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 five 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  *
  30  * Globals used:
  31  *    objects:   $wgLang, $wgContLang
  32  *
  33  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  34  *
  35  * settings:
  36  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  37  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  38  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  39  *  $wgMaxArticleSize*
  40  *
  41  *  * only within ParserOptions
  42  * </pre>
  43  *
  44  * @ingroup Parser
  45  */
  46 class Parser
  47 {
  48         /**
  49          * Update this version number when the ParserOutput format
  50          * changes in an incompatible way, so the parser cache
  51          * can automatically discard old data.
  52          */
  53         const VERSION = '1.6.4';
  54
  55         # Flags for Parser::setFunctionHook
  56         # Also available as global constants from Defines.php
  57         const SFH_NO_HASH = 1;
  58         const SFH_OBJECT_ARGS = 2;
  59
  60         # Constants needed for external link processing
  61         # Everything except bracket, space, or control characters
  62         const EXT_LINK_URL_CLASS = '[^][<>"\\x00-\\x20\\x7F]';
  63         const EXT_IMAGE_REGEX = '/^(http:\/\/|https:\/\/)([^][<>"\\x00-\\x20\\x7F]+)
  64                 \\/([A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]+)\\.((?i)gif|png|jpg|jpeg)$/Sx';
  65
  66         // State constants for the definition list colon extraction
  67         const COLON_STATE_TEXT = 0;
  68         const COLON_STATE_TAG = 1;
  69         const COLON_STATE_TAGSTART = 2;
  70         const COLON_STATE_CLOSETAG = 3;
  71         const COLON_STATE_TAGSLASH = 4;
  72         const COLON_STATE_COMMENT = 5;
  73         const COLON_STATE_COMMENTDASH = 6;
  74         const COLON_STATE_COMMENTDASHDASH = 7;
  75
  76         // Flags for preprocessToDom
  77         const PTD_FOR_INCLUSION = 1;
  78
  79         // Allowed values for $this->mOutputType
  80         // Parameter to startExternalParse().
  81         const OT_HTML = 1;
  82         const OT_WIKI = 2;
  83         const OT_PREPROCESS = 3;
  84         const OT_MSG = 3;
  85
  86         // Marker Suffix needs to be accessible staticly.
  87         const MARKER_SUFFIX = "-QINU\x7f";
  88
  89         /**#@+
  90          * @private
  91          */
  92         # Persistent:
  93         var $mTagHooks, $mTransparentTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables,
  94                 $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerIndex, $mPreprocessor,
  95                 $mExtLinkBracketedRegex, $mDefaultStripList, $mVarCache, $mConf;
  96
  97
  98         # Cleared with clearState():
  99         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 100         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 101         var $mInterwikiLinkHolders, $mLinkHolders;
 102         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 103         var $mTplExpandCache; // empty-frame expansion cache
 104         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 105         var $mExpensiveFunctionCount; // number of expensive parser function calls
 106
 107         # Temporary
 108         # These are variables reset at least once per parse regardless of $clearState
 109         var $mOptions,      // ParserOptions object
 110                 $mTitle,        // Title context, used for self-link rendering and similar things
 111                 $mOutputType,   // Output type, one of the OT_xxx constants
 112                 $ot,            // Shortcut alias, see setOutputType()
 113                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 114                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 115                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 116
 117         /**#@-*/
 118
 119         /**
 120          * Constructor
 121          *
 122          * @public
 123          */
 124         function __construct( $conf = array() ) {
 125                 $this->mConf = $conf;
 126                 $this->mTagHooks = array();
 127                 $this->mTransparentTagHooks = array();
 128                 $this->mFunctionHooks = array();
 129                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 130                 $this->mDefaultStripList = $this->mStripList = array( 'nowiki', 'gallery' );
 131                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 132                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 133                 $this->mVarCache = array();
 134                 if ( isset( $conf['preprocessorClass'] ) ) {
 135                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 136                 } elseif ( class_exists( 'DOMDocument' ) ) {
 137                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 138                 } else {
 139                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 140                 }
 141                 $this->mMarkerIndex = 0;
 142                 $this->mFirstCall = true;
 143         }
 144
 145         /**
 146          * Do various kinds of initialisation on the first call of the parser
 147          */
 148         function firstCallInit() {
 149                 if ( !$this->mFirstCall ) {
 150                         return;
 151                 }
 152                 $this->mFirstCall = false;
 153
 154                 wfProfileIn( __METHOD__ );
 155
 156                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 157                 CoreParserFunctions::register( $this );
 158                 $this->initialiseVariables();
 159
 160                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 161                 wfProfileOut( __METHOD__ );
 162         }
 163
 164         /**
 165          * Clear Parser state
 166          *
 167          * @private
 168          */
 169         function clearState() {
 170                 wfProfileIn( __METHOD__ );
 171                 if ( $this->mFirstCall ) {
 172                         $this->firstCallInit();
 173                 }
 174                 $this->mOutput = new ParserOutput;
 175                 $this->mAutonumber = 0;
 176                 $this->mLastSection = '';
 177                 $this->mDTopen = false;
 178                 $this->mIncludeCount = array();
 179                 $this->mStripState = new StripState;
 180                 $this->mArgStack = false;
 181                 $this->mInPre = false;
 182                 $this->mInterwikiLinkHolders = array(
 183                         'texts' => array(),
 184                         'titles' => array()
 185                 );
 186                 $this->mLinkHolders = array(
 187                         'namespaces' => array(),
 188                         'dbkeys' => array(),
 189                         'queries' => array(),
 190                         'texts' => array(),
 191                         'titles' => array()
 192                 );
 193                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 194
 195                 /**
 196                  * Prefix for temporary replacement strings for the multipass parser.
 197                  * \x07 should never appear in input as it's disallowed in XML.
 198                  * Using it at the front also gives us a little extra robustness
 199                  * since it shouldn't match when butted up against identifier-like
 200                  * string constructs.
 201                  *
 202                  * Must not consist of all title characters, or else it will change
 203                  * the behaviour of <nowiki> in a link.
 204                  */
 205                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 206                 # Changed to \x7f to allow XML double-parsing -- TS
 207                 $this->mUniqPrefix = "\x7fUNIQ" . Parser::getRandomString();
 208
 209
 210                 # Clear these on every parse, bug 4549
 211                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 212
 213                 $this->mShowToc = true;
 214                 $this->mForceTocPosition = false;
 215                 $this->mIncludeSizes = array(
 216                         'post-expand' => 0,
 217                         'arg' => 0,
 218                 );
 219                 $this->mPPNodeCount = 0;
 220                 $this->mDefaultSort = false;
 221                 $this->mHeadings = array();
 222                 $this->mDoubleUnderscores = array();
 223                 $this->mExpensiveFunctionCount = 0;
 224
 225                 # Fix cloning
 226                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 227                         $this->mPreprocessor = null;
 228                 }
 229
 230                 wfRunHooks( 'ParserClearState', array( &$this ) );
 231                 wfProfileOut( __METHOD__ );
 232         }
 233
 234         function setOutputType( $ot ) {
 235                 $this->mOutputType = $ot;
 236                 // Shortcut alias
 237                 $this->ot = array(
 238                         'html' => $ot == self::OT_HTML,
 239                         'wiki' => $ot == self::OT_WIKI,
 240                         'pre' => $ot == self::OT_PREPROCESS,
 241                 );
 242         }
 243
 244         /**
 245          * Set the context title
 246          */
 247         function setTitle( $t ) {
 248                 if ( !$t || $t instanceof FakeTitle ) {
 249                         $t = Title::newFromText( 'NO TITLE' );
 250                 }
 251                 if ( strval( $t->getFragment() ) !== '' ) {
 252                         # Strip the fragment to avoid various odd effects
 253                         $this->mTitle = clone $t;
 254                         $this->mTitle->setFragment( '' );
 255                 } else {
 256                         $this->mTitle = $t;
 257                 }
 258         }
 259
 260         /**
 261          * Accessor for mUniqPrefix.
 262          *
 263          * @public
 264          */
 265         function uniqPrefix() {
 266                 if( !isset( $this->mUniqPrefix ) ) {
 267                         // @fixme this is probably *horribly wrong*
 268                         // LanguageConverter seems to want $wgParser's uniqPrefix, however
 269                         // if this is called for a parser cache hit, the parser may not
 270                         // have ever been initialized in the first place.
 271                         // Not really sure what the heck is supposed to be going on here.
 272                         return '';
 273                         //throw new MWException( "Accessing uninitialized mUniqPrefix" );
 274                 }
 275                 return $this->mUniqPrefix;
 276         }
 277
 278         /**
 279          * Convert wikitext to HTML
 280          * Do not call this function recursively.
 281          *
 282          * @param string $text Text we want to parse
 283          * @param Title &$title A title object
 284          * @param array $options
 285          * @param boolean $linestart
 286          * @param boolean $clearState
 287          * @param int $revid number to pass in {{REVISIONID}}
 288          * @return ParserOutput a ParserOutput
 289          */
 290         public function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 291                 /**
 292                  * First pass--just handle <nowiki> sections, pass the rest off
 293                  * to internalParse() which does all the real work.
 294                  */
 295
 296                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 297                 $fname = 'Parser::parse-' . wfGetCaller();
 298                 wfProfileIn( __METHOD__ );
 299                 wfProfileIn( $fname );
 300
 301                 if ( $clearState ) {
 302                         $this->clearState();
 303                 }
 304
 305                 $this->mOptions = $options;
 306                 $this->setTitle( $title );
 307                 $oldRevisionId = $this->mRevisionId;
 308                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 309                 if( $revid !== null ) {
 310                         $this->mRevisionId = $revid;
 311                         $this->mRevisionTimestamp = null;
 312                 }
 313                 $this->setOutputType( self::OT_HTML );
 314                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 315                 # No more strip!
 316                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 317                 $text = $this->internalParse( $text );
 318                 $text = $this->mStripState->unstripGeneral( $text );
 319
 320                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 321                 $fixtags = array(
 322                         # french spaces, last one Guillemet-left
 323                         # only if there is something before the space
 324                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 325                         # french spaces, Guillemet-right
 326                         '/(\\302\\253) /' => '\\1&nbsp;',
 327                         '/&nbsp;(!\s*important)/' => ' \\1', #Beware of CSS magic word !important, bug #11874.
 328                 );
 329                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 330
 331                 # only once and last
 332                 $text = $this->doBlockLevels( $text, $linestart );
 333
 334                 $this->replaceLinkHolders( $text );
 335
 336                 # the position of the parserConvert() call should not be changed. it
 337                 # assumes that the links are all replaced and the only thing left
 338                 # is the <nowiki> mark.
 339                 # Side-effects: this calls $this->mOutput->setTitleText()
 340                 $text = $wgContLang->parserConvert( $text, $this );
 341
 342                 $text = $this->mStripState->unstripNoWiki( $text );
 343
 344                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 345
 346 //!JF Move to its own function
 347
 348                 $uniq_prefix = $this->mUniqPrefix;
 349                 $matches = array();
 350                 $elements = array_keys( $this->mTransparentTagHooks );
 351                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 352
 353                 foreach( $matches as $marker => $data ) {
 354                         list( $element, $content, $params, $tag ) = $data;
 355                         $tagName = strtolower( $element );
 356                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 357                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 358                                         array( $content, $params, $this ) );
 359                         } else {
 360                                 $output = $tag;
 361                         }
 362                         $this->mStripState->general->setPair( $marker, $output );
 363                 }
 364                 $text = $this->mStripState->unstripGeneral( $text );
 365
 366                 $text = Sanitizer::normalizeCharReferences( $text );
 367
 368                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 369                         $text = Parser::tidy($text);
 370                 } else {
 371                         # attempt to sanitize at least some nesting problems
 372                         # (bug #2702 and quite a few others)
 373                         $tidyregs = array(
 374                                 # ''Something [http://www.cool.com cool''] -->
 375                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 376                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 377                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 378                                 # fix up an anchor inside another anchor, only
 379                                 # at least for a single single nested link (bug 3695)
 380                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 381                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 382                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 383                                 # contains a div, so fix it up here; replace
 384                                 # div with escaped text
 385                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 386                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 387                                 # remove empty italic or bold tag pairs, some
 388                                 # introduced by rules above
 389                                 '/<([bi])><\/\\1>/' => '',
 390                         );
 391
 392                         $text = preg_replace(
 393                                 array_keys( $tidyregs ),
 394                                 array_values( $tidyregs ),
 395                                 $text );
 396                 }
 397                 global $wgExpensiveParserFunctionLimit;
 398                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 399                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 400                 }
 401
 402                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 403
 404                 # Information on include size limits, for the benefit of users who try to skirt them
 405                 if ( $this->mOptions->getEnableLimitReport() ) {
 406                         global $wgExpensiveParserFunctionLimit;
 407                         $max = $this->mOptions->getMaxIncludeSize();
 408                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 409                         $limitReport =
 410                                 "NewPP limit report\n" .
 411                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 412                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 413                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 414                                 $PFreport;
 415                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 416                         $text .= "\n<!-- \n$limitReport-->\n";
 417                 }
 418                 $this->mOutput->setText( $text );
 419                 $this->mRevisionId = $oldRevisionId;
 420                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 421                 wfProfileOut( $fname );
 422                 wfProfileOut( __METHOD__ );
 423
 424                 return $this->mOutput;
 425         }
 426
 427         /**
 428          * Recursive parser entry point that can be called from an extension tag
 429          * hook.
 430          */
 431         function recursiveTagParse( $text ) {
 432                 wfProfileIn( __METHOD__ );
 433                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 434                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 435                 $text = $this->internalParse( $text );
 436                 wfProfileOut( __METHOD__ );
 437                 return $text;
 438         }
 439
 440         /**
 441          * Expand templates and variables in the text, producing valid, static wikitext.
 442          * Also removes comments.
 443          */
 444         function preprocess( $text, $title, $options, $revid = null ) {
 445                 wfProfileIn( __METHOD__ );
 446                 $this->clearState();
 447                 $this->setOutputType( self::OT_PREPROCESS );
 448                 $this->mOptions = $options;
 449                 $this->setTitle( $title );
 450                 if( $revid !== null ) {
 451                         $this->mRevisionId = $revid;
 452                 }
 453                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 454                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 455                 $text = $this->replaceVariables( $text );
 456                 $text = $this->mStripState->unstripBoth( $text );
 457                 wfProfileOut( __METHOD__ );
 458                 return $text;
 459         }
 460
 461         /**
 462          * Get a random string
 463          *
 464          * @private
 465          * @static
 466          */
 467         function getRandomString() {
 468                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 469         }
 470
 471         function &getTitle() { return $this->mTitle; }
 472         function getOptions() { return $this->mOptions; }
 473         function getRevisionId() { return $this->mRevisionId; }
 474
 475         function getFunctionLang() {
 476                 global $wgLang, $wgContLang;
 477
 478                 $target = $this->mOptions->getTargetLanguage();
 479                 if ( $target !== null ) {
 480                         return $target;
 481                 } else {
 482                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 483                 }
 484         }
 485
 486         /**
 487          * Get a preprocessor object
 488          */
 489         function getPreprocessor() {
 490                 if ( !isset( $this->mPreprocessor ) ) {
 491                         $class = $this->mPreprocessorClass;
 492                         $this->mPreprocessor = new $class( $this );
 493                 }
 494                 return $this->mPreprocessor;
 495         }
 496
 497         /**
 498          * Replaces all occurrences of HTML-style comments and the given tags
 499          * in the text with a random marker and returns the next text. The output
 500          * parameter $matches will be an associative array filled with data in
 501          * the form:
 502          *   'UNIQ-xxxxx' => array(
 503          *     'element',
 504          *     'tag content',
 505          *     array( 'param' => 'x' ),
 506          *     '<element param="x">tag content</element>' ) )
 507          *
 508          * @param $elements list of element names. Comments are always extracted.
 509          * @param $text Source text string.
 510          * @param $uniq_prefix
 511          *
 512          * @public
 513          * @static
 514          */
 515         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 516                 static $n = 1;
 517                 $stripped = '';
 518                 $matches = array();
 519
 520                 $taglist = implode( '|', $elements );
 521                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 522
 523                 while ( '' != $text ) {
 524                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 525                         $stripped .= $p[0];
 526                         if( count( $p ) < 5 ) {
 527                                 break;
 528                         }
 529                         if( count( $p ) > 5 ) {
 530                                 // comment
 531                                 $element    = $p[4];
 532                                 $attributes = '';
 533                                 $close      = '';
 534                                 $inside     = $p[5];
 535                         } else {
 536                                 // tag
 537                                 $element    = $p[1];
 538                                 $attributes = $p[2];
 539                                 $close      = $p[3];
 540                                 $inside     = $p[4];
 541                         }
 542
 543                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . self::MARKER_SUFFIX;
 544                         $stripped .= $marker;
 545
 546                         if ( $close === '/>' ) {
 547                                 // Empty element tag, <tag />
 548                                 $content = null;
 549                                 $text = $inside;
 550                                 $tail = null;
 551                         } else {
 552                                 if( $element == '!--' ) {
 553                                         $end = '/(-->)/';
 554                                 } else {
 555                                         $end = "/(<\\/$element\\s*>)/i";
 556                                 }
 557                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 558                                 $content = $q[0];
 559                                 if( count( $q ) < 3 ) {
 560                                         # No end tag -- let it run out to the end of the text.
 561                                         $tail = '';
 562                                         $text = '';
 563                                 } else {
 564                                         $tail = $q[1];
 565                                         $text = $q[2];
 566                                 }
 567                         }
 568
 569                         $matches[$marker] = array( $element,
 570                                 $content,
 571                                 Sanitizer::decodeTagAttributes( $attributes ),
 572                                 "<$element$attributes$close$content$tail" );
 573                 }
 574                 return $stripped;
 575         }
 576
 577         /**
 578          * Get a list of strippable XML-like elements
 579          */
 580         function getStripList() {
 581                 global $wgRawHtml;
 582                 $elements = $this->mStripList;
 583                 if( $wgRawHtml ) {
 584                         $elements[] = 'html';
 585                 }
 586                 if( $this->mOptions->getUseTeX() ) {
 587                         $elements[] = 'math';
 588                 }
 589                 return $elements;
 590         }
 591
 592         /**
 593          * @deprecated use replaceVariables
 594          */
 595         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 596                 return $text;
 597         }
 598
 599         /**
 600          * Restores pre, math, and other extensions removed by strip()
 601          *
 602          * always call unstripNoWiki() after this one
 603          * @private
 604          * @deprecated use $this->mStripState->unstrip()
 605          */
 606         function unstrip( $text, $state ) {
 607                 return $state->unstripGeneral( $text );
 608         }
 609
 610         /**
 611          * Always call this after unstrip() to preserve the order
 612          *
 613          * @private
 614          * @deprecated use $this->mStripState->unstrip()
 615          */
 616         function unstripNoWiki( $text, $state ) {
 617                 return $state->unstripNoWiki( $text );
 618         }
 619
 620         /**
 621          * @deprecated use $this->mStripState->unstripBoth()
 622          */
 623         function unstripForHTML( $text ) {
 624                 return $this->mStripState->unstripBoth( $text );
 625         }
 626
 627         /**
 628          * Add an item to the strip state
 629          * Returns the unique tag which must be inserted into the stripped text
 630          * The tag will be replaced with the original text in unstrip()
 631          *
 632          * @private
 633          */
 634         function insertStripItem( $text ) {
 635                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 636                 $this->mMarkerIndex++;
 637                 $this->mStripState->general->setPair( $rnd, $text );
 638                 return $rnd;
 639         }
 640
 641         /**
 642          * Interface with html tidy, used if $wgUseTidy = true.
 643          * If tidy isn't able to correct the markup, the original will be
 644          * returned in all its glory with a warning comment appended.
 645          *
 646          * Either the external tidy program or the in-process tidy extension
 647          * will be used depending on availability. Override the default
 648          * $wgTidyInternal setting to disable the internal if it's not working.
 649          *
 650          * @param string $text Hideous HTML input
 651          * @return string Corrected HTML output
 652          * @public
 653          * @static
 654          */
 655         function tidy( $text ) {
 656                 global $wgTidyInternal;
 657                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 658 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 659 '<head><title>test</title></head><body>'.$text.'</body></html>';
 660                 if( $wgTidyInternal ) {
 661                         $correctedtext = Parser::internalTidy( $wrappedtext );
 662                 } else {
 663                         $correctedtext = Parser::externalTidy( $wrappedtext );
 664                 }
 665                 if( is_null( $correctedtext ) ) {
 666                         wfDebug( "Tidy error detected!\n" );
 667                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 668                 }
 669                 return $correctedtext;
 670         }
 671
 672         /**
 673          * Spawn an external HTML tidy process and get corrected markup back from it.
 674          *
 675          * @private
 676          * @static
 677          */
 678         function externalTidy( $text ) {
 679                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 680                 $fname = 'Parser::externalTidy';
 681                 wfProfileIn( $fname );
 682
 683                 $cleansource = '';
 684                 $opts = ' -utf8';
 685
 686                 $descriptorspec = array(
 687                         0 => array('pipe', 'r'),
 688                         1 => array('pipe', 'w'),
 689                         2 => array('file', wfGetNull(), 'a')
 690                 );
 691                 $pipes = array();
 692                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 693                 if (is_resource($process)) {
 694                         // Theoretically, this style of communication could cause a deadlock
 695                         // here. If the stdout buffer fills up, then writes to stdin could
 696                         // block. This doesn't appear to happen with tidy, because tidy only
 697                         // writes to stdout after it's finished reading from stdin. Search
 698                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 699                         fwrite($pipes[0], $text);
 700                         fclose($pipes[0]);
 701                         while (!feof($pipes[1])) {
 702                                 $cleansource .= fgets($pipes[1], 1024);
 703                         }
 704                         fclose($pipes[1]);
 705                         proc_close($process);
 706                 }
 707
 708                 wfProfileOut( $fname );
 709
 710                 if( $cleansource == '' && $text != '') {
 711                         // Some kind of error happened, so we couldn't get the corrected text.
 712                         // Just give up; we'll use the source text and append a warning.
 713                         return null;
 714                 } else {
 715                         return $cleansource;
 716                 }
 717         }
 718
 719         /**
 720          * Use the HTML tidy PECL extension to use the tidy library in-process,
 721          * saving the overhead of spawning a new process.
 722          *
 723          * 'pear install tidy' should be able to compile the extension module.
 724          *
 725          * @private
 726          * @static
 727          */
 728         function internalTidy( $text ) {
 729                 global $wgTidyConf, $IP, $wgDebugTidy;
 730                 $fname = 'Parser::internalTidy';
 731                 wfProfileIn( $fname );
 732
 733                 $tidy = new tidy;
 734                 $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 735                 $tidy->cleanRepair();
 736                 if( $tidy->getStatus() == 2 ) {
 737                         // 2 is magic number for fatal error
 738                         // http://www.php.net/manual/en/function.tidy-get-status.php
 739                         $cleansource = null;
 740                 } else {
 741                         $cleansource = tidy_get_output( $tidy );
 742                 }
 743                 if ( $wgDebugTidy && $tidy->getStatus() > 0 ) {
 744                         $cleansource .= "<!--\nTidy reports:\n" .
 745                                 str_replace( '-->', '--&gt;', $tidy->errorBuffer ) .
 746                                 "\n-->";
 747                 }
 748
 749                 wfProfileOut( $fname );
 750                 return $cleansource;
 751         }
 752
 753         /**
 754          * parse the wiki syntax used to render tables
 755          *
 756          * @private
 757          */
 758         function doTableStuff ( $text ) {
 759                 $fname = 'Parser::doTableStuff';
 760                 wfProfileIn( $fname );
 761
 762                 $lines = explode ( "\n" , $text );
 763                 $td_history = array (); // Is currently a td tag open?
 764                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 765                 $tr_history = array (); // Is currently a tr tag open?
 766                 $tr_attributes = array (); // history of tr attributes
 767                 $has_opened_tr = array(); // Did this table open a <tr> element?
 768                 $indent_level = 0; // indent level of the table
 769                 foreach ( $lines as $key => $line )
 770                 {
 771                         $line = trim ( $line );
 772
 773                         if( $line == '' ) { // empty line, go to next line
 774                                 continue;
 775                         }
 776                         $first_character = $line{0};
 777                         $matches = array();
 778
 779                         if ( preg_match( '/^(:*)\{\|(.*)$/' , $line , $matches ) ) {
 780                                 // First check if we are starting a new table
 781                                 $indent_level = strlen( $matches[1] );
 782
 783                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 784                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 785
 786                                 $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 787                                 array_push ( $td_history , false );
 788                                 array_push ( $last_tag_history , '' );
 789                                 array_push ( $tr_history , false );
 790                                 array_push ( $tr_attributes , '' );
 791                                 array_push ( $has_opened_tr , false );
 792                         } else if ( count ( $td_history ) == 0 ) {
 793                                 // Don't do any of the following
 794                                 continue;
 795                         } else if ( substr ( $line , 0 , 2 ) == '|}' ) {
 796                                 // We are ending a table
 797                                 $line = '</table>' . substr ( $line , 2 );
 798                                 $last_tag = array_pop ( $last_tag_history );
 799
 800                                 if ( !array_pop ( $has_opened_tr ) ) {
 801                                         $line = "<tr><td></td></tr>{$line}";
 802                                 }
 803
 804                                 if ( array_pop ( $tr_history ) ) {
 805                                         $line = "</tr>{$line}";
 806                                 }
 807
 808                                 if ( array_pop ( $td_history ) ) {
 809                                         $line = "</{$last_tag}>{$line}";
 810                                 }
 811                                 array_pop ( $tr_attributes );
 812                                 $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );
 813                         } else if ( substr ( $line , 0 , 2 ) == '|-' ) {
 814                                 // Now we have a table row
 815                                 $line = preg_replace( '#^\|-+#', '', $line );
 816
 817                                 // Whats after the tag is now only attributes
 818                                 $attributes = $this->mStripState->unstripBoth( $line );
 819                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 820                                 array_pop ( $tr_attributes );
 821                                 array_push ( $tr_attributes , $attributes );
 822
 823                                 $line = '';
 824                                 $last_tag = array_pop ( $last_tag_history );
 825                                 array_pop ( $has_opened_tr );
 826                                 array_push ( $has_opened_tr , true );
 827
 828                                 if ( array_pop ( $tr_history ) ) {
 829                                         $line = '</tr>';
 830                                 }
 831
 832                                 if ( array_pop ( $td_history ) ) {
 833                                         $line = "</{$last_tag}>{$line}";
 834                                 }
 835
 836                                 $lines[$key] = $line;
 837                                 array_push ( $tr_history , false );
 838                                 array_push ( $td_history , false );
 839                                 array_push ( $last_tag_history , '' );
 840                         }
 841                         else if ( $first_character == '|' || $first_character == '!' || substr ( $line , 0 , 2 )  == '|+' ) {
 842                                 // This might be cell elements, td, th or captions
 843                                 if ( substr ( $line , 0 , 2 ) == '|+' ) {
 844                                         $first_character = '+';
 845                                         $line = substr ( $line , 1 );
 846                                 }
 847
 848                                 $line = substr ( $line , 1 );
 849
 850                                 if ( $first_character == '!' ) {
 851                                         $line = str_replace ( '!!' , '||' , $line );
 852                                 }
 853
 854                                 // Split up multiple cells on the same line.
 855                                 // FIXME : This can result in improper nesting of tags processed
 856                                 // by earlier parser steps, but should avoid splitting up eg
 857                                 // attribute values containing literal "||".
 858                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 859
 860                                 $lines[$key] = '';
 861
 862                                 // Loop through each table cell
 863                                 foreach ( $cells as $cell )
 864                                 {
 865                                         $previous = '';
 866                                         if ( $first_character != '+' )
 867                                         {
 868                                                 $tr_after = array_pop ( $tr_attributes );
 869                                                 if ( !array_pop ( $tr_history ) ) {
 870                                                         $previous = "<tr{$tr_after}>\n";
 871                                                 }
 872                                                 array_push ( $tr_history , true );
 873                                                 array_push ( $tr_attributes , '' );
 874                                                 array_pop ( $has_opened_tr );
 875                                                 array_push ( $has_opened_tr , true );
 876                                         }
 877
 878                                         $last_tag = array_pop ( $last_tag_history );
 879
 880                                         if ( array_pop ( $td_history ) ) {
 881                                                 $previous = "</{$last_tag}>{$previous}";
 882                                         }
 883
 884                                         if ( $first_character == '|' ) {
 885                                                 $last_tag = 'td';
 886                                         } else if ( $first_character == '!' ) {
 887                                                 $last_tag = 'th';
 888                                         } else if ( $first_character == '+' ) {
 889                                                 $last_tag = 'caption';
 890                                         } else {
 891                                                 $last_tag = '';
 892                                         }
 893
 894                                         array_push ( $last_tag_history , $last_tag );
 895
 896                                         // A cell could contain both parameters and data
 897                                         $cell_data = explode ( '|' , $cell , 2 );
 898
 899                                         // Bug 553: Note that a '|' inside an invalid link should not
 900                                         // be mistaken as delimiting cell parameters
 901                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 902                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 903                                         } else if ( count ( $cell_data ) == 1 )
 904                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 905                                         else {
 906                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 907                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 908                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 909                                         }
 910
 911                                         $lines[$key] .= $cell;
 912                                         array_push ( $td_history , true );
 913                                 }
 914                         }
 915                 }
 916
 917                 // Closing open td, tr && table
 918                 while ( count ( $td_history ) > 0 )
 919                 {
 920                         if ( array_pop ( $td_history ) ) {
 921                                 $lines[] = '</td>' ;
 922                         }
 923                         if ( array_pop ( $tr_history ) ) {
 924                                 $lines[] = '</tr>' ;
 925                         }
 926                         if ( !array_pop ( $has_opened_tr ) ) {
 927                                 $lines[] = "<tr><td></td></tr>" ;
 928                         }
 929
 930                         $lines[] = '</table>' ;
 931                 }
 932
 933                 $output = implode ( "\n" , $lines ) ;
 934
 935                 // special case: don't return empty table
 936                 if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {
 937                         $output = '';
 938                 }
 939
 940                 wfProfileOut( $fname );
 941
 942                 return $output;
 943         }
 944
 945         /**
 946          * Helper function for parse() that transforms wiki markup into
 947          * HTML. Only called for $mOutputType == self::OT_HTML.
 948          *
 949          * @private
 950          */
 951         function internalParse( $text ) {
 952                 $isMain = true;
 953                 $fname = 'Parser::internalParse';
 954                 wfProfileIn( $fname );
 955
 956                 # Hook to suspend the parser in this state
 957                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 958                         wfProfileOut( $fname );
 959                         return $text ;
 960                 }
 961
 962                 $text = $this->replaceVariables( $text );
 963                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 964                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 965
 966                 // Tables need to come after variable replacement for things to work
 967                 // properly; putting them before other transformations should keep
 968                 // exciting things like link expansions from showing up in surprising
 969                 // places.
 970                 $text = $this->doTableStuff( $text );
 971
 972                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 973
 974                 $text = $this->doDoubleUnderscore( $text );
 975                 $text = $this->doHeadings( $text );
 976                 if($this->mOptions->getUseDynamicDates()) {
 977                         $df = DateFormatter::getInstance();
 978                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 979                 }
 980                 $text = $this->doAllQuotes( $text );
 981                 $text = $this->replaceInternalLinks( $text );
 982                 $text = $this->replaceExternalLinks( $text );
 983
 984                 # replaceInternalLinks may sometimes leave behind
 985                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 986                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 987
 988                 $text = $this->doMagicLinks( $text );
 989                 $text = $this->formatHeadings( $text, $isMain );
 990
 991                 wfProfileOut( $fname );
 992                 return $text;
 993         }
 994
 995         /**
 996          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 997          * magic external links.
 998          *
 999          * @private
1000          */
1001         function doMagicLinks( $text ) {
1002                 wfProfileIn( __METHOD__ );
1003                 $text = preg_replace_callback(
1004                         '!(?:                           # Start cases
1005                             <a.*?</a> |                 # Skip link text
1006                             <.*?> |                     # Skip stuff inside HTML elements
1007                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
1008                             ISBN\s+(\b                  # ISBN, capture number as m[2]
1009                                      (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1010                                      (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1011                                      [0-9Xx]                 # check digit
1012                                    \b)
1013                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1014                 wfProfileOut( __METHOD__ );
1015                 return $text;
1016         }
1017
1018         function magicLinkCallback( $m ) {
1019                 if ( substr( $m[0], 0, 1 ) == '<' ) {
1020                         # Skip HTML element
1021                         return $m[0];
1022                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
1023                         $isbn = $m[2];
1024                         $num = strtr( $isbn, array(
1025                                 '-' => '',
1026                                 ' ' => '',
1027                                 'x' => 'X',
1028                         ));
1029                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1030                         $text = '<a href="' .
1031                                 $titleObj->escapeLocalUrl() .
1032                                 "\" class=\"internal\">ISBN $isbn</a>";
1033                 } else {
1034                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
1035                                 $keyword = 'RFC';
1036                                 $urlmsg = 'rfcurl';
1037                                 $id = $m[1];
1038                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
1039                                 $keyword = 'PMID';
1040                                 $urlmsg = 'pubmedurl';
1041                                 $id = $m[1];
1042                         } else {
1043                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1044                                         substr($m[0], 0, 20 ) . '"' );
1045                         }
1046
1047                         $url = wfMsg( $urlmsg, $id);
1048                         $sk = $this->mOptions->getSkin();
1049                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1050                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1051                 }
1052                 return $text;
1053         }
1054
1055         /**
1056          * Parse headers and return html
1057          *
1058          * @private
1059          */
1060         function doHeadings( $text ) {
1061                 $fname = 'Parser::doHeadings';
1062                 wfProfileIn( $fname );
1063                 for ( $i = 6; $i >= 1; --$i ) {
1064                         $h = str_repeat( '=', $i );
1065                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1066                           "<h$i>\\1</h$i>", $text );
1067                 }
1068                 wfProfileOut( $fname );
1069                 return $text;
1070         }
1071
1072         /**
1073          * Replace single quotes with HTML markup
1074          * @private
1075          * @return string the altered text
1076          */
1077         function doAllQuotes( $text ) {
1078                 $fname = 'Parser::doAllQuotes';
1079                 wfProfileIn( $fname );
1080                 $outtext = '';
1081                 $lines = explode( "\n", $text );
1082                 foreach ( $lines as $line ) {
1083                         $outtext .= $this->doQuotes ( $line ) . "\n";
1084                 }
1085                 $outtext = substr($outtext, 0,-1);
1086                 wfProfileOut( $fname );
1087                 return $outtext;
1088         }
1089
1090         /**
1091          * Helper function for doAllQuotes()
1092          */
1093         public function doQuotes( $text ) {
1094                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1095                 if ( count( $arr ) == 1 )
1096                         return $text;
1097                 else
1098                 {
1099                         # First, do some preliminary work. This may shift some apostrophes from
1100                         # being mark-up to being text. It also counts the number of occurrences
1101                         # of bold and italics mark-ups.
1102                         $i = 0;
1103                         $numbold = 0;
1104                         $numitalics = 0;
1105                         foreach ( $arr as $r )
1106                         {
1107                                 if ( ( $i % 2 ) == 1 )
1108                                 {
1109                                         # If there are ever four apostrophes, assume the first is supposed to
1110                                         # be text, and the remaining three constitute mark-up for bold text.
1111                                         if ( strlen( $arr[$i] ) == 4 )
1112                                         {
1113                                                 $arr[$i-1] .= "'";
1114                                                 $arr[$i] = "'''";
1115                                         }
1116                                         # If there are more than 5 apostrophes in a row, assume they're all
1117                                         # text except for the last 5.
1118                                         else if ( strlen( $arr[$i] ) > 5 )
1119                                         {
1120                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1121                                                 $arr[$i] = "'''''";
1122                                         }
1123                                         # Count the number of occurrences of bold and italics mark-ups.
1124                                         # We are not counting sequences of five apostrophes.
1125                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1126                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1127                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1128                                 }
1129                                 $i++;
1130                         }
1131
1132                         # If there is an odd number of both bold and italics, it is likely
1133                         # that one of the bold ones was meant to be an apostrophe followed
1134                         # by italics. Which one we cannot know for certain, but it is more
1135                         # likely to be one that has a single-letter word before it.
1136                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1137                         {
1138                                 $i = 0;
1139                                 $firstsingleletterword = -1;
1140                                 $firstmultiletterword = -1;
1141                                 $firstspace = -1;
1142                                 foreach ( $arr as $r )
1143                                 {
1144                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1145                                         {
1146                                                 $x1 = substr ($arr[$i-1], -1);
1147                                                 $x2 = substr ($arr[$i-1], -2, 1);
1148                                                 if ($x1 == ' ') {
1149                                                         if ($firstspace == -1) $firstspace = $i;
1150                                                 } else if ($x2 == ' ') {
1151                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1152                                                 } else {
1153                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1154                                                 }
1155                                         }
1156                                         $i++;
1157                                 }
1158
1159                                 # If there is a single-letter word, use it!
1160                                 if ($firstsingleletterword > -1)
1161                                 {
1162                                         $arr [ $firstsingleletterword ] = "''";
1163                                         $arr [ $firstsingleletterword-1 ] .= "'";
1164                                 }
1165                                 # If not, but there's a multi-letter word, use that one.
1166                                 else if ($firstmultiletterword > -1)
1167                                 {
1168                                         $arr [ $firstmultiletterword ] = "''";
1169                                         $arr [ $firstmultiletterword-1 ] .= "'";
1170                                 }
1171                                 # ... otherwise use the first one that has neither.
1172                                 # (notice that it is possible for all three to be -1 if, for example,
1173                                 # there is only one pentuple-apostrophe in the line)
1174                                 else if ($firstspace > -1)
1175                                 {
1176                                         $arr [ $firstspace ] = "''";
1177                                         $arr [ $firstspace-1 ] .= "'";
1178                                 }
1179                         }
1180
1181                         # Now let's actually convert our apostrophic mush to HTML!
1182                         $output = '';
1183                         $buffer = '';
1184                         $state = '';
1185                         $i = 0;
1186                         foreach ($arr as $r)
1187                         {
1188                                 if (($i % 2) == 0)
1189                                 {
1190                                         if ($state == 'both')
1191                                                 $buffer .= $r;
1192                                         else
1193                                                 $output .= $r;
1194                                 }
1195                                 else
1196                                 {
1197                                         if (strlen ($r) == 2)
1198                                         {
1199                                                 if ($state == 'i')
1200                                                 { $output .= '</i>'; $state = ''; }
1201                                                 else if ($state == 'bi')
1202                                                 { $output .= '</i>'; $state = 'b'; }
1203                                                 else if ($state == 'ib')
1204                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1205                                                 else if ($state == 'both')
1206                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1207                                                 else # $state can be 'b' or ''
1208                                                 { $output .= '<i>'; $state .= 'i'; }
1209                                         }
1210                                         else if (strlen ($r) == 3)
1211                                         {
1212                                                 if ($state == 'b')
1213                                                 { $output .= '</b>'; $state = ''; }
1214                                                 else if ($state == 'bi')
1215                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1216                                                 else if ($state == 'ib')
1217                                                 { $output .= '</b>'; $state = 'i'; }
1218                                                 else if ($state == 'both')
1219                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1220                                                 else # $state can be 'i' or ''
1221                                                 { $output .= '<b>'; $state .= 'b'; }
1222                                         }
1223                                         else if (strlen ($r) == 5)
1224                                         {
1225                                                 if ($state == 'b')
1226                                                 { $output .= '</b><i>'; $state = 'i'; }
1227                                                 else if ($state == 'i')
1228                                                 { $output .= '</i><b>'; $state = 'b'; }
1229                                                 else if ($state == 'bi')
1230                                                 { $output .= '</i></b>'; $state = ''; }
1231                                                 else if ($state == 'ib')
1232                                                 { $output .= '</b></i>'; $state = ''; }
1233                                                 else if ($state == 'both')
1234                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1235                                                 else # ($state == '')
1236                                                 { $buffer = ''; $state = 'both'; }
1237                                         }
1238                                 }
1239                                 $i++;
1240                         }
1241                         # Now close all remaining tags.  Notice that the order is important.
1242                         if ($state == 'b' || $state == 'ib')
1243                                 $output .= '</b>';
1244                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1245                                 $output .= '</i>';
1246                         if ($state == 'bi')
1247                                 $output .= '</b>';
1248                         # There might be lonely ''''', so make sure we have a buffer
1249                         if ($state == 'both' && $buffer)
1250                                 $output .= '<b><i>'.$buffer.'</i></b>';
1251                         return $output;
1252                 }
1253         }
1254
1255         /**
1256          * Replace external links
1257          *
1258          * Note: this is all very hackish and the order of execution matters a lot.
1259          * Make sure to run maintenance/parserTests.php if you change this code.
1260          *
1261          * @private
1262          */
1263         function replaceExternalLinks( $text ) {
1264                 global $wgContLang;
1265                 $fname = 'Parser::replaceExternalLinks';
1266                 wfProfileIn( $fname );
1267
1268                 $sk = $this->mOptions->getSkin();
1269
1270                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1271
1272                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1273
1274                 $i = 0;
1275                 while ( $i<count( $bits ) ) {
1276                         $url = $bits[$i++];
1277                         $protocol = $bits[$i++];
1278                         $text = $bits[$i++];
1279                         $trail = $bits[$i++];
1280
1281                         # The characters '<' and '>' (which were escaped by
1282                         # removeHTMLtags()) should not be included in
1283                         # URLs, per RFC 2396.
1284                         $m2 = array();
1285                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1286                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1287                                 $url = substr($url, 0, $m2[0][1]);
1288                         }
1289
1290                         # If the link text is an image URL, replace it with an <img> tag
1291                         # This happened by accident in the original parser, but some people used it extensively
1292                         $img = $this->maybeMakeExternalImage( $text );
1293                         if ( $img !== false ) {
1294                                 $text = $img;
1295                         }
1296
1297                         $dtrail = '';
1298
1299                         # Set linktype for CSS - if URL==text, link is essentially free
1300                         $linktype = ($text == $url) ? 'free' : 'text';
1301
1302                         # No link text, e.g. [http://domain.tld/some.link]
1303                         if ( $text == '' ) {
1304                                 # Autonumber if allowed. See bug #5918
1305                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1306                                         $text = '[' . ++$this->mAutonumber . ']';
1307                                         $linktype = 'autonumber';
1308                                 } else {
1309                                         # Otherwise just use the URL
1310                                         $text = htmlspecialchars( $url );
1311                                         $linktype = 'free';
1312                                 }
1313                         } else {
1314                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1315                                 # Check for trail
1316                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1317                         }
1318
1319                         $text = $wgContLang->markNoConversion($text);
1320
1321                         $url = Sanitizer::cleanUrl( $url );
1322
1323                         # Process the trail (i.e. everything after this link up until start of the next link),
1324                         # replacing any non-bracketed links
1325                         $trail = $this->replaceFreeExternalLinks( $trail );
1326
1327                         # Use the encoded URL
1328                         # This means that users can paste URLs directly into the text
1329                         # Funny characters like &ouml; aren't valid in URLs anyway
1330                         # This was changed in August 2004
1331                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1332
1333                         # Register link in the output object.
1334                         # Replace unnecessary URL escape codes with the referenced character
1335                         # This prevents spammers from hiding links from the filters
1336                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1337                         $this->mOutput->addExternalLink( $pasteurized );
1338                 }
1339
1340                 wfProfileOut( $fname );
1341                 return $s;
1342         }
1343
1344         /**
1345          * Replace anything that looks like a URL with a link
1346          * @private
1347          */
1348         function replaceFreeExternalLinks( $text ) {
1349                 global $wgContLang;
1350                 $fname = 'Parser::replaceFreeExternalLinks';
1351                 wfProfileIn( $fname );
1352
1353                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1354                 $s = array_shift( $bits );
1355                 $i = 0;
1356
1357                 $sk = $this->mOptions->getSkin();
1358
1359                 while ( $i < count( $bits ) ){
1360                         $protocol = $bits[$i++];
1361                         $remainder = $bits[$i++];
1362
1363                         $m = array();
1364                         if ( preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1365                                 # Found some characters after the protocol that look promising
1366                                 $url = $protocol . $m[1];
1367                                 $trail = $m[2];
1368
1369                                 # special case: handle urls as url args:
1370                                 # http://www.example.com/foo?=http://www.example.com/bar
1371                                 if(strlen($trail) == 0 &&
1372                                         isset($bits[$i]) &&
1373                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1374                                         preg_match( '/^('.self::EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1375                                 {
1376                                         # add protocol, arg
1377                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1378                                         $i += 2;
1379                                         $trail = $m[2];
1380                                 }
1381
1382                                 # The characters '<' and '>' (which were escaped by
1383                                 # removeHTMLtags()) should not be included in
1384                                 # URLs, per RFC 2396.
1385                                 $m2 = array();
1386                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1387                                         $trail = substr($url, $m2[0][1]) . $trail;
1388                                         $url = substr($url, 0, $m2[0][1]);
1389                                 }
1390
1391                                 # Move trailing punctuation to $trail
1392                                 $sep = ',;\.:!?';
1393                                 # If there is no left bracket, then consider right brackets fair game too
1394                                 if ( strpos( $url, '(' ) === false ) {
1395                                         $sep .= ')';
1396                                 }
1397
1398                                 $numSepChars = strspn( strrev( $url ), $sep );
1399                                 if ( $numSepChars ) {
1400                                         $trail = substr( $url, -$numSepChars ) . $trail;
1401                                         $url = substr( $url, 0, -$numSepChars );
1402                                 }
1403
1404                                 $url = Sanitizer::cleanUrl( $url );
1405
1406                                 # Is this an external image?
1407                                 $text = $this->maybeMakeExternalImage( $url );
1408                                 if ( $text === false ) {
1409                                         # Not an image, make a link
1410                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1411                                         # Register it in the output object...
1412                                         # Replace unnecessary URL escape codes with their equivalent characters
1413                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1414                                         $this->mOutput->addExternalLink( $pasteurized );
1415                                 }
1416                                 $s .= $text . $trail;
1417                         } else {
1418                                 $s .= $protocol . $remainder;
1419                         }
1420                 }
1421                 wfProfileOut( $fname );
1422                 return $s;
1423         }
1424
1425         /**
1426          * Replace unusual URL escape codes with their equivalent characters
1427          * @param string
1428          * @return string
1429          * @static
1430          * @todo  This can merge genuinely required bits in the path or query string,
1431          *        breaking legit URLs. A proper fix would treat the various parts of
1432          *        the URL differently; as a workaround, just use the output for
1433          *        statistical records, not for actual linking/output.
1434          */
1435         static function replaceUnusualEscapes( $url ) {
1436                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1437                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1438         }
1439
1440         /**
1441          * Callback function used in replaceUnusualEscapes().
1442          * Replaces unusual URL escape codes with their equivalent character
1443          * @static
1444          * @private
1445          */
1446         private static function replaceUnusualEscapesCallback( $matches ) {
1447                 $char = urldecode( $matches[0] );
1448                 $ord = ord( $char );
1449                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1450                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1451                         // No, shouldn't be escaped
1452                         return $char;
1453                 } else {
1454                         // Yes, leave it escaped
1455                         return $matches[0];
1456                 }
1457         }
1458
1459         /**
1460          * make an image if it's allowed, either through the global
1461          * option or through the exception
1462          * @private
1463          */
1464         function maybeMakeExternalImage( $url ) {
1465                 $sk = $this->mOptions->getSkin();
1466                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1467                 $imagesexception = !empty($imagesfrom);
1468                 $text = false;
1469                 if ( $this->mOptions->getAllowExternalImages()
1470                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1471                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1472                                 # Image found
1473                                 $text = $sk->makeExternalImage( $url );
1474                         }
1475                 }
1476                 return $text;
1477         }
1478
1479         /**
1480          * Process [[ ]] wikilinks
1481          *
1482          * @private
1483          */
1484         function replaceInternalLinks( $s ) {
1485                 global $wgContLang;
1486                 static $fname = 'Parser::replaceInternalLinks' ;
1487
1488                 wfProfileIn( $fname );
1489
1490                 wfProfileIn( $fname.'-setup' );
1491                 static $tc = FALSE;
1492                 # the % is needed to support urlencoded titles as well
1493                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1494
1495                 $sk = $this->mOptions->getSkin();
1496
1497                 #split the entire text string on occurences of [[
1498                 $a = explode( '[[', ' ' . $s );
1499                 #get the first element (all text up to first [[), and remove the space we added
1500                 $s = array_shift( $a );
1501                 $s = substr( $s, 1 );
1502
1503                 # Match a link having the form [[namespace:link|alternate]]trail
1504                 static $e1 = FALSE;
1505                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1506                 # Match cases where there is no "]]", which might still be images
1507                 static $e1_img = FALSE;
1508                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1509
1510                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1511                 $e2 = null;
1512                 if ( $useLinkPrefixExtension ) {
1513                         # Match the end of a line for a word that's not followed by whitespace,
1514                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1515                         $e2 = wfMsgForContent( 'linkprefix' );
1516                 }
1517
1518                 if( is_null( $this->mTitle ) ) {
1519                         wfProfileOut( $fname );
1520                         wfProfileOut( $fname.'-setup' );
1521                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1522                 }
1523                 $nottalk = !$this->mTitle->isTalkPage();
1524
1525                 if ( $useLinkPrefixExtension ) {
1526                         $m = array();
1527                         if ( preg_match( $e2, $s, $m ) ) {
1528                                 $first_prefix = $m[2];
1529                         } else {
1530                                 $first_prefix = false;
1531                         }
1532                 } else {
1533                         $prefix = '';
1534                 }
1535
1536                 if($wgContLang->hasVariants()) {
1537                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1538                 } else {
1539                         $selflink = array($this->mTitle->getPrefixedText());
1540                 }
1541                 $useSubpages = $this->areSubpagesAllowed();
1542                 wfProfileOut( $fname.'-setup' );
1543
1544                 # Loop for each link
1545                 for ($k = 0; isset( $a[$k] ); $k++) {
1546                         $line = $a[$k];
1547                         if ( $useLinkPrefixExtension ) {
1548                                 wfProfileIn( $fname.'-prefixhandling' );
1549                                 if ( preg_match( $e2, $s, $m ) ) {
1550                                         $prefix = $m[2];
1551                                         $s = $m[1];
1552                                 } else {
1553                                         $prefix='';
1554                                 }
1555                                 # first link
1556                                 if($first_prefix) {
1557                                         $prefix = $first_prefix;
1558                                         $first_prefix = false;
1559                                 }
1560                                 wfProfileOut( $fname.'-prefixhandling' );
1561                         }
1562
1563                         $might_be_img = false;
1564
1565                         wfProfileIn( "$fname-e1" );
1566                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1567                                 $text = $m[2];
1568                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1569                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1570                                 # the real problem is with the $e1 regex
1571                                 # See bug 1300.
1572                                 #
1573                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1574                                 # and no image is in sight. See bug 2095.
1575                                 #
1576                                 if( $text !== '' &&
1577                                         substr( $m[3], 0, 1 ) === ']' &&
1578                                         strpos($text, '[') !== false
1579                                 )
1580                                 {
1581                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1582                                         $m[3] = substr( $m[3], 1 );
1583                                 }
1584                                 # fix up urlencoded title texts
1585                                 if( strpos( $m[1], '%' ) !== false ) {
1586                                         # Should anchors '#' also be rejected?
1587                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1588                                 }
1589                                 $trail = $m[3];
1590                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1591                                 $might_be_img = true;
1592                                 $text = $m[2];
1593                                 if ( strpos( $m[1], '%' ) !== false ) {
1594                                         $m[1] = urldecode($m[1]);
1595                                 }
1596                                 $trail = "";
1597                         } else { # Invalid form; output directly
1598                                 $s .= $prefix . '[[' . $line ;
1599                                 wfProfileOut( "$fname-e1" );
1600                                 continue;
1601                         }
1602                         wfProfileOut( "$fname-e1" );
1603                         wfProfileIn( "$fname-misc" );
1604
1605                         # Don't allow internal links to pages containing
1606                         # PROTO: where PROTO is a valid URL protocol; these
1607                         # should be external links.
1608                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1609                                 $s .= $prefix . '[[' . $line ;
1610                                 wfProfileOut( "$fname-misc" );
1611                                 continue;
1612                         }
1613
1614                         # Make subpage if necessary
1615                         if( $useSubpages ) {
1616                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1617                         } else {
1618                                 $link = $m[1];
1619                         }
1620
1621                         $noforce = (substr($m[1], 0, 1) != ':');
1622                         if (!$noforce) {
1623                                 # Strip off leading ':'
1624                                 $link = substr($link, 1);
1625                         }
1626
1627                         wfProfileOut( "$fname-misc" );
1628                         wfProfileIn( "$fname-title" );
1629                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1630                         if( !$nt ) {
1631                                 $s .= $prefix . '[[' . $line;
1632                                 wfProfileOut( "$fname-title" );
1633                                 continue;
1634                         }
1635
1636                         $ns = $nt->getNamespace();
1637                         $iw = $nt->getInterWiki();
1638                         wfProfileOut( "$fname-title" );
1639
1640                         if ($might_be_img) { # if this is actually an invalid link
1641                                 wfProfileIn( "$fname-might_be_img" );
1642                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1643                                         $found = false;
1644                                         while (isset ($a[$k+1]) ) {
1645                                                 #look at the next 'line' to see if we can close it there
1646                                                 $spliced = array_splice( $a, $k + 1, 1 );
1647                                                 $next_line = array_shift( $spliced );
1648                                                 $m = explode( ']]', $next_line, 3 );
1649                                                 if ( count( $m ) == 3 ) {
1650                                                         # the first ]] closes the inner link, the second the image
1651                                                         $found = true;
1652                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1653                                                         $trail = $m[2];
1654                                                         break;
1655                                                 } elseif ( count( $m ) == 2 ) {
1656                                                         #if there's exactly one ]] that's fine, we'll keep looking
1657                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1658                                                 } else {
1659                                                         #if $next_line is invalid too, we need look no further
1660                                                         $text .= '[[' . $next_line;
1661                                                         break;
1662                                                 }
1663                                         }
1664                                         if ( !$found ) {
1665                                                 # we couldn't find the end of this imageLink, so output it raw
1666                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1667                                                 $text = $this->replaceInternalLinks($text);
1668                                                 $s .= "{$prefix}[[$link|$text";
1669                                                 # note: no $trail, because without an end, there *is* no trail
1670                                                 wfProfileOut( "$fname-might_be_img" );
1671                                                 continue;
1672                                         }
1673                                 } else { #it's not an image, so output it raw
1674                                         $s .= "{$prefix}[[$link|$text";
1675                                         # note: no $trail, because without an end, there *is* no trail
1676                                         wfProfileOut( "$fname-might_be_img" );
1677                                         continue;
1678                                 }
1679                                 wfProfileOut( "$fname-might_be_img" );
1680                         }
1681
1682                         $wasblank = ( '' == $text );
1683                         if( $wasblank ) $text = $link;
1684
1685                         # Link not escaped by : , create the various objects
1686                         if( $noforce ) {
1687
1688                                 # Interwikis
1689                                 wfProfileIn( "$fname-interwiki" );
1690                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1691                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1692                                         $s = rtrim($s . $prefix);
1693                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1694                                         wfProfileOut( "$fname-interwiki" );
1695                                         continue;
1696                                 }
1697                                 wfProfileOut( "$fname-interwiki" );
1698
1699                                 if ( $ns == NS_IMAGE ) {
1700                                         wfProfileIn( "$fname-image" );
1701                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1702                                                 # recursively parse links inside the image caption
1703                                                 # actually, this will parse them in any other parameters, too,
1704                                                 # but it might be hard to fix that, and it doesn't matter ATM
1705                                                 $text = $this->replaceExternalLinks($text);
1706                                                 $text = $this->replaceInternalLinks($text);
1707
1708                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1709                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1710                                                 $this->mOutput->addImage( $nt->getDBkey() );
1711
1712                                                 wfProfileOut( "$fname-image" );
1713                                                 continue;
1714                                         } else {
1715                                                 # We still need to record the image's presence on the page
1716                                                 $this->mOutput->addImage( $nt->getDBkey() );
1717                                         }
1718                                         wfProfileOut( "$fname-image" );
1719
1720                                 }
1721
1722                                 if ( $ns == NS_CATEGORY ) {
1723                                         wfProfileIn( "$fname-category" );
1724                                         $s = rtrim($s . "\n"); # bug 87
1725
1726                                         if ( $wasblank ) {
1727                                                 $sortkey = $this->getDefaultSort();
1728                                         } else {
1729                                                 $sortkey = $text;
1730                                         }
1731                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1732                                         $sortkey = str_replace( "\n", '', $sortkey );
1733                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1734                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1735
1736                                         /**
1737                                          * Strip the whitespace Category links produce, see bug 87
1738                                          * @todo We might want to use trim($tmp, "\n") here.
1739                                          */
1740                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1741
1742                                         wfProfileOut( "$fname-category" );
1743                                         continue;
1744                                 }
1745                         }
1746
1747                         # Self-link checking
1748                         if( $nt->getFragment() === '' ) {
1749                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1750                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1751                                         continue;
1752                                 }
1753                         }
1754
1755                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1756                         if( $ns == NS_MEDIA ) {
1757                                 # Give extensions a chance to select the file revision for us
1758                                 $skip = $time = false;
1759                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1760                                 if ( $skip ) {
1761                                         $link = $sk->makeLinkObj( $nt );
1762                                 } else {
1763                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1764                                 }
1765                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1766                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1767                                 $this->mOutput->addImage( $nt->getDBkey() );
1768                                 continue;
1769                         } elseif( $ns == NS_SPECIAL ) {
1770                                 if( SpecialPage::exists( $nt->getDBkey() ) ) {
1771                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1772                                 } else {
1773                                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1774                                 }
1775                                 continue;
1776                         } elseif( $ns == NS_IMAGE ) {
1777                                 $img = wfFindFile( $nt );
1778                                 if( $img ) {
1779                                         // Force a blue link if the file exists; may be a remote
1780                                         // upload on the shared repository, and we want to see its
1781                                         // auto-generated page.
1782                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1783                                         $this->mOutput->addLink( $nt );
1784                                         continue;
1785                                 }
1786                         }
1787                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1788                 }
1789                 wfProfileOut( $fname );
1790                 return $s;
1791         }
1792
1793         /**
1794          * Make a link placeholder. The text returned can be later resolved to a real link with
1795          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1796          * parsing of interwiki links, and secondly to allow all existence checks and
1797          * article length checks (for stub links) to be bundled into a single query.
1798          *
1799          */
1800         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1801                 wfProfileIn( __METHOD__ );
1802                 if ( ! is_object($nt) ) {
1803                         # Fail gracefully
1804                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1805                 } else {
1806                         # Separate the link trail from the rest of the link
1807                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1808
1809                         if ( $nt->isExternal() ) {
1810                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1811                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1812                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1813                         } else {
1814                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1815                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1816                                 $this->mLinkHolders['queries'][] = $query;
1817                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1818                                 $this->mLinkHolders['titles'][] = $nt;
1819
1820                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1821                         }
1822                 }
1823                 wfProfileOut( __METHOD__ );
1824                 return $retVal;
1825         }
1826
1827         /**
1828          * Render a forced-blue link inline; protect against double expansion of
1829          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1830          * Since this little disaster has to split off the trail text to avoid
1831          * breaking URLs in the following text without breaking trails on the
1832          * wiki links, it's been made into a horrible function.
1833          *
1834          * @param Title $nt
1835          * @param string $text
1836          * @param string $query
1837          * @param string $trail
1838          * @param string $prefix
1839          * @return string HTML-wikitext mix oh yuck
1840          */
1841         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1842                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1843                 $sk = $this->mOptions->getSkin();
1844                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1845                 return $this->armorLinks( $link ) . $trail;
1846         }
1847
1848         /**
1849          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1850          * going to go through further parsing steps before inline URL expansion.
1851          *
1852          * In particular this is important when using action=render, which causes
1853          * full URLs to be included.
1854          *
1855          * Oh man I hate our multi-layer parser!
1856          *
1857          * @param string more-or-less HTML
1858          * @return string less-or-more HTML with NOPARSE bits
1859          */
1860         function armorLinks( $text ) {
1861                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1862                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1863         }
1864
1865         /**
1866          * Return true if subpage links should be expanded on this page.
1867          * @return bool
1868          */
1869         function areSubpagesAllowed() {
1870                 # Some namespaces don't allow subpages
1871                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1872         }
1873
1874         /**
1875          * Handle link to subpage if necessary
1876          * @param string $target the source of the link
1877          * @param string &$text the link text, modified as necessary
1878          * @return string the full name of the link
1879          * @private
1880          */
1881         function maybeDoSubpageLink($target, &$text) {
1882                 # Valid link forms:
1883                 # Foobar -- normal
1884                 # :Foobar -- override special treatment of prefix (images, language links)
1885                 # /Foobar -- convert to CurrentPage/Foobar
1886                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1887                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1888                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1889
1890                 $fname = 'Parser::maybeDoSubpageLink';
1891                 wfProfileIn( $fname );
1892                 $ret = $target; # default return value is no change
1893
1894                 # Some namespaces don't allow subpages,
1895                 # so only perform processing if subpages are allowed
1896                 if( $this->areSubpagesAllowed() ) {
1897                         $hash = strpos( $target, '#' );
1898                         if( $hash !== false ) {
1899                                 $suffix = substr( $target, $hash );
1900                                 $target = substr( $target, 0, $hash );
1901                         } else {
1902                                 $suffix = '';
1903                         }
1904                         # bug 7425
1905                         $target = trim( $target );
1906                         # Look at the first character
1907                         if( $target != '' && $target{0} == '/' ) {
1908                                 # / at end means we don't want the slash to be shown
1909                                 $m = array();
1910                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1911                                 if( $trailingSlashes ) {
1912                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1913                                 } else {
1914                                         $noslash = substr( $target, 1 );
1915                                 }
1916
1917                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1918                                 if( '' === $text ) {
1919                                         $text = $target . $suffix;
1920                                 } # this might be changed for ugliness reasons
1921                         } else {
1922                                 # check for .. subpage backlinks
1923                                 $dotdotcount = 0;
1924                                 $nodotdot = $target;
1925                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1926                                         ++$dotdotcount;
1927                                         $nodotdot = substr( $nodotdot, 3 );
1928                                 }
1929                                 if($dotdotcount > 0) {
1930                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1931                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1932                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1933                                                 # / at the end means don't show full path
1934                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1935                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1936                                                         if( '' === $text ) {
1937                                                                 $text = $nodotdot . $suffix;
1938                                                         }
1939                                                 }
1940                                                 $nodotdot = trim( $nodotdot );
1941                                                 if( $nodotdot != '' ) {
1942                                                         $ret .= '/' . $nodotdot;
1943                                                 }
1944                                                 $ret .= $suffix;
1945                                         }
1946                                 }
1947                         }
1948                 }
1949
1950                 wfProfileOut( $fname );
1951                 return $ret;
1952         }
1953
1954         /**#@+
1955          * Used by doBlockLevels()
1956          * @private
1957          */
1958         /* private */ function closeParagraph() {
1959                 $result = '';
1960                 if ( '' != $this->mLastSection ) {
1961                         $result = '</' . $this->mLastSection  . ">\n";
1962                 }
1963                 $this->mInPre = false;
1964                 $this->mLastSection = '';
1965                 return $result;
1966         }
1967         # getCommon() returns the length of the longest common substring
1968         # of both arguments, starting at the beginning of both.
1969         #
1970         /* private */ function getCommon( $st1, $st2 ) {
1971                 $fl = strlen( $st1 );
1972                 $shorter = strlen( $st2 );
1973                 if ( $fl < $shorter ) { $shorter = $fl; }
1974
1975                 for ( $i = 0; $i < $shorter; ++$i ) {
1976                         if ( $st1{$i} != $st2{$i} ) { break; }
1977                 }
1978                 return $i;
1979         }
1980         # These next three functions open, continue, and close the list
1981         # element appropriate to the prefix character passed into them.
1982         #
1983         /* private */ function openList( $char ) {
1984                 $result = $this->closeParagraph();
1985
1986                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1987                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1988                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1989                 else if ( ';' == $char ) {
1990                         $result .= '<dl><dt>';
1991                         $this->mDTopen = true;
1992                 }
1993                 else { $result = '<!-- ERR 1 -->'; }
1994
1995                 return $result;
1996         }
1997
1998         /* private */ function nextItem( $char ) {
1999                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
2000                 else if ( ':' == $char || ';' == $char ) {
2001                         $close = '</dd>';
2002                         if ( $this->mDTopen ) { $close = '</dt>'; }
2003                         if ( ';' == $char ) {
2004                                 $this->mDTopen = true;
2005                                 return $close . '<dt>';
2006                         } else {
2007                                 $this->mDTopen = false;
2008                                 return $close . '<dd>';
2009                         }
2010                 }
2011                 return '<!-- ERR 2 -->';
2012         }
2013
2014         /* private */ function closeList( $char ) {
2015                 if ( '*' == $char ) { $text = '</li></ul>'; }
2016                 else if ( '#' == $char ) { $text = '</li></ol>'; }
2017                 else if ( ':' == $char ) {
2018                         if ( $this->mDTopen ) {
2019                                 $this->mDTopen = false;
2020                                 $text = '</dt></dl>';
2021                         } else {
2022                                 $text = '</dd></dl>';
2023                         }
2024                 }
2025                 else {  return '<!-- ERR 3 -->'; }
2026                 return $text."\n";
2027         }
2028         /**#@-*/
2029
2030         /**
2031          * Make lists from lines starting with ':', '*', '#', etc.
2032          *
2033          * @private
2034          * @return string the lists rendered as HTML
2035          */
2036         function doBlockLevels( $text, $linestart ) {
2037                 $fname = 'Parser::doBlockLevels';
2038                 wfProfileIn( $fname );
2039
2040                 # Parsing through the text line by line.  The main thing
2041                 # happening here is handling of block-level elements p, pre,
2042                 # and making lists from lines starting with * # : etc.
2043                 #
2044                 $textLines = explode( "\n", $text );
2045
2046                 $lastPrefix = $output = '';
2047                 $this->mDTopen = $inBlockElem = false;
2048                 $prefixLength = 0;
2049                 $paragraphStack = false;
2050
2051                 if ( !$linestart ) {
2052                         $output .= array_shift( $textLines );
2053                 }
2054                 foreach ( $textLines as $oLine ) {
2055                         $lastPrefixLength = strlen( $lastPrefix );
2056                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2057                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2058                         if ( !$this->mInPre ) {
2059                                 # Multiple prefixes may abut each other for nested lists.
2060                                 $prefixLength = strspn( $oLine, '*#:;' );
2061                                 $pref = substr( $oLine, 0, $prefixLength );
2062
2063                                 # eh?
2064                                 $pref2 = str_replace( ';', ':', $pref );
2065                                 $t = substr( $oLine, $prefixLength );
2066                                 $this->mInPre = !empty($preOpenMatch);
2067                         } else {
2068                                 # Don't interpret any other prefixes in preformatted text
2069                                 $prefixLength = 0;
2070                                 $pref = $pref2 = '';
2071                                 $t = $oLine;
2072                         }
2073
2074                         # List generation
2075                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2076                                 # Same as the last item, so no need to deal with nesting or opening stuff
2077                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2078                                 $paragraphStack = false;
2079
2080                                 if ( substr( $pref, -1 ) == ';') {
2081                                         # The one nasty exception: definition lists work like this:
2082                                         # ; title : definition text
2083                                         # So we check for : in the remainder text to split up the
2084                                         # title and definition, without b0rking links.
2085                                         $term = $t2 = '';
2086                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2087                                                 $t = $t2;
2088                                                 $output .= $term . $this->nextItem( ':' );
2089                                         }
2090                                 }
2091                         } elseif( $prefixLength || $lastPrefixLength ) {
2092                                 # Either open or close a level...
2093                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2094                                 $paragraphStack = false;
2095
2096                                 while( $commonPrefixLength < $lastPrefixLength ) {
2097                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2098                                         --$lastPrefixLength;
2099                                 }
2100                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2101                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2102                                 }
2103                                 while ( $prefixLength > $commonPrefixLength ) {
2104                                         $char = substr( $pref, $commonPrefixLength, 1 );
2105                                         $output .= $this->openList( $char );
2106
2107                                         if ( ';' == $char ) {
2108                                                 # FIXME: This is dupe of code above
2109                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2110                                                         $t = $t2;
2111                                                         $output .= $term . $this->nextItem( ':' );
2112                                                 }
2113                                         }
2114                                         ++$commonPrefixLength;
2115                                 }
2116                                 $lastPrefix = $pref2;
2117                         }
2118                         if( 0 == $prefixLength ) {
2119                                 wfProfileIn( "$fname-paragraph" );
2120                                 # No prefix (not in list)--go to paragraph mode
2121                                 // XXX: use a stack for nestable elements like span, table and div
2122                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2123                                 $closematch = preg_match(
2124                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2125                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2126                                 if ( $openmatch or $closematch ) {
2127                                         $paragraphStack = false;
2128                                         # TODO bug 5718: paragraph closed
2129                                         $output .= $this->closeParagraph();
2130                                         if ( $preOpenMatch and !$preCloseMatch ) {
2131                                                 $this->mInPre = true;
2132                                         }
2133                                         if ( $closematch ) {
2134                                                 $inBlockElem = false;
2135                                         } else {
2136                                                 $inBlockElem = true;
2137                                         }
2138                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2139                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2140                                                 // pre
2141                                                 if ($this->mLastSection != 'pre') {
2142                                                         $paragraphStack = false;
2143                                                         $output .= $this->closeParagraph().'<pre>';
2144                                                         $this->mLastSection = 'pre';
2145                                                 }
2146                                                 $t = substr( $t, 1 );
2147                                         } else {
2148                                                 // paragraph
2149                                                 if ( '' == trim($t) ) {
2150                                                         if ( $paragraphStack ) {
2151                                                                 $output .= $paragraphStack.'<br />';
2152                                                                 $paragraphStack = false;
2153                                                                 $this->mLastSection = 'p';
2154                                                         } else {
2155                                                                 if ($this->mLastSection != 'p' ) {
2156                                                                         $output .= $this->closeParagraph();
2157                                                                         $this->mLastSection = '';
2158                                                                         $paragraphStack = '<p>';
2159                                                                 } else {
2160                                                                         $paragraphStack = '</p><p>';
2161                                                                 }
2162                                                         }
2163                                                 } else {
2164                                                         if ( $paragraphStack ) {
2165                                                                 $output .= $paragraphStack;
2166                                                                 $paragraphStack = false;
2167                                                                 $this->mLastSection = 'p';
2168                                                         } else if ($this->mLastSection != 'p') {
2169                                                                 $output .= $this->closeParagraph().'<p>';
2170                                                                 $this->mLastSection = 'p';
2171                                                         }
2172                                                 }
2173                                         }
2174                                 }
2175                                 wfProfileOut( "$fname-paragraph" );
2176                         }
2177                         // somewhere above we forget to get out of pre block (bug 785)
2178                         if($preCloseMatch && $this->mInPre) {
2179                                 $this->mInPre = false;
2180                         }
2181                         if ($paragraphStack === false) {
2182                                 $output .= $t."\n";
2183                         }
2184                 }
2185                 while ( $prefixLength ) {
2186                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2187                         --$prefixLength;
2188                 }
2189                 if ( '' != $this->mLastSection ) {
2190                         $output .= '</' . $this->mLastSection . '>';
2191                         $this->mLastSection = '';
2192                 }
2193
2194                 wfProfileOut( $fname );
2195                 return $output;
2196         }
2197
2198         /**
2199          * Split up a string on ':', ignoring any occurences inside tags
2200          * to prevent illegal overlapping.
2201          * @param string $str the string to split
2202          * @param string &$before set to everything before the ':'
2203          * @param string &$after set to everything after the ':'
2204          * return string the position of the ':', or false if none found
2205          */
2206         function findColonNoLinks($str, &$before, &$after) {
2207                 $fname = 'Parser::findColonNoLinks';
2208                 wfProfileIn( $fname );
2209
2210                 $pos = strpos( $str, ':' );
2211                 if( $pos === false ) {
2212                         // Nothing to find!
2213                         wfProfileOut( $fname );
2214                         return false;
2215                 }
2216
2217                 $lt = strpos( $str, '<' );
2218                 if( $lt === false || $lt > $pos ) {
2219                         // Easy; no tag nesting to worry about
2220                         $before = substr( $str, 0, $pos );
2221                         $after = substr( $str, $pos+1 );
2222                         wfProfileOut( $fname );
2223                         return $pos;
2224                 }
2225
2226                 // Ugly state machine to walk through avoiding tags.
2227                 $state = self::COLON_STATE_TEXT;
2228                 $stack = 0;
2229                 $len = strlen( $str );
2230                 for( $i = 0; $i < $len; $i++ ) {
2231                         $c = $str{$i};
2232
2233                         switch( $state ) {
2234                         // (Using the number is a performance hack for common cases)
2235                         case 0: // self::COLON_STATE_TEXT:
2236                                 switch( $c ) {
2237                                 case "<":
2238                                         // Could be either a <start> tag or an </end> tag
2239                                         $state = self::COLON_STATE_TAGSTART;
2240                                         break;
2241                                 case ":":
2242                                         if( $stack == 0 ) {
2243                                                 // We found it!
2244                                                 $before = substr( $str, 0, $i );
2245                                                 $after = substr( $str, $i + 1 );
2246                                                 wfProfileOut( $fname );
2247                                                 return $i;
2248                                         }
2249                                         // Embedded in a tag; don't break it.
2250                                         break;
2251                                 default:
2252                                         // Skip ahead looking for something interesting
2253                                         $colon = strpos( $str, ':', $i );
2254                                         if( $colon === false ) {
2255                                                 // Nothing else interesting
2256                                                 wfProfileOut( $fname );
2257                                                 return false;
2258                                         }
2259                                         $lt = strpos( $str, '<', $i );
2260                                         if( $stack === 0 ) {
2261                                                 if( $lt === false || $colon < $lt ) {
2262                                                         // We found it!
2263                                                         $before = substr( $str, 0, $colon );
2264                                                         $after = substr( $str, $colon + 1 );
2265                                                         wfProfileOut( $fname );
2266                                                         return $i;
2267                                                 }
2268                                         }
2269                                         if( $lt === false ) {
2270                                                 // Nothing else interesting to find; abort!
2271                                                 // We're nested, but there's no close tags left. Abort!
2272                                                 break 2;
2273                                         }
2274                                         // Skip ahead to next tag start
2275                                         $i = $lt;
2276                                         $state = self::COLON_STATE_TAGSTART;
2277                                 }
2278                                 break;
2279                         case 1: // self::COLON_STATE_TAG:
2280                                 // In a <tag>
2281                                 switch( $c ) {
2282                                 case ">":
2283                                         $stack++;
2284                                         $state = self::COLON_STATE_TEXT;
2285                                         break;
2286                                 case "/":
2287                                         // Slash may be followed by >?
2288                                         $state = self::COLON_STATE_TAGSLASH;
2289                                         break;
2290                                 default:
2291                                         // ignore
2292                                 }
2293                                 break;
2294                         case 2: // self::COLON_STATE_TAGSTART:
2295                                 switch( $c ) {
2296                                 case "/":
2297                                         $state = self::COLON_STATE_CLOSETAG;
2298                                         break;
2299                                 case "!":
2300                                         $state = self::COLON_STATE_COMMENT;
2301                                         break;
2302                                 case ">":
2303                                         // Illegal early close? This shouldn't happen D:
2304                                         $state = self::COLON_STATE_TEXT;
2305                                         break;
2306                                 default:
2307                                         $state = self::COLON_STATE_TAG;
2308                                 }
2309                                 break;
2310                         case 3: // self::COLON_STATE_CLOSETAG:
2311                                 // In a </tag>
2312                                 if( $c == ">" ) {
2313                                         $stack--;
2314                                         if( $stack < 0 ) {
2315                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2316                                                 wfProfileOut( $fname );
2317                                                 return false;
2318                                         }
2319                                         $state = self::COLON_STATE_TEXT;
2320                                 }
2321                                 break;
2322                         case self::COLON_STATE_TAGSLASH:
2323                                 if( $c == ">" ) {
2324                                         // Yes, a self-closed tag <blah/>
2325                                         $state = self::COLON_STATE_TEXT;
2326                                 } else {
2327                                         // Probably we're jumping the gun, and this is an attribute
2328                                         $state = self::COLON_STATE_TAG;
2329                                 }
2330                                 break;
2331                         case 5: // self::COLON_STATE_COMMENT:
2332                                 if( $c == "-" ) {
2333                                         $state = self::COLON_STATE_COMMENTDASH;
2334                                 }
2335                                 break;
2336                         case self::COLON_STATE_COMMENTDASH:
2337                                 if( $c == "-" ) {
2338                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2339                                 } else {
2340                                         $state = self::COLON_STATE_COMMENT;
2341                                 }
2342                                 break;
2343                         case self::COLON_STATE_COMMENTDASHDASH:
2344                                 if( $c == ">" ) {
2345                                         $state = self::COLON_STATE_TEXT;
2346                                 } else {
2347                                         $state = self::COLON_STATE_COMMENT;
2348                                 }
2349                                 break;
2350                         default:
2351                                 throw new MWException( "State machine error in $fname" );
2352                         }
2353                 }
2354                 if( $stack > 0 ) {
2355                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2356                         return false;
2357                 }
2358                 wfProfileOut( $fname );
2359                 return false;
2360         }
2361
2362         /**
2363          * Return value of a magic variable (like PAGENAME)
2364          *
2365          * @private
2366          */
2367         function getVariableValue( $index ) {
2368                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2369
2370                 /**
2371                  * Some of these require message or data lookups and can be
2372                  * expensive to check many times.
2373                  */
2374                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2375                         if ( isset( $this->mVarCache[$index] ) ) {
2376                                 return $this->mVarCache[$index];
2377                         }
2378                 }
2379
2380                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2381                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2382
2383                 # Use the time zone
2384                 global $wgLocaltimezone;
2385                 if ( isset( $wgLocaltimezone ) ) {
2386                         $oldtz = getenv( 'TZ' );
2387                         putenv( 'TZ='.$wgLocaltimezone );
2388                 }
2389
2390                 wfSuppressWarnings(); // E_STRICT system time bitching
2391                 $localTimestamp = date( 'YmdHis', $ts );
2392                 $localMonth = date( 'm', $ts );
2393                 $localMonthName = date( 'n', $ts );
2394                 $localDay = date( 'j', $ts );
2395                 $localDay2 = date( 'd', $ts );
2396                 $localDayOfWeek = date( 'w', $ts );
2397                 $localWeek = date( 'W', $ts );
2398                 $localYear = date( 'Y', $ts );
2399                 $localHour = date( 'H', $ts );
2400                 if ( isset( $wgLocaltimezone ) ) {
2401                         putenv( 'TZ='.$oldtz );
2402                 }
2403                 wfRestoreWarnings();
2404
2405                 switch ( $index ) {
2406                         case 'currentmonth':
2407                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2408                         case 'currentmonthname':
2409                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2410                         case 'currentmonthnamegen':
2411                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2412                         case 'currentmonthabbrev':
2413                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2414                         case 'currentday':
2415                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2416                         case 'currentday2':
2417                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2418                         case 'localmonth':
2419                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localMonth );
2420                         case 'localmonthname':
2421                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( $localMonthName );
2422                         case 'localmonthnamegen':
2423                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2424                         case 'localmonthabbrev':
2425                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2426                         case 'localday':
2427                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay );
2428                         case 'localday2':
2429                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay2 );
2430                         case 'pagename':
2431                                 return wfEscapeWikiText( $this->mTitle->getText() );
2432                         case 'pagenamee':
2433                                 return $this->mTitle->getPartialURL();
2434                         case 'fullpagename':
2435                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2436                         case 'fullpagenamee':
2437                                 return $this->mTitle->getPrefixedURL();
2438                         case 'subpagename':
2439                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2440                         case 'subpagenamee':
2441                                 return $this->mTitle->getSubpageUrlForm();
2442                         case 'basepagename':
2443                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2444                         case 'basepagenamee':
2445                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2446                         case 'talkpagename':
2447                                 if( $this->mTitle->canTalk() ) {
2448                                         $talkPage = $this->mTitle->getTalkPage();
2449                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2450                                 } else {
2451                                         return '';
2452                                 }
2453                         case 'talkpagenamee':
2454                                 if( $this->mTitle->canTalk() ) {
2455                                         $talkPage = $this->mTitle->getTalkPage();
2456                                         return $talkPage->getPrefixedUrl();
2457                                 } else {
2458                                         return '';
2459                                 }
2460                         case 'subjectpagename':
2461                                 $subjPage = $this->mTitle->getSubjectPage();
2462                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2463                         case 'subjectpagenamee':
2464                                 $subjPage = $this->mTitle->getSubjectPage();
2465                                 return $subjPage->getPrefixedUrl();
2466                         case 'revisionid':
2467                                 // Let the edit saving system know we should parse the page
2468                                 // *after* a revision ID has been assigned.
2469                                 $this->mOutput->setFlag( 'vary-revision' );
2470                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2471                                 return $this->mRevisionId;
2472                         case 'revisionday':
2473                                 // Let the edit saving system know we should parse the page
2474                                 // *after* a revision ID has been assigned. This is for null edits.
2475                                 $this->mOutput->setFlag( 'vary-revision' );
2476                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2477                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2478                         case 'revisionday2':
2479                                 // Let the edit saving system know we should parse the page
2480                                 // *after* a revision ID has been assigned. This is for null edits.
2481                                 $this->mOutput->setFlag( 'vary-revision' );
2482                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2483                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2484                         case 'revisionmonth':
2485                                 // Let the edit saving system know we should parse the page
2486                                 // *after* a revision ID has been assigned. This is for null edits.
2487                                 $this->mOutput->setFlag( 'vary-revision' );
2488                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2489                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2490                         case 'revisionyear':
2491                                 // Let the edit saving system know we should parse the page
2492                                 // *after* a revision ID has been assigned. This is for null edits.
2493                                 $this->mOutput->setFlag( 'vary-revision' );
2494                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2495                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2496                         case 'revisiontimestamp':
2497                                 // Let the edit saving system know we should parse the page
2498                                 // *after* a revision ID has been assigned. This is for null edits.
2499                                 $this->mOutput->setFlag( 'vary-revision' );
2500                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2501                                 return $this->getRevisionTimestamp();
2502                         case 'namespace':
2503                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2504                         case 'namespacee':
2505                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2506                         case 'talkspace':
2507                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2508                         case 'talkspacee':
2509                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2510                         case 'subjectspace':
2511                                 return $this->mTitle->getSubjectNsText();
2512                         case 'subjectspacee':
2513                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2514                         case 'currentdayname':
2515                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2516                         case 'currentyear':
2517                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2518                         case 'currenttime':
2519                                 return $this->mVarCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2520                         case 'currenthour':
2521                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2522                         case 'currentweek':
2523                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2524                                 // int to remove the padding
2525                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2526                         case 'currentdow':
2527                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2528                         case 'localdayname':
2529                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2530                         case 'localyear':
2531                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localYear, true );
2532                         case 'localtime':
2533                                 return $this->mVarCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2534                         case 'localhour':
2535                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localHour, true );
2536                         case 'localweek':
2537                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2538                                 // int to remove the padding
2539                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2540                         case 'localdow':
2541                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2542                         case 'numberofarticles':
2543                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2544                         case 'numberoffiles':
2545                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2546                         case 'numberofusers':
2547                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2548                         case 'numberofpages':
2549                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2550                         case 'numberofadmins':
2551                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::numberingroup('sysop') );
2552                         case 'numberofedits':
2553                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
2554                         case 'currenttimestamp':
2555                                 return $this->mVarCache[$index] = wfTimestamp( TS_MW, $ts );
2556                         case 'localtimestamp':
2557                                 return $this->mVarCache[$index] = $localTimestamp;
2558                         case 'currentversion':
2559                                 return $this->mVarCache[$index] = SpecialVersion::getVersion();
2560                         case 'sitename':
2561                                 return $wgSitename;
2562                         case 'server':
2563                                 return $wgServer;
2564                         case 'servername':
2565                                 return $wgServerName;
2566                         case 'scriptpath':
2567                                 return $wgScriptPath;
2568                         case 'directionmark':
2569                                 return $wgContLang->getDirMark();
2570                         case 'contentlanguage':
2571                                 global $wgContLanguageCode;
2572                                 return $wgContLanguageCode;
2573                         default:
2574                                 $ret = null;
2575                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret ) ) )
2576                                         return $ret;
2577                                 else
2578                                         return null;
2579                 }
2580         }
2581
2582         /**
2583          * initialise the magic variables (like CURRENTMONTHNAME)
2584          *
2585          * @private
2586          */
2587         function initialiseVariables() {
2588                 $fname = 'Parser::initialiseVariables';
2589                 wfProfileIn( $fname );
2590                 $variableIDs = MagicWord::getVariableIDs();
2591
2592                 $this->mVariables = new MagicWordArray( $variableIDs );
2593                 wfProfileOut( $fname );
2594         }
2595
2596         /**
2597          * Preprocess some wikitext and return the document tree.
2598          * This is the ghost of replace_variables().
2599          *
2600          * @param string $text The text to parse
2601          * @param integer flags Bitwise combination of:
2602          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2603          *                                     included. Default is to assume a direct page view.
2604          *
2605          * The generated DOM tree must depend only on the input text and the flags.
2606          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2607          *
2608          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2609          * change in the DOM tree for a given text, must be passed through the section identifier
2610          * in the section edit link and thus back to extractSections().
2611          *
2612          * The output of this function is currently only cached in process memory, but a persistent
2613          * cache may be implemented at a later date which takes further advantage of these strict
2614          * dependency requirements.
2615          *
2616          * @private
2617          */
2618         function preprocessToDom ( $text, $flags = 0 ) {
2619                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2620                 return $dom;
2621         }
2622
2623         /*
2624          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2625          */
2626         public static function splitWhitespace( $s ) {
2627                 $ltrimmed = ltrim( $s );
2628                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2629                 $trimmed = rtrim( $ltrimmed );
2630                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2631                 if ( $diff > 0 ) {
2632                         $w2 = substr( $ltrimmed, -$diff );
2633                 } else {
2634                         $w2 = '';
2635                 }
2636                 return array( $w1, $trimmed, $w2 );
2637         }
2638
2639         /**
2640          * Replace magic variables, templates, and template arguments
2641          * with the appropriate text. Templates are substituted recursively,
2642          * taking care to avoid infinite loops.
2643          *
2644          * Note that the substitution depends on value of $mOutputType:
2645          *  self::OT_WIKI: only {{subst:}} templates
2646          *  self::OT_PREPROCESS: templates but not extension tags
2647          *  self::OT_HTML: all templates and extension tags
2648          *
2649          * @param string $tex The text to transform
2650          * @param PPFrame $frame Object describing the arguments passed to the template.
2651          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2652          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2653          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2654          * @private
2655          */
2656         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2657                 # Prevent too big inclusions
2658                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2659                         return $text;
2660                 }
2661
2662                 $fname = __METHOD__;
2663                 wfProfileIn( $fname );
2664
2665                 if ( $frame === false ) {
2666                         $frame = $this->getPreprocessor()->newFrame();
2667                 } elseif ( !( $frame instanceof PPFrame ) ) {
2668                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2669                         $frame = $this->getPreprocessor()->newCustomFrame($frame);
2670                 }
2671
2672                 $dom = $this->preprocessToDom( $text );
2673                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2674                 $text = $frame->expand( $dom, $flags );
2675
2676                 wfProfileOut( $fname );
2677                 return $text;
2678         }
2679
2680         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2681         static function createAssocArgs( $args ) {
2682                 $assocArgs = array();
2683                 $index = 1;
2684                 foreach( $args as $arg ) {
2685                         $eqpos = strpos( $arg, '=' );
2686                         if ( $eqpos === false ) {
2687                                 $assocArgs[$index++] = $arg;
2688                         } else {
2689                                 $name = trim( substr( $arg, 0, $eqpos ) );
2690                                 $value = trim( substr( $arg, $eqpos+1 ) );
2691                                 if ( $value === false ) {
2692                                         $value = '';
2693                                 }
2694                                 if ( $name !== false ) {
2695                                         $assocArgs[$name] = $value;
2696                                 }
2697                         }
2698                 }
2699
2700                 return $assocArgs;
2701         }
2702
2703         /**
2704          * Warn the user when a parser limitation is reached
2705          * Will warn at most once the user per limitation type
2706          *
2707          * @param string $limitationType, should be one of:
2708          *   'expensive-parserfunction' (corresponding messages: 'expensive-parserfunction-warning', 'expensive-parserfunction-category')
2709          *   'post-expand-template-argument' (corresponding messages: 'post-expand-template-argument-warning', 'post-expand-template-argument-category')
2710          *   'post-expand-template-inclusion' (corresponding messages: 'post-expand-template-inclusion-warning', 'post-expand-template-inclusion-category')
2711          * @params int $current, $max When an explicit limit has been
2712          *       exceeded, provide the values (optional)
2713          */
2714         function limitationWarn( $limitationType, $current=null, $max=null) {
2715                 $msgName = $limitationType . '-warning';
2716                 //does no harm if $current and $max are present but are unnecessary for the message
2717                 $warning = wfMsg( $msgName, $current, $max);
2718                 $this->mOutput->addWarning( $warning );
2719                 $cat = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( $limitationType . '-category' ) );
2720                 if ( $cat ) {
2721                         $this->mOutput->addCategory( $cat->getDBkey(), $this->getDefaultSort() );
2722                 }
2723         }
2724
2725         /**
2726          * Return the text of a template, after recursively
2727          * replacing any variables or templates within the template.
2728          *
2729          * @param array $piece The parts of the template
2730          *  $piece['title']: the title, i.e. the part before the |
2731          *  $piece['parts']: the parameter array
2732          *  $piece['lineStart']: whether the brace was at the start of a line
2733          * @param PPFrame The current frame, contains template arguments
2734          * @return string the text of the template
2735          * @private
2736          */
2737         function braceSubstitution( $piece, $frame ) {
2738                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
2739                 $fname = __METHOD__;
2740                 wfProfileIn( $fname );
2741                 wfProfileIn( __METHOD__.'-setup' );
2742
2743                 # Flags
2744                 $found = false;             # $text has been filled
2745                 $nowiki = false;            # wiki markup in $text should be escaped
2746                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2747                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2748                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2749                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2750
2751                 # Title object, where $text came from
2752                 $title = NULL;
2753
2754                 # $part1 is the bit before the first |, and must contain only title characters.
2755                 # Various prefixes will be stripped from it later.
2756                 $titleWithSpaces = $frame->expand( $piece['title'] );
2757                 $part1 = trim( $titleWithSpaces );
2758                 $titleText = false;
2759
2760                 # Original title text preserved for various purposes
2761                 $originalTitle = $part1;
2762
2763                 # $args is a list of argument nodes, starting from index 0, not including $part1
2764                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2765                 wfProfileOut( __METHOD__.'-setup' );
2766
2767                 # SUBST
2768                 wfProfileIn( __METHOD__.'-modifiers' );
2769                 if ( !$found ) {
2770                         $mwSubst = MagicWord::get( 'subst' );
2771                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2772                                 # One of two possibilities is true:
2773                                 # 1) Found SUBST but not in the PST phase
2774                                 # 2) Didn't find SUBST and in the PST phase
2775                                 # In either case, return without further processing
2776                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2777                                 $isLocalObj = true;
2778                                 $found = true;
2779                         }
2780                 }
2781
2782                 # Variables
2783                 if ( !$found && $args->getLength() == 0 ) {
2784                         $id = $this->mVariables->matchStartToEnd( $part1 );
2785                         if ( $id !== false ) {
2786                                 $text = $this->getVariableValue( $id );
2787                                 if (MagicWord::getCacheTTL($id)>-1)
2788                                         $this->mOutput->mContainsOldMagic = true;
2789                                 $found = true;
2790                         }
2791                 }
2792
2793                 # MSG, MSGNW and RAW
2794                 if ( !$found ) {
2795                         # Check for MSGNW:
2796                         $mwMsgnw = MagicWord::get( 'msgnw' );
2797                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2798                                 $nowiki = true;
2799                         } else {
2800                                 # Remove obsolete MSG:
2801                                 $mwMsg = MagicWord::get( 'msg' );
2802                                 $mwMsg->matchStartAndRemove( $part1 );
2803                         }
2804
2805                         # Check for RAW:
2806                         $mwRaw = MagicWord::get( 'raw' );
2807                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2808                                 $forceRawInterwiki = true;
2809                         }
2810                 }
2811                 wfProfileOut( __METHOD__.'-modifiers' );
2812
2813                 # Parser functions
2814                 if ( !$found ) {
2815                         wfProfileIn( __METHOD__ . '-pfunc' );
2816
2817                         $colonPos = strpos( $part1, ':' );
2818                         if ( $colonPos !== false ) {
2819                                 # Case sensitive functions
2820                                 $function = substr( $part1, 0, $colonPos );
2821                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2822                                         $function = $this->mFunctionSynonyms[1][$function];
2823                                 } else {
2824                                         # Case insensitive functions
2825                                         $function = strtolower( $function );
2826                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2827                                                 $function = $this->mFunctionSynonyms[0][$function];
2828                                         } else {
2829                                                 $function = false;
2830                                         }
2831                                 }
2832                                 if ( $function ) {
2833                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2834                                         $initialArgs = array( &$this );
2835                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2836                                         if ( $flags & SFH_OBJECT_ARGS ) {
2837                                                 # Add a frame parameter, and pass the arguments as an array
2838                                                 $allArgs = $initialArgs;
2839                                                 $allArgs[] = $frame;
2840                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2841                                                         $funcArgs[] = $args->item( $i );
2842                                                 }
2843                                                 $allArgs[] = $funcArgs;
2844                                         } else {
2845                                                 # Convert arguments to plain text
2846                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2847                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2848                                                 }
2849                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2850                                         }
2851
2852                                         # Workaround for PHP bug 35229 and similar
2853                                         if ( !is_callable( $callback ) ) {
2854                                                 throw new MWException( "Tag hook for $name is not callable\n" );
2855                                         }
2856                                         $result = call_user_func_array( $callback, $allArgs );
2857                                         $found = true;
2858                                         $noparse = true;
2859                                         $preprocessFlags = 0;
2860
2861                                         if ( is_array( $result ) ) {
2862                                                 if ( isset( $result[0] ) ) {
2863                                                         $text = $result[0];
2864                                                         unset( $result[0] );
2865                                                 }
2866
2867                                                 // Extract flags into the local scope
2868                                                 // This allows callers to set flags such as nowiki, found, etc.
2869                                                 extract( $result );
2870                                         } else {
2871                                                 $text = $result;
2872                                         }
2873                                         if ( !$noparse ) {
2874                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
2875                                                 $isChildObj = true;
2876                                         }
2877                                 }
2878                         }
2879                         wfProfileOut( __METHOD__ . '-pfunc' );
2880                 }
2881
2882                 # Finish mangling title and then check for loops.
2883                 # Set $title to a Title object and $titleText to the PDBK
2884                 if ( !$found ) {
2885                         $ns = NS_TEMPLATE;
2886                         # Split the title into page and subpage
2887                         $subpage = '';
2888                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2889                         if ($subpage !== '') {
2890                                 $ns = $this->mTitle->getNamespace();
2891                         }
2892                         $title = Title::newFromText( $part1, $ns );
2893                         if ( $title ) {
2894                                 $titleText = $title->getPrefixedText();
2895                                 # Check for language variants if the template is not found
2896                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
2897                                         $wgContLang->findVariantLink($part1, $title);
2898                                 }
2899                                 # Do infinite loop check
2900                                 if ( !$frame->loopCheck( $title ) ) {
2901                                         $found = true;
2902                                         $text = "<span class=\"error\">Template loop detected: [[$titleText]]</span>";
2903                                         wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
2904                                 }
2905                                 # Do recursion depth check
2906                                 $limit = $this->mOptions->getMaxTemplateDepth();
2907                                 if ( $frame->depth >= $limit ) {
2908                                         $found = true;
2909                                         $text = "<span class=\"error\">Template recursion depth limit exceeded ($limit)</span>";
2910                                 }
2911                         }
2912                 }
2913
2914                 # Load from database
2915                 if ( !$found && $title ) {
2916                         wfProfileIn( __METHOD__ . '-loadtpl' );
2917                         if ( !$title->isExternal() ) {
2918                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
2919                                         $text = SpecialPage::capturePath( $title );
2920                                         if ( is_string( $text ) ) {
2921                                                 $found = true;
2922                                                 $isHTML = true;
2923                                                 $this->disableCache();
2924                                         }
2925                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
2926                                         $found = false; //access denied
2927                                         wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );
2928                                 } else {
2929                                         list( $text, $title ) = $this->getTemplateDom( $title );
2930                                         if ( $text !== false ) {
2931                                                 $found = true;
2932                                                 $isChildObj = true;
2933                                         }
2934                                 }
2935
2936                                 # If the title is valid but undisplayable, make a link to it
2937                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2938                                         $text = "[[:$titleText]]";
2939                                         $found = true;
2940                                 }
2941                         } elseif ( $title->isTrans() ) {
2942                                 // Interwiki transclusion
2943                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
2944                                         $text = $this->interwikiTransclude( $title, 'render' );
2945                                         $isHTML = true;
2946                                 } else {
2947                                         $text = $this->interwikiTransclude( $title, 'raw' );
2948                                         // Preprocess it like a template
2949                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
2950                                         $isChildObj = true;
2951                                 }
2952                                 $found = true;
2953                         }
2954                         wfProfileOut( __METHOD__ . '-loadtpl' );
2955                 }
2956
2957                 # If we haven't found text to substitute by now, we're done
2958                 # Recover the source wikitext and return it
2959                 if ( !$found ) {
2960                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2961                         wfProfileOut( $fname );
2962                         return array( 'object' => $text );
2963                 }
2964
2965                 # Expand DOM-style return values in a child frame
2966                 if ( $isChildObj ) {
2967                         # Clean up argument array
2968                         $newFrame = $frame->newChild( $args, $title );
2969
2970                         if ( $nowiki ) {
2971                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
2972                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
2973                                 # Expansion is eligible for the empty-frame cache
2974                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
2975                                         $text = $this->mTplExpandCache[$titleText];
2976                                 } else {
2977                                         $text = $newFrame->expand( $text );
2978                                         $this->mTplExpandCache[$titleText] = $text;
2979                                 }
2980                         } else {
2981                                 # Uncached expansion
2982                                 $text = $newFrame->expand( $text );
2983                         }
2984                 }
2985                 if ( $isLocalObj && $nowiki ) {
2986                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
2987                         $isLocalObj = false;
2988                 }
2989
2990                 # Replace raw HTML by a placeholder
2991                 # Add a blank line preceding, to prevent it from mucking up
2992                 # immediately preceding headings
2993                 if ( $isHTML ) {
2994                         $text = "\n\n" . $this->insertStripItem( $text );
2995                 }
2996                 # Escape nowiki-style return values
2997                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2998                         $text = wfEscapeWikiText( $text );
2999                 }
3000                 # Bug 529: if the template begins with a table or block-level
3001                 # element, it should be treated as beginning a new line.
3002                 # This behaviour is somewhat controversial.
3003                 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3004                         $text = "\n" . $text;
3005                 }
3006
3007                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3008                         # Error, oversize inclusion
3009                         $text = "[[$originalTitle]]" .
3010                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3011                         $this->limitationWarn( 'post-expand-template-inclusion' );
3012                 }
3013
3014                 if ( $isLocalObj ) {
3015                         $ret = array( 'object' => $text );
3016                 } else {
3017                         $ret = array( 'text' => $text );
3018                 }
3019
3020                 wfProfileOut( $fname );
3021                 return $ret;
3022         }
3023
3024         /**
3025          * Get the semi-parsed DOM representation of a template with a given title,
3026          * and its redirect destination title. Cached.
3027          */
3028         function getTemplateDom( $title ) {
3029                 $cacheTitle = $title;
3030                 $titleText = $title->getPrefixedDBkey();
3031
3032                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3033                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3034                         $title = Title::makeTitle( $ns, $dbk );
3035                         $titleText = $title->getPrefixedDBkey();
3036                 }
3037                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3038                         return array( $this->mTplDomCache[$titleText], $title );
3039                 }
3040
3041                 // Cache miss, go to the database
3042                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3043
3044                 if ( $text === false ) {
3045                         $this->mTplDomCache[$titleText] = false;
3046                         return array( false, $title );
3047                 }
3048
3049                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3050                 $this->mTplDomCache[ $titleText ] = $dom;
3051
3052                 if (! $title->equals($cacheTitle)) {
3053                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3054                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3055                 }
3056
3057                 return array( $dom, $title );
3058         }
3059
3060         /**
3061          * Fetch the unparsed text of a template and register a reference to it.
3062          */
3063         function fetchTemplateAndTitle( $title ) {
3064                 $templateCb = $this->mOptions->getTemplateCallback();
3065                 $stuff = call_user_func( $templateCb, $title, $this );
3066                 $text = $stuff['text'];
3067                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3068                 if ( isset( $stuff['deps'] ) ) {
3069                         foreach ( $stuff['deps'] as $dep ) {
3070                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3071                         }
3072                 }
3073                 return array($text,$finalTitle);
3074         }
3075
3076         function fetchTemplate( $title ) {
3077                 $rv = $this->fetchTemplateAndTitle($title);
3078                 return $rv[0];
3079         }
3080
3081         /**
3082          * Static function to get a template
3083          * Can be overridden via ParserOptions::setTemplateCallback().
3084          */
3085         static function statelessFetchTemplate( $title, $parser=false ) {
3086                 $text = $skip = false;
3087                 $finalTitle = $title;
3088                 $deps = array();
3089
3090                 // Loop to fetch the article, with up to 1 redirect
3091                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3092                         # Give extensions a chance to select the revision instead
3093                         $id = false; // Assume current
3094                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3095
3096                         if( $skip ) {
3097                                 $text = false;
3098                                 $deps[] = array(
3099                                         'title' => $title,
3100                                         'page_id' => $title->getArticleID(),
3101                                         'rev_id' => null );
3102                                 break;
3103                         }
3104                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3105                         $rev_id = $rev ? $rev->getId() : 0;
3106                         // If there is no current revision, there is no page
3107                         if( $id === false && !$rev ) {
3108                                 $linkCache = LinkCache::singleton();
3109                                 $linkCache->addBadLinkObj( $title );
3110                         }
3111
3112                         $deps[] = array(
3113                                 'title' => $title,
3114                                 'page_id' => $title->getArticleID(),
3115                                 'rev_id' => $rev_id );
3116
3117                         if( $rev ) {
3118                                 $text = $rev->getText();
3119                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3120                                 global $wgLang;
3121                                 $message = $wgLang->lcfirst( $title->getText() );
3122                                 $text = wfMsgForContentNoTrans( $message );
3123                                 if( wfEmptyMsg( $message, $text ) ) {
3124                                         $text = false;
3125                                         break;
3126                                 }
3127                         } else {
3128                                 break;
3129                         }
3130                         if ( $text === false ) {
3131                                 break;
3132                         }
3133                         // Redirect?
3134                         $finalTitle = $title;
3135                         $title = Title::newFromRedirect( $text );
3136                 }
3137                 return array(
3138                         'text' => $text,
3139                         'finalTitle' => $finalTitle,
3140                         'deps' => $deps );
3141         }
3142
3143         /**
3144          * Transclude an interwiki link.
3145          */
3146         function interwikiTransclude( $title, $action ) {
3147                 global $wgEnableScaryTranscluding;
3148
3149                 if (!$wgEnableScaryTranscluding)
3150                         return wfMsg('scarytranscludedisabled');
3151
3152                 $url = $title->getFullUrl( "action=$action" );
3153
3154                 if (strlen($url) > 255)
3155                         return wfMsg('scarytranscludetoolong');
3156                 return $this->fetchScaryTemplateMaybeFromCache($url);
3157         }
3158
3159         function fetchScaryTemplateMaybeFromCache($url) {
3160                 global $wgTranscludeCacheExpiry;
3161                 $dbr = wfGetDB(DB_SLAVE);
3162                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3163                                 array('tc_url' => $url));
3164                 if ($obj) {
3165                         $time = $obj->tc_time;
3166                         $text = $obj->tc_contents;
3167                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3168                                 return $text;
3169                         }
3170                 }
3171
3172                 $text = Http::get($url);
3173                 if (!$text)
3174                         return wfMsg('scarytranscludefailed', $url);
3175
3176                 $dbw = wfGetDB(DB_MASTER);
3177                 $dbw->replace('transcache', array('tc_url'), array(
3178                         'tc_url' => $url,
3179                         'tc_time' => time(),
3180                         'tc_contents' => $text));
3181                 return $text;
3182         }
3183
3184
3185         /**
3186          * Triple brace replacement -- used for template arguments
3187          * @private
3188          */
3189         function argSubstitution( $piece, $frame ) {
3190                 wfProfileIn( __METHOD__ );
3191
3192                 $error = false;
3193                 $parts = $piece['parts'];
3194                 $nameWithSpaces = $frame->expand( $piece['title'] );
3195                 $argName = trim( $nameWithSpaces );
3196                 $object = false;
3197                 $text = $frame->getArgument( $argName );
3198                 if (  $text === false && $parts->getLength() > 0
3199                   && (
3200                     $this->ot['html']
3201                     || $this->ot['pre']
3202                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3203                   )
3204                 ) {
3205                         # No match in frame, use the supplied default
3206                         $object = $parts->item( 0 )->getChildren();
3207                 }
3208                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3209                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3210                         $this->limitationWarn( 'post-expand-template-argument' );
3211                 }
3212
3213                 if ( $text === false && $object === false ) {
3214                         # No match anywhere
3215                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3216                 }
3217                 if ( $error !== false ) {
3218                         $text .= $error;
3219                 }
3220                 if ( $object !== false ) {
3221                         $ret = array( 'object' => $object );
3222                 } else {
3223                         $ret = array( 'text' => $text );
3224                 }
3225
3226                 wfProfileOut( __METHOD__ );
3227                 return $ret;
3228         }
3229
3230         /**
3231          * Return the text to be used for a given extension tag.
3232          * This is the ghost of strip().
3233          *
3234          * @param array $params Associative array of parameters:
3235          *     name       PPNode for the tag name
3236          *     attr       PPNode for unparsed text where tag attributes are thought to be
3237          *     attributes Optional associative array of parsed attributes
3238          *     inner      Contents of extension element
3239          *     noClose    Original text did not have a close tag
3240          * @param PPFrame $frame
3241          */
3242         function extensionSubstitution( $params, $frame ) {
3243                 global $wgRawHtml, $wgContLang;
3244
3245                 $name = $frame->expand( $params['name'] );
3246                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3247                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3248
3249                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $this->mMarkerIndex++) . self::MARKER_SUFFIX;
3250
3251                 if ( $this->ot['html'] ) {
3252                         $name = strtolower( $name );
3253
3254                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3255                         if ( isset( $params['attributes'] ) ) {
3256                                 $attributes = $attributes + $params['attributes'];
3257                         }
3258                         switch ( $name ) {
3259                                 case 'html':
3260                                         if( $wgRawHtml ) {
3261                                                 $output = $content;
3262                                                 break;
3263                                         } else {
3264                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3265                                         }
3266                                 case 'nowiki':
3267                                         $output = Xml::escapeTagsOnly( $content );
3268                                         break;
3269                                 case 'math':
3270                                         $output = $wgContLang->armourMath(
3271                                                 MathRenderer::renderMath( $content, $attributes ) );
3272                                         break;
3273                                 case 'gallery':
3274                                         $output = $this->renderImageGallery( $content, $attributes );
3275                                         break;
3276                                 default:
3277                                         if( isset( $this->mTagHooks[$name] ) ) {
3278                                                 # Workaround for PHP bug 35229 and similar
3279                                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3280                                                         throw new MWException( "Tag hook for $name is not callable\n" );
3281                                                 }
3282                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3283                                                         array( $content, $attributes, $this ) );
3284                                         } else {
3285                                                 $output = '<span class="error">Invalid tag extension name: ' .
3286                                                         htmlspecialchars( $name ) . '</span>';
3287                                         }
3288                         }
3289                 } else {
3290                         if ( is_null( $attrText ) ) {
3291                                 $attrText = '';
3292                         }
3293                         if ( isset( $params['attributes'] ) ) {
3294                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3295                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3296                                                 htmlspecialchars( $attrValue ) . '"';
3297                                 }
3298                         }
3299                         if ( $content === null ) {
3300                                 $output = "<$name$attrText/>";
3301                         } else {
3302                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3303                                 $output = "<$name$attrText>$content$close";
3304                         }
3305                 }
3306
3307                 if ( $name == 'html' || $name == 'nowiki' ) {
3308                         $this->mStripState->nowiki->setPair( $marker, $output );
3309                 } else {
3310                         $this->mStripState->general->setPair( $marker, $output );
3311                 }
3312                 return $marker;
3313         }
3314
3315         /**
3316          * Increment an include size counter
3317          *
3318          * @param string $type The type of expansion
3319          * @param integer $size The size of the text
3320          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3321          */
3322         function incrementIncludeSize( $type, $size ) {
3323                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3324                         return false;
3325                 } else {
3326                         $this->mIncludeSizes[$type] += $size;
3327                         return true;
3328                 }
3329         }
3330
3331         /**
3332          * Increment the expensive function count
3333          *
3334          * @return boolean False if the limit has been exceeded
3335          */
3336         function incrementExpensiveFunctionCount() {
3337                 global $wgExpensiveParserFunctionLimit;
3338                 $this->mExpensiveFunctionCount++;
3339                 if($this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit) {
3340                         return true;
3341                 }
3342                 return false;
3343         }
3344
3345         /**
3346          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3347          * Fills $this->mDoubleUnderscores, returns the modified text
3348          */
3349         function doDoubleUnderscore( $text ) {
3350                 // The position of __TOC__ needs to be recorded
3351                 $mw = MagicWord::get( 'toc' );
3352                 if( $mw->match( $text ) ) {
3353                         $this->mShowToc = true;
3354                         $this->mForceTocPosition = true;
3355
3356                         // Set a placeholder. At the end we'll fill it in with the TOC.
3357                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3358
3359                         // Only keep the first one.
3360                         $text = $mw->replace( '', $text );
3361                 }
3362
3363                 // Now match and remove the rest of them
3364                 $mwa = MagicWord::getDoubleUnderscoreArray();
3365                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3366
3367                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3368                         $this->mOutput->mNoGallery = true;
3369                 }
3370                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3371                         $this->mShowToc = false;
3372                 }
3373                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3374                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3375
3376                         $containerCategory = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( 'hidden-category-category' ) );
3377                         if ( $containerCategory ) {
3378                                 $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3379                         } else {
3380                                 wfDebug( __METHOD__.": [[MediaWiki:hidden-category-category]] is not a valid title!\n" );
3381                         }
3382                 }
3383                 # (bug 8068) Allow control over whether robots index a page.
3384                 #
3385                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3386                 # is not desirable, the last one on the page should win.
3387                 if( isset( $this->mDoubleUnderscores['noindex'] ) ) {
3388                         $this->mOutput->setIndexPolicy( 'noindex' );
3389                 } elseif( isset( $this->mDoubleUnderscores['index'] ) ) {
3390                         $this->mOutput->setIndexPolicy( 'index' );
3391                 }
3392
3393                 return $text;
3394         }
3395
3396         /**
3397          * This function accomplishes several tasks:
3398          * 1) Auto-number headings if that option is enabled
3399          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3400          * 3) Add a Table of contents on the top for users who have enabled the option
3401          * 4) Auto-anchor headings
3402          *
3403          * It loops through all headlines, collects the necessary data, then splits up the
3404          * string and re-inserts the newly formatted headlines.
3405          *
3406          * @param string $text
3407          * @param boolean $isMain
3408          * @private
3409          */
3410         function formatHeadings( $text, $isMain=true ) {
3411                 global $wgMaxTocLevel, $wgContLang;
3412
3413                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3414                 if( !$this->mTitle->quickUserCan( 'edit' ) ) {
3415                         $showEditLink = 0;
3416                 } else {
3417                         $showEditLink = $this->mOptions->getEditSection();
3418                 }
3419
3420                 # Inhibit editsection links if requested in the page
3421                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3422                         $showEditLink = 0;
3423                 }
3424
3425                 # Get all headlines for numbering them and adding funky stuff like [edit]
3426                 # links - this is for later, but we need the number of headlines right now
3427                 $matches = array();
3428                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3429
3430                 # if there are fewer than 4 headlines in the article, do not show TOC
3431                 # unless it's been explicitly enabled.
3432                 $enoughToc = $this->mShowToc &&
3433                         (($numMatches >= 4) || $this->mForceTocPosition);
3434
3435                 # Allow user to stipulate that a page should have a "new section"
3436                 # link added via __NEWSECTIONLINK__
3437                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3438                         $this->mOutput->setNewSection( true );
3439                 }
3440
3441                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3442                 # override above conditions and always show TOC above first header
3443                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3444                         $this->mShowToc = true;
3445                         $enoughToc = true;
3446                 }
3447
3448                 # We need this to perform operations on the HTML
3449                 $sk = $this->mOptions->getSkin();
3450
3451                 # headline counter
3452                 $headlineCount = 0;
3453                 $numVisible = 0;
3454
3455                 # Ugh .. the TOC should have neat indentation levels which can be
3456                 # passed to the skin functions. These are determined here
3457                 $toc = '';
3458                 $full = '';
3459                 $head = array();
3460                 $sublevelCount = array();
3461                 $levelCount = array();
3462                 $toclevel = 0;
3463                 $level = 0;
3464                 $prevlevel = 0;
3465                 $toclevel = 0;
3466                 $prevtoclevel = 0;
3467                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3468                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3469                 $tocraw = array();
3470
3471                 foreach( $matches[3] as $headline ) {
3472                         $isTemplate = false;
3473                         $titleText = false;
3474                         $sectionIndex = false;
3475                         $numbering = '';
3476                         $markerMatches = array();
3477                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3478                                 $serial = $markerMatches[1];
3479                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3480                                 $isTemplate = ($titleText != $baseTitleText);
3481                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3482                         }
3483
3484                         if( $toclevel ) {
3485                                 $prevlevel = $level;
3486                                 $prevtoclevel = $toclevel;
3487                         }
3488                         $level = $matches[1][$headlineCount];
3489
3490                         if( $doNumberHeadings || $enoughToc ) {
3491
3492                                 if ( $level > $prevlevel ) {
3493                                         # Increase TOC level
3494                                         $toclevel++;
3495                                         $sublevelCount[$toclevel] = 0;
3496                                         if( $toclevel<$wgMaxTocLevel ) {
3497                                                 $prevtoclevel = $toclevel;
3498                                                 $toc .= $sk->tocIndent();
3499                                                 $numVisible++;
3500                                         }
3501                                 }
3502                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3503                                         # Decrease TOC level, find level to jump to
3504
3505                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3506                                                 # Can only go down to level 1
3507                                                 $toclevel = 1;
3508                                         } else {
3509                                                 for ($i = $toclevel; $i > 0; $i--) {
3510                                                         if ( $levelCount[$i] == $level ) {
3511                                                                 # Found last matching level
3512                                                                 $toclevel = $i;
3513                                                                 break;
3514                                                         }
3515                                                         elseif ( $levelCount[$i] < $level ) {
3516                                                                 # Found first matching level below current level
3517                                                                 $toclevel = $i + 1;
3518                                                                 break;
3519                                                         }
3520                                                 }
3521                                         }
3522                                         if( $toclevel<$wgMaxTocLevel ) {
3523                                                 if($prevtoclevel < $wgMaxTocLevel) {
3524                                                         # Unindent only if the previous toc level was shown :p
3525                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3526                                                         $prevtoclevel = $toclevel;
3527                                                 } else {
3528                                                         $toc .= $sk->tocLineEnd();
3529                                                 }
3530                                         }
3531                                 }
3532                                 else {
3533                                         # No change in level, end TOC line
3534                                         if( $toclevel<$wgMaxTocLevel ) {
3535                                                 $toc .= $sk->tocLineEnd();
3536                                         }
3537                                 }
3538
3539                                 $levelCount[$toclevel] = $level;
3540
3541                                 # count number of headlines for each level
3542                                 @$sublevelCount[$toclevel]++;
3543                                 $dot = 0;
3544                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3545                                         if( !empty( $sublevelCount[$i] ) ) {
3546                                                 if( $dot ) {
3547                                                         $numbering .= '.';
3548                                                 }
3549                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3550                                                 $dot = 1;
3551                                         }
3552                                 }
3553                         }
3554
3555                         # The safe header is a version of the header text safe to use for links
3556                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3557                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3558
3559                         # Remove link placeholders by the link text.
3560                         #     <!--LINK number-->
3561                         # turns into
3562                         #     link text with suffix
3563                         $safeHeadline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3564                                                             "\$this->mLinkHolders['texts'][\$1]",
3565                                                             $safeHeadline );
3566                         $safeHeadline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3567                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3568                                                             $safeHeadline );
3569
3570                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3571                         $tocline = preg_replace(
3572                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3573                                 array( '',                          '<$1>'),
3574                                 $safeHeadline
3575                         );
3576                         $tocline = trim( $tocline );
3577
3578                         # For the anchor, strip out HTML-y stuff period
3579                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3580                         $safeHeadline = trim( $safeHeadline );
3581
3582                         # Save headline for section edit hint before it's escaped
3583                         $headlineHint = $safeHeadline;
3584                         $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3585                         # HTML names must be case-insensitively unique (bug 10721)
3586                         $arrayKey = strtolower( $safeHeadline );
3587
3588                         # count how many in assoc. array so we can track dupes in anchors
3589                         isset( $refers[$arrayKey] ) ? $refers[$arrayKey]++ : $refers[$arrayKey] = 1;
3590                         $refcount[$headlineCount] = $refers[$arrayKey];
3591
3592                         # Don't number the heading if it is the only one (looks silly)
3593                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3594                                 # the two are different if the line contains a link
3595                                 $headline=$numbering . ' ' . $headline;
3596                         }
3597
3598                         # Create the anchor for linking from the TOC to the section
3599                         $anchor = $safeHeadline;
3600                         if($refcount[$headlineCount] > 1 ) {
3601                                 $anchor .= '_' . $refcount[$headlineCount];
3602                         }
3603                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3604                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3605                                 $tocraw[] = array( 'toclevel' => $toclevel, 'level' => $level, 'line' => $tocline, 'number' => $numbering );
3606                         }
3607                         # give headline the correct <h#> tag
3608                         if( $showEditLink && $sectionIndex !== false ) {
3609                                 if( $isTemplate ) {
3610                                         # Put a T flag in the section identifier, to indicate to extractSections()
3611                                         # that sections inside <includeonly> should be counted.
3612                                         $editlink = $sk->editSectionLinkForOther($titleText, "T-$sectionIndex");
3613                                 } else {
3614                                         $editlink = $sk->editSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3615                                 }
3616                         } else {
3617                                 $editlink = '';
3618                         }
3619                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
3620
3621                         $headlineCount++;
3622                 }
3623
3624                 $this->mOutput->setSections( $tocraw );
3625
3626                 # Never ever show TOC if no headers
3627                 if( $numVisible < 1 ) {
3628                         $enoughToc = false;
3629                 }
3630
3631                 if( $enoughToc ) {
3632                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3633                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3634                         }
3635                         $toc = $sk->tocList( $toc );
3636                 }
3637
3638                 # split up and insert constructed headlines
3639
3640                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3641                 $i = 0;
3642
3643                 foreach( $blocks as $block ) {
3644                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3645                                 # This is the [edit] link that appears for the top block of text when
3646                                 # section editing is enabled
3647
3648                                 # Disabled because it broke block formatting
3649                                 # For example, a bullet point in the top line
3650                                 # $full .= $sk->editSectionLink(0);
3651                         }
3652                         $full .= $block;
3653                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3654                                 # Top anchor now in skin
3655                                 $full = $full.$toc;
3656                         }
3657
3658                         if( !empty( $head[$i] ) ) {
3659                                 $full .= $head[$i];
3660                         }
3661                         $i++;
3662                 }
3663                 if( $this->mForceTocPosition ) {
3664                         return str_replace( '<!--MWTOC-->', $toc, $full );
3665                 } else {
3666                         return $full;
3667                 }
3668         }
3669
3670         /**
3671          * Transform wiki markup when saving a page by doing \r\n -> \n
3672          * conversion, substitting signatures, {{subst:}} templates, etc.
3673          *
3674          * @param string $text the text to transform
3675          * @param Title &$title the Title object for the current article
3676          * @param User &$user the User object describing the current user
3677          * @param ParserOptions $options parsing options
3678          * @param bool $clearState whether to clear the parser state first
3679          * @return string the altered wiki markup
3680          * @public
3681          */
3682         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3683                 $this->mOptions = $options;
3684                 $this->setTitle( $title );
3685                 $this->setOutputType( self::OT_WIKI );
3686
3687                 if ( $clearState ) {
3688                         $this->clearState();
3689                 }
3690
3691                 $pairs = array(
3692                         "\r\n" => "\n",
3693                 );
3694                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3695                 $text = $this->pstPass2( $text, $user );
3696                 $text = $this->mStripState->unstripBoth( $text );
3697                 return $text;
3698         }
3699
3700         /**
3701          * Pre-save transform helper function
3702          * @private
3703          */
3704         function pstPass2( $text, $user ) {
3705                 global $wgContLang, $wgLocaltimezone;
3706
3707                 /* Note: This is the timestamp saved as hardcoded wikitext to
3708                  * the database, we use $wgContLang here in order to give
3709                  * everyone the same signature and use the default one rather
3710                  * than the one selected in each user's preferences.
3711                  *
3712                  * (see also bug 12815)
3713                  */
3714                 $ts = $this->mOptions->getTimestamp();
3715                 $tz = wfMsgForContent( 'timezone-utc' );
3716                 if ( isset( $wgLocaltimezone ) ) {
3717                         $unixts = wfTimestamp( TS_UNIX, $ts );
3718                         $oldtz = getenv( 'TZ' );
3719                         putenv( 'TZ='.$wgLocaltimezone );
3720                         $ts = date( 'YmdHis', $unixts );
3721                         $tz = date( 'T', $unixts );  # might vary on DST changeover!
3722                         putenv( 'TZ='.$oldtz );
3723                 }
3724
3725                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tz)";
3726
3727                 # Variable replacement
3728                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3729                 $text = $this->replaceVariables( $text );
3730
3731                 # Signatures
3732                 $sigText = $this->getUserSig( $user );
3733                 $text = strtr( $text, array(
3734                         '~~~~~' => $d,
3735                         '~~~~' => "$sigText $d",
3736                         '~~~' => $sigText
3737                 ) );
3738
3739                 # Context links: [[|name]] and [[name (context)|]]
3740                 #
3741                 global $wgLegalTitleChars;
3742                 $tc = "[$wgLegalTitleChars]";
3743                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
3744
3745                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3746                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3747                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3748
3749                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3750                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3751                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3752
3753                 $t = $this->mTitle->getText();
3754                 $m = array();
3755                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3756                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3757                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3758                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3759                 } else {
3760                         # if there's no context, don't bother duplicating the title
3761                         $text = preg_replace( $p2, '[[\\1]]', $text );
3762                 }
3763
3764                 # Trim trailing whitespace
3765                 $text = rtrim( $text );
3766
3767                 return $text;
3768         }
3769
3770         /**
3771          * Fetch the user's signature text, if any, and normalize to
3772          * validated, ready-to-insert wikitext.
3773          *
3774          * @param User $user
3775          * @return string
3776          * @private
3777          */
3778         function getUserSig( &$user ) {
3779                 global $wgMaxSigChars;
3780
3781                 $username = $user->getName();
3782                 $nickname = $user->getOption( 'nickname' );
3783                 $nickname = $nickname === '' ? $username : $nickname;
3784
3785                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
3786                         $nickname = $username;
3787                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
3788                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
3789                         # Sig. might contain markup; validate this
3790                         if( $this->validateSig( $nickname ) !== false ) {
3791                                 # Validated; clean up (if needed) and return it
3792                                 return $this->cleanSig( $nickname, true );
3793                         } else {
3794                                 # Failed to validate; fall back to the default
3795                                 $nickname = $username;
3796                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3797                         }
3798                 }
3799
3800                 // Make sure nickname doesnt get a sig in a sig
3801                 $nickname = $this->cleanSigInSig( $nickname );
3802
3803                 # If we're still here, make it a link to the user page
3804                 $userText = wfEscapeWikiText( $username );
3805                 $nickText = wfEscapeWikiText( $nickname );
3806                 if ( $user->isAnon() )  {
3807                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
3808                 } else {
3809                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
3810                 }
3811         }
3812
3813         /**
3814          * Check that the user's signature contains no bad XML
3815          *
3816          * @param string $text
3817          * @return mixed An expanded string, or false if invalid.
3818          */
3819         function validateSig( $text ) {
3820                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3821         }
3822
3823         /**
3824          * Clean up signature text
3825          *
3826          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3827          * 2) Substitute all transclusions
3828          *
3829          * @param string $text
3830          * @param $parsing Whether we're cleaning (preferences save) or parsing
3831          * @return string Signature text
3832          */
3833         function cleanSig( $text, $parsing = false ) {
3834                 if ( !$parsing ) {
3835                         global $wgTitle;
3836                         $this->clearState();
3837                         $this->setTitle( $wgTitle );
3838                         $this->mOptions = new ParserOptions;
3839                         $this->setOutputType = self::OT_PREPROCESS;
3840                 }
3841
3842                 # FIXME: regex doesn't respect extension tags or nowiki
3843                 #  => Move this logic to braceSubstitution()
3844                 $substWord = MagicWord::get( 'subst' );
3845                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3846                 $substText = '{{' . $substWord->getSynonym( 0 );
3847
3848                 $text = preg_replace( $substRegex, $substText, $text );
3849                 $text = $this->cleanSigInSig( $text );
3850                 $dom = $this->preprocessToDom( $text );
3851                 $frame = $this->getPreprocessor()->newFrame();
3852                 $text = $frame->expand( $dom );
3853
3854                 if ( !$parsing ) {
3855                         $text = $this->mStripState->unstripBoth( $text );
3856                 }
3857
3858                 return $text;
3859         }
3860
3861         /**
3862          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3863          * @param string $text
3864          * @return string Signature text with /~{3,5}/ removed
3865          */
3866         function cleanSigInSig( $text ) {
3867                 $text = preg_replace( '/~{3,5}/', '', $text );
3868                 return $text;
3869         }
3870
3871         /**
3872          * Set up some variables which are usually set up in parse()
3873          * so that an external function can call some class members with confidence
3874          * @public
3875          */
3876         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3877                 $this->setTitle( $title );
3878                 $this->mOptions = $options;
3879                 $this->setOutputType( $outputType );
3880                 if ( $clearState ) {
3881                         $this->clearState();
3882                 }
3883         }
3884
3885         /**
3886          * Wrapper for preprocess()
3887          *
3888          * @param string $text the text to preprocess
3889          * @param ParserOptions $options  options
3890          * @return string
3891          * @public
3892          */
3893         function transformMsg( $text, $options ) {
3894                 global $wgTitle;
3895                 static $executing = false;
3896
3897                 $fname = "Parser::transformMsg";
3898
3899                 # Guard against infinite recursion
3900                 if ( $executing ) {
3901                         return $text;
3902                 }
3903                 $executing = true;
3904
3905                 wfProfileIn($fname);
3906                 $text = $this->preprocess( $text, $wgTitle, $options );
3907
3908                 $executing = false;
3909                 wfProfileOut($fname);
3910                 return $text;
3911         }
3912
3913         /**
3914          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3915          * The callback should have the following form:
3916          *    function myParserHook( $text, $params, &$parser ) { ... }
3917          *
3918          * Transform and return $text. Use $parser for any required context, e.g. use
3919          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3920          *
3921          * @public
3922          *
3923          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3924          * @param mixed $callback The callback function (and object) to use for the tag
3925          *
3926          * @return The old value of the mTagHooks array associated with the hook
3927          */
3928         function setHook( $tag, $callback ) {
3929                 $tag = strtolower( $tag );
3930                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
3931                 $this->mTagHooks[$tag] = $callback;
3932                 if( !in_array( $tag, $this->mStripList ) ) {
3933                         $this->mStripList[] = $tag;
3934                 }
3935
3936                 return $oldVal;
3937         }
3938
3939         function setTransparentTagHook( $tag, $callback ) {
3940                 $tag = strtolower( $tag );
3941                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
3942                 $this->mTransparentTagHooks[$tag] = $callback;
3943
3944                 return $oldVal;
3945         }
3946
3947         /**
3948          * Remove all tag hooks
3949          */
3950         function clearTagHooks() {
3951                 $this->mTagHooks = array();
3952                 $this->mStripList = $this->mDefaultStripList;
3953         }
3954
3955         /**
3956          * Create a function, e.g. {{sum:1|2|3}}
3957          * The callback function should have the form:
3958          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3959          *
3960          * Or with SFH_OBJECT_ARGS:
3961          *    function myParserFunction( $parser, $frame, $args ) { ... }
3962          *
3963          * The callback may either return the text result of the function, or an array with the text
3964          * in element 0, and a number of flags in the other elements. The names of the flags are
3965          * specified in the keys. Valid flags are:
3966          *   found                     The text returned is valid, stop processing the template. This
3967          *                             is on by default.
3968          *   nowiki                    Wiki markup in the return value should be escaped
3969          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3970          *
3971          * @public
3972          *
3973          * @param string $id The magic word ID
3974          * @param mixed $callback The callback function (and object) to use
3975          * @param integer $flags a combination of the following flags:
3976          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
3977          *
3978          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
3979          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
3980          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
3981          *     the arguments, and to control the way they are expanded.
3982          *
3983          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
3984          *     arguments, for instance:
3985          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
3986          *
3987          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
3988          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
3989          *     working if/when this is changed.
3990          *
3991          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
3992          *     expansion.
3993          *
3994          *     Please read the documentation in includes/parser/Preprocessor.php for more information
3995          *     about the methods available in PPFrame and PPNode.
3996          *
3997          * @return The old callback function for this name, if any
3998          */
3999         function setFunctionHook( $id, $callback, $flags = 0 ) {
4000                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4001                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4002
4003                 # Add to function cache
4004                 $mw = MagicWord::get( $id );
4005                 if( !$mw )
4006                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
4007
4008                 $synonyms = $mw->getSynonyms();
4009                 $sensitive = intval( $mw->isCaseSensitive() );
4010
4011                 foreach ( $synonyms as $syn ) {
4012                         # Case
4013                         if ( !$sensitive ) {
4014                                 $syn = strtolower( $syn );
4015                         }
4016                         # Add leading hash
4017                         if ( !( $flags & SFH_NO_HASH ) ) {
4018                                 $syn = '#' . $syn;
4019                         }
4020                         # Remove trailing colon
4021                         if ( substr( $syn, -1, 1 ) == ':' ) {
4022                                 $syn = substr( $syn, 0, -1 );
4023                         }
4024                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4025                 }
4026                 return $oldVal;
4027         }
4028
4029         /**
4030          * Get all registered function hook identifiers
4031          *
4032          * @return array
4033          */
4034         function getFunctionHooks() {
4035                 return array_keys( $this->mFunctionHooks );
4036         }
4037
4038         /**
4039          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4040          * Placeholders created in Skin::makeLinkObj()
4041          * Returns an array of link CSS classes, indexed by PDBK.
4042          * $options is a bit field, RLH_FOR_UPDATE to select for update
4043          */
4044         function replaceLinkHolders( &$text, $options = 0 ) {
4045                 global $wgUser;
4046                 global $wgContLang;
4047
4048                 $fname = 'Parser::replaceLinkHolders';
4049                 wfProfileIn( $fname );
4050
4051                 $pdbks = array();
4052                 $colours = array();
4053                 $linkcolour_ids = array();
4054                 $sk = $this->mOptions->getSkin();
4055                 $linkCache = LinkCache::singleton();
4056
4057                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
4058                         wfProfileIn( $fname.'-check' );
4059                         $dbr = wfGetDB( DB_SLAVE );
4060                         $page = $dbr->tableName( 'page' );
4061                         $threshold = $wgUser->getOption('stubthreshold');
4062
4063                         # Sort by namespace
4064                         asort( $this->mLinkHolders['namespaces'] );
4065
4066                         # Generate query
4067                         $query = false;
4068                         $current = null;
4069                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4070                                 # Make title object
4071                                 $title = $this->mLinkHolders['titles'][$key];
4072
4073                                 # Skip invalid entries.
4074                                 # Result will be ugly, but prevents crash.
4075                                 if ( is_null( $title ) ) {
4076                                         continue;
4077                                 }
4078                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
4079
4080                                 # Check if it's a static known link, e.g. interwiki
4081                                 if ( $title->isAlwaysKnown() ) {
4082                                         $colours[$pdbk] = '';
4083                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
4084                                         $colours[$pdbk] = '';
4085                                         $this->mOutput->addLink( $title, $id );
4086                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
4087                                         $colours[$pdbk] = 'new';
4088                                 } elseif ( $title->getNamespace() == NS_SPECIAL && !SpecialPage::exists( $pdbk ) ) {
4089                                         $colours[$pdbk] = 'new';
4090                                 } else {
4091                                         # Not in the link cache, add it to the query
4092                                         if ( !isset( $current ) ) {
4093                                                 $current = $ns;
4094                                                 $query =  "SELECT page_id, page_namespace, page_title, page_is_redirect, page_len";
4095                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
4096                                         } elseif ( $current != $ns ) {
4097                                                 $current = $ns;
4098                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
4099                                         } else {
4100                                                 $query .= ', ';
4101                                         }
4102
4103                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
4104                                 }
4105                         }
4106                         if ( $query ) {
4107                                 $query .= '))';
4108                                 if ( $options & RLH_FOR_UPDATE ) {
4109                                         $query .= ' FOR UPDATE';
4110                                 }
4111
4112                                 $res = $dbr->query( $query, $fname );
4113
4114                                 # Fetch data and form into an associative array
4115                                 # non-existent = broken
4116                                 while ( $s = $dbr->fetchObject($res) ) {
4117                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4118                                         $pdbk = $title->getPrefixedDBkey();
4119                                         $linkCache->addGoodLinkObj( $s->page_id, $title, $s->page_len, $s->page_is_redirect );
4120                                         $this->mOutput->addLink( $title, $s->page_id );
4121                                         $colours[$pdbk] = $sk->getLinkColour( $title, $threshold );
4122                                         //add id to the extension todolist
4123                                         $linkcolour_ids[$s->page_id] = $pdbk;
4124                                 }
4125                                 //pass an array of page_ids to an extension
4126                                 wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4127                         }
4128                         wfProfileOut( $fname.'-check' );
4129
4130                         # Do a second query for different language variants of links and categories
4131                         if($wgContLang->hasVariants()){
4132                                 $linkBatch = new LinkBatch();
4133                                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
4134                                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
4135                                 $varCategories = array(); // category replacements oldDBkey => newDBkey
4136
4137                                 $categories = $this->mOutput->getCategoryLinks();
4138
4139                                 // Add variants of links to link batch
4140                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4141                                         $title = $this->mLinkHolders['titles'][$key];
4142                                         if ( is_null( $title ) )
4143                                                 continue;
4144
4145                                         $pdbk = $title->getPrefixedDBkey();
4146                                         $titleText = $title->getText();
4147
4148                                         // generate all variants of the link title text
4149                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
4150
4151                                         // if link was not found (in first query), add all variants to query
4152                                         if ( !isset($colours[$pdbk]) ){
4153                                                 foreach($allTextVariants as $textVariant){
4154                                                         if($textVariant != $titleText){
4155                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
4156                                                                 if(is_null($variantTitle)) continue;
4157                                                                 $linkBatch->addObj( $variantTitle );
4158                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4159                                                         }
4160                                                 }
4161                                         }
4162                                 }
4163
4164                                 // process categories, check if a category exists in some variant
4165                                 foreach( $categories as $category ){
4166                                         $variants = $wgContLang->convertLinkToAllVariants($category);
4167                                         foreach($variants as $variant){
4168                                                 if($variant != $category){
4169                                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
4170                                                         if(is_null($variantTitle)) continue;
4171                                                         $linkBatch->addObj( $variantTitle );
4172                                                         $categoryMap[$variant] = $category;
4173                                                 }
4174                                         }
4175                                 }
4176
4177
4178                                 if(!$linkBatch->isEmpty()){
4179                                         // construct query
4180                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4181
4182                                         $variantQuery =  "SELECT page_id, page_namespace, page_title, page_is_redirect, page_len";
4183
4184                                         $variantQuery .= " FROM $page WHERE $titleClause";
4185                                         if ( $options & RLH_FOR_UPDATE ) {
4186                                                 $variantQuery .= ' FOR UPDATE';
4187                                         }
4188
4189                                         $varRes = $dbr->query( $variantQuery, $fname );
4190
4191                                         // for each found variants, figure out link holders and replace
4192                                         while ( $s = $dbr->fetchObject($varRes) ) {
4193
4194                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4195                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4196                                                 $vardbk = $variantTitle->getDBkey();
4197
4198                                                 $holderKeys = array();
4199                                                 if(isset($variantMap[$varPdbk])){
4200                                                         $holderKeys = $variantMap[$varPdbk];
4201                                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle, $s->page_len, $s->page_is_redirect );
4202                                                         $this->mOutput->addLink( $variantTitle, $s->page_id );
4203                                                 }
4204
4205                                                 // loop over link holders
4206                                                 foreach($holderKeys as $key){
4207                                                         $title = $this->mLinkHolders['titles'][$key];
4208                                                         if ( is_null( $title ) ) continue;
4209
4210                                                         $pdbk = $title->getPrefixedDBkey();
4211
4212                                                         if(!isset($colours[$pdbk])){
4213                                                                 // found link in some of the variants, replace the link holder data
4214                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4215                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4216
4217                                                                 // set pdbk and colour
4218                                                                 $pdbks[$key] = $varPdbk;
4219                                                                 $colours[$varPdbk] = $sk->getLinkColour( $variantTitle, $threshold );
4220                                                                 $linkcolour_ids[$s->page_id] = $pdbk;
4221                                                         }
4222                                                         wfRunHooks( 'GetLinkColours', array( $linkcolour_ids, &$colours ) );
4223                                                 }
4224
4225                                                 // check if the object is a variant of a category
4226                                                 if(isset($categoryMap[$vardbk])){
4227                                                         $oldkey = $categoryMap[$vardbk];
4228                                                         if($oldkey != $vardbk)
4229                                                                 $varCategories[$oldkey]=$vardbk;
4230                                                 }
4231                                         }
4232
4233                                         // rebuild the categories in original order (if there are replacements)
4234                                         if(count($varCategories)>0){
4235                                                 $newCats = array();
4236                                                 $originalCats = $this->mOutput->getCategories();
4237                                                 foreach($originalCats as $cat => $sortkey){
4238                                                         // make the replacement
4239                                                         if( array_key_exists($cat,$varCategories) )
4240                                                                 $newCats[$varCategories[$cat]] = $sortkey;
4241                                                         else $newCats[$cat] = $sortkey;
4242                                                 }
4243                                                 $this->mOutput->setCategoryLinks($newCats);
4244                                         }
4245                                 }
4246                         }
4247
4248                         # Construct search and replace arrays
4249                         wfProfileIn( $fname.'-construct' );
4250                         $replacePairs = array();
4251                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4252                                 $pdbk = $pdbks[$key];
4253                                 $searchkey = "<!--LINK $key-->";
4254                                 $title = $this->mLinkHolders['titles'][$key];
4255                                 if ( !isset( $colours[$pdbk] ) || $colours[$pdbk] == 'new' ) {
4256                                         $linkCache->addBadLinkObj( $title );
4257                                         $colours[$pdbk] = 'new';
4258                                         $this->mOutput->addLink( $title, 0 );
4259                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4260                                                                         $this->mLinkHolders['texts'][$key],
4261                                                                         $this->mLinkHolders['queries'][$key] );
4262                                 } else {
4263                                         $replacePairs[$searchkey] = $sk->makeColouredLinkObj( $title, $colours[$pdbk],
4264                                                                         $this->mLinkHolders['texts'][$key],
4265                                                                         $this->mLinkHolders['queries'][$key] );
4266                                 }
4267                         }
4268                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4269                         wfProfileOut( $fname.'-construct' );
4270
4271                         # Do the thing
4272                         wfProfileIn( $fname.'-replace' );
4273                         $text = preg_replace_callback(
4274                                 '/(<!--LINK .*?-->)/',
4275                                 $replacer->cb(),
4276                                 $text);
4277
4278                         wfProfileOut( $fname.'-replace' );
4279                 }
4280
4281                 # Now process interwiki link holders
4282                 # This is quite a bit simpler than internal links
4283                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4284                         wfProfileIn( $fname.'-interwiki' );
4285                         # Make interwiki link HTML
4286                         $replacePairs = array();
4287                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4288                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4289                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4290                         }
4291                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4292
4293                         $text = preg_replace_callback(
4294                                 '/<!--IWLINK (.*?)-->/',
4295                                 $replacer->cb(),
4296                                 $text );
4297                         wfProfileOut( $fname.'-interwiki' );
4298                 }
4299
4300                 wfProfileOut( $fname );
4301                 return $colours;
4302         }
4303
4304         /**
4305          * Replace <!--LINK--> link placeholders with plain text of links
4306          * (not HTML-formatted).
4307          * @param string $text
4308          * @return string
4309          */
4310         function replaceLinkHoldersText( $text ) {
4311                 $fname = 'Parser::replaceLinkHoldersText';
4312                 wfProfileIn( $fname );
4313
4314                 $text = preg_replace_callback(
4315                         '/<!--(LINK|IWLINK) (.*?)-->/',
4316                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4317                         $text );
4318
4319                 wfProfileOut( $fname );
4320                 return $text;
4321         }
4322
4323         /**
4324          * @param array $matches
4325          * @return string
4326          * @private
4327          */
4328         function replaceLinkHoldersTextCallback( $matches ) {
4329                 $type = $matches[1];
4330                 $key  = $matches[2];
4331                 if( $type == 'LINK' ) {
4332                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4333                                 return $this->mLinkHolders['texts'][$key];
4334                         }
4335                 } elseif( $type == 'IWLINK' ) {
4336                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4337                                 return $this->mInterwikiLinkHolders['texts'][$key];
4338                         }
4339                 }
4340                 return $matches[0];
4341         }
4342
4343         /**
4344          * Tag hook handler for 'pre'.
4345          */
4346         function renderPreTag( $text, $attribs ) {
4347                 // Backwards-compatibility hack
4348                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4349
4350                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4351                 return wfOpenElement( 'pre', $attribs ) .
4352                         Xml::escapeTagsOnly( $content ) .
4353                         '</pre>';
4354         }
4355
4356         /**
4357          * Renders an image gallery from a text with one line per image.
4358          * text labels may be given by using |-style alternative text. E.g.
4359          *   Image:one.jpg|The number "1"
4360          *   Image:tree.jpg|A tree
4361          * given as text will return the HTML of a gallery with two images,
4362          * labeled 'The number "1"' and
4363          * 'A tree'.
4364          */
4365         function renderImageGallery( $text, $params ) {
4366                 $ig = new ImageGallery();
4367                 $ig->setContextTitle( $this->mTitle );
4368                 $ig->setShowBytes( false );
4369                 $ig->setShowFilename( false );
4370                 $ig->setParser( $this );
4371                 $ig->setHideBadImages();
4372                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4373                 $ig->useSkin( $this->mOptions->getSkin() );
4374                 $ig->mRevisionId = $this->mRevisionId;
4375
4376                 if( isset( $params['caption'] ) ) {
4377                         $caption = $params['caption'];
4378                         $caption = htmlspecialchars( $caption );
4379                         $caption = $this->replaceInternalLinks( $caption );
4380                         $ig->setCaptionHtml( $caption );
4381                 }
4382                 if( isset( $params['perrow'] ) ) {
4383                         $ig->setPerRow( $params['perrow'] );
4384                 }
4385                 if( isset( $params['widths'] ) ) {
4386                         $ig->setWidths( $params['widths'] );
4387                 }
4388                 if( isset( $params['heights'] ) ) {
4389                         $ig->setHeights( $params['heights'] );
4390                 }
4391
4392                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4393
4394                 $lines = explode( "\n", $text );
4395                 foreach ( $lines as $line ) {
4396                         # match lines like these:
4397                         # Image:someimage.jpg|This is some image
4398                         $matches = array();
4399                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4400                         # Skip empty lines
4401                         if ( count( $matches ) == 0 ) {
4402                                 continue;
4403                         }
4404
4405                         if ( strpos( $matches[0], '%' ) !== false )
4406                                 $matches[1] = urldecode( $matches[1] );
4407                         $tp = Title::newFromText( $matches[1] );
4408                         $nt =& $tp;
4409                         if( is_null( $nt ) ) {
4410                                 # Bogus title. Ignore these so we don't bomb out later.
4411                                 continue;
4412                         }
4413                         if ( isset( $matches[3] ) ) {
4414                                 $label = $matches[3];
4415                         } else {
4416                                 $label = '';
4417                         }
4418
4419                         $html = $this->recursiveTagParse( trim( $label ) );
4420
4421                         $ig->add( $nt, $html );
4422
4423                         # Only add real images (bug #5586)
4424                         if ( $nt->getNamespace() == NS_IMAGE ) {
4425                                 $this->mOutput->addImage( $nt->getDBkey() );
4426                         }
4427                 }
4428                 return $ig->toHTML();
4429         }
4430
4431         function getImageParams( $handler ) {
4432                 if ( $handler ) {
4433                         $handlerClass = get_class( $handler );
4434                 } else {
4435                         $handlerClass = '';
4436                 }
4437                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4438                         // Initialise static lists
4439                         static $internalParamNames = array(
4440                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4441                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4442                                         'bottom', 'text-bottom' ),
4443                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4444                                         'upright', 'border' ),
4445                         );
4446                         static $internalParamMap;
4447                         if ( !$internalParamMap ) {
4448                                 $internalParamMap = array();
4449                                 foreach ( $internalParamNames as $type => $names ) {
4450                                         foreach ( $names as $name ) {
4451                                                 $magicName = str_replace( '-', '_', "img_$name" );
4452                                                 $internalParamMap[$magicName] = array( $type, $name );
4453                                         }
4454                                 }
4455                         }
4456
4457                         // Add handler params
4458                         $paramMap = $internalParamMap;
4459                         if ( $handler ) {
4460                                 $handlerParamMap = $handler->getParamMap();
4461                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4462                                         $paramMap[$magic] = array( 'handler', $paramName );
4463                                 }
4464                         }
4465                         $this->mImageParams[$handlerClass] = $paramMap;
4466                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4467                 }
4468                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4469         }
4470
4471         /**
4472          * Parse image options text and use it to make an image
4473          */
4474         function makeImage( $title, $options ) {
4475                 # Check if the options text is of the form "options|alt text"
4476                 # Options are:
4477                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4478                 #  * left               no resizing, just left align. label is used for alt= only
4479                 #  * right              same, but right aligned
4480                 #  * none               same, but not aligned
4481                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4482                 #  * center             center the image
4483                 #  * framed             Keep original image size, no magnify-button.
4484                 #  * frameless          like 'thumb' but without a frame. Keeps user preferences for width
4485                 #  * upright            reduce width for upright images, rounded to full __0 px
4486                 #  * border             draw a 1px border around the image
4487                 # vertical-align values (no % or length right now):
4488                 #  * baseline
4489                 #  * sub
4490                 #  * super
4491                 #  * top
4492                 #  * text-top
4493                 #  * middle
4494                 #  * bottom
4495                 #  * text-bottom
4496
4497                 $parts = array_map( 'trim', explode( '|', $options) );
4498                 $sk = $this->mOptions->getSkin();
4499
4500                 # Give extensions a chance to select the file revision for us
4501                 $skip = $time = $descQuery = false;
4502                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4503
4504                 if ( $skip ) {
4505                         return $sk->makeLinkObj( $title );
4506                 }
4507
4508                 # Get parameter map
4509                 $file = wfFindFile( $title, $time );
4510                 $handler = $file ? $file->getHandler() : false;
4511
4512                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4513
4514                 # Process the input parameters
4515                 $caption = '';
4516                 $params = array( 'frame' => array(), 'handler' => array(),
4517                         'horizAlign' => array(), 'vertAlign' => array() );
4518                 foreach( $parts as $part ) {
4519                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4520                         $validated = false;
4521                         if( isset( $paramMap[$magicName] ) ) {
4522                                 list( $type, $paramName ) = $paramMap[$magicName];
4523
4524                                 // Special case; width and height come in one variable together
4525                                 if( $type == 'handler' && $paramName == 'width' ) {
4526                                         $m = array();
4527                                         # (bug 13500) In both cases (width/height and width only),
4528                                         # permit trailing "px" for backward compatibility.
4529                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4530                                                 $width = intval( $m[1] );
4531                                                 $height = intval( $m[2] );
4532                                                 if ( $handler->validateParam( 'width', $width ) ) {
4533                                                         $params[$type]['width'] = $width;
4534                                                         $validated = true;
4535                                                 }
4536                                                 if ( $handler->validateParam( 'height', $height ) ) {
4537                                                         $params[$type]['height'] = $height;
4538                                                         $validated = true;
4539                                                 }
4540                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4541                                                 $width = intval( $value );
4542                                                 if ( $handler->validateParam( 'width', $width ) ) {
4543                                                         $params[$type]['width'] = $width;
4544                                                         $validated = true;
4545                                                 }
4546                                         } // else no validation -- bug 13436
4547                                 } else {
4548                                         if ( $type == 'handler' ) {
4549                                                 # Validate handler parameter
4550                                                 $validated = $handler->validateParam( $paramName, $value );
4551                                         } else {
4552                                                 # Validate internal parameters
4553                                                 switch( $paramName ) {
4554                                                 case "manualthumb":
4555                                                         /// @fixme - possibly check validity here?
4556                                                         /// downstream behavior seems odd with missing manual thumbs.
4557                                                         $validated = true;
4558                                                         break;
4559                                                 default:
4560                                                         // Most other things appear to be empty or numeric...
4561                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4562                                                 }
4563                                         }
4564
4565                                         if ( $validated ) {
4566                                                 $params[$type][$paramName] = $value;
4567                                         }
4568                                 }
4569                         }
4570                         if ( !$validated ) {
4571                                 $caption = $part;
4572                         }
4573                 }
4574
4575                 # Process alignment parameters
4576                 if ( $params['horizAlign'] ) {
4577                         $params['frame']['align'] = key( $params['horizAlign'] );
4578                 }
4579                 if ( $params['vertAlign'] ) {
4580                         $params['frame']['valign'] = key( $params['vertAlign'] );
4581                 }
4582
4583                 # Strip bad stuff out of the alt text
4584                 $alt = $this->replaceLinkHoldersText( $caption );
4585
4586                 # make sure there are no placeholders in thumbnail attributes
4587                 # that are later expanded to html- so expand them now and
4588                 # remove the tags
4589                 $alt = $this->mStripState->unstripBoth( $alt );
4590                 $alt = Sanitizer::stripAllTags( $alt );
4591
4592                 $params['frame']['alt'] = $alt;
4593                 $params['frame']['caption'] = $caption;
4594
4595                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4596
4597                 # Linker does the rest
4598                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4599
4600                 # Give the handler a chance to modify the parser object
4601                 if ( $handler ) {
4602                         $handler->parserTransformHook( $this, $file );
4603                 }
4604
4605                 return $ret;
4606         }
4607
4608         /**
4609          * Set a flag in the output object indicating that the content is dynamic and
4610          * shouldn't be cached.
4611          */
4612         function disableCache() {
4613                 wfDebug( "Parser output marked as uncacheable.\n" );
4614                 $this->mOutput->mCacheTime = -1;
4615         }
4616
4617         /**#@+
4618          * Callback from the Sanitizer for expanding items found in HTML attribute
4619          * values, so they can be safely tested and escaped.
4620          * @param string $text
4621          * @param PPFrame $frame
4622          * @return string
4623          * @private
4624          */
4625         function attributeStripCallback( &$text, $frame = false ) {
4626                 $text = $this->replaceVariables( $text, $frame );
4627                 $text = $this->mStripState->unstripBoth( $text );
4628                 return $text;
4629         }
4630
4631         /**#@-*/
4632
4633         /**#@+
4634          * Accessor/mutator
4635          */
4636         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4637         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4638         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4639         /**#@-*/
4640
4641         /**#@+
4642          * Accessor
4643          */
4644         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4645         /**#@-*/
4646
4647
4648         /**
4649          * Break wikitext input into sections, and either pull or replace
4650          * some particular section's text.
4651          *
4652          * External callers should use the getSection and replaceSection methods.
4653          *
4654          * @param string $text Page wikitext
4655          * @param string $section A section identifier string of the form:
4656          *   <flag1> - <flag2> - ... - <section number>
4657          *
4658          * Currently the only recognised flag is "T", which means the target section number
4659          * was derived during a template inclusion parse, in other words this is a template
4660          * section edit link. If no flags are given, it was an ordinary section edit link.
4661          * This flag is required to avoid a section numbering mismatch when a section is
4662          * enclosed by <includeonly> (bug 6563).
4663          *
4664          * The section number 0 pulls the text before the first heading; other numbers will
4665          * pull the given section along with its lower-level subsections. If the section is
4666          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4667          *
4668          * @param string $mode One of "get" or "replace"
4669          * @param string $newText Replacement text for section data.
4670          * @return string for "get", the extracted section text.
4671          *                for "replace", the whole page with the section replaced.
4672          */
4673         private function extractSections( $text, $section, $mode, $newText='' ) {
4674                 global $wgTitle;
4675                 $this->clearState();
4676                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4677                 $this->mOptions = new ParserOptions;
4678                 $this->setOutputType( self::OT_WIKI );
4679                 $outText = '';
4680                 $frame = $this->getPreprocessor()->newFrame();
4681
4682                 // Process section extraction flags
4683                 $flags = 0;
4684                 $sectionParts = explode( '-', $section );
4685                 $sectionIndex = array_pop( $sectionParts );
4686                 foreach ( $sectionParts as $part ) {
4687                         if ( $part == 'T' ) {
4688                                 $flags |= self::PTD_FOR_INCLUSION;
4689                         }
4690                 }
4691                 // Preprocess the text
4692                 $root = $this->preprocessToDom( $text, $flags );
4693
4694                 // <h> nodes indicate section breaks
4695                 // They can only occur at the top level, so we can find them by iterating the root's children
4696                 $node = $root->getFirstChild();
4697
4698                 // Find the target section
4699                 if ( $sectionIndex == 0 ) {
4700                         // Section zero doesn't nest, level=big
4701                         $targetLevel = 1000;
4702                 } else {
4703             while ( $node ) {
4704                 if ( $node->getName() == 'h' ) {
4705                     $bits = $node->splitHeading();
4706                                         if ( $bits['i'] == $sectionIndex ) {
4707                                         $targetLevel = $bits['level'];
4708                                                 break;
4709                                         }
4710                                 }
4711                                 if ( $mode == 'replace' ) {
4712                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4713                                 }
4714                                 $node = $node->getNextSibling();
4715                         }
4716                 }
4717
4718                 if ( !$node ) {
4719                         // Not found
4720                         if ( $mode == 'get' ) {
4721                                 return $newText;
4722                         } else {
4723                                 return $text;
4724                         }
4725                 }
4726
4727                 // Find the end of the section, including nested sections
4728                 do {
4729                         if ( $node->getName() == 'h' ) {
4730                                 $bits = $node->splitHeading();
4731                                 $curLevel = $bits['level'];
4732                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4733                                         break;
4734                                 }
4735                         }
4736                         if ( $mode == 'get' ) {
4737                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4738                         }
4739                         $node = $node->getNextSibling();
4740                 } while ( $node );
4741
4742                 // Write out the remainder (in replace mode only)
4743                 if ( $mode == 'replace' ) {
4744                         // Output the replacement text
4745                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4746                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4747                         $outText .= $newText . "\n\n";
4748                         while ( $node ) {
4749                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4750                                 $node = $node->getNextSibling();
4751                         }
4752                 }
4753
4754                 if ( is_string( $outText ) ) {
4755                         // Re-insert stripped tags
4756                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4757                 }
4758
4759                 return $outText;
4760         }
4761
4762         /**
4763          * This function returns the text of a section, specified by a number ($section).
4764          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4765          * the first section before any such heading (section 0).
4766          *
4767          * If a section contains subsections, these are also returned.
4768          *
4769          * @param string $text text to look in
4770          * @param string $section section identifier
4771          * @param string $deftext default to return if section is not found
4772          * @return string text of the requested section
4773          */
4774         public function getSection( $text, $section, $deftext='' ) {
4775                 return $this->extractSections( $text, $section, "get", $deftext );
4776         }
4777
4778         public function replaceSection( $oldtext, $section, $text ) {
4779                 return $this->extractSections( $oldtext, $section, "replace", $text );
4780         }
4781
4782         /**
4783          * Get the timestamp associated with the current revision, adjusted for
4784          * the default server-local timestamp
4785          */
4786         function getRevisionTimestamp() {
4787                 if ( is_null( $this->mRevisionTimestamp ) ) {
4788                         wfProfileIn( __METHOD__ );
4789                         global $wgContLang;
4790                         $dbr = wfGetDB( DB_SLAVE );
4791                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4792                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4793
4794                         // Normalize timestamp to internal MW format for timezone processing.
4795                         // This has the added side-effect of replacing a null value with
4796                         // the current time, which gives us more sensible behavior for
4797                         // previews.
4798                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4799
4800                         // The cryptic '' timezone parameter tells to use the site-default
4801                         // timezone offset instead of the user settings.
4802                         //
4803                         // Since this value will be saved into the parser cache, served
4804                         // to other users, and potentially even used inside links and such,
4805                         // it needs to be consistent for all visitors.
4806                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4807
4808                         wfProfileOut( __METHOD__ );
4809                 }
4810                 return $this->mRevisionTimestamp;
4811         }
4812
4813         /**
4814          * Mutator for $mDefaultSort
4815          *
4816          * @param $sort New value
4817          */
4818         public function setDefaultSort( $sort ) {
4819                 $this->mDefaultSort = $sort;
4820         }
4821
4822         /**
4823          * Accessor for $mDefaultSort
4824          * Will use the title/prefixed title if none is set
4825          *
4826          * @return string
4827          */
4828         public function getDefaultSort() {
4829                 if( $this->mDefaultSort !== false ) {
4830                         return $this->mDefaultSort;
4831                 } else {
4832                         return $this->mTitle->getNamespace() == NS_CATEGORY
4833                                         ? $this->mTitle->getText()
4834                                         : $this->mTitle->getPrefixedText();
4835                 }
4836         }
4837
4838         /**
4839          * Try to guess the section anchor name based on a wikitext fragment
4840          * presumably extracted from a heading, for example "Header" from
4841          * "== Header ==".
4842          */
4843         public function guessSectionNameFromWikiText( $text ) {
4844                 # Strip out wikitext links(they break the anchor)
4845                 $text = $this->stripSectionName( $text );
4846                 $headline = Sanitizer::decodeCharReferences( $text );
4847                 # strip out HTML
4848                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
4849                 $headline = trim( $headline );
4850                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
4851                 $replacearray = array(
4852                         '%3A' => ':',
4853                         '%' => '.'
4854                 );
4855                 return str_replace(
4856                         array_keys( $replacearray ),
4857                         array_values( $replacearray ),
4858                         $sectionanchor );
4859         }
4860
4861         /**
4862          * Strips a text string of wikitext for use in a section anchor
4863          *
4864          * Accepts a text string and then removes all wikitext from the
4865          * string and leaves only the resultant text (i.e. the result of
4866          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
4867          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
4868          * to create valid section anchors by mimicing the output of the
4869          * parser when headings are parsed.
4870          *
4871          * @param $text string Text string to be stripped of wikitext
4872          * for use in a Section anchor
4873          * @return Filtered text string
4874          */
4875         public function stripSectionName( $text ) {
4876                 # Strip internal link markup
4877                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
4878                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
4879
4880                 # Strip external link markup (FIXME: Not Tolerant to blank link text
4881                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
4882                 # on how many empty links there are on the page - need to figure that out.
4883                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
4884
4885                 # Parse wikitext quotes (italics & bold)
4886                 $text = $this->doQuotes($text);
4887
4888                 # Strip HTML tags
4889                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
4890                 return $text;
4891         }
4892
4893         function srvus( $text ) {
4894                 return $this->testSrvus( $text, $this->mOutputType );
4895         }
4896
4897         /**
4898          * strip/replaceVariables/unstrip for preprocessor regression testing
4899          */
4900         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
4901                 $this->clearState();
4902                 if ( ! ( $title instanceof Title ) ) {
4903                         $title = Title::newFromText( $title );
4904                 }
4905                 $this->mTitle = $title;
4906                 $this->mOptions = $options;
4907                 $this->setOutputType( $outputType );
4908                 $text = $this->replaceVariables( $text );
4909                 $text = $this->mStripState->unstripBoth( $text );
4910                 $text = Sanitizer::removeHTMLtags( $text );
4911                 return $text;
4912         }
4913
4914         function testPst( $text, $title, $options ) {
4915                 global $wgUser;
4916                 if ( ! ( $title instanceof Title ) ) {
4917                         $title = Title::newFromText( $title );
4918                 }
4919                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
4920         }
4921
4922         function testPreprocess( $text, $title, $options ) {
4923                 if ( ! ( $title instanceof Title ) ) {
4924                         $title = Title::newFromText( $title );
4925                 }
4926                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
4927         }
4928
4929         function markerSkipCallback( $s, $callback ) {
4930                 $i = 0;
4931                 $out = '';
4932                 while ( $i < strlen( $s ) ) {
4933                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
4934                         if ( $markerStart === false ) {
4935                                 $out .= call_user_func( $callback, substr( $s, $i ) );
4936                                 break;
4937                         } else {
4938                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
4939                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
4940                                 if ( $markerEnd === false ) {
4941                                         $out .= substr( $s, $markerStart );
4942                                         break;
4943                                 } else {
4944                                         $markerEnd += strlen( self::MARKER_SUFFIX );
4945                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
4946                                         $i = $markerEnd;
4947                                 }
4948                         }
4949                 }
4950                 return $out;
4951         }
4952 }
4953
4954 /**
4955  * @todo document, briefly.
4956  * @ingroup Parser
4957  */
4958 class StripState {
4959         var $general, $nowiki;
4960
4961         function __construct() {
4962                 $this->general = new ReplacementArray;
4963                 $this->nowiki = new ReplacementArray;
4964         }
4965
4966         function unstripGeneral( $text ) {
4967                 wfProfileIn( __METHOD__ );
4968                 do {
4969                         $oldText = $text;
4970                         $text = $this->general->replace( $text );
4971                 } while ( $text != $oldText );
4972                 wfProfileOut( __METHOD__ );
4973                 return $text;
4974         }
4975
4976         function unstripNoWiki( $text ) {
4977                 wfProfileIn( __METHOD__ );
4978                 do {
4979                         $oldText = $text;
4980                         $text = $this->nowiki->replace( $text );
4981                 } while ( $text != $oldText );
4982                 wfProfileOut( __METHOD__ );
4983                 return $text;
4984         }
4985
4986         function unstripBoth( $text ) {
4987                 wfProfileIn( __METHOD__ );
4988                 do {
4989                         $oldText = $text;
4990                         $text = $this->general->replace( $text );
4991                         $text = $this->nowiki->replace( $text );
4992                 } while ( $text != $oldText );
4993                 wfProfileOut( __METHOD__ );
4994                 return $text;
4995         }
4996 }
4997
4998 /**
4999  * @todo document, briefly.
5000  * @ingroup Parser
5001  */
5002 class OnlyIncludeReplacer {
5003         var $output = '';
5004
5005         function replace( $matches ) {
5006                 if ( substr( $matches[1], -1 ) == "\n" ) {
5007                         $this->output .= substr( $matches[1], 0, -1 );
5008                 } else {
5009                         $this->output .= $matches[1];
5010                 }
5011         }
5012 }