $sublevelCount[$toclevel] = 0;
3637                                 if( $toclevel<$wgMaxTocLevel ) {
3638                                         $prevtoclevel = $toclevel;
3639                                         $toc .= $sk->tocIndent();
3640                                         $numVisible++;
3641                                 }
3642                         }
3643                         elseif ( $level < $prevlevel && $toclevel > 1 ) {
3644                                 # Decrease TOC level, find level to jump to
3645
3646                                 for ($i = $toclevel; $i > 0; $i--) {
3647                                         if ( $levelCount[$i] == $level ) {
3648                                                 # Found last matching level
3649                                                 $toclevel = $i;
3650                                                 break;
3651                                         }
3652                                         elseif ( $levelCount[$i] < $level ) {
3653                                                 # Found first matching level below current level
3654                                                 $toclevel = $i + 1;
3655                                                 break;
3656                                         }
3657                                 }
3658                                 if( $i == 0 ) $toclevel = 1;
3659                                 if( $toclevel<$wgMaxTocLevel ) {
3660                                         if($prevtoclevel < $wgMaxTocLevel) {
3661                                                 # Unindent only if the previous toc level was shown :p
3662                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3663                                                 $prevtoclevel = $toclevel;
3664                                         } else {
3665                                                 $toc .= $sk->tocLineEnd();
3666                                         }
3667                                 }
3668                         }
3669                         else {
3670                                 # No change in level, end TOC line
3671                                 if( $toclevel<$wgMaxTocLevel ) {
3672                                         $toc .= $sk->tocLineEnd();
3673                                 }
3674                         }
3675
3676                         $levelCount[$toclevel] = $level;
3677
3678                         # count number of headlines for each level
3679                         @$sublevelCount[$toclevel]++;
3680                         $dot = 0;
3681                         for( $i = 1; $i <= $toclevel; $i++ ) {
3682                                 if( !empty( $sublevelCount[$i] ) ) {
3683                                         if( $dot ) {
3684                                                 $numbering .= '.';
3685                                         }
3686                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3687                                         $dot = 1;
3688                                 }
3689                         }
3690
3691                         # The safe header is a version of the header text safe to use for links
3692                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3693                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3694
3695                         # Remove link placeholders by the link text.
3696                         #     <!--LINK number-->
3697                         # turns into
3698                         #     link text with suffix
3699                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3700
3701                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3702                         $tocline = preg_replace(
3703                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3704                                 array( '',                          '<$1>'),
3705                                 $safeHeadline
3706                         );
3707                         $tocline = trim( $tocline );
3708
3709                         # For the anchor, strip out HTML-y stuff period
3710                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3711                         $safeHeadline = preg_replace( '/[ _]+/', ' ', $safeHeadline );
3712                         $safeHeadline = trim( $safeHeadline );
3713
3714                         # Save headline for section edit hint before it's escaped
3715                         $headlineHint = $safeHeadline;
3716
3717                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
3718                                 # For reverse compatibility, provide an id that's
3719                                 # HTML4-compatible, like we used to.
3720                                 #
3721                                 # It may be worth noting, academically, that it's possible for
3722                                 # the legacy anchor to conflict with a non-legacy headline
3723                                 # anchor on the page.  In this case likely the "correct" thing
3724                                 # would be to either drop the legacy anchors or make sure
3725                                 # they're numbered first.  However, this would require people
3726                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3727                                 # manually, so let's not bother worrying about it.
3728                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3729                                         array( 'noninitial', 'legacy' ) );
3730                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3731
3732                                 if ( $legacyHeadline == $safeHeadline ) {
3733                                         # No reason to have both (in fact, we can't)
3734                                         $legacyHeadline = false;
3735                                 }
3736                         } else {
3737                                 $legacyHeadline = false;
3738                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3739                                         'noninitial' );
3740                         }
3741
3742                         # HTML names must be case-insensitively unique (bug 10721).  FIXME:
3743                         # Does this apply to Unicode characters?  Because we aren't
3744                         # handling those here.
3745                         $arrayKey = strtolower( $safeHeadline );
3746                         if ( $legacyHeadline === false ) {
3747                                 $legacyArrayKey = false;
3748                         } else {
3749                                 $legacyArrayKey = strtolower( $legacyHeadline );
3750                         }
3751
3752                         # count how many in assoc. array so we can track dupes in anchors
3753                         if ( isset( $refers[$arrayKey] ) ) {
3754                                 $refers[$arrayKey]++;
3755                         } else {
3756                                 $refers[$arrayKey] = 1;
3757                         }
3758                         if ( isset( $refers[$legacyArrayKey] ) ) {
3759                                 $refers[$legacyArrayKey]++;
3760                         } else {
3761                                 $refers[$legacyArrayKey] = 1;
3762                         }
3763
3764                         # Don't number the heading if it is the only one (looks silly)
3765                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3766                                 # the two are different if the line contains a link
3767                                 $headline = $numbering . ' ' . $headline;
3768                         }
3769
3770                         # Create the anchor for linking from the TOC to the section
3771                         $anchor = $safeHeadline;
3772                         $legacyAnchor = $legacyHeadline;
3773                         if ( $refers[$arrayKey] > 1 ) {
3774                                 $anchor .= '_' . $refers[$arrayKey];
3775                         }
3776                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3777                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3778                         }
3779                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3780                                 $toc .= $sk->tocLine($anchor, $tocline,
3781                                         $numbering, $toclevel, ($isTemplate ? false : $sectionIndex));
3782                         }
3783
3784                         # Add the section to the section tree
3785                         # Find the DOM node for this header
3786                         while ( $node && !$isTemplate ) {
3787                                 if ( $node->getName() === 'h' ) {
3788                                         $bits = $node->splitHeading();
3789                                         if ( $bits['i'] == $sectionIndex )
3790                                                 break;
3791                                 }
3792                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
3793                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
3794                                 $node = $node->getNextSibling();
3795                         }
3796                         $tocraw[] = array(
3797                                 'toclevel' => $toclevel,
3798                                 'level' => $level,
3799                                 'line' => $tocline,
3800                                 'number' => $numbering,
3801                                 'index' => ($isTemplate ? 'T-' : '' ) . $sectionIndex,
3802                                 'fromtitle' => $titleText,
3803                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
3804                                 'anchor' => $anchor,
3805                         );
3806
3807                         # give headline the correct <h#> tag
3808                         if( $showEditLink && $sectionIndex !== false ) {
3809                                 if( $isTemplate ) {
3810                                         # Put a T flag in the section identifier, to indicate to extractSections()
3811                                         # that sections inside <includeonly> should be counted.
3812                                         $editlink = $sk->doEditSectionLink(Title::newFromText( $titleText ), "T-$sectionIndex");
3813                                 } else {
3814                                         $editlink = $sk->doEditSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3815                                 }
3816                         } else {
3817                                 $editlink = '';
3818                         }
3819                         $head[$headlineCount] = $sk->makeHeadline( $level,
3820                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3821                                 $editlink, $legacyAnchor );
3822
3823                         $headlineCount++;
3824                 }
3825
3826                 $this->setOutputType( $oldType );
3827
3828                 # Never ever show TOC if no headers
3829                 if( $numVisible < 1 ) {
3830                         $enoughToc = false;
3831                 }
3832
3833                 if( $enoughToc ) {
3834                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3835                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3836                         }
3837                         $toc = $sk->tocList( $toc );
3838                         $this->mOutput->setTOCHTML( $toc );
3839                 }
3840
3841                 if ( $isMain ) {
3842                         $this->mOutput->setSections( $tocraw );
3843                 }
3844
3845                 # split up and insert constructed headlines
3846
3847                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3848                 $i = 0;
3849
3850                 foreach( $blocks as $block ) {
3851                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3852                                 # This is the [edit] link that appears for the top block of text when
3853                                 # section editing is enabled
3854
3855                                 # Disabled because it broke block formatting
3856                                 # For example, a bullet point in the top line
3857                                 # $full .= $sk->editSectionLink(0);
3858                         }
3859                         $full .= $block;
3860                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3861                                 # Top anchor now in skin
3862                                 $full = $full.$toc;
3863                         }
3864
3865                         if( !empty( $head[$i] ) ) {
3866                                 $full .= $head[$i];
3867                         }
3868                         $i++;
3869                 }
3870                 if( $this->mForceTocPosition ) {
3871                         return str_replace( '<!--MWTOC-->', $toc, $full );
3872                 } else {
3873                         return $full;
3874                 }
3875         }
3876
3877         /**
3878          * Merge $tree2 into $tree1 by replacing the section with index
3879          * $section in $tree1 and its descendants with the sections in $tree2.
3880          * Note that in the returned section tree, only the 'index' and
3881          * 'byteoffset' fields are guaranteed to be correct.
3882          * @param $tree1 array Section tree from ParserOutput::getSectons()
3883         *  @param $tree2 array Section tree
3884          * @param $section int Section index
3885          * @param $title Title Title both section trees come from
3886          * @param $len2 int Length of the original wikitext for $tree2
3887          * @return array Merged section tree
3888          */
3889         public static function mergeSectionTrees( $tree1, $tree2, $section, $title, $len2 ) {
3890                 global $wgContLang;
3891                 $newTree = array();
3892                 $targetLevel = false;
3893                 $merged = false;
3894                 $lastLevel = 1;
3895                 $nextIndex = 1;
3896                 $numbering = array( 0 );
3897                 $titletext = $title->getPrefixedDBkey();
3898                 foreach ( $tree1 as $s ) {
3899                         if ( $targetLevel !== false ) {
3900                                 if ( $s['level'] <= $targetLevel )
3901                                         // We've skipped enough
3902                                         $targetLevel = false;
3903                                 else
3904                                         continue;
3905                         }
3906                         if ( $s['index'] != $section ||
3907                                         $s['fromtitle'] != $titletext ) {
3908                                 self::incrementNumbering( $numbering,
3909                                         $s['toclevel'], $lastLevel );
3910
3911                                 // Rewrite index, byteoffset and number
3912                                 if ( $s['fromtitle'] == $titletext ) {
3913                                         $s['index'] = $nextIndex++;
3914                                         if ( $merged )
3915                                                 $s['byteoffset'] += $len2;
3916                                 }
3917                                 $s['number']  = implode( '.', array_map(
3918                                         array( $wgContLang, 'formatnum' ),
3919                                         $numbering ) );
3920                                 $lastLevel = $s['toclevel'];
3921                                 $newTree[] = $s;
3922                         } else {
3923                                 // We're at $section
3924                                 // Insert sections from $tree2 here
3925                                 foreach ( $tree2 as $s2 ) {
3926                                         // Rewrite the fields in $s2
3927                                         // before inserting it
3928                                         $s2['toclevel'] += $s['toclevel'] - 1;
3929                                         $s2['level'] += $s['level'] - 1;
3930                                         $s2['index'] = $nextIndex++;
3931                                         $s2['byteoffset'] += $s['byteoffset'];
3932
3933                                         self::incrementNumbering( $numbering,
3934                                                 $s2['toclevel'], $lastLevel );
3935                                         $s2['number']  = implode( '.', array_map(
3936                                                 array( $wgContLang, 'formatnum' ),
3937                                                 $numbering ) );
3938                                         $lastLevel = $s2['toclevel'];
3939                                         $newTree[] = $s2;
3940                                 }
3941                                 // Skip all descendants of $section in $tree1
3942                                 $targetLevel = $s['level'];
3943                                 $merged = true;
3944                         }
3945                 }
3946                 return $newTree;
3947         }
3948
3949         /**
3950          * Increment a section number. Helper function for mergeSectionTrees()
3951          * @param $number array Array representing a section number
3952          * @param $level int Current TOC level (depth)
3953          * @param $lastLevel int Level of previous TOC entry
3954          */
3955         private static function incrementNumbering( &$number, $level, $lastLevel ) {
3956                 if ( $level > $lastLevel )
3957                         $number[$level - 1] = 1;
3958                 else if ( $level < $lastLevel ) {
3959                         foreach ( $number as $key => $unused )
3960                                 if ( $key >= $level )
3961                                         unset( $number[$key] );
3962                         $number[$level - 1]++;
3963                 } else
3964                         $number[$level - 1]++;
3965         }
3966
3967         /**
3968          * Transform wiki markup when saving a page by doing \r\n -> \n
3969          * conversion, substitting signatures, {{subst:}} templates, etc.
3970          *
3971          * @param string $text the text to transform
3972          * @param Title &$title the Title object for the current article
3973          * @param User $user the User object describing the current user
3974          * @param ParserOptions $options parsing options
3975          * @param bool $clearState whether to clear the parser state first
3976          * @return string the altered wiki markup
3977          * @public
3978          */
3979         function preSaveTransform( $text, Title $title, $user, $options, $clearState = true ) {
3980                 $this->mOptions = $options;
3981                 $this->setTitle( $title );
3982                 $this->setOutputType( self::OT_WIKI );
3983
3984                 if ( $clearState ) {
3985                         $this->clearState();
3986                 }
3987
3988                 $pairs = array(
3989                         "\r\n" => "\n",
3990                 );
3991                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3992                 $text = $this->pstPass2( $text, $user );
3993                 $text = $this->mStripState->unstripBoth( $text );
3994                 return $text;
3995         }
3996
3997         /**
3998          * Pre-save transform helper function
3999          * @private
4000          */
4001         function pstPass2( $text, $user ) {
4002                 global $wgContLang, $wgLocaltimezone;
4003
4004                 /* Note: This is the timestamp saved as hardcoded wikitext to
4005                  * the database, we use $wgContLang here in order to give
4006                  * everyone the same signature and use the default one rather
4007                  * than the one selected in each user's preferences.
4008                  *
4009                  * (see also bug 12815)
4010                  */
4011                 $ts = $this->mOptions->getTimestamp();
4012                 if ( isset( $wgLocaltimezone ) ) {
4013                         $tz = $wgLocaltimezone;
4014                 } else {
4015                         $tz = date_default_timezone_get();
4016                 }
4017
4018                 $unixts = wfTimestamp( TS_UNIX, $ts );
4019                 $oldtz = date_default_timezone_get();
4020                 date_default_timezone_set( $tz );
4021                 $ts = date( 'YmdHis', $unixts );
4022                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
4023
4024                 /* Allow translation of timezones trough wiki. date() can return
4025                  * whatever crap the system uses, localised or not, so we cannot
4026                  * ship premade translations.
4027                  */
4028                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
4029                 $value = wfMsgForContent( $key );
4030                 if ( !wfEmptyMsg( $key, $value ) ) $tzMsg = $value;
4031
4032                 date_default_timezone_set( $oldtz );
4033
4034                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
4035
4036                 # Variable replacement
4037                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
4038                 $text = $this->replaceVariables( $text );
4039
4040                 # Signatures
4041                 $sigText = $this->getUserSig( $user );
4042                 $text = strtr( $text, array(
4043                         '~~~~~' => $d,
4044                         '~~~~' => "$sigText $d",
4045                         '~~~' => $sigText
4046                 ) );
4047
4048                 # Links of the form [[|<blah>]] or [[<blah>|]] perform pipe tricks
4049                 # Note this only allows the # in the position it works.
4050                 global $wgLegalTitleChars;
4051                 $pipeTrickRe = "/\[\[(?:(\\|)([$wgLegalTitleChars]+)|([#$wgLegalTitleChars]+)\\|)\]\]/";
4052                 $text = preg_replace_callback( $pipeTrickRe, array( $this, 'pstPipeTrickCallback' ), $text );
4053
4054                 # Trim trailing whitespace
4055                 $text = rtrim( $text );
4056
4057                 return $text;
4058         }
4059
4060         /**
4061          * Called from pstPass2 to perform the pipe trick on links.
4062          * Original was either [[|text]] or [[link|]]
4063          *
4064          * @param Array ("|" or "", text, link) $m
4065          */
4066         function pstPipeTrickCallback( $m )
4067         {
4068                 if( $m[1] ) { # [[|<blah>]]
4069                         $text = $m[2];
4070                         $link = $this->getPipeTrickLink( $text );
4071                 } else { # [[<blah>|]]
4072                         $link = $m[3];
4073                         $text = $this->getPipeTrickText( $link );
4074                 }
4075
4076                 return $link === $text ? "[[$link]]" : "[[$link|$text]]";
4077         }
4078
4079         /**
4080          * Fetch the user's signature text, if any, and normalize to
4081          * validated, ready-to-insert wikitext.
4082          * If you have pre-fetched the nickname or the fancySig option, you can
4083          * specify them here to save a database query.
4084          *
4085          * @param User $user
4086          * @return string
4087          */
4088         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4089                 global $wgMaxSigChars;
4090
4091                 $username = $user->getName();
4092
4093                 // If not given, retrieve from the user object.
4094                 if ( $nickname === false )
4095                         $nickname = $user->getOption( 'nickname' );
4096
4097                 if ( is_null( $fancySig ) )
4098                         $fancySig = $user->getBoolOption( 'fancysig' );
4099
4100                 $nickname = $nickname == null ? $username : $nickname;
4101
4102                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4103                         $nickname = $username;
4104                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4105                 } elseif( $fancySig !== false ) {
4106                         # Sig. might contain markup; validate this
4107                         if( $this->validateSig( $nickname ) !== false ) {
4108                                 # Validated; clean up (if needed) and return it
4109                                 return $this->cleanSig( $nickname, true );
4110                         } else {
4111                                 # Failed to validate; fall back to the default
4112                                 $nickname = $username;
4113                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4114                         }
4115                 }
4116
4117                 // Make sure nickname doesnt get a sig in a sig
4118                 $nickname = $this->cleanSigInSig( $nickname );
4119
4120                 # If we're still here, make it a link to the user page
4121                 $userText = wfEscapeWikiText( $username );
4122                 $nickText = wfEscapeWikiText( $nickname );
4123                 if ( $user->isAnon() )  {
4124                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4125                 } else {
4126                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4127                 }
4128         }
4129
4130         /**
4131          * Check that the user's signature contains no bad XML
4132          *
4133          * @param string $text
4134          * @return mixed An expanded string, or false if invalid.
4135          */
4136         function validateSig( $text ) {
4137                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4138         }
4139
4140         /**
4141          * Clean up signature text
4142          *
4143          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4144          * 2) Substitute all transclusions
4145          *
4146          * @param string $text
4147          * @param $parsing Whether we're cleaning (preferences save) or parsing
4148          * @return string Signature text
4149          */
4150         function cleanSig( $text, $parsing = false ) {
4151                 if ( !$parsing ) {
4152                         global $wgTitle;
4153                         $this->clearState();
4154                         $this->setTitle( $wgTitle );
4155                         $this->mOptions = new ParserOptions;
4156                         $this->setOutputType = self::OT_PREPROCESS;
4157                 }
4158
4159                 # Option to disable this feature
4160                 if ( !$this->mOptions->getCleanSignatures() ) {
4161                         return $text;
4162                 }
4163
4164                 # FIXME: regex doesn't respect extension tags or nowiki
4165                 #  => Move this logic to braceSubstitution()
4166                 $substWord = MagicWord::get( 'subst' );
4167                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4168                 $substText = '{{' . $substWord->getSynonym( 0 );
4169
4170                 $text = preg_replace( $substRegex, $substText, $text );
4171                 $text = $this->cleanSigInSig( $text );
4172                 $dom = $this->preprocessToDom( $text );
4173                 $frame = $this->getPreprocessor()->newFrame();
4174                 $text = $frame->expand( $dom );
4175
4176                 if ( !$parsing ) {
4177                         $text = $this->mStripState->unstripBoth( $text );
4178                 }
4179
4180                 return $text;
4181         }
4182
4183         /**
4184          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4185          * @param string $text
4186          * @return string Signature text with /~{3,5}/ removed
4187          */
4188         function cleanSigInSig( $text ) {
4189                 $text = preg_replace( '/~{3,5}/', '', $text );
4190                 return $text;
4191         }
4192
4193         /**
4194          * Set up some variables which are usually set up in parse()
4195          * so that an external function can call some class members with confidence
4196          * @public
4197          */
4198         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4199                 $this->setTitle( $title );
4200                 $this->mOptions = $options;
4201                 $this->setOutputType( $outputType );
4202                 if ( $clearState ) {
4203                         $this->clearState();
4204                 }
4205         }
4206
4207         /**
4208          * Wrapper for preprocess()
4209          *
4210          * @param string $text the text to preprocess
4211          * @param ParserOptions $options  options
4212          * @return string
4213          * @public
4214          */
4215         function transformMsg( $text, $options ) {
4216                 global $wgTitle;
4217                 static $executing = false;
4218
4219                 # Guard against infinite recursion
4220                 if ( $executing ) {
4221                         return $text;
4222                 }
4223                 $executing = true;
4224
4225                 wfProfileIn(__METHOD__);
4226                 $text = $this->preprocess( $text, $wgTitle, $options );
4227
4228                 $executing = false;
4229                 wfProfileOut(__METHOD__);
4230                 return $text;
4231         }
4232
4233         /**
4234          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4235          * The callback should have the following form:
4236          *    function myParserHook( $text, $params, &$parser ) { ... }
4237          *
4238          * Transform and return $text. Use $parser for any required context, e.g. use
4239          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4240          *
4241          * @public
4242          *
4243          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4244          * @param mixed $callback The callback function (and object) to use for the tag
4245          *
4246          * @return The old value of the mTagHooks array associated with the hook
4247          */
4248         function setHook( $tag, $callback ) {
4249                 $tag = strtolower( $tag );
4250                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4251                 $this->mTagHooks[$tag] = $callback;
4252                 if( !in_array( $tag, $this->mStripList ) ) {
4253                         $this->mStripList[] = $tag;
4254                 }
4255
4256                 return $oldVal;
4257         }
4258
4259         /* An old work-around for bug 2257 - deprecated 2010-02-13 */
4260         function setTransparentTagHook( $tag, $callback ) {
4261                 wfDeprecated( __METHOD__ );
4262                 $tag = strtolower( $tag );
4263                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4264                 $this->mTransparentTagHooks[$tag] = $callback;
4265
4266                 return $oldVal;
4267         }
4268
4269         /**
4270          * Remove all tag hooks
4271          */
4272         function clearTagHooks() {
4273                 $this->mTagHooks = array();
4274                 $this->mStripList = $this->mDefaultStripList;
4275         }
4276
4277         /**
4278          * Create a function, e.g. {{sum:1|2|3}}
4279          * The callback function should have the form:
4280          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4281          *
4282          * Or with SFH_OBJECT_ARGS:
4283          *    function myParserFunction( $parser, $frame, $args ) { ... }
4284          *
4285          * The callback may either return the text result of the function, or an array with the text
4286          * in element 0, and a number of flags in the other elements. The names of the flags are
4287          * specified in the keys. Valid flags are:
4288          *   found                     The text returned is valid, stop processing the template. This
4289          *                             is on by default.
4290          *   nowiki                    Wiki markup in the return value should be escaped
4291          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4292          *
4293          * @public
4294          *
4295          * @param string $id The magic word ID
4296          * @param mixed $callback The callback function (and object) to use
4297          * @param integer $flags a combination of the following flags:
4298          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4299          *
4300          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4301          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4302          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4303          *     the arguments, and to control the way they are expanded.
4304          *
4305          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4306          *     arguments, for instance:
4307          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4308          *
4309          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4310          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4311          *     working if/when this is changed.
4312          *
4313          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4314          *     expansion.
4315          *
4316          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4317          *     about the methods available in PPFrame and PPNode.
4318          *
4319          * @return The old callback function for this name, if any
4320          */
4321         function setFunctionHook( $id, $callback, $flags = 0 ) {
4322                 global $wgContLang;
4323
4324                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4325                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4326
4327                 # Add to function cache
4328                 $mw = MagicWord::get( $id );
4329                 if( !$mw )
4330                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4331
4332                 $synonyms = $mw->getSynonyms();
4333                 $sensitive = intval( $mw->isCaseSensitive() );
4334
4335                 foreach ( $synonyms as $syn ) {
4336                         # Case
4337                         if ( !$sensitive ) {
4338                                 $syn = $wgContLang->lc( $syn );
4339                         }
4340                         # Add leading hash
4341                         if ( !( $flags & SFH_NO_HASH ) ) {
4342                                 $syn = '#' . $syn;
4343                         }
4344                         # Remove trailing colon
4345                         if ( substr( $syn, -1, 1 ) === ':' ) {
4346                                 $syn = substr( $syn, 0, -1 );
4347                         }
4348                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4349                 }
4350                 return $oldVal;
4351         }
4352
4353         /**
4354          * Get all registered function hook identifiers
4355          *
4356          * @return array
4357          */
4358         function getFunctionHooks() {
4359                 return array_keys( $this->mFunctionHooks );
4360         }
4361
4362         /**
4363          * Create a tag function, e.g. <test>some stuff</test>.
4364          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4365          * Unlike parser functions, their content is not preprocessed.
4366          */
4367         function setFunctionTagHook( $tag, $callback, $flags ) {
4368                 $tag = strtolower( $tag );
4369                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4370                         $this->mFunctionTagHooks[$tag] : null;
4371                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4372
4373                 if( !in_array( $tag, $this->mStripList ) ) {
4374                         $this->mStripList[] = $tag;
4375                 }
4376
4377                 return $old;
4378         }
4379
4380         /**
4381          * FIXME: update documentation. makeLinkObj() is deprecated.
4382          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4383          * Placeholders created in Skin::makeLinkObj()
4384          * Returns an array of link CSS classes, indexed by PDBK.
4385          */
4386         function replaceLinkHolders( &$text, $options = 0 ) {
4387                 return $this->mLinkHolders->replace( $text );
4388         }
4389
4390         /**
4391          * Replace <!--LINK--> link placeholders with plain text of links
4392          * (not HTML-formatted).
4393          * @param string $text
4394          * @return string
4395          */
4396         function replaceLinkHoldersText( $text ) {
4397                 return $this->mLinkHolders->replaceText( $text );
4398         }
4399
4400         /**
4401          * Renders an image gallery from a text with one line per image.
4402          * text labels may be given by using |-style alternative text. E.g.
4403          *   Image:one.jpg|The number "1"
4404          *   Image:tree.jpg|A tree
4405          * given as text will return the HTML of a gallery with two images,
4406          * labeled 'The number "1"' and
4407          * 'A tree'.
4408          */
4409         function renderImageGallery( $text, $params ) {
4410                 $ig = new ImageGallery();
4411                 $ig->setContextTitle( $this->mTitle );
4412                 $ig->setShowBytes( false );
4413                 $ig->setShowFilename( false );
4414                 $ig->setParser( $this );
4415                 $ig->setHideBadImages();
4416                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4417                 $ig->useSkin( $this->mOptions->getSkin() );
4418                 $ig->mRevisionId = $this->mRevisionId;
4419
4420                 if( isset( $params['caption'] ) ) {
4421                         $caption = $params['caption'];
4422                         $caption = htmlspecialchars( $caption );
4423                         $caption = $this->replaceInternalLinks( $caption );
4424                         $ig->setCaptionHtml( $caption );
4425                 }
4426                 if( isset( $params['perrow'] ) ) {
4427                         $ig->setPerRow( $params['perrow'] );
4428                 }
4429                 if( isset( $params['widths'] ) ) {
4430                         $ig->setWidths( $params['widths'] );
4431                 }
4432                 if( isset( $params['heights'] ) ) {
4433                         $ig->setHeights( $params['heights'] );
4434                 }
4435
4436                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4437
4438                 $lines = StringUtils::explode( "\n", $text );
4439                 foreach ( $lines as $line ) {
4440                         # match lines like these:
4441                         # Image:someimage.jpg|This is some image
4442                         $matches = array();
4443                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4444                         # Skip empty lines
4445                         if ( count( $matches ) == 0 ) {
4446                                 continue;
4447                         }
4448
4449                         if ( strpos( $matches[0], '%' ) !== false )
4450                                 $matches[1] = urldecode( $matches[1] );
4451                         $tp = Title::newFromText( $matches[1]/*, NS_FILE*/ );
4452                         $nt =& $tp;
4453                         if( is_null( $nt ) ) {
4454                                 # Bogus title. Ignore these so we don't bomb out later.
4455                                 continue;
4456                         }
4457                         if ( isset( $matches[3] ) ) {
4458                                 $label = $matches[3];
4459                         } else {
4460                                 $label = '';
4461                         }
4462
4463                         $html = $this->recursiveTagParse( trim( $label ) );
4464
4465                         $ig->add( $nt, $html );
4466
4467                         # Only add real images (bug #5586)
4468                         if ( $nt->getNamespace() == NS_FILE ) {
4469                                 $this->mOutput->addImage( $nt->getDBkey() );
4470                         }
4471                 }
4472                 return $ig->toHTML();
4473         }
4474
4475         function getImageParams( $handler ) {
4476                 if ( $handler ) {
4477                         $handlerClass = get_class( $handler );
4478                 } else {
4479                         $handlerClass = '';
4480                 }
4481                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4482                         // Initialise static lists
4483                         static $internalParamNames = array(
4484                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4485                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4486                                         'bottom', 'text-bottom' ),
4487                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4488                                         'upright', 'border', 'link', 'alt' ),
4489                         );
4490                         static $internalParamMap;
4491                         if ( !$internalParamMap ) {
4492                                 $internalParamMap = array();
4493                                 foreach ( $internalParamNames as $type => $names ) {
4494                                         foreach ( $names as $name ) {
4495                                                 $magicName = str_replace( '-', '_', "img_$name" );
4496                                                 $internalParamMap[$magicName] = array( $type, $name );
4497                                         }
4498                                 }
4499                         }
4500
4501                         // Add handler params
4502                         $paramMap = $internalParamMap;
4503                         if ( $handler ) {
4504                                 $handlerParamMap = $handler->getParamMap();
4505                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4506                                         $paramMap[$magic] = array( 'handler', $paramName );
4507                                 }
4508                         }
4509                         $this->mImageParams[$handlerClass] = $paramMap;
4510                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4511                 }
4512                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4513         }
4514
4515         /**
4516          * Parse image options text and use it to make an image
4517          * @param Title $title
4518          * @param string $options
4519          * @param LinkHolderArray $holders
4520          */
4521         function makeImage( $title, $options, $holders = false ) {
4522                 # Check if the options text is of the form "options|alt text"
4523                 # Options are:
4524                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4525                 #  * left       no resizing, just left align. label is used for alt= only
4526                 #  * right      same, but right aligned
4527                 #  * none       same, but not aligned
4528                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4529                 #  * center     center the image
4530                 #  * frame      Keep original image size, no magnify-button.
4531                 #  * framed     Same as "frame"
4532                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4533                 #  * upright    reduce width for upright images, rounded to full __0 px
4534                 #  * border     draw a 1px border around the image
4535                 #  * alt        Text for HTML alt attribute (defaults to empty)
4536                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4537                 # vertical-align values (no % or length right now):
4538                 #  * baseline
4539                 #  * sub
4540                 #  * super
4541                 #  * top
4542                 #  * text-top
4543                 #  * middle
4544                 #  * bottom
4545                 #  * text-bottom
4546
4547                 $parts = StringUtils::explode( "|", $options );
4548                 $sk = $this->mOptions->getSkin();
4549
4550                 # Give extensions a chance to select the file revision for us
4551                 $skip = $time = $descQuery = false;
4552                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4553
4554                 if ( $skip ) {
4555                         return $sk->link( $title );
4556                 }
4557
4558                 # Get the file
4559                 $imagename = $title->getDBkey();
4560                 $file = wfFindFile( $title, array( 'time' => $time ) );
4561                 # Get parameter map
4562                 $handler = $file ? $file->getHandler() : false;
4563
4564                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4565
4566                 # Process the input parameters
4567                 $caption = '';
4568                 $params = array( 'frame' => array(), 'handler' => array(),
4569                         'horizAlign' => array(), 'vertAlign' => array() );
4570                 foreach( $parts as $part ) {
4571                         $part = trim( $part );
4572                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4573                         $validated = false;
4574                         if( isset( $paramMap[$magicName] ) ) {
4575                                 list( $type, $paramName ) = $paramMap[$magicName];
4576
4577                                 // Special case; width and height come in one variable together
4578                                 if( $type === 'handler' && $paramName === 'width' ) {
4579                                         $m = array();
4580                                         # (bug 13500) In both cases (width/height and width only),
4581                                         # permit trailing "px" for backward compatibility.
4582                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4583                                                 $width = intval( $m[1] );
4584                                                 $height = intval( $m[2] );
4585                                                 if ( $handler->validateParam( 'width', $width ) ) {
4586                                                         $params[$type]['width'] = $width;
4587                                                         $validated = true;
4588                                                 }
4589                                                 if ( $handler->validateParam( 'height', $height ) ) {
4590                                                         $params[$type]['height'] = $height;
4591                                                         $validated = true;
4592                                                 }
4593                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4594                                                 $width = intval( $value );
4595                                                 if ( $handler->validateParam( 'width', $width ) ) {
4596                                                         $params[$type]['width'] = $width;
4597                                                         $validated = true;
4598                                                 }
4599                                         } // else no validation -- bug 13436
4600                                 } else {
4601                                         if ( $type === 'handler' ) {
4602                                                 # Validate handler parameter
4603                                                 $validated = $handler->validateParam( $paramName, $value );
4604                                         } else {
4605                                                 # Validate internal parameters
4606                                                 switch( $paramName ) {
4607                                                 case 'manualthumb':
4608                                                 case 'alt':
4609                                                         // @todo Fixme: possibly check validity here for
4610                                                         // manualthumb? downstream behavior seems odd with
4611                                                         // missing manual thumbs.
4612                                                         $validated = true;
4613                                                         $value = $this->stripAltText( $value, $holders );
4614                                                         break;
4615                                                 case 'link':
4616                                                         $chars = self::EXT_LINK_URL_CLASS;
4617                                                         $prots = $this->mUrlProtocols;
4618                                                         if ( $value === '' ) {
4619                                                                 $paramName = 'no-link';
4620                                                                 $value = true;
4621                                                                 $validated = true;
4622                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4623                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4624                                                                         $paramName = 'link-url';
4625                                                                         $this->mOutput->addExternalLink( $value );
4626                                                                         $validated = true;
4627                                                                 }
4628                                                         } else {
4629                                                                 $linkTitle = Title::newFromText( $value );
4630                                                                 if ( $linkTitle ) {
4631                                                                         $paramName = 'link-title';
4632                                                                         $value = $linkTitle;
4633                                                                         $this->mOutput->addLink( $linkTitle );
4634                                                                         $validated = true;
4635                                                                 }
4636                                                         }
4637                                                         break;
4638                                                 default:
4639                                                         // Most other things appear to be empty or numeric...
4640                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4641                                                 }
4642                                         }
4643
4644                                         if ( $validated ) {
4645                                                 $params[$type][$paramName] = $value;
4646                                         }
4647                                 }
4648                         }
4649                         if ( !$validated ) {
4650                                 $caption = $part;
4651                         }
4652                 }
4653
4654                 # Process alignment parameters
4655                 if ( $params['horizAlign'] ) {
4656                         $params['frame']['align'] = key( $params['horizAlign'] );
4657                 }
4658                 if ( $params['vertAlign'] ) {
4659                         $params['frame']['valign'] = key( $params['vertAlign'] );
4660                 }
4661
4662                 $params['frame']['caption'] = $caption;
4663
4664                 # Will the image be presented in a frame, with the caption below?
4665                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4666                                  isset( $params['frame']['framed'] ) ||
4667                                  isset( $params['frame']['thumbnail'] ) ||
4668                                  isset( $params['frame']['manualthumb'] );
4669
4670                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4671                 # came to also set the caption, ordinary text after the image -- which
4672                 # makes no sense, because that just repeats the text multiple times in
4673                 # screen readers.  It *also* came to set the title attribute.
4674                 #
4675                 # Now that we have an alt attribute, we should not set the alt text to
4676                 # equal the caption: that's worse than useless, it just repeats the
4677                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4678                 # use the unnamed parameter for alt text as well, just for the time be-
4679                 # ing, if the unnamed param is set and the alt param is not.
4680                 #
4681                 # For the future, we need to figure out if we want to tweak this more,
4682                 # e.g., introducing a title= parameter for the title; ignoring the un-
4683                 # named parameter entirely for images without a caption; adding an ex-
4684                 # plicit caption= parameter and preserving the old magic unnamed para-
4685                 # meter for BC; ...
4686                 if ( $imageIsFramed ) { # Framed image
4687                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4688                                 # No caption or alt text, add the filename as the alt text so
4689                                 # that screen readers at least get some description of the image
4690                                 $params['frame']['alt'] = $title->getText();
4691                         }
4692                         # Do not set $params['frame']['title'] because tooltips don't make sense
4693                         # for framed images
4694                 } else { # Inline image
4695                         if ( !isset( $params['frame']['alt'] ) ) {
4696                                 # No alt text, use the "caption" for the alt text
4697                                 if ( $caption !== '') {
4698                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4699                                 } else {
4700                                         # No caption, fall back to using the filename for the
4701                                         # alt text
4702                                         $params['frame']['alt'] = $title->getText();
4703                                 }
4704                         }
4705                         # Use the "caption" for the tooltip text
4706                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4707                 }
4708
4709                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4710
4711                 # Linker does the rest
4712                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4713
4714                 # Give the handler a chance to modify the parser object
4715                 if ( $handler ) {
4716                         $handler->parserTransformHook( $this, $file );
4717                 }
4718
4719                 return $ret;
4720         }
4721
4722         protected function stripAltText( $caption, $holders ) {
4723                 # Strip bad stuff out of the title (tooltip).  We can't just use
4724                 # replaceLinkHoldersText() here, because if this function is called
4725                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4726                 if ( $holders ) {
4727                         $tooltip = $holders->replaceText( $caption );
4728                 } else {
4729                         $tooltip = $this->replaceLinkHoldersText( $caption );
4730                 }
4731
4732                 # make sure there are no placeholders in thumbnail attributes
4733                 # that are later expanded to html- so expand them now and
4734                 # remove the tags
4735                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4736                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4737
4738                 return $tooltip;
4739         }
4740
4741         /**
4742          * Set a flag in the output object indicating that the content is dynamic and
4743          * shouldn't be cached.
4744          */
4745         function disableCache() {
4746                 wfDebug( "Parser output marked as uncacheable.\n" );
4747                 $this->mOutput->mCacheTime = -1;
4748         }
4749
4750         /**#@+
4751          * Callback from the Sanitizer for expanding items found in HTML attribute
4752          * values, so they can be safely tested and escaped.
4753          * @param string $text
4754          * @param PPFrame $frame
4755          * @return string
4756          * @private
4757          */
4758         function attributeStripCallback( &$text, $frame = false ) {
4759                 $text = $this->replaceVariables( $text, $frame );
4760                 $text = $this->mStripState->unstripBoth( $text );
4761                 return $text;
4762         }
4763
4764         /**#@-*/
4765
4766         /**#@+
4767          * Accessor/mutator
4768          */
4769         function Title( $x = null ) { return wfSetVar( $this->mTitle, $x ); }
4770         function Options( $x = null ) { return wfSetVar( $this->mOptions, $x ); }
4771         function OutputType( $x = null ) { return wfSetVar( $this->mOutputType, $x ); }
4772         /**#@-*/
4773
4774         /**#@+
4775          * Accessor
4776          */
4777         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4778         /**#@-*/
4779
4780
4781         /**
4782          * Break wikitext input into sections, and either pull or replace
4783          * some particular section's text.
4784          *
4785          * External callers should use the getSection and replaceSection methods.
4786          *
4787          * @param string $text Page wikitext
4788          * @param string $section A section identifier string of the form:
4789          *   <flag1> - <flag2> - ... - <section number>
4790          *
4791          * Currently the only recognised flag is "T", which means the target section number
4792          * was derived during a template inclusion parse, in other words this is a template
4793          * section edit link. If no flags are given, it was an ordinary section edit link.
4794          * This flag is required to avoid a section numbering mismatch when a section is
4795          * enclosed by <includeonly> (bug 6563).
4796          *
4797          * The section number 0 pulls the text before the first heading; other numbers will
4798          * pull the given section along with its lower-level subsections. If the section is
4799          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4800          *
4801          * @param string $mode One of "get" or "replace"
4802          * @param string $newText Replacement text for section data.
4803          * @return string for "get", the extracted section text.
4804          *                for "replace", the whole page with the section replaced.
4805          */
4806         private function extractSections( $text, $section, $mode, $newText='' ) {
4807                 global $wgTitle;
4808                 $this->clearState();
4809                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4810                 $this->mOptions = new ParserOptions;
4811                 $this->setOutputType( self::OT_WIKI );
4812                 $outText = '';
4813                 $frame = $this->getPreprocessor()->newFrame();
4814
4815                 // Process section extraction flags
4816                 $flags = 0;
4817                 $sectionParts = explode( '-', $section );
4818                 $sectionIndex = array_pop( $sectionParts );
4819                 foreach ( $sectionParts as $part ) {
4820                         if ( $part === 'T' ) {
4821                                 $flags |= self::PTD_FOR_INCLUSION;
4822                         }
4823                 }
4824                 // Preprocess the text
4825                 $root = $this->preprocessToDom( $text, $flags );
4826
4827                 // <h> nodes indicate section breaks
4828                 // They can only occur at the top level, so we can find them by iterating the root's children
4829                 $node = $root->getFirstChild();
4830
4831                 // Find the target section
4832                 if ( $sectionIndex == 0 ) {
4833                         // Section zero doesn't nest, level=big
4834                         $targetLevel = 1000;
4835                 } else {
4836             while ( $node ) {
4837                 if ( $node->getName() === 'h' ) {
4838                     $bits = $node->splitHeading();
4839                                         if ( $bits['i'] == $sectionIndex ) {
4840                                         $targetLevel = $bits['level'];
4841                                                 break;
4842                                         }
4843                                 }
4844                                 if ( $mode === 'replace' ) {
4845                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4846                                 }
4847                                 $node = $node->getNextSibling();
4848                         }
4849                 }
4850
4851                 if ( !$node ) {
4852                         // Not found
4853                         if ( $mode === 'get' ) {
4854                                 return $newText;
4855                         } else {
4856                                 return $text;
4857                         }
4858                 }
4859
4860                 // Find the end of the section, including nested sections
4861                 do {
4862                         if ( $node->getName() === 'h' ) {
4863                                 $bits = $node->splitHeading();
4864                                 $curLevel = $bits['level'];
4865                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4866                                         break;
4867                                 }
4868                         }
4869                         if ( $mode === 'get' ) {
4870                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4871                         }
4872                         $node = $node->getNextSibling();
4873                 } while ( $node );
4874
4875                 // Write out the remainder (in replace mode only)
4876                 if ( $mode === 'replace' ) {
4877                         // Output the replacement text
4878                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4879                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4880                         // Only add trailing whitespace if there is newText
4881                         if($newText != "") {
4882                                 $outText .= $newText . "\n\n";
4883                         }
4884
4885                         while ( $node ) {
4886                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4887                                 $node = $node->getNextSibling();
4888                         }
4889                 }
4890
4891                 if ( is_string( $outText ) ) {
4892                         // Re-insert stripped tags
4893                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4894                 }
4895
4896                 return $outText;
4897         }
4898
4899         /**
4900          * This function returns the text of a section, specified by a number ($section).
4901          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4902          * the first section before any such heading (section 0).
4903          *
4904          * If a section contains subsections, these are also returned.
4905          *
4906          * @param string $text text to look in
4907          * @param string $section section identifier
4908          * @param string $deftext default to return if section is not found
4909          * @return string text of the requested section
4910          */
4911         public function getSection( $text, $section, $deftext='' ) {
4912                 return $this->extractSections( $text, $section, "get", $deftext );
4913         }
4914
4915         public function replaceSection( $oldtext, $section, $text ) {
4916                 return $this->extractSections( $oldtext, $section, "replace", $text );
4917         }
4918
4919         /**
4920          * Get the timestamp associated with the current revision, adjusted for
4921          * the default server-local timestamp
4922          */
4923         function getRevisionTimestamp() {
4924                 if ( is_null( $this->mRevisionTimestamp ) ) {
4925                         wfProfileIn( __METHOD__ );
4926                         global $wgContLang;
4927                         $dbr = wfGetDB( DB_SLAVE );
4928                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4929                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4930
4931                         // Normalize timestamp to internal MW format for timezone processing.
4932                         // This has the added side-effect of replacing a null value with
4933                         // the current time, which gives us more sensible behavior for
4934                         // previews.
4935                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4936
4937                         // The cryptic '' timezone parameter tells to use the site-default
4938                         // timezone offset instead of the user settings.
4939                         //
4940                         // Since this value will be saved into the parser cache, served
4941                         // to other users, and potentially even used inside links and such,
4942                         // it needs to be consistent for all visitors.
4943                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4944
4945                         wfProfileOut( __METHOD__ );
4946                 }
4947                 return $this->mRevisionTimestamp;
4948         }
4949
4950         /**
4951          * Get the name of the user that edited the last revision
4952          */
4953         function getRevisionUser() {
4954                 // if this template is subst: the revision id will be blank,
4955                 // so just use the current user's name
4956                 if( $this->mRevisionId ) {
4957                         $revision = Revision::newFromId( $this->mRevisionId );
4958                         $revuser = $revision->getUserText();
4959                 } else {
4960                         global $wgUser;
4961                         $revuser = $wgUser->getName();
4962                 }
4963                 return $revuser;
4964         }
4965
4966         /**
4967          * Mutator for $mDefaultSort
4968          *
4969          * @param $sort New value
4970          */
4971         public function setDefaultSort( $sort ) {
4972                 $this->mDefaultSort = $sort;
4973         }
4974
4975         /**
4976          * Accessor for $mDefaultSort
4977          * Will use the title/prefixed title if none is set
4978          *
4979          * @return string
4980          */
4981         public function getDefaultSort() {
4982                 global $wgCategoryPrefixedDefaultSortkey;
4983                 if( $this->mDefaultSort !== false ) {
4984                         return $this->mDefaultSort;
4985                 } elseif ($this->mTitle->getNamespace() == NS_CATEGORY ||
4986                         !$wgCategoryPrefixedDefaultSortkey) {
4987                         return $this->mTitle->getText();
4988                 } else {
4989                         return $this->mTitle->getPrefixedText();
4990                 }
4991         }
4992
4993         /**
4994          * Accessor for $mDefaultSort
4995          * Unlike getDefaultSort(), will return false if none is set
4996          *
4997          * @return string or false
4998          */
4999         public function getCustomDefaultSort() {
5000                 return $this->mDefaultSort;
5001         }
5002
5003         /**
5004          * Try to guess the section anchor name based on a wikitext fragment
5005          * presumably extracted from a heading, for example "Header" from
5006          * "== Header ==".
5007          */
5008         public function guessSectionNameFromWikiText( $text ) {
5009                 # Strip out wikitext links(they break the anchor)
5010                 $text = $this->stripSectionName( $text );
5011                 $headline = Sanitizer::decodeCharReferences( $text );
5012                 # strip out HTML
5013                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
5014                 $headline = trim( $headline );
5015                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
5016                 $replacearray = array(
5017                         '%3A' => ':',
5018                         '%' => '.'
5019                 );
5020                 return str_replace(
5021                         array_keys( $replacearray ),
5022                         array_values( $replacearray ),
5023                         $sectionanchor );
5024         }
5025
5026         /**
5027          * Strips a text string of wikitext for use in a section anchor
5028          *
5029          * Accepts a text string and then removes all wikitext from the
5030          * string and leaves only the resultant text (i.e. the result of
5031          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
5032          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
5033          * to create valid section anchors by mimicing the output of the
5034          * parser when headings are parsed.
5035          *
5036          * @param $text string Text string to be stripped of wikitext
5037          * for use in a Section anchor
5038          * @return Filtered text string
5039          */
5040         public function stripSectionName( $text ) {
5041                 # Strip internal link markup
5042                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
5043                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
5044
5045                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5046                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5047                 # on how many empty links there are on the page - need to figure that out.
5048                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
5049
5050                 # Parse wikitext quotes (italics & bold)
5051                 $text = $this->doQuotes($text);
5052
5053                 # Strip HTML tags
5054                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5055                 return $text;
5056         }
5057
5058         function srvus( $text ) {
5059                 return $this->testSrvus( $text, $this->mOutputType );
5060         }
5061
5062         /**
5063          * strip/replaceVariables/unstrip for preprocessor regression testing
5064          */
5065         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
5066                 $this->clearState();
5067                 if ( ! ( $title instanceof Title ) ) {
5068                         $title = Title::newFromText( $title );
5069                 }
5070                 $this->mTitle = $title;
5071                 $this->mOptions = $options;
5072                 $this->setOutputType( $outputType );
5073                 $text = $this->replaceVariables( $text );
5074                 $text = $this->mStripState->unstripBoth( $text );
5075                 $text = Sanitizer::removeHTMLtags( $text );
5076                 return $text;
5077         }
5078
5079         function testPst( $text, $title, $options ) {
5080                 global $wgUser;
5081                 if ( ! ( $title instanceof Title ) ) {
5082                         $title = Title::newFromText( $title );
5083                 }
5084                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5085         }
5086
5087         function testPreprocess( $text, $title, $options ) {
5088                 if ( ! ( $title instanceof Title ) ) {
5089                         $title = Title::newFromText( $title );
5090                 }
5091                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5092         }
5093
5094         function markerSkipCallback( $s, $callback ) {
5095                 $i = 0;
5096                 $out = '';
5097                 while ( $i < strlen( $s ) ) {
5098                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5099                         if ( $markerStart === false ) {
5100                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5101                                 break;
5102                         } else {
5103                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5104                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5105                                 if ( $markerEnd === false ) {
5106                                         $out .= substr( $s, $markerStart );
5107                                         break;
5108                                 } else {
5109                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5110                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5111                                         $i = $markerEnd;
5112                                 }
5113                         }
5114                 }
5115                 return $out;
5116         }
5117
5118         function serialiseHalfParsedText( $text ) {
5119                 $data = array();
5120                 $data['text'] = $text;
5121
5122                 // First, find all strip markers, and store their
5123                 //  data in an array.
5124                 $stripState = new StripState;
5125                 $pos = 0;
5126                 while( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) ) && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) ) {
5127                         $end_pos += strlen( self::MARKER_SUFFIX );
5128                         $marker = substr( $text, $start_pos, $end_pos-$start_pos );
5129
5130                         if ( !empty( $this->mStripState->general->data[$marker] ) ) {
5131                                 $replaceArray = $stripState->general;
5132                                 $stripText = $this->mStripState->general->data[$marker];
5133                         } elseif ( !empty( $this->mStripState->nowiki->data[$marker] ) ) {
5134                                 $replaceArray = $stripState->nowiki;
5135                                 $stripText = $this->mStripState->nowiki->data[$marker];
5136                         } else {
5137                                 throw new MWException( "Hanging strip marker: '$marker'." );
5138                         }
5139
5140                         $replaceArray->setPair( $marker, $stripText );
5141                         $pos = $end_pos;
5142                 }
5143                 $data['stripstate'] = $stripState;
5144
5145                 // Now, find all of our links, and store THEIR
5146                 //  data in an array! :)
5147                 $links = array( 'internal' => array(), 'interwiki' => array() );
5148                 $pos = 0;
5149
5150                 // Internal links
5151                 while( ( $start_pos = strpos( $text, '<!--LINK ', $pos ) ) ) {
5152                         list( $ns, $trail ) = explode( ':', substr( $text, $start_pos + strlen( '<!--LINK ' ) ), 2 );
5153
5154                         $ns = trim($ns);
5155                         if (empty( $links['internal'][$ns] )) {
5156                                 $links['internal'][$ns] = array();
5157                         }
5158
5159                         $key = trim( substr( $trail, 0, strpos( $trail, '-->' ) ) );
5160                         $links['internal'][$ns][] = $this->mLinkHolders->internals[$ns][$key];
5161                         $pos = $start_pos + strlen( "<!--LINK $ns:$key-->" );
5162                 }
5163
5164                 $pos = 0;
5165
5166                 // Interwiki links
5167                 while( ( $start_pos = strpos( $text, '<!--IWLINK ', $pos ) ) ) {
5168                         $data = substr( $text, $start_pos );
5169                         $key = trim( substr( $data, 0, strpos( $data, '-->' ) ) );
5170                         $links['interwiki'][] = $this->mLinkHolders->interwiki[$key];
5171                         $pos = $start_pos + strlen( "<!--IWLINK $key-->" );
5172                 }
5173
5174                 $data['linkholder'] = $links;
5175
5176                 return $data;
5177         }
5178
5179         function unserialiseHalfParsedText( $data, $intPrefix = null /* Unique identifying prefix */ ) {
5180                 if (!$intPrefix)
5181                         $intPrefix = $this->getRandomString();
5182
5183                 // First, extract the strip state.
5184                 $stripState = $data['stripstate'];
5185                 $this->mStripState->general->merge( $stripState->general );
5186                 $this->mStripState->nowiki->merge( $stripState->nowiki );
5187
5188                 // Now, extract the text, and renumber links
5189                 $text = $data['text'];
5190                 $links = $data['linkholder'];
5191
5192                 // Internal...
5193                 foreach( $links['internal'] as $ns => $nsLinks ) {
5194                         foreach( $nsLinks as $key => $entry ) {
5195                                 $newKey = $intPrefix . '-' . $key;
5196                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
5197
5198                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
5199                         }
5200                 }
5201
5202                 // Interwiki...
5203                 foreach( $links['interwiki'] as $key => $entry ) {
5204                         $newKey = "$intPrefix-$key";
5205                         $this->mLinkHolders->interwikis[$newKey] = $entry;
5206
5207                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
5208                 }
5209
5210                 // Should be good to go.
5211                 return $text;
5212         }
5213 }
5214
5215 /**
5216  * @todo document, briefly.
5217  * @ingroup Parser
5218  */
5219 class StripState {
5220         var $general, $nowiki;
5221
5222         function __construct() {
5223                 $this->general = new ReplacementArray;
5224                 $this->nowiki = new ReplacementArray;
5225         }
5226
5227         function unstripGeneral( $text ) {
5228                 wfProfileIn( __METHOD__ );
5229                 do {
5230                         $oldText = $text;
5231                         $text = $this->general->replace( $text );
5232                 } while ( $text !== $oldText );
5233                 wfProfileOut( __METHOD__ );
5234                 return $text;
5235         }
5236
5237         function unstripNoWiki( $text ) {
5238                 wfProfileIn( __METHOD__ );
5239                 do {
5240                         $oldText = $text;
5241                         $text = $this->nowiki->replace( $text );
5242                 } while ( $text !== $oldText );
5243                 wfProfileOut( __METHOD__ );
5244                 return $text;
5245         }
5246
5247         function unstripBoth( $text ) {
5248                 wfProfileIn( __METHOD__ );
5249                 do {
5250                         $oldText = $text;
5251                         $text = $this->general->replace( $text );
5252                         $text = $this->nowiki->replace( $text );
5253                 } while ( $text !== $oldText );
5254                 wfProfileOut( __METHOD__ );
5255                 return $text;
5256         }
5257 }
5258
5259 /**
5260  * @todo document, briefly.
5261  * @ingroup Parser
5262  */
5263 class OnlyIncludeReplacer {
5264         var $output = '';
5265
5266         function replace( $matches ) {
5267                 if ( substr( $matches[1], -1 ) === "\n" ) {
5268                         $this->output .= substr( $matches[1], 0, -1 );
5269                 } else {
5270                         $this->output .= $matches[1];
5271                 }
5272         }
5273 }