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, $mUrlProtocols, $mDefaultStripList, $mVarCache, $mConf;
  96
  97
  98         # Cleared with clearState():
  99         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 100         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 101         var $mLinkHolders, $mLinkID;
 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', 'poem' );
 131                 $this->mUrlProtocols = wfUrlProtocols();
 132                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 133                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 134                 $this->mVarCache = array();
 135                 if ( isset( $conf['preprocessorClass'] ) ) {
 136                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 137                 } elseif ( extension_loaded( 'domxml' ) ) {
 138                         // PECL extension that conflicts with the core DOM extension (bug 13770)
 139                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 140                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 141                 } elseif ( extension_loaded( 'dom' ) ) {
 142                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 143                 } else {
 144                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 145                 }
 146                 $this->mMarkerIndex = 0;
 147                 $this->mFirstCall = true;
 148         }
 149
 150         /**
 151          * Reduce memory usage to reduce the impact of circular references
 152          */
 153         function __destruct() {
 154                 if ( isset( $this->mLinkHolders ) ) {
 155                         $this->mLinkHolders->__destruct();
 156                 }
 157                 foreach ( $this as $name => $value ) {
 158                         unset( $this->$name );
 159                 }
 160         }
 161
 162         /**
 163          * Do various kinds of initialisation on the first call of the parser
 164          */
 165         function firstCallInit() {
 166                 if ( !$this->mFirstCall ) {
 167                         return;
 168                 }
 169                 $this->mFirstCall = false;
 170
 171                 wfProfileIn( __METHOD__ );
 172
 173                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 174                 CoreParserFunctions::register( $this );
 175                 $this->initialiseVariables();
 176
 177                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 178                 wfProfileOut( __METHOD__ );
 179         }
 180
 181         /**
 182          * Clear Parser state
 183          *
 184          * @private
 185          */
 186         function clearState() {
 187                 wfProfileIn( __METHOD__ );
 188                 if ( $this->mFirstCall ) {
 189                         $this->firstCallInit();
 190                 }
 191                 $this->mOutput = new ParserOutput;
 192                 $this->mAutonumber = 0;
 193                 $this->mLastSection = '';
 194                 $this->mDTopen = false;
 195                 $this->mIncludeCount = array();
 196                 $this->mStripState = new StripState;
 197                 $this->mArgStack = false;
 198                 $this->mInPre = false;
 199                 $this->mLinkHolders = new LinkHolderArray( $this );
 200                 $this->mLinkID = 0;
 201                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 202
 203                 /**
 204                  * Prefix for temporary replacement strings for the multipass parser.
 205                  * \x07 should never appear in input as it's disallowed in XML.
 206                  * Using it at the front also gives us a little extra robustness
 207                  * since it shouldn't match when butted up against identifier-like
 208                  * string constructs.
 209                  *
 210                  * Must not consist of all title characters, or else it will change
 211                  * the behaviour of <nowiki> in a link.
 212                  */
 213                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 214                 # Changed to \x7f to allow XML double-parsing -- TS
 215                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 216
 217
 218                 # Clear these on every parse, bug 4549
 219                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 220
 221                 $this->mShowToc = true;
 222                 $this->mForceTocPosition = false;
 223                 $this->mIncludeSizes = array(
 224                         'post-expand' => 0,
 225                         'arg' => 0,
 226                 );
 227                 $this->mPPNodeCount = 0;
 228                 $this->mDefaultSort = false;
 229                 $this->mHeadings = array();
 230                 $this->mDoubleUnderscores = array();
 231                 $this->mExpensiveFunctionCount = 0;
 232
 233                 # Fix cloning
 234                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 235                         $this->mPreprocessor = null;
 236                 }
 237
 238                 wfRunHooks( 'ParserClearState', array( &$this ) );
 239                 wfProfileOut( __METHOD__ );
 240         }
 241
 242         function setOutputType( $ot ) {
 243                 $this->mOutputType = $ot;
 244                 // Shortcut alias
 245                 $this->ot = array(
 246                         'html' => $ot == self::OT_HTML,
 247                         'wiki' => $ot == self::OT_WIKI,
 248                         'pre' => $ot == self::OT_PREPROCESS,
 249                 );
 250         }
 251
 252         /**
 253          * Set the context title
 254          */
 255         function setTitle( $t ) {
 256                 if ( !$t || $t instanceof FakeTitle ) {
 257                         $t = Title::newFromText( 'NO TITLE' );
 258                 }
 259                 if ( strval( $t->getFragment() ) !== '' ) {
 260                         # Strip the fragment to avoid various odd effects
 261                         $this->mTitle = clone $t;
 262                         $this->mTitle->setFragment( '' );
 263                 } else {
 264                         $this->mTitle = $t;
 265                 }
 266         }
 267
 268         /**
 269          * Accessor for mUniqPrefix.
 270          *
 271          * @public
 272          */
 273         function uniqPrefix() {
 274                 if( !isset( $this->mUniqPrefix ) ) {
 275                         // @fixme this is probably *horribly wrong*
 276                         // LanguageConverter seems to want $wgParser's uniqPrefix, however
 277                         // if this is called for a parser cache hit, the parser may not
 278                         // have ever been initialized in the first place.
 279                         // Not really sure what the heck is supposed to be going on here.
 280                         return '';
 281                         //throw new MWException( "Accessing uninitialized mUniqPrefix" );
 282                 }
 283                 return $this->mUniqPrefix;
 284         }
 285
 286         /**
 287          * Convert wikitext to HTML
 288          * Do not call this function recursively.
 289          *
 290          * @param $text String: text we want to parse
 291          * @param $title A title object
 292          * @param $options ParserOptions
 293          * @param $linestart boolean
 294          * @param $clearState boolean
 295          * @param $revid Int: number to pass in {{REVISIONID}}
 296          * @return ParserOutput a ParserOutput
 297          */
 298         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 299                 /**
 300                  * First pass--just handle <nowiki> sections, pass the rest off
 301                  * to internalParse() which does all the real work.
 302                  */
 303
 304                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 305                 $fname = __METHOD__.'-' . wfGetCaller();
 306                 wfProfileIn( __METHOD__ );
 307                 wfProfileIn( $fname );
 308
 309                 if ( $clearState ) {
 310                         $this->clearState();
 311                 }
 312
 313                 $this->mOptions = $options;
 314                 $this->setTitle( $title );
 315                 $oldRevisionId = $this->mRevisionId;
 316                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 317                 if( $revid !== null ) {
 318                         $this->mRevisionId = $revid;
 319                         $this->mRevisionTimestamp = null;
 320                 }
 321                 $this->setOutputType( self::OT_HTML );
 322                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 323                 # No more strip!
 324                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 325                 $text = $this->internalParse( $text );
 326                 $text = $this->mStripState->unstripGeneral( $text );
 327
 328                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 329                 $fixtags = array(
 330                         # french spaces, last one Guillemet-left
 331                         # only if there is something before the space
 332                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 333                         # french spaces, Guillemet-right
 334                         '/(\\302\\253) /' => '\\1&nbsp;',
 335                         '/&nbsp;(!\s*important)/' => ' \\1', #Beware of CSS magic word !important, bug #11874.
 336                 );
 337                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 338
 339                 $text = $this->doBlockLevels( $text, $linestart );
 340
 341                 $this->replaceLinkHolders( $text );
 342
 343                 # the position of the parserConvert() call should not be changed. it
 344                 # assumes that the links are all replaced and the only thing left
 345                 # is the <nowiki> mark.
 346                 # Side-effects: this calls $this->mOutput->setTitleText()
 347                 $text = $wgContLang->parserConvert( $text, $this );
 348
 349                 $text = $this->mStripState->unstripNoWiki( $text );
 350
 351                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 352
 353 //!JF Move to its own function
 354
 355                 $uniq_prefix = $this->mUniqPrefix;
 356                 $matches = array();
 357                 $elements = array_keys( $this->mTransparentTagHooks );
 358                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 359
 360                 foreach( $matches as $marker => $data ) {
 361                         list( $element, $content, $params, $tag ) = $data;
 362                         $tagName = strtolower( $element );
 363                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 364                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 365                                         array( $content, $params, $this ) );
 366                         } else {
 367                                 $output = $tag;
 368                         }
 369                         $this->mStripState->general->setPair( $marker, $output );
 370                 }
 371                 $text = $this->mStripState->unstripGeneral( $text );
 372
 373                 $text = Sanitizer::normalizeCharReferences( $text );
 374
 375                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 376                         $text = self::tidy($text);
 377                 } else {
 378                         # attempt to sanitize at least some nesting problems
 379                         # (bug #2702 and quite a few others)
 380                         $tidyregs = array(
 381                                 # ''Something [http://www.cool.com cool''] -->
 382                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 383                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 384                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 385                                 # fix up an anchor inside another anchor, only
 386                                 # at least for a single single nested link (bug 3695)
 387                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 388                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 389                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 390                                 # contains a div, so fix it up here; replace
 391                                 # div with escaped text
 392                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 393                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 394                                 # remove empty italic or bold tag pairs, some
 395                                 # introduced by rules above
 396                                 '/<([bi])><\/\\1>/' => '',
 397                         );
 398
 399                         $text = preg_replace(
 400                                 array_keys( $tidyregs ),
 401                                 array_values( $tidyregs ),
 402                                 $text );
 403                 }
 404                 global $wgExpensiveParserFunctionLimit;
 405                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 406                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 407                 }
 408
 409                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 410
 411                 # Information on include size limits, for the benefit of users who try to skirt them
 412                 if ( $this->mOptions->getEnableLimitReport() ) {
 413                         global $wgExpensiveParserFunctionLimit;
 414                         $max = $this->mOptions->getMaxIncludeSize();
 415                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 416                         $limitReport =
 417                                 "NewPP limit report\n" .
 418                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 419                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 420                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 421                                 $PFreport;
 422                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 423                         $text .= "\n<!-- \n$limitReport-->\n";
 424                 }
 425                 $this->mOutput->setText( $text );
 426                 $this->mRevisionId = $oldRevisionId;
 427                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 428                 wfProfileOut( $fname );
 429                 wfProfileOut( __METHOD__ );
 430
 431                 return $this->mOutput;
 432         }
 433
 434         /**
 435          * Recursive parser entry point that can be called from an extension tag
 436          * hook.
 437          */
 438         function recursiveTagParse( $text ) {
 439                 wfProfileIn( __METHOD__ );
 440                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 441                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 442                 $text = $this->internalParse( $text );
 443                 wfProfileOut( __METHOD__ );
 444                 return $text;
 445         }
 446
 447         /**
 448          * Expand templates and variables in the text, producing valid, static wikitext.
 449          * Also removes comments.
 450          */
 451         function preprocess( $text, $title, $options, $revid = null ) {
 452                 wfProfileIn( __METHOD__ );
 453                 $this->clearState();
 454                 $this->setOutputType( self::OT_PREPROCESS );
 455                 $this->mOptions = $options;
 456                 $this->setTitle( $title );
 457                 if( $revid !== null ) {
 458                         $this->mRevisionId = $revid;
 459                 }
 460                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 461                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 462                 $text = $this->replaceVariables( $text );
 463                 $text = $this->mStripState->unstripBoth( $text );
 464                 wfProfileOut( __METHOD__ );
 465                 return $text;
 466         }
 467
 468         /**
 469          * Get a random string
 470          *
 471          * @private
 472          * @static
 473          */
 474         function getRandomString() {
 475                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 476         }
 477
 478         function &getTitle() { return $this->mTitle; }
 479         function getOptions() { return $this->mOptions; }
 480         function getRevisionId() { return $this->mRevisionId; }
 481         function getOutput() { return $this->mOutput; }
 482         function nextLinkID() { return $this->mLinkID++; }
 483
 484         function getFunctionLang() {
 485                 global $wgLang, $wgContLang;
 486
 487                 $target = $this->mOptions->getTargetLanguage();
 488                 if ( $target !== null ) {
 489                         return $target;
 490                 } else {
 491                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 492                 }
 493         }
 494
 495         /**
 496          * Get a preprocessor object
 497          */
 498         function getPreprocessor() {
 499                 if ( !isset( $this->mPreprocessor ) ) {
 500                         $class = $this->mPreprocessorClass;
 501                         $this->mPreprocessor = new $class( $this );
 502                 }
 503                 return $this->mPreprocessor;
 504         }
 505
 506         /**
 507          * Replaces all occurrences of HTML-style comments and the given tags
 508          * in the text with a random marker and returns the next text. The output
 509          * parameter $matches will be an associative array filled with data in
 510          * the form:
 511          *   'UNIQ-xxxxx' => array(
 512          *     'element',
 513          *     'tag content',
 514          *     array( 'param' => 'x' ),
 515          *     '<element param="x">tag content</element>' ) )
 516          *
 517          * @param $elements list of element names. Comments are always extracted.
 518          * @param $text Source text string.
 519          * @param $uniq_prefix
 520          *
 521          * @public
 522          * @static
 523          */
 524         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 525                 static $n = 1;
 526                 $stripped = '';
 527                 $matches = array();
 528
 529                 $taglist = implode( '|', $elements );
 530                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 531
 532                 while ( '' != $text ) {
 533                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 534                         $stripped .= $p[0];
 535                         if( count( $p ) < 5 ) {
 536                                 break;
 537                         }
 538                         if( count( $p ) > 5 ) {
 539                                 // comment
 540                                 $element    = $p[4];
 541                                 $attributes = '';
 542                                 $close      = '';
 543                                 $inside     = $p[5];
 544                         } else {
 545                                 // tag
 546                                 $element    = $p[1];
 547                                 $attributes = $p[2];
 548                                 $close      = $p[3];
 549                                 $inside     = $p[4];
 550                         }
 551
 552                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . self::MARKER_SUFFIX;
 553                         $stripped .= $marker;
 554
 555                         if ( $close === '/>' ) {
 556                                 // Empty element tag, <tag />
 557                                 $content = null;
 558                                 $text = $inside;
 559                                 $tail = null;
 560                         } else {
 561                                 if( $element === '!--' ) {
 562                                         $end = '/(-->)/';
 563                                 } else {
 564                                         $end = "/(<\\/$element\\s*>)/i";
 565                                 }
 566                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 567                                 $content = $q[0];
 568                                 if( count( $q ) < 3 ) {
 569                                         # No end tag -- let it run out to the end of the text.
 570                                         $tail = '';
 571                                         $text = '';
 572                                 } else {
 573                                         $tail = $q[1];
 574                                         $text = $q[2];
 575                                 }
 576                         }
 577
 578                         $matches[$marker] = array( $element,
 579                                 $content,
 580                                 Sanitizer::decodeTagAttributes( $attributes ),
 581                                 "<$element$attributes$close$content$tail" );
 582                 }
 583                 return $stripped;
 584         }
 585
 586         /**
 587          * Get a list of strippable XML-like elements
 588          */
 589         function getStripList() {
 590                 global $wgRawHtml;
 591                 $elements = $this->mStripList;
 592                 if( $wgRawHtml ) {
 593                         $elements[] = 'html';
 594                 }
 595                 if( $this->mOptions->getUseTeX() ) {
 596                         $elements[] = 'math';
 597                 }
 598                 return $elements;
 599         }
 600
 601         /**
 602          * @deprecated use replaceVariables
 603          */
 604         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 605                 return $text;
 606         }
 607
 608         /**
 609          * Restores pre, math, and other extensions removed by strip()
 610          *
 611          * always call unstripNoWiki() after this one
 612          * @private
 613          * @deprecated use $this->mStripState->unstrip()
 614          */
 615         function unstrip( $text, $state ) {
 616                 return $state->unstripGeneral( $text );
 617         }
 618
 619         /**
 620          * Always call this after unstrip() to preserve the order
 621          *
 622          * @private
 623          * @deprecated use $this->mStripState->unstrip()
 624          */
 625         function unstripNoWiki( $text, $state ) {
 626                 return $state->unstripNoWiki( $text );
 627         }
 628
 629         /**
 630          * @deprecated use $this->mStripState->unstripBoth()
 631          */
 632         function unstripForHTML( $text ) {
 633                 return $this->mStripState->unstripBoth( $text );
 634         }
 635
 636         /**
 637          * Add an item to the strip state
 638          * Returns the unique tag which must be inserted into the stripped text
 639          * The tag will be replaced with the original text in unstrip()
 640          *
 641          * @private
 642          */
 643         function insertStripItem( $text ) {
 644                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 645                 $this->mMarkerIndex++;
 646                 $this->mStripState->general->setPair( $rnd, $text );
 647                 return $rnd;
 648         }
 649
 650         /**
 651          * Interface with html tidy, used if $wgUseTidy = true.
 652          * If tidy isn't able to correct the markup, the original will be
 653          * returned in all its glory with a warning comment appended.
 654          *
 655          * Either the external tidy program or the in-process tidy extension
 656          * will be used depending on availability. Override the default
 657          * $wgTidyInternal setting to disable the internal if it's not working.
 658          *
 659          * @param string $text Hideous HTML input
 660          * @return string Corrected HTML output
 661          * @public
 662          * @static
 663          */
 664         function tidy( $text ) {
 665                 global $wgTidyInternal;
 666                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 667 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 668 '<head><title>test</title></head><body>'.$text.'</body></html>';
 669                 if( $wgTidyInternal ) {
 670                         $correctedtext = self::internalTidy( $wrappedtext );
 671                 } else {
 672                         $correctedtext = self::externalTidy( $wrappedtext );
 673                 }
 674                 if( is_null( $correctedtext ) ) {
 675                         wfDebug( "Tidy error detected!\n" );
 676                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 677                 }
 678                 return $correctedtext;
 679         }
 680
 681         /**
 682          * Spawn an external HTML tidy process and get corrected markup back from it.
 683          *
 684          * @private
 685          * @static
 686          */
 687         function externalTidy( $text ) {
 688                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 689                 wfProfileIn( __METHOD__ );
 690
 691                 $cleansource = '';
 692                 $opts = ' -utf8';
 693
 694                 $descriptorspec = array(
 695                         0 => array('pipe', 'r'),
 696                         1 => array('pipe', 'w'),
 697                         2 => array('file', wfGetNull(), 'a')
 698                 );
 699                 $pipes = array();
 700                 if( function_exists('proc_open') ) {
 701                         $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 702                         if (is_resource($process)) {
 703                                 // Theoretically, this style of communication could cause a deadlock
 704                                 // here. If the stdout buffer fills up, then writes to stdin could
 705                                 // block. This doesn't appear to happen with tidy, because tidy only
 706                                 // writes to stdout after it's finished reading from stdin. Search
 707                                 // for tidyParseStdin and tidySaveStdout in console/tidy.c
 708                                 fwrite($pipes[0], $text);
 709                                 fclose($pipes[0]);
 710                                 while (!feof($pipes[1])) {
 711                                         $cleansource .= fgets($pipes[1], 1024);
 712                                 }
 713                                 fclose($pipes[1]);
 714                                 proc_close($process);
 715                         }
 716                 }
 717
 718                 wfProfileOut( __METHOD__ );
 719
 720                 if( $cleansource == '' && $text != '') {
 721                         // Some kind of error happened, so we couldn't get the corrected text.
 722                         // Just give up; we'll use the source text and append a warning.
 723                         return null;
 724                 } else {
 725                         return $cleansource;
 726                 }
 727         }
 728
 729         /**
 730          * Use the HTML tidy PECL extension to use the tidy library in-process,
 731          * saving the overhead of spawning a new process.
 732          *
 733          * 'pear install tidy' should be able to compile the extension module.
 734          *
 735          * @private
 736          * @static
 737          */
 738         function internalTidy( $text ) {
 739                 global $wgTidyConf, $IP, $wgDebugTidy;
 740                 wfProfileIn( __METHOD__ );
 741
 742                 $tidy = new tidy;
 743                 $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 744                 $tidy->cleanRepair();
 745                 if( $tidy->getStatus() == 2 ) {
 746                         // 2 is magic number for fatal error
 747                         // http://www.php.net/manual/en/function.tidy-get-status.php
 748                         $cleansource = null;
 749                 } else {
 750                         $cleansource = tidy_get_output( $tidy );
 751                 }
 752                 if ( $wgDebugTidy && $tidy->getStatus() > 0 ) {
 753                         $cleansource .= "<!--\nTidy reports:\n" .
 754                                 str_replace( '-->', '--&gt;', $tidy->errorBuffer ) .
 755                                 "\n-->";
 756                 }
 757
 758                 wfProfileOut( __METHOD__ );
 759                 return $cleansource;
 760         }
 761
 762         /**
 763          * parse the wiki syntax used to render tables
 764          *
 765          * @private
 766          */
 767         function doTableStuff ( $text ) {
 768                 wfProfileIn( __METHOD__ );
 769
 770                 $lines = StringUtils::explode( "\n", $text );
 771                 $out = '';
 772                 $td_history = array (); // Is currently a td tag open?
 773                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 774                 $tr_history = array (); // Is currently a tr tag open?
 775                 $tr_attributes = array (); // history of tr attributes
 776                 $has_opened_tr = array(); // Did this table open a <tr> element?
 777                 $indent_level = 0; // indent level of the table
 778
 779                 foreach ( $lines as $outLine ) {
 780                         $line = trim( $outLine );
 781
 782                         if( $line == '' ) { // empty line, go to next line
 783                                 $out .= $outLine."\n";
 784                                 continue;
 785                         }
 786                         $first_character = $line[0];
 787                         $matches = array();
 788
 789                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 790                                 // First check if we are starting a new table
 791                                 $indent_level = strlen( $matches[1] );
 792
 793                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 794                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 795
 796                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 797                                 array_push ( $td_history , false );
 798                                 array_push ( $last_tag_history , '' );
 799                                 array_push ( $tr_history , false );
 800                                 array_push ( $tr_attributes , '' );
 801                                 array_push ( $has_opened_tr , false );
 802                         } else if ( count ( $td_history ) == 0 ) {
 803                                 // Don't do any of the following
 804                                 $out .= $outLine."\n";
 805                                 continue;
 806                         } else if ( substr ( $line , 0 , 2 ) === '|}' ) {
 807                                 // We are ending a table
 808                                 $line = '</table>' . substr ( $line , 2 );
 809                                 $last_tag = array_pop ( $last_tag_history );
 810
 811                                 if ( !array_pop ( $has_opened_tr ) ) {
 812                                         $line = "<tr><td></td></tr>{$line}";
 813                                 }
 814
 815                                 if ( array_pop ( $tr_history ) ) {
 816                                         $line = "</tr>{$line}";
 817                                 }
 818
 819                                 if ( array_pop ( $td_history ) ) {
 820                                         $line = "</{$last_tag}>{$line}";
 821                                 }
 822                                 array_pop ( $tr_attributes );
 823                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 824                         } else if ( substr ( $line , 0 , 2 ) === '|-' ) {
 825                                 // Now we have a table row
 826                                 $line = preg_replace( '#^\|-+#', '', $line );
 827
 828                                 // Whats after the tag is now only attributes
 829                                 $attributes = $this->mStripState->unstripBoth( $line );
 830                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 831                                 array_pop ( $tr_attributes );
 832                                 array_push ( $tr_attributes , $attributes );
 833
 834                                 $line = '';
 835                                 $last_tag = array_pop ( $last_tag_history );
 836                                 array_pop ( $has_opened_tr );
 837                                 array_push ( $has_opened_tr , true );
 838
 839                                 if ( array_pop ( $tr_history ) ) {
 840                                         $line = '</tr>';
 841                                 }
 842
 843                                 if ( array_pop ( $td_history ) ) {
 844                                         $line = "</{$last_tag}>{$line}";
 845                                 }
 846
 847                                 $outLine = $line;
 848                                 array_push ( $tr_history , false );
 849                                 array_push ( $td_history , false );
 850                                 array_push ( $last_tag_history , '' );
 851                         }
 852                         else if ( $first_character === '|' || $first_character === '!' || substr ( $line , 0 , 2 )  === '|+' ) {
 853                                 // This might be cell elements, td, th or captions
 854                                 if ( substr ( $line , 0 , 2 ) === '|+' ) {
 855                                         $first_character = '+';
 856                                         $line = substr ( $line , 1 );
 857                                 }
 858
 859                                 $line = substr ( $line , 1 );
 860
 861                                 if ( $first_character === '!' ) {
 862                                         $line = str_replace ( '!!' , '||' , $line );
 863                                 }
 864
 865                                 // Split up multiple cells on the same line.
 866                                 // FIXME : This can result in improper nesting of tags processed
 867                                 // by earlier parser steps, but should avoid splitting up eg
 868                                 // attribute values containing literal "||".
 869                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 870
 871                                 $outLine = '';
 872
 873                                 // Loop through each table cell
 874                                 foreach ( $cells as $cell )
 875                                 {
 876                                         $previous = '';
 877                                         if ( $first_character !== '+' )
 878                                         {
 879                                                 $tr_after = array_pop ( $tr_attributes );
 880                                                 if ( !array_pop ( $tr_history ) ) {
 881                                                         $previous = "<tr{$tr_after}>\n";
 882                                                 }
 883                                                 array_push ( $tr_history , true );
 884                                                 array_push ( $tr_attributes , '' );
 885                                                 array_pop ( $has_opened_tr );
 886                                                 array_push ( $has_opened_tr , true );
 887                                         }
 888
 889                                         $last_tag = array_pop ( $last_tag_history );
 890
 891                                         if ( array_pop ( $td_history ) ) {
 892                                                 $previous = "</{$last_tag}>{$previous}";
 893                                         }
 894
 895                                         if ( $first_character === '|' ) {
 896                                                 $last_tag = 'td';
 897                                         } else if ( $first_character === '!' ) {
 898                                                 $last_tag = 'th';
 899                                         } else if ( $first_character === '+' ) {
 900                                                 $last_tag = 'caption';
 901                                         } else {
 902                                                 $last_tag = '';
 903                                         }
 904
 905                                         array_push ( $last_tag_history , $last_tag );
 906
 907                                         // A cell could contain both parameters and data
 908                                         $cell_data = explode ( '|' , $cell , 2 );
 909
 910                                         // Bug 553: Note that a '|' inside an invalid link should not
 911                                         // be mistaken as delimiting cell parameters
 912                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 913                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 914                                         } else if ( count ( $cell_data ) == 1 )
 915                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 916                                         else {
 917                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 918                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 919                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 920                                         }
 921
 922                                         $outLine .= $cell;
 923                                         array_push ( $td_history , true );
 924                                 }
 925                         }
 926                         $out .= $outLine . "\n";
 927                 }
 928
 929                 // Closing open td, tr && table
 930                 while ( count ( $td_history ) > 0 )
 931                 {
 932                         if ( array_pop ( $td_history ) ) {
 933                                 $out .= "</td>\n";
 934                         }
 935                         if ( array_pop ( $tr_history ) ) {
 936                                 $out .= "</tr>\n";
 937                         }
 938                         if ( !array_pop ( $has_opened_tr ) ) {
 939                                 $out .= "<tr><td></td></tr>\n" ;
 940                         }
 941
 942                         $out .= "</table>\n";
 943                 }
 944
 945                 // Remove trailing line-ending (b/c)
 946                 if ( substr( $out, -1 ) === "\n" ) {
 947                         $out = substr( $out, 0, -1 );
 948                 }
 949
 950                 // special case: don't return empty table
 951                 if( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 952                         $out = '';
 953                 }
 954
 955                 wfProfileOut( __METHOD__ );
 956
 957                 return $out;
 958         }
 959
 960         /**
 961          * Helper function for parse() that transforms wiki markup into
 962          * HTML. Only called for $mOutputType == self::OT_HTML.
 963          *
 964          * @private
 965          */
 966         function internalParse( $text ) {
 967                 $isMain = true;
 968                 wfProfileIn( __METHOD__ );
 969
 970                 # Hook to suspend the parser in this state
 971                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 972                         wfProfileOut( __METHOD__ );
 973                         return $text ;
 974                 }
 975
 976                 $text = $this->replaceVariables( $text );
 977                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 978                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 979
 980                 // Tables need to come after variable replacement for things to work
 981                 // properly; putting them before other transformations should keep
 982                 // exciting things like link expansions from showing up in surprising
 983                 // places.
 984                 $text = $this->doTableStuff( $text );
 985
 986                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 987
 988                 $text = $this->doDoubleUnderscore( $text );
 989                 $text = $this->doHeadings( $text );
 990                 if($this->mOptions->getUseDynamicDates()) {
 991                         $df = DateFormatter::getInstance();
 992                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 993                 }
 994                 $text = $this->doAllQuotes( $text );
 995                 $text = $this->replaceInternalLinks( $text );
 996                 $text = $this->replaceExternalLinks( $text );
 997
 998                 # replaceInternalLinks may sometimes leave behind
 999                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1000                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
1001
1002                 $text = $this->doMagicLinks( $text );
1003                 $text = $this->formatHeadings( $text, $isMain );
1004
1005                 wfProfileOut( __METHOD__ );
1006                 return $text;
1007         }
1008
1009         /**
1010          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1011          * magic external links.
1012          *
1013          * DML
1014          * @private
1015          */
1016         function doMagicLinks( $text ) {
1017                 wfProfileIn( __METHOD__ );
1018                 $prots = $this->mUrlProtocols;
1019                 $urlChar = self::EXT_LINK_URL_CLASS;
1020                 $text = preg_replace_callback(
1021                         '!(?:                           # Start cases
1022                                 (<a.*?</a>) |               # m[1]: Skip link text
1023                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
1024                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
1025                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
1026                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
1027                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1028                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1029                                     [0-9Xx]                 # check digit
1030                                     \b)
1031                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1032                 wfProfileOut( __METHOD__ );
1033                 return $text;
1034         }
1035
1036         function magicLinkCallback( $m ) {
1037                 if ( isset( $m[1] ) && strval( $m[1] ) !== '' ) {
1038                         # Skip anchor
1039                         return $m[0];
1040                 } elseif ( isset( $m[2] ) && strval( $m[2] ) !== '' ) {
1041                         # Skip HTML element
1042                         return $m[0];
1043                 } elseif ( isset( $m[3] ) && strval( $m[3] ) !== '' ) {
1044                         # Free external link
1045                         return $this->makeFreeExternalLink( $m[0] );
1046                 } elseif ( isset( $m[4] ) && strval( $m[4] ) !== '' ) {
1047                         # RFC or PMID
1048                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
1049                                 $keyword = 'RFC';
1050                                 $urlmsg = 'rfcurl';
1051                                 $id = $m[4];
1052                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
1053                                 $keyword = 'PMID';
1054                                 $urlmsg = 'pubmedurl';
1055                                 $id = $m[4];
1056                         } else {
1057                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1058                                         substr($m[0], 0, 20 ) . '"' );
1059                         }
1060                         $url = wfMsg( $urlmsg, $id);
1061                         $sk = $this->mOptions->getSkin();
1062                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1063                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1064                 } elseif ( isset( $m[5] ) && strval( $m[5] ) !== '' ) {
1065                         # ISBN
1066                         $isbn = $m[5];
1067                         $num = strtr( $isbn, array(
1068                                 '-' => '',
1069                                 ' ' => '',
1070                                 'x' => 'X',
1071                         ));
1072                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1073                         return'<a href="' .
1074                                 $titleObj->escapeLocalUrl() .
1075                                 "\" class=\"internal\">ISBN $isbn</a>";
1076                 } else {
1077                         return $m[0];
1078                 }
1079         }
1080
1081         /**
1082          * Make a free external link, given a user-supplied URL
1083          * @return HTML
1084          * @private
1085          */
1086         function makeFreeExternalLink( $url ) {
1087                 global $wgContLang;
1088                 wfProfileIn( __METHOD__ );
1089
1090                 $sk = $this->mOptions->getSkin();
1091                 $trail = '';
1092
1093                 # The characters '<' and '>' (which were escaped by
1094                 # removeHTMLtags()) should not be included in
1095                 # URLs, per RFC 2396.
1096                 $m2 = array();
1097                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1098                         $trail = substr($url, $m2[0][1]) . $trail;
1099                         $url = substr($url, 0, $m2[0][1]);
1100                 }
1101
1102                 # Move trailing punctuation to $trail
1103                 $sep = ',;\.:!?';
1104                 # If there is no left bracket, then consider right brackets fair game too
1105                 if ( strpos( $url, '(' ) === false ) {
1106                         $sep .= ')';
1107                 }
1108
1109                 $numSepChars = strspn( strrev( $url ), $sep );
1110                 if ( $numSepChars ) {
1111                         $trail = substr( $url, -$numSepChars ) . $trail;
1112                         $url = substr( $url, 0, -$numSepChars );
1113                 }
1114
1115                 $url = Sanitizer::cleanUrl( $url );
1116
1117                 # Is this an external image?
1118                 $text = $this->maybeMakeExternalImage( $url );
1119                 if ( $text === false ) {
1120                         # Not an image, make a link
1121                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1122                                 $this->getExternalLinkAttribs() );
1123                         # Register it in the output object...
1124                         # Replace unnecessary URL escape codes with their equivalent characters
1125                         $pasteurized = self::replaceUnusualEscapes( $url );
1126                         $this->mOutput->addExternalLink( $pasteurized );
1127                 }
1128                 wfProfileOut( __METHOD__ );
1129                 return $text . $trail;
1130         }
1131
1132
1133         /**
1134          * Parse headers and return html
1135          *
1136          * @private
1137          */
1138         function doHeadings( $text ) {
1139                 wfProfileIn( __METHOD__ );
1140                 for ( $i = 6; $i >= 1; --$i ) {
1141                         $h = str_repeat( '=', $i );
1142                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1143                           "<h$i>\\1</h$i>", $text );
1144                 }
1145                 wfProfileOut( __METHOD__ );
1146                 return $text;
1147         }
1148
1149         /**
1150          * Replace single quotes with HTML markup
1151          * @private
1152          * @return string the altered text
1153          */
1154         function doAllQuotes( $text ) {
1155                 wfProfileIn( __METHOD__ );
1156                 $outtext = '';
1157                 $lines = StringUtils::explode( "\n", $text );
1158                 foreach ( $lines as $line ) {
1159                         $outtext .= $this->doQuotes( $line ) . "\n";
1160                 }
1161                 $outtext = substr($outtext, 0,-1);
1162                 wfProfileOut( __METHOD__ );
1163                 return $outtext;
1164         }
1165
1166         /**
1167          * Helper function for doAllQuotes()
1168          */
1169         public function doQuotes( $text ) {
1170                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1171                 if ( count( $arr ) == 1 )
1172                         return $text;
1173                 else
1174                 {
1175                         # First, do some preliminary work. This may shift some apostrophes from
1176                         # being mark-up to being text. It also counts the number of occurrences
1177                         # of bold and italics mark-ups.
1178                         $i = 0;
1179                         $numbold = 0;
1180                         $numitalics = 0;
1181                         foreach ( $arr as $r )
1182                         {
1183                                 if ( ( $i % 2 ) == 1 )
1184                                 {
1185                                         # If there are ever four apostrophes, assume the first is supposed to
1186                                         # be text, and the remaining three constitute mark-up for bold text.
1187                                         if ( strlen( $arr[$i] ) == 4 )
1188                                         {
1189                                                 $arr[$i-1] .= "'";
1190                                                 $arr[$i] = "'''";
1191                                         }
1192                                         # If there are more than 5 apostrophes in a row, assume they're all
1193                                         # text except for the last 5.
1194                                         else if ( strlen( $arr[$i] ) > 5 )
1195                                         {
1196                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1197                                                 $arr[$i] = "'''''";
1198                                         }
1199                                         # Count the number of occurrences of bold and italics mark-ups.
1200                                         # We are not counting sequences of five apostrophes.
1201                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1202                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1203                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1204                                 }
1205                                 $i++;
1206                         }
1207
1208                         # If there is an odd number of both bold and italics, it is likely
1209                         # that one of the bold ones was meant to be an apostrophe followed
1210                         # by italics. Which one we cannot know for certain, but it is more
1211                         # likely to be one that has a single-letter word before it.
1212                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1213                         {
1214                                 $i = 0;
1215                                 $firstsingleletterword = -1;
1216                                 $firstmultiletterword = -1;
1217                                 $firstspace = -1;
1218                                 foreach ( $arr as $r )
1219                                 {
1220                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1221                                         {
1222                                                 $x1 = substr ($arr[$i-1], -1);
1223                                                 $x2 = substr ($arr[$i-1], -2, 1);
1224                                                 if ($x1 === ' ') {
1225                                                         if ($firstspace == -1) $firstspace = $i;
1226                                                 } else if ($x2 === ' ') {
1227                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1228                                                 } else {
1229                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1230                                                 }
1231                                         }
1232                                         $i++;
1233                                 }
1234
1235                                 # If there is a single-letter word, use it!
1236                                 if ($firstsingleletterword > -1)
1237                                 {
1238                                         $arr [ $firstsingleletterword ] = "''";
1239                                         $arr [ $firstsingleletterword-1 ] .= "'";
1240                                 }
1241                                 # If not, but there's a multi-letter word, use that one.
1242                                 else if ($firstmultiletterword > -1)
1243                                 {
1244                                         $arr [ $firstmultiletterword ] = "''";
1245                                         $arr [ $firstmultiletterword-1 ] .= "'";
1246                                 }
1247                                 # ... otherwise use the first one that has neither.
1248                                 # (notice that it is possible for all three to be -1 if, for example,
1249                                 # there is only one pentuple-apostrophe in the line)
1250                                 else if ($firstspace > -1)
1251                                 {
1252                                         $arr [ $firstspace ] = "''";
1253                                         $arr [ $firstspace-1 ] .= "'";
1254                                 }
1255                         }
1256
1257                         # Now let's actually convert our apostrophic mush to HTML!
1258                         $output = '';
1259                         $buffer = '';
1260                         $state = '';
1261                         $i = 0;
1262                         foreach ($arr as $r)
1263                         {
1264                                 if (($i % 2) == 0)
1265                                 {
1266                                         if ($state === 'both')
1267                                                 $buffer .= $r;
1268                                         else
1269                                                 $output .= $r;
1270                                 }
1271                                 else
1272                                 {
1273                                         if (strlen ($r) == 2)
1274                                         {
1275                                                 if ($state === 'i')
1276                                                 { $output .= '</i>'; $state = ''; }
1277                                                 else if ($state === 'bi')
1278                                                 { $output .= '</i>'; $state = 'b'; }
1279                                                 else if ($state === 'ib')
1280                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1281                                                 else if ($state === 'both')
1282                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1283                                                 else # $state can be 'b' or ''
1284                                                 { $output .= '<i>'; $state .= 'i'; }
1285                                         }
1286                                         else if (strlen ($r) == 3)
1287                                         {
1288                                                 if ($state === 'b')
1289                                                 { $output .= '</b>'; $state = ''; }
1290                                                 else if ($state === 'bi')
1291                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1292                                                 else if ($state === 'ib')
1293                                                 { $output .= '</b>'; $state = 'i'; }
1294                                                 else if ($state === 'both')
1295                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1296                                                 else # $state can be 'i' or ''
1297                                                 { $output .= '<b>'; $state .= 'b'; }
1298                                         }
1299                                         else if (strlen ($r) == 5)
1300                                         {
1301                                                 if ($state === 'b')
1302                                                 { $output .= '</b><i>'; $state = 'i'; }
1303                                                 else if ($state === 'i')
1304                                                 { $output .= '</i><b>'; $state = 'b'; }
1305                                                 else if ($state === 'bi')
1306                                                 { $output .= '</i></b>'; $state = ''; }
1307                                                 else if ($state === 'ib')
1308                                                 { $output .= '</b></i>'; $state = ''; }
1309                                                 else if ($state === 'both')
1310                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1311                                                 else # ($state == '')
1312                                                 { $buffer = ''; $state = 'both'; }
1313                                         }
1314                                 }
1315                                 $i++;
1316                         }
1317                         # Now close all remaining tags.  Notice that the order is important.
1318                         if ($state === 'b' || $state === 'ib')
1319                                 $output .= '</b>';
1320                         if ($state === 'i' || $state === 'bi' || $state === 'ib')
1321                                 $output .= '</i>';
1322                         if ($state === 'bi')
1323                                 $output .= '</b>';
1324                         # There might be lonely ''''', so make sure we have a buffer
1325                         if ($state === 'both' && $buffer)
1326                                 $output .= '<b><i>'.$buffer.'</i></b>';
1327                         return $output;
1328                 }
1329         }
1330
1331         /**
1332          * Replace external links (REL)
1333          *
1334          * Note: this is all very hackish and the order of execution matters a lot.
1335          * Make sure to run maintenance/parserTests.php if you change this code.
1336          *
1337          * @private
1338          */
1339         function replaceExternalLinks( $text ) {
1340                 global $wgContLang;
1341                 wfProfileIn( __METHOD__ );
1342
1343                 $sk = $this->mOptions->getSkin();
1344
1345                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1346                 $s = array_shift( $bits );
1347
1348                 $i = 0;
1349                 while ( $i<count( $bits ) ) {
1350                         $url = $bits[$i++];
1351                         $protocol = $bits[$i++];
1352                         $text = $bits[$i++];
1353                         $trail = $bits[$i++];
1354
1355                         # The characters '<' and '>' (which were escaped by
1356                         # removeHTMLtags()) should not be included in
1357                         # URLs, per RFC 2396.
1358                         $m2 = array();
1359                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1360                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1361                                 $url = substr($url, 0, $m2[0][1]);
1362                         }
1363
1364                         # If the link text is an image URL, replace it with an <img> tag
1365                         # This happened by accident in the original parser, but some people used it extensively
1366                         $img = $this->maybeMakeExternalImage( $text );
1367                         if ( $img !== false ) {
1368                                 $text = $img;
1369                         }
1370
1371                         $dtrail = '';
1372
1373                         # Set linktype for CSS - if URL==text, link is essentially free
1374                         $linktype = ($text === $url) ? 'free' : 'text';
1375
1376                         # No link text, e.g. [http://domain.tld/some.link]
1377                         if ( $text == '' ) {
1378                                 # Autonumber if allowed. See bug #5918
1379                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1380                                         $langObj = $this->getFunctionLang();
1381                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1382                                         $linktype = 'autonumber';
1383                                 } else {
1384                                         # Otherwise just use the URL
1385                                         $text = htmlspecialchars( $url );
1386                                         $linktype = 'free';
1387                                 }
1388                         } else {
1389                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1390                                 # Check for trail
1391                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1392                         }
1393
1394                         $text = $wgContLang->markNoConversion($text);
1395
1396                         $url = Sanitizer::cleanUrl( $url );
1397
1398                         if ( $this->mOptions->mExternalLinkTarget ) {
1399                                 $attribs = array( 'target' => $this->mOptions->mExternalLinkTarget );
1400                         } else {
1401                                 $attribs = array();
1402                         }
1403
1404                         # Use the encoded URL
1405                         # This means that users can paste URLs directly into the text
1406                         # Funny characters like &ouml; aren't valid in URLs anyway
1407                         # This was changed in August 2004
1408                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->getExternalLinkAttribs() )
1409                                 . $dtrail . $trail;
1410
1411                         # Register link in the output object.
1412                         # Replace unnecessary URL escape codes with the referenced character
1413                         # This prevents spammers from hiding links from the filters
1414                         $pasteurized = self::replaceUnusualEscapes( $url );
1415                         $this->mOutput->addExternalLink( $pasteurized );
1416                 }
1417
1418                 wfProfileOut( __METHOD__ );
1419                 return $s;
1420         }
1421
1422         function getExternalLinkAttribs() {
1423                 $attribs = array();
1424                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1425                 $ns = $this->mTitle->getNamespace();
1426                 if( $wgNoFollowLinks && !in_array($ns, $wgNoFollowNsExceptions) ) {
1427                         $attribs['rel'] = 'nofollow';
1428                 }
1429                 if ( $this->mOptions->getExternalLinkTarget() ) {
1430                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1431                 }
1432                 return $attribs;
1433         }
1434
1435
1436         /**
1437          * Replace unusual URL escape codes with their equivalent characters
1438          * @param string
1439          * @return string
1440          * @static
1441          * @todo  This can merge genuinely required bits in the path or query string,
1442          *        breaking legit URLs. A proper fix would treat the various parts of
1443          *        the URL differently; as a workaround, just use the output for
1444          *        statistical records, not for actual linking/output.
1445          */
1446         static function replaceUnusualEscapes( $url ) {
1447                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1448                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1449         }
1450
1451         /**
1452          * Callback function used in replaceUnusualEscapes().
1453          * Replaces unusual URL escape codes with their equivalent character
1454          * @static
1455          * @private
1456          */
1457         private static function replaceUnusualEscapesCallback( $matches ) {
1458                 $char = urldecode( $matches[0] );
1459                 $ord = ord( $char );
1460                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1461                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1462                         // No, shouldn't be escaped
1463                         return $char;
1464                 } else {
1465                         // Yes, leave it escaped
1466                         return $matches[0];
1467                 }
1468         }
1469
1470         /**
1471          * make an image if it's allowed, either through the global
1472          * option, through the exception, or through the on-wiki whitelist
1473          * @private
1474          */
1475         function maybeMakeExternalImage( $url ) {
1476                 $sk = $this->mOptions->getSkin();
1477                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1478                 $imagesexception = !empty($imagesfrom);
1479                 $text = false;
1480                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1481                 if( $imagesexception && is_array( $imagesfrom ) ) {
1482                         $imagematch = false;
1483                         foreach( $imagesfrom as $match ) {
1484                                 if( strpos( $url, $match ) === 0 ) {
1485                                         $imagematch = true;
1486                                         break;
1487                                 }
1488                         }
1489                 } elseif( $imagesexception ) {
1490                         $imagematch = (strpos( $url, $imagesfrom ) === 0);
1491                 } else {
1492                         $imagematch = false;
1493                 }
1494                 if ( $this->mOptions->getAllowExternalImages()
1495                      || ( $imagesexception && $imagematch ) ) {
1496                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1497                                 # Image found
1498                                 $text = $sk->makeExternalImage( $url );
1499                         }
1500                 }
1501                 if( !$text && $this->mOptions->getEnableImageWhitelist()
1502                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1503                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1504                         foreach( $whitelist as $entry ) {
1505                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1506                                 if( strpos( $entry, '#' ) === 0 || $entry === '' )
1507                                         continue;
1508                                 if( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1509                                         # Image matches a whitelist entry
1510                                         $text = $sk->makeExternalImage( $url );
1511                                         break;
1512                                 }
1513                         }
1514                 }
1515                 return $text;
1516         }
1517
1518         /**
1519          * Process [[ ]] wikilinks
1520          * @return processed text
1521          *
1522          * @private
1523          */
1524         function replaceInternalLinks( $s ) {
1525                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1526                 return $s;
1527         }
1528
1529         /**
1530          * Process [[ ]] wikilinks (RIL)
1531          * @return LinkHolderArray
1532          *
1533          * @private
1534          */
1535         function replaceInternalLinks2( &$s ) {
1536                 global $wgContLang;
1537
1538                 wfProfileIn( __METHOD__ );
1539
1540                 wfProfileIn( __METHOD__.'-setup' );
1541                 static $tc = FALSE, $e1, $e1_img;
1542                 # the % is needed to support urlencoded titles as well
1543                 if ( !$tc ) {
1544                         $tc = Title::legalChars() . '#%';
1545                         # Match a link having the form [[namespace:link|alternate]]trail
1546                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1547                         # Match cases where there is no "]]", which might still be images
1548                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1549                 }
1550
1551                 $sk = $this->mOptions->getSkin();
1552                 $holders = new LinkHolderArray( $this );
1553
1554                 #split the entire text string on occurences of [[
1555                 $a = StringUtils::explode( '[[', ' ' . $s );
1556                 #get the first element (all text up to first [[), and remove the space we added
1557                 $s = $a->current();
1558                 $a->next();
1559                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1560                 $s = substr( $s, 1 );
1561
1562                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1563                 $e2 = null;
1564                 if ( $useLinkPrefixExtension ) {
1565                         # Match the end of a line for a word that's not followed by whitespace,
1566                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1567                         $e2 = wfMsgForContent( 'linkprefix' );
1568                 }
1569
1570                 if( is_null( $this->mTitle ) ) {
1571                         wfProfileOut( __METHOD__.'-setup' );
1572                         wfProfileOut( __METHOD__ );
1573                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1574                 }
1575                 $nottalk = !$this->mTitle->isTalkPage();
1576
1577                 if ( $useLinkPrefixExtension ) {
1578                         $m = array();
1579                         if ( preg_match( $e2, $s, $m ) ) {
1580                                 $first_prefix = $m[2];
1581                         } else {
1582                                 $first_prefix = false;
1583                         }
1584                 } else {
1585                         $prefix = '';
1586                 }
1587
1588                 if($wgContLang->hasVariants()) {
1589                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1590                 } else {
1591                         $selflink = array($this->mTitle->getPrefixedText());
1592                 }
1593                 $useSubpages = $this->areSubpagesAllowed();
1594                 wfProfileOut( __METHOD__.'-setup' );
1595
1596                 # Loop for each link
1597                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1598                         # Check for excessive memory usage
1599                         if ( $holders->isBig() ) {
1600                                 # Too big
1601                                 # Do the existence check, replace the link holders and clear the array
1602                                 $holders->replace( $s );
1603                                 $holders->clear();
1604                         }
1605
1606                         if ( $useLinkPrefixExtension ) {
1607                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1608                                 if ( preg_match( $e2, $s, $m ) ) {
1609                                         $prefix = $m[2];
1610                                         $s = $m[1];
1611                                 } else {
1612                                         $prefix='';
1613                                 }
1614                                 # first link
1615                                 if($first_prefix) {
1616                                         $prefix = $first_prefix;
1617                                         $first_prefix = false;
1618                                 }
1619                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1620                         }
1621
1622                         $might_be_img = false;
1623
1624                         wfProfileIn( __METHOD__."-e1" );
1625                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1626                                 $text = $m[2];
1627                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1628                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1629                                 # the real problem is with the $e1 regex
1630                                 # See bug 1300.
1631                                 #
1632                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1633                                 # and no image is in sight. See bug 2095.
1634                                 #
1635                                 if( $text !== '' &&
1636                                         substr( $m[3], 0, 1 ) === ']' &&
1637                                         strpos($text, '[') !== false
1638                                 )
1639                                 {
1640                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1641                                         $m[3] = substr( $m[3], 1 );
1642                                 }
1643                                 # fix up urlencoded title texts
1644                                 if( strpos( $m[1], '%' ) !== false ) {
1645                                         # Should anchors '#' also be rejected?
1646                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1647                                 }
1648                                 $trail = $m[3];
1649                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1650                                 $might_be_img = true;
1651                                 $text = $m[2];
1652                                 if ( strpos( $m[1], '%' ) !== false ) {
1653                                         $m[1] = urldecode($m[1]);
1654                                 }
1655                                 $trail = "";
1656                         } else { # Invalid form; output directly
1657                                 $s .= $prefix . '[[' . $line ;
1658                                 wfProfileOut( __METHOD__."-e1" );
1659                                 continue;
1660                         }
1661                         wfProfileOut( __METHOD__."-e1" );
1662                         wfProfileIn( __METHOD__."-misc" );
1663
1664                         # Don't allow internal links to pages containing
1665                         # PROTO: where PROTO is a valid URL protocol; these
1666                         # should be external links.
1667                         if (preg_match('/^\b(?:' . wfUrlProtocols() . ')/', $m[1])) {
1668                                 $s .= $prefix . '[[' . $line ;
1669                                 wfProfileOut( __METHOD__."-misc" );
1670                                 continue;
1671                         }
1672
1673                         # Make subpage if necessary
1674                         if( $useSubpages ) {
1675                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1676                         } else {
1677                                 $link = $m[1];
1678                         }
1679
1680                         $noforce = (substr($m[1], 0, 1) !== ':');
1681                         if (!$noforce) {
1682                                 # Strip off leading ':'
1683                                 $link = substr($link, 1);
1684                         }
1685
1686                         wfProfileOut( __METHOD__."-misc" );
1687                         wfProfileIn( __METHOD__."-title" );
1688                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1689                         if( !$nt ) {
1690                                 $s .= $prefix . '[[' . $line;
1691                                 wfProfileOut( __METHOD__."-title" );
1692                                 continue;
1693                         }
1694
1695                         $ns = $nt->getNamespace();
1696                         $iw = $nt->getInterWiki();
1697                         wfProfileOut( __METHOD__."-title" );
1698
1699                         if ($might_be_img) { # if this is actually an invalid link
1700                                 wfProfileIn( __METHOD__."-might_be_img" );
1701                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1702                                         $found = false;
1703                                         while ( true ) {
1704                                                 #look at the next 'line' to see if we can close it there
1705                                                 $a->next();
1706                                                 $next_line = $a->current();
1707                                                 if ( $next_line === false || $next_line === null ) {
1708                                                         break;
1709                                                 }
1710                                                 $m = explode( ']]', $next_line, 3 );
1711                                                 if ( count( $m ) == 3 ) {
1712                                                         # the first ]] closes the inner link, the second the image
1713                                                         $found = true;
1714                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1715                                                         $trail = $m[2];
1716                                                         break;
1717                                                 } elseif ( count( $m ) == 2 ) {
1718                                                         #if there's exactly one ]] that's fine, we'll keep looking
1719                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1720                                                 } else {
1721                                                         #if $next_line is invalid too, we need look no further
1722                                                         $text .= '[[' . $next_line;
1723                                                         break;
1724                                                 }
1725                                         }
1726                                         if ( !$found ) {
1727                                                 # we couldn't find the end of this imageLink, so output it raw
1728                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1729                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1730                                                 $s .= "{$prefix}[[$link|$text";
1731                                                 # note: no $trail, because without an end, there *is* no trail
1732                                                 wfProfileOut( __METHOD__."-might_be_img" );
1733                                                 continue;
1734                                         }
1735                                 } else { #it's not an image, so output it raw
1736                                         $s .= "{$prefix}[[$link|$text";
1737                                         # note: no $trail, because without an end, there *is* no trail
1738                                         wfProfileOut( __METHOD__."-might_be_img" );
1739                                         continue;
1740                                 }
1741                                 wfProfileOut( __METHOD__."-might_be_img" );
1742                         }
1743
1744                         $wasblank = ( '' == $text );
1745                         if( $wasblank ) $text = $link;
1746
1747                         # Link not escaped by : , create the various objects
1748                         if( $noforce ) {
1749
1750                                 # Interwikis
1751                                 wfProfileIn( __METHOD__."-interwiki" );
1752                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1753                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1754                                         $s = rtrim($s . $prefix);
1755                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1756                                         wfProfileOut( __METHOD__."-interwiki" );
1757                                         continue;
1758                                 }
1759                                 wfProfileOut( __METHOD__."-interwiki" );
1760
1761                                 if ( $ns == NS_IMAGE ) {
1762                                         wfProfileIn( __METHOD__."-image" );
1763                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1764                                                 # recursively parse links inside the image caption
1765                                                 # actually, this will parse them in any other parameters, too,
1766                                                 # but it might be hard to fix that, and it doesn't matter ATM
1767                                                 $text = $this->replaceExternalLinks($text);
1768                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1769
1770                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1771                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1772                                         }
1773                                         $this->mOutput->addImage( $nt->getDBkey() );
1774                                         wfProfileOut( __METHOD__."-image" );
1775                                         continue;
1776
1777                                 }
1778
1779                                 if ( $ns == NS_CATEGORY ) {
1780                                         wfProfileIn( __METHOD__."-category" );
1781                                         $s = rtrim($s . "\n"); # bug 87
1782
1783                                         if ( $wasblank ) {
1784                                                 $sortkey = $this->getDefaultSort();
1785                                         } else {
1786                                                 $sortkey = $text;
1787                                         }
1788                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1789                                         $sortkey = str_replace( "\n", '', $sortkey );
1790                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1791                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1792
1793                                         /**
1794                                          * Strip the whitespace Category links produce, see bug 87
1795                                          * @todo We might want to use trim($tmp, "\n") here.
1796                                          */
1797                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1798
1799                                         wfProfileOut( __METHOD__."-category" );
1800                                         continue;
1801                                 }
1802                         }
1803
1804                         # Self-link checking
1805                         if( $nt->getFragment() === '' && $nt->getNamespace() != NS_SPECIAL ) {
1806                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1807                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1808                                         continue;
1809                                 }
1810                         }
1811
1812                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1813                         if( $ns == NS_MEDIA ) {
1814                                 # Give extensions a chance to select the file revision for us
1815                                 $skip = $time = false;
1816                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1817                                 if ( $skip ) {
1818                                         $link = $sk->link( $nt );
1819                                 } else {
1820                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1821                                 }
1822                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1823                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1824                                 $this->mOutput->addImage( $nt->getDBkey() );
1825                                 continue;
1826                         } elseif( $ns == NS_SPECIAL ) {
1827                                 if( SpecialPage::exists( $nt->getDBkey() ) ) {
1828                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1829                                 } else {
1830                                         $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1831                                 }
1832                                 continue;
1833                         } elseif( $ns == NS_IMAGE ) {
1834                                 $img = wfFindFile( $nt );
1835                                 if( $img ) {
1836                                         // Force a blue link if the file exists; may be a remote
1837                                         // upload on the shared repository, and we want to see its
1838                                         // auto-generated page.
1839                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1840                                         $this->mOutput->addLink( $nt );
1841                                         continue;
1842                                 }
1843                         }
1844                         $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1845                 }
1846                 wfProfileOut( __METHOD__ );
1847                 return $holders;
1848         }
1849
1850         /**
1851          * Make a link placeholder. The text returned can be later resolved to a real link with
1852          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1853          * parsing of interwiki links, and secondly to allow all existence checks and
1854          * article length checks (for stub links) to be bundled into a single query.
1855          *
1856          * @deprecated
1857          */
1858         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1859                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1860         }
1861
1862         /**
1863          * Render a forced-blue link inline; protect against double expansion of
1864          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1865          * Since this little disaster has to split off the trail text to avoid
1866          * breaking URLs in the following text without breaking trails on the
1867          * wiki links, it's been made into a horrible function.
1868          *
1869          * @param Title $nt
1870          * @param string $text
1871          * @param string $query
1872          * @param string $trail
1873          * @param string $prefix
1874          * @return string HTML-wikitext mix oh yuck
1875          */
1876         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1877                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1878                 $sk = $this->mOptions->getSkin();
1879                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1880                 return $this->armorLinks( $link ) . $trail;
1881         }
1882
1883         /**
1884          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1885          * going to go through further parsing steps before inline URL expansion.
1886          *
1887          * Not needed quite as much as it used to be since free links are a bit
1888          * more sensible these days. But bracketed links are still an issue.
1889          *
1890          * @param string more-or-less HTML
1891          * @return string less-or-more HTML with NOPARSE bits
1892          */
1893         function armorLinks( $text ) {
1894                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1895                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1896         }
1897
1898         /**
1899          * Return true if subpage links should be expanded on this page.
1900          * @return bool
1901          */
1902         function areSubpagesAllowed() {
1903                 # Some namespaces don't allow subpages
1904                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1905         }
1906
1907         /**
1908          * Handle link to subpage if necessary
1909          * @param string $target the source of the link
1910          * @param string &$text the link text, modified as necessary
1911          * @return string the full name of the link
1912          * @private
1913          */
1914         function maybeDoSubpageLink($target, &$text) {
1915                 # Valid link forms:
1916                 # Foobar -- normal
1917                 # :Foobar -- override special treatment of prefix (images, language links)
1918                 # /Foobar -- convert to CurrentPage/Foobar
1919                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1920                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1921                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1922
1923                 wfProfileIn( __METHOD__ );
1924                 $ret = $target; # default return value is no change
1925
1926                 # Some namespaces don't allow subpages,
1927                 # so only perform processing if subpages are allowed
1928                 if( $this->areSubpagesAllowed() ) {
1929                         $hash = strpos( $target, '#' );
1930                         if( $hash !== false ) {
1931                                 $suffix = substr( $target, $hash );
1932                                 $target = substr( $target, 0, $hash );
1933                         } else {
1934                                 $suffix = '';
1935                         }
1936                         # bug 7425
1937                         $target = trim( $target );
1938                         # Look at the first character
1939                         if( $target != '' && $target{0} === '/' ) {
1940                                 # / at end means we don't want the slash to be shown
1941                                 $m = array();
1942                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1943                                 if( $trailingSlashes ) {
1944                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1945                                 } else {
1946                                         $noslash = substr( $target, 1 );
1947                                 }
1948
1949                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash) . $suffix;
1950                                 if( '' === $text ) {
1951                                         $text = $target . $suffix;
1952                                 } # this might be changed for ugliness reasons
1953                         } else {
1954                                 # check for .. subpage backlinks
1955                                 $dotdotcount = 0;
1956                                 $nodotdot = $target;
1957                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1958                                         ++$dotdotcount;
1959                                         $nodotdot = substr( $nodotdot, 3 );
1960                                 }
1961                                 if($dotdotcount > 0) {
1962                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1963                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1964                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1965                                                 # / at the end means don't show full path
1966                                                 if( substr( $nodotdot, -1, 1 ) === '/' ) {
1967                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1968                                                         if( '' === $text ) {
1969                                                                 $text = $nodotdot . $suffix;
1970                                                         }
1971                                                 }
1972                                                 $nodotdot = trim( $nodotdot );
1973                                                 if( $nodotdot != '' ) {
1974                                                         $ret .= '/' . $nodotdot;
1975                                                 }
1976                                                 $ret .= $suffix;
1977                                         }
1978                                 }
1979                         }
1980                 }
1981
1982                 wfProfileOut( __METHOD__ );
1983                 return $ret;
1984         }
1985
1986         /**#@+
1987          * Used by doBlockLevels()
1988          * @private
1989          */
1990         /* private */ function closeParagraph() {
1991                 $result = '';
1992                 if ( '' != $this->mLastSection ) {
1993                         $result = '</' . $this->mLastSection  . ">\n";
1994                 }
1995                 $this->mInPre = false;
1996                 $this->mLastSection = '';
1997                 return $result;
1998         }
1999         # getCommon() returns the length of the longest common substring
2000         # of both arguments, starting at the beginning of both.
2001         #
2002         /* private */ function getCommon( $st1, $st2 ) {
2003                 $fl = strlen( $st1 );
2004                 $shorter = strlen( $st2 );
2005                 if ( $fl < $shorter ) { $shorter = $fl; }
2006
2007                 for ( $i = 0; $i < $shorter; ++$i ) {
2008                         if ( $st1{$i} != $st2{$i} ) { break; }
2009                 }
2010                 return $i;
2011         }
2012         # These next three functions open, continue, and close the list
2013         # element appropriate to the prefix character passed into them.
2014         #
2015         /* private */ function openList( $char ) {
2016                 $result = $this->closeParagraph();
2017
2018                 if ( '*' === $char ) { $result .= '<ul><li>'; }
2019                 else if ( '#' === $char ) { $result .= '<ol><li>'; }
2020                 else if ( ':' === $char ) { $result .= '<dl><dd>'; }
2021                 else if ( ';' === $char ) {
2022                         $result .= '<dl><dt>';
2023                         $this->mDTopen = true;
2024                 }
2025                 else { $result = '<!-- ERR 1 -->'; }
2026
2027                 return $result;
2028         }
2029
2030         /* private */ function nextItem( $char ) {
2031                 if ( '*' === $char || '#' === $char ) { return '</li><li>'; }
2032                 else if ( ':' === $char || ';' === $char ) {
2033                         $close = '</dd>';
2034                         if ( $this->mDTopen ) { $close = '</dt>'; }
2035                         if ( ';' === $char ) {
2036                                 $this->mDTopen = true;
2037                                 return $close . '<dt>';
2038                         } else {
2039                                 $this->mDTopen = false;
2040                                 return $close . '<dd>';
2041                         }
2042                 }
2043                 return '<!-- ERR 2 -->';
2044         }
2045
2046         /* private */ function closeList( $char ) {
2047                 if ( '*' === $char ) { $text = '</li></ul>'; }
2048                 else if ( '#' === $char ) { $text = '</li></ol>'; }
2049                 else if ( ':' === $char ) {
2050                         if ( $this->mDTopen ) {
2051                                 $this->mDTopen = false;
2052                                 $text = '</dt></dl>';
2053                         } else {
2054                                 $text = '</dd></dl>';
2055                         }
2056                 }
2057                 else {  return '<!-- ERR 3 -->'; }
2058                 return $text."\n";
2059         }
2060         /**#@-*/
2061
2062         /**
2063          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
2064          *
2065          * @private
2066          * @return string the lists rendered as HTML
2067          */
2068         function doBlockLevels( $text, $linestart ) {
2069                 wfProfileIn( __METHOD__ );
2070
2071                 # Parsing through the text line by line.  The main thing
2072                 # happening here is handling of block-level elements p, pre,
2073                 # and making lists from lines starting with * # : etc.
2074                 #
2075                 $textLines = StringUtils::explode( "\n", $text );
2076
2077                 $lastPrefix = $output = '';
2078                 $this->mDTopen = $inBlockElem = false;
2079                 $prefixLength = 0;
2080                 $paragraphStack = false;
2081
2082                 foreach ( $textLines as $oLine ) {
2083                         # Fix up $linestart
2084                         if ( !$linestart ) {
2085                                 $output .= $oLine;
2086                                 $linestart = true;
2087                                 continue;
2088                         }
2089
2090                         $lastPrefixLength = strlen( $lastPrefix );
2091                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2092                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2093                         if ( !$this->mInPre ) {
2094                                 # Multiple prefixes may abut each other for nested lists.
2095                                 $prefixLength = strspn( $oLine, '*#:;' );
2096                                 $prefix = substr( $oLine, 0, $prefixLength );
2097
2098                                 # eh?
2099                                 $prefix2 = str_replace( ';', ':', $prefix );
2100                                 $t = substr( $oLine, $prefixLength );
2101                                 $this->mInPre = (bool)$preOpenMatch;
2102                         } else {
2103                                 # Don't interpret any other prefixes in preformatted text
2104                                 $prefixLength = 0;
2105                                 $prefix = $prefix2 = '';
2106                                 $t = $oLine;
2107                         }
2108
2109                         # List generation
2110                         if( $prefixLength && $lastPrefix === $prefix2 ) {
2111                                 # Same as the last item, so no need to deal with nesting or opening stuff
2112                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2113                                 $paragraphStack = false;
2114
2115                                 if ( substr( $prefix, -1 ) === ';') {
2116                                         # The one nasty exception: definition lists work like this:
2117                                         # ; title : definition text
2118                                         # So we check for : in the remainder text to split up the
2119                                         # title and definition, without b0rking links.
2120                                         $term = $t2 = '';
2121                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2122                                                 $t = $t2;
2123                                                 $output .= $term . $this->nextItem( ':' );
2124                                         }
2125                                 }
2126                         } elseif( $prefixLength || $lastPrefixLength ) {
2127                                 # Either open or close a level...
2128                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2129                                 $paragraphStack = false;
2130
2131                                 while( $commonPrefixLength < $lastPrefixLength ) {
2132                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2133                                         --$lastPrefixLength;
2134                                 }
2135                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2136                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2137                                 }
2138                                 while ( $prefixLength > $commonPrefixLength ) {
2139                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2140                                         $output .= $this->openList( $char );
2141
2142                                         if ( ';' === $char ) {
2143                                                 # FIXME: This is dupe of code above
2144                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2145                                                         $t = $t2;
2146                                                         $output .= $term . $this->nextItem( ':' );
2147                                                 }
2148                                         }
2149                                         ++$commonPrefixLength;
2150                                 }
2151                                 $lastPrefix = $prefix2;
2152                         }
2153                         if( 0 == $prefixLength ) {
2154                                 wfProfileIn( __METHOD__."-paragraph" );
2155                                 # No prefix (not in list)--go to paragraph mode
2156                                 // XXX: use a stack for nestable elements like span, table and div
2157                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2158                                 $closematch = preg_match(
2159                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2160                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2161                                 if ( $openmatch or $closematch ) {
2162                                         $paragraphStack = false;
2163                                         # TODO bug 5718: paragraph closed
2164                                         $output .= $this->closeParagraph();
2165                                         if ( $preOpenMatch and !$preCloseMatch ) {
2166                                                 $this->mInPre = true;
2167                                         }
2168                                         if ( $closematch ) {
2169                                                 $inBlockElem = false;
2170                                         } else {
2171                                                 $inBlockElem = true;
2172                                         }
2173                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2174                                         if ( ' ' == $t{0} and ( $this->mLastSection === 'pre' or trim($t) != '' ) ) {
2175                                                 // pre
2176                                                 if ($this->mLastSection !== 'pre') {
2177                                                         $paragraphStack = false;
2178                                                         $output .= $this->closeParagraph().'<pre>';
2179                                                         $this->mLastSection = 'pre';
2180                                                 }
2181                                                 $t = substr( $t, 1 );
2182                                         } else {
2183                                                 // paragraph
2184                                                 if ( '' == trim($t) ) {
2185                                                         if ( $paragraphStack ) {
2186                                                                 $output .= $paragraphStack.'<br />';
2187                                                                 $paragraphStack = false;
2188                                                                 $this->mLastSection = 'p';
2189                                                         } else {
2190                                                                 if ($this->mLastSection !== 'p' ) {
2191                                                                         $output .= $this->closeParagraph();
2192                                                                         $this->mLastSection = '';
2193                                                                         $paragraphStack = '<p>';
2194                                                                 } else {
2195                                                                         $paragraphStack = '</p><p>';
2196                                                                 }
2197                                                         }
2198                                                 } else {
2199                                                         if ( $paragraphStack ) {
2200                                                                 $output .= $paragraphStack;
2201                                                                 $paragraphStack = false;
2202                                                                 $this->mLastSection = 'p';
2203                                                         } else if ($this->mLastSection !== 'p') {
2204                                                                 $output .= $this->closeParagraph().'<p>';
2205                                                                 $this->mLastSection = 'p';
2206                                                         }
2207                                                 }
2208                                         }
2209                                 }
2210                                 wfProfileOut( __METHOD__."-paragraph" );
2211                         }
2212                         // somewhere above we forget to get out of pre block (bug 785)
2213                         if($preCloseMatch && $this->mInPre) {
2214                                 $this->mInPre = false;
2215                         }
2216                         if ($paragraphStack === false) {
2217                                 $output .= $t."\n";
2218                         }
2219                 }
2220                 while ( $prefixLength ) {
2221                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2222                         --$prefixLength;
2223                 }
2224                 if ( '' != $this->mLastSection ) {
2225                         $output .= '</' . $this->mLastSection . '>';
2226                         $this->mLastSection = '';
2227                 }
2228
2229                 wfProfileOut( __METHOD__ );
2230                 return $output;
2231         }
2232
2233         /**
2234          * Split up a string on ':', ignoring any occurences inside tags
2235          * to prevent illegal overlapping.
2236          * @param string $str the string to split
2237          * @param string &$before set to everything before the ':'
2238          * @param string &$after set to everything after the ':'
2239          * return string the position of the ':', or false if none found
2240          */
2241         function findColonNoLinks($str, &$before, &$after) {
2242                 wfProfileIn( __METHOD__ );
2243
2244                 $pos = strpos( $str, ':' );
2245                 if( $pos === false ) {
2246                         // Nothing to find!
2247                         wfProfileOut( __METHOD__ );
2248                         return false;
2249                 }
2250
2251                 $lt = strpos( $str, '<' );
2252                 if( $lt === false || $lt > $pos ) {
2253                         // Easy; no tag nesting to worry about
2254                         $before = substr( $str, 0, $pos );
2255                         $after = substr( $str, $pos+1 );
2256                         wfProfileOut( __METHOD__ );
2257                         return $pos;
2258                 }
2259
2260                 // Ugly state machine to walk through avoiding tags.
2261                 $state = self::COLON_STATE_TEXT;
2262                 $stack = 0;
2263                 $len = strlen( $str );
2264                 for( $i = 0; $i < $len; $i++ ) {
2265                         $c = $str{$i};
2266
2267                         switch( $state ) {
2268                         // (Using the number is a performance hack for common cases)
2269                         case 0: // self::COLON_STATE_TEXT:
2270                                 switch( $c ) {
2271                                 case "<":
2272                                         // Could be either a <start> tag or an </end> tag
2273                                         $state = self::COLON_STATE_TAGSTART;
2274                                         break;
2275                                 case ":":
2276                                         if( $stack == 0 ) {
2277                                                 // We found it!
2278                                                 $before = substr( $str, 0, $i );
2279                                                 $after = substr( $str, $i + 1 );
2280                                                 wfProfileOut( __METHOD__ );
2281                                                 return $i;
2282                                         }
2283                                         // Embedded in a tag; don't break it.
2284                                         break;
2285                                 default:
2286                                         // Skip ahead looking for something interesting
2287                                         $colon = strpos( $str, ':', $i );
2288                                         if( $colon === false ) {
2289                                                 // Nothing else interesting
2290                                                 wfProfileOut( __METHOD__ );
2291                                                 return false;
2292                                         }
2293                                         $lt = strpos( $str, '<', $i );
2294                                         if( $stack === 0 ) {
2295                                                 if( $lt === false || $colon < $lt ) {
2296                                                         // We found it!
2297                                                         $before = substr( $str, 0, $colon );
2298                                                         $after = substr( $str, $colon + 1 );
2299                                                         wfProfileOut( __METHOD__ );
2300                                                         return $i;
2301                                                 }
2302                                         }
2303                                         if( $lt === false ) {
2304                                                 // Nothing else interesting to find; abort!
2305                                                 // We're nested, but there's no close tags left. Abort!
2306                                                 break 2;
2307                                         }
2308                                         // Skip ahead to next tag start
2309                                         $i = $lt;
2310                                         $state = self::COLON_STATE_TAGSTART;
2311                                 }
2312                                 break;
2313                         case 1: // self::COLON_STATE_TAG:
2314                                 // In a <tag>
2315                                 switch( $c ) {
2316                                 case ">":
2317                                         $stack++;
2318                                         $state = self::COLON_STATE_TEXT;
2319                                         break;
2320                                 case "/":
2321                                         // Slash may be followed by >?
2322                                         $state = self::COLON_STATE_TAGSLASH;
2323                                         break;
2324                                 default:
2325                                         // ignore
2326                                 }
2327                                 break;
2328                         case 2: // self::COLON_STATE_TAGSTART:
2329                                 switch( $c ) {
2330                                 case "/":
2331                                         $state = self::COLON_STATE_CLOSETAG;
2332                                         break;
2333                                 case "!":
2334                                         $state = self::COLON_STATE_COMMENT;
2335                                         break;
2336                                 case ">":
2337                                         // Illegal early close? This shouldn't happen D:
2338                                         $state = self::COLON_STATE_TEXT;
2339                                         break;
2340                                 default:
2341                                         $state = self::COLON_STATE_TAG;
2342                                 }
2343                                 break;
2344                         case 3: // self::COLON_STATE_CLOSETAG:
2345                                 // In a </tag>
2346                                 if( $c === ">" ) {
2347                                         $stack--;
2348                                         if( $stack < 0 ) {
2349                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2350                                                 wfProfileOut( __METHOD__ );
2351                                                 return false;
2352                                         }
2353                                         $state = self::COLON_STATE_TEXT;
2354                                 }
2355                                 break;
2356                         case self::COLON_STATE_TAGSLASH:
2357                                 if( $c === ">" ) {
2358                                         // Yes, a self-closed tag <blah/>
2359                                         $state = self::COLON_STATE_TEXT;
2360                                 } else {
2361                                         // Probably we're jumping the gun, and this is an attribute
2362                                         $state = self::COLON_STATE_TAG;
2363                                 }
2364                                 break;
2365                         case 5: // self::COLON_STATE_COMMENT:
2366                                 if( $c === "-" ) {
2367                                         $state = self::COLON_STATE_COMMENTDASH;
2368                                 }
2369                                 break;
2370                         case self::COLON_STATE_COMMENTDASH:
2371                                 if( $c === "-" ) {
2372                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2373                                 } else {
2374                                         $state = self::COLON_STATE_COMMENT;
2375                                 }
2376                                 break;
2377                         case self::COLON_STATE_COMMENTDASHDASH:
2378                                 if( $c === ">" ) {
2379                                         $state = self::COLON_STATE_TEXT;
2380                                 } else {
2381                                         $state = self::COLON_STATE_COMMENT;
2382                                 }
2383                                 break;
2384                         default:
2385                                 throw new MWException( "State machine error in " . __METHOD__ );
2386                         }
2387                 }
2388                 if( $stack > 0 ) {
2389                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2390                         return false;
2391                 }
2392                 wfProfileOut( __METHOD__ );
2393                 return false;
2394         }
2395
2396         /**
2397          * Return value of a magic variable (like PAGENAME)
2398          *
2399          * @private
2400          */
2401         function getVariableValue( $index ) {
2402                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2403
2404                 /**
2405                  * Some of these require message or data lookups and can be
2406                  * expensive to check many times.
2407                  */
2408                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2409                         if ( isset( $this->mVarCache[$index] ) ) {
2410                                 return $this->mVarCache[$index];
2411                         }
2412                 }
2413
2414                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2415                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2416
2417                 # Use the time zone
2418                 global $wgLocaltimezone;
2419                 if ( isset( $wgLocaltimezone ) ) {
2420                         $oldtz = getenv( 'TZ' );
2421                         putenv( 'TZ='.$wgLocaltimezone );
2422                 }
2423
2424                 wfSuppressWarnings(); // E_STRICT system time bitching
2425                 $localTimestamp = date( 'YmdHis', $ts );
2426                 $localMonth = date( 'm', $ts );
2427                 $localMonthName = date( 'n', $ts );
2428                 $localDay = date( 'j', $ts );
2429                 $localDay2 = date( 'd', $ts );
2430                 $localDayOfWeek = date( 'w', $ts );
2431                 $localWeek = date( 'W', $ts );
2432                 $localYear = date( 'Y', $ts );
2433                 $localHour = date( 'H', $ts );
2434                 if ( isset( $wgLocaltimezone ) ) {
2435                         putenv( 'TZ='.$oldtz );
2436                 }
2437                 wfRestoreWarnings();
2438
2439                 switch ( $index ) {
2440                         case 'currentmonth':
2441                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2442                         case 'currentmonthname':
2443                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2444                         case 'currentmonthnamegen':
2445                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2446                         case 'currentmonthabbrev':
2447                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2448                         case 'currentday':
2449                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2450                         case 'currentday2':
2451                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2452                         case 'localmonth':
2453                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localMonth );
2454                         case 'localmonthname':
2455                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( $localMonthName );
2456                         case 'localmonthnamegen':
2457                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2458                         case 'localmonthabbrev':
2459                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2460                         case 'localday':
2461                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay );
2462                         case 'localday2':
2463                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay2 );
2464                         case 'pagename':
2465                                 return wfEscapeWikiText( $this->mTitle->getText() );
2466                         case 'pagenamee':
2467                                 return $this->mTitle->getPartialURL();
2468                         case 'fullpagename':
2469                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2470                         case 'fullpagenamee':
2471                                 return $this->mTitle->getPrefixedURL();
2472                         case 'subpagename':
2473                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2474                         case 'subpagenamee':
2475                                 return $this->mTitle->getSubpageUrlForm();
2476                         case 'basepagename':
2477                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2478                         case 'basepagenamee':
2479                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2480                         case 'talkpagename':
2481                                 if( $this->mTitle->canTalk() ) {
2482                                         $talkPage = $this->mTitle->getTalkPage();
2483                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2484                                 } else {
2485                                         return '';
2486                                 }
2487                         case 'talkpagenamee':
2488                                 if( $this->mTitle->canTalk() ) {
2489                                         $talkPage = $this->mTitle->getTalkPage();
2490                                         return $talkPage->getPrefixedUrl();
2491                                 } else {
2492                                         return '';
2493                                 }
2494                         case 'subjectpagename':
2495                                 $subjPage = $this->mTitle->getSubjectPage();
2496                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2497                         case 'subjectpagenamee':
2498                                 $subjPage = $this->mTitle->getSubjectPage();
2499                                 return $subjPage->getPrefixedUrl();
2500                         case 'revisionid':
2501                                 // Let the edit saving system know we should parse the page
2502                                 // *after* a revision ID has been assigned.
2503                                 $this->mOutput->setFlag( 'vary-revision' );
2504                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2505                                 return $this->mRevisionId;
2506                         case 'revisionday':
2507                                 // Let the edit saving system know we should parse the page
2508                                 // *after* a revision ID has been assigned. This is for null edits.
2509                                 $this->mOutput->setFlag( 'vary-revision' );
2510                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2511                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2512                         case 'revisionday2':
2513                                 // Let the edit saving system know we should parse the page
2514                                 // *after* a revision ID has been assigned. This is for null edits.
2515                                 $this->mOutput->setFlag( 'vary-revision' );
2516                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2517                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2518                         case 'revisionmonth':
2519                                 // Let the edit saving system know we should parse the page
2520                                 // *after* a revision ID has been assigned. This is for null edits.
2521                                 $this->mOutput->setFlag( 'vary-revision' );
2522                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2523                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2524                         case 'revisionyear':
2525                                 // Let the edit saving system know we should parse the page
2526                                 // *after* a revision ID has been assigned. This is for null edits.
2527                                 $this->mOutput->setFlag( 'vary-revision' );
2528                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2529                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2530                         case 'revisiontimestamp':
2531                                 // Let the edit saving system know we should parse the page
2532                                 // *after* a revision ID has been assigned. This is for null edits.
2533                                 $this->mOutput->setFlag( 'vary-revision' );
2534                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2535                                 return $this->getRevisionTimestamp();
2536                         case 'namespace':
2537                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2538                         case 'namespacee':
2539                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2540                         case 'talkspace':
2541                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2542                         case 'talkspacee':
2543                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2544                         case 'subjectspace':
2545                                 return $this->mTitle->getSubjectNsText();
2546                         case 'subjectspacee':
2547                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2548                         case 'currentdayname':
2549                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2550                         case 'currentyear':
2551                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2552                         case 'currenttime':
2553                                 return $this->mVarCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2554                         case 'currenthour':
2555                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2556                         case 'currentweek':
2557                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2558                                 // int to remove the padding
2559                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2560                         case 'currentdow':
2561                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2562                         case 'localdayname':
2563                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2564                         case 'localyear':
2565                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localYear, true );
2566                         case 'localtime':
2567                                 return $this->mVarCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2568                         case 'localhour':
2569                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localHour, true );
2570                         case 'localweek':
2571                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2572                                 // int to remove the padding
2573                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2574                         case 'localdow':
2575                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2576                         case 'numberofarticles':
2577                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2578                         case 'numberoffiles':
2579                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2580                         case 'numberofusers':
2581                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2582                         case 'numberofpages':
2583                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2584                         case 'numberofadmins':
2585                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::numberingroup('sysop') );
2586                         case 'numberofedits':
2587                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
2588                         case 'currenttimestamp':
2589                                 return $this->mVarCache[$index] = wfTimestamp( TS_MW, $ts );
2590                         case 'localtimestamp':
2591                                 return $this->mVarCache[$index] = $localTimestamp;
2592                         case 'currentversion':
2593                                 return $this->mVarCache[$index] = SpecialVersion::getVersion();
2594                         case 'sitename':
2595                                 return $wgSitename;
2596                         case 'server':
2597                                 return $wgServer;
2598                         case 'servername':
2599                                 return $wgServerName;
2600                         case 'scriptpath':
2601                                 return $wgScriptPath;
2602                         case 'directionmark':
2603                                 return $wgContLang->getDirMark();
2604                         case 'contentlanguage':
2605                                 global $wgContLanguageCode;
2606                                 return $wgContLanguageCode;
2607                         default:
2608                                 $ret = null;
2609                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret ) ) )
2610                                         return $ret;
2611                                 else
2612                                         return null;
2613                 }
2614         }
2615
2616         /**
2617          * initialise the magic variables (like CURRENTMONTHNAME)
2618          *
2619          * @private
2620          */
2621         function initialiseVariables() {
2622                 wfProfileIn( __METHOD__ );
2623                 $variableIDs = MagicWord::getVariableIDs();
2624
2625                 $this->mVariables = new MagicWordArray( $variableIDs );
2626                 wfProfileOut( __METHOD__ );
2627         }
2628
2629         /**
2630          * Preprocess some wikitext and return the document tree.
2631          * This is the ghost of replace_variables().
2632          *
2633          * @param string $text The text to parse
2634          * @param integer flags Bitwise combination of:
2635          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2636          *                                     included. Default is to assume a direct page view.
2637          *
2638          * The generated DOM tree must depend only on the input text and the flags.
2639          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2640          *
2641          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2642          * change in the DOM tree for a given text, must be passed through the section identifier
2643          * in the section edit link and thus back to extractSections().
2644          *
2645          * The output of this function is currently only cached in process memory, but a persistent
2646          * cache may be implemented at a later date which takes further advantage of these strict
2647          * dependency requirements.
2648          *
2649          * @private
2650          */
2651         function preprocessToDom ( $text, $flags = 0 ) {
2652                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2653                 return $dom;
2654         }
2655
2656         /*
2657          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2658          */
2659         public static function splitWhitespace( $s ) {
2660                 $ltrimmed = ltrim( $s );
2661                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2662                 $trimmed = rtrim( $ltrimmed );
2663                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2664                 if ( $diff > 0 ) {
2665                         $w2 = substr( $ltrimmed, -$diff );
2666                 } else {
2667                         $w2 = '';
2668                 }
2669                 return array( $w1, $trimmed, $w2 );
2670         }
2671
2672         /**
2673          * Replace magic variables, templates, and template arguments
2674          * with the appropriate text. Templates are substituted recursively,
2675          * taking care to avoid infinite loops.
2676          *
2677          * Note that the substitution depends on value of $mOutputType:
2678          *  self::OT_WIKI: only {{subst:}} templates
2679          *  self::OT_PREPROCESS: templates but not extension tags
2680          *  self::OT_HTML: all templates and extension tags
2681          *
2682          * @param string $tex The text to transform
2683          * @param PPFrame $frame Object describing the arguments passed to the template.
2684          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2685          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2686          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2687          * @private
2688          */
2689         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2690                 # Prevent too big inclusions
2691                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2692                         return $text;
2693                 }
2694
2695                 wfProfileIn( __METHOD__ );
2696
2697                 if ( $frame === false ) {
2698                         $frame = $this->getPreprocessor()->newFrame();
2699                 } elseif ( !( $frame instanceof PPFrame ) ) {
2700                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2701                         $frame = $this->getPreprocessor()->newCustomFrame($frame);
2702                 }
2703
2704                 $dom = $this->preprocessToDom( $text );
2705                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2706                 $text = $frame->expand( $dom, $flags );
2707
2708                 wfProfileOut( __METHOD__ );
2709                 return $text;
2710         }
2711
2712         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2713         static function createAssocArgs( $args ) {
2714                 $assocArgs = array();
2715                 $index = 1;
2716                 foreach( $args as $arg ) {
2717                         $eqpos = strpos( $arg, '=' );
2718                         if ( $eqpos === false ) {
2719                                 $assocArgs[$index++] = $arg;
2720                         } else {
2721                                 $name = trim( substr( $arg, 0, $eqpos ) );
2722                                 $value = trim( substr( $arg, $eqpos+1 ) );
2723                                 if ( $value === false ) {
2724                                         $value = '';
2725                                 }
2726                                 if ( $name !== false ) {
2727                                         $assocArgs[$name] = $value;
2728                                 }
2729                         }
2730                 }
2731
2732                 return $assocArgs;
2733         }
2734
2735         /**
2736          * Warn the user when a parser limitation is reached
2737          * Will warn at most once the user per limitation type
2738          *
2739          * @param string $limitationType, should be one of:
2740          *   'expensive-parserfunction' (corresponding messages: 'expensive-parserfunction-warning', 'expensive-parserfunction-category')
2741          *   'post-expand-template-argument' (corresponding messages: 'post-expand-template-argument-warning', 'post-expand-template-argument-category')
2742          *   'post-expand-template-inclusion' (corresponding messages: 'post-expand-template-inclusion-warning', 'post-expand-template-inclusion-category')
2743          * @params int $current, $max When an explicit limit has been
2744          *       exceeded, provide the values (optional)
2745          */
2746         function limitationWarn( $limitationType, $current=null, $max=null) {
2747                 $msgName = $limitationType . '-warning';
2748                 //does no harm if $current and $max are present but are unnecessary for the message
2749                 $warning = wfMsgExt( $msgName, array( 'parsemag', 'escape' ), $current, $max );
2750                 $this->mOutput->addWarning( $warning );
2751                 $cat = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( $limitationType . '-category' ) );
2752                 if ( $cat ) {
2753                         $this->mOutput->addCategory( $cat->getDBkey(), $this->getDefaultSort() );
2754                 }
2755         }
2756
2757         /**
2758          * Return the text of a template, after recursively
2759          * replacing any variables or templates within the template.
2760          *
2761          * @param array $piece The parts of the template
2762          *  $piece['title']: the title, i.e. the part before the |
2763          *  $piece['parts']: the parameter array
2764          *  $piece['lineStart']: whether the brace was at the start of a line
2765          * @param PPFrame The current frame, contains template arguments
2766          * @return string the text of the template
2767          * @private
2768          */
2769         function braceSubstitution( $piece, $frame ) {
2770                 global $wgContLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
2771                 wfProfileIn( __METHOD__ );
2772                 wfProfileIn( __METHOD__.'-setup' );
2773
2774                 # Flags
2775                 $found = false;             # $text has been filled
2776                 $nowiki = false;            # wiki markup in $text should be escaped
2777                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2778                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2779                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2780                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2781
2782                 # Title object, where $text came from
2783                 $title = NULL;
2784
2785                 # $part1 is the bit before the first |, and must contain only title characters.
2786                 # Various prefixes will be stripped from it later.
2787                 $titleWithSpaces = $frame->expand( $piece['title'] );
2788                 $part1 = trim( $titleWithSpaces );
2789                 $titleText = false;
2790
2791                 # Original title text preserved for various purposes
2792                 $originalTitle = $part1;
2793
2794                 # $args is a list of argument nodes, starting from index 0, not including $part1
2795                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2796                 wfProfileOut( __METHOD__.'-setup' );
2797
2798                 # SUBST
2799                 wfProfileIn( __METHOD__.'-modifiers' );
2800                 if ( !$found ) {
2801                         $mwSubst = MagicWord::get( 'subst' );
2802                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2803                                 # One of two possibilities is true:
2804                                 # 1) Found SUBST but not in the PST phase
2805                                 # 2) Didn't find SUBST and in the PST phase
2806                                 # In either case, return without further processing
2807                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2808                                 $isLocalObj = true;
2809                                 $found = true;
2810                         }
2811                 }
2812
2813                 # Variables
2814                 if ( !$found && $args->getLength() == 0 ) {
2815                         $id = $this->mVariables->matchStartToEnd( $part1 );
2816                         if ( $id !== false ) {
2817                                 $text = $this->getVariableValue( $id );
2818                                 if (MagicWord::getCacheTTL($id)>-1)
2819                                         $this->mOutput->mContainsOldMagic = true;
2820                                 $found = true;
2821                         }
2822                 }
2823
2824                 # MSG, MSGNW and RAW
2825                 if ( !$found ) {
2826                         # Check for MSGNW:
2827                         $mwMsgnw = MagicWord::get( 'msgnw' );
2828                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2829                                 $nowiki = true;
2830                         } else {
2831                                 # Remove obsolete MSG:
2832                                 $mwMsg = MagicWord::get( 'msg' );
2833                                 $mwMsg->matchStartAndRemove( $part1 );
2834                         }
2835
2836                         # Check for RAW:
2837                         $mwRaw = MagicWord::get( 'raw' );
2838                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2839                                 $forceRawInterwiki = true;
2840                         }
2841                 }
2842                 wfProfileOut( __METHOD__.'-modifiers' );
2843
2844                 # Parser functions
2845                 if ( !$found ) {
2846                         wfProfileIn( __METHOD__ . '-pfunc' );
2847
2848                         $colonPos = strpos( $part1, ':' );
2849                         if ( $colonPos !== false ) {
2850                                 # Case sensitive functions
2851                                 $function = substr( $part1, 0, $colonPos );
2852                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2853                                         $function = $this->mFunctionSynonyms[1][$function];
2854                                 } else {
2855                                         # Case insensitive functions
2856                                         $function = strtolower( $function );
2857                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2858                                                 $function = $this->mFunctionSynonyms[0][$function];
2859                                         } else {
2860                                                 $function = false;
2861                                         }
2862                                 }
2863                                 if ( $function ) {
2864                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2865                                         $initialArgs = array( &$this );
2866                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2867                                         if ( $flags & SFH_OBJECT_ARGS ) {
2868                                                 # Add a frame parameter, and pass the arguments as an array
2869                                                 $allArgs = $initialArgs;
2870                                                 $allArgs[] = $frame;
2871                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2872                                                         $funcArgs[] = $args->item( $i );
2873                                                 }
2874                                                 $allArgs[] = $funcArgs;
2875                                         } else {
2876                                                 # Convert arguments to plain text
2877                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2878                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2879                                                 }
2880                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2881                                         }
2882
2883                                         # Workaround for PHP bug 35229 and similar
2884                                         if ( !is_callable( $callback ) ) {
2885                                                 throw new MWException( "Tag hook for $function is not callable\n" );
2886                                         }
2887                                         $result = call_user_func_array( $callback, $allArgs );
2888                                         $found = true;
2889                                         $noparse = true;
2890                                         $preprocessFlags = 0;
2891
2892                                         if ( is_array( $result ) ) {
2893                                                 if ( isset( $result[0] ) ) {
2894                                                         $text = $result[0];
2895                                                         unset( $result[0] );
2896                                                 }
2897
2898                                                 // Extract flags into the local scope
2899                                                 // This allows callers to set flags such as nowiki, found, etc.
2900                                                 extract( $result );
2901                                         } else {
2902                                                 $text = $result;
2903                                         }
2904                                         if ( !$noparse ) {
2905                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
2906                                                 $isChildObj = true;
2907                                         }
2908                                 }
2909                         }
2910                         wfProfileOut( __METHOD__ . '-pfunc' );
2911                 }
2912
2913                 # Finish mangling title and then check for loops.
2914                 # Set $title to a Title object and $titleText to the PDBK
2915                 if ( !$found ) {
2916                         $ns = NS_TEMPLATE;
2917                         # Split the title into page and subpage
2918                         $subpage = '';
2919                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2920                         if ($subpage !== '') {
2921                                 $ns = $this->mTitle->getNamespace();
2922                         }
2923                         $title = Title::newFromText( $part1, $ns );
2924                         if ( $title ) {
2925                                 $titleText = $title->getPrefixedText();
2926                                 # Check for language variants if the template is not found
2927                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
2928                                         $wgContLang->findVariantLink( $part1, $title, true );
2929                                 }
2930                                 # Do infinite loop check
2931                                 if ( !$frame->loopCheck( $title ) ) {
2932                                         $found = true;
2933                                         $text = "<span class=\"error\">Template loop detected: [[$titleText]]</span>";
2934                                         wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
2935                                 }
2936                                 # Do recursion depth check
2937                                 $limit = $this->mOptions->getMaxTemplateDepth();
2938                                 if ( $frame->depth >= $limit ) {
2939                                         $found = true;
2940                                         $text = "<span class=\"error\">Template recursion depth limit exceeded ($limit)</span>";
2941                                 }
2942                         }
2943                 }
2944
2945                 # Load from database
2946                 if ( !$found && $title ) {
2947                         wfProfileIn( __METHOD__ . '-loadtpl' );
2948                         if ( !$title->isExternal() ) {
2949                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
2950                                         $text = SpecialPage::capturePath( $title );
2951                                         if ( is_string( $text ) ) {
2952                                                 $found = true;
2953                                                 $isHTML = true;
2954                                                 $this->disableCache();
2955                                         }
2956                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
2957                                         $found = false; //access denied
2958                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
2959                                 } else {
2960                                         list( $text, $title ) = $this->getTemplateDom( $title );
2961                                         if ( $text !== false ) {
2962                                                 $found = true;
2963                                                 $isChildObj = true;
2964                                         }
2965                                 }
2966
2967                                 # If the title is valid but undisplayable, make a link to it
2968                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2969                                         $text = "[[:$titleText]]";
2970                                         $found = true;
2971                                 }
2972                         } elseif ( $title->isTrans() ) {
2973                                 // Interwiki transclusion
2974                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
2975                                         $text = $this->interwikiTransclude( $title, 'render' );
2976                                         $isHTML = true;
2977                                 } else {
2978                                         $text = $this->interwikiTransclude( $title, 'raw' );
2979                                         // Preprocess it like a template
2980                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
2981                                         $isChildObj = true;
2982                                 }
2983                                 $found = true;
2984                         }
2985                         wfProfileOut( __METHOD__ . '-loadtpl' );
2986                 }
2987
2988                 # If we haven't found text to substitute by now, we're done
2989                 # Recover the source wikitext and return it
2990                 if ( !$found ) {
2991                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2992                         wfProfileOut( __METHOD__ );
2993                         return array( 'object' => $text );
2994                 }
2995
2996                 # Expand DOM-style return values in a child frame
2997                 if ( $isChildObj ) {
2998                         # Clean up argument array
2999                         $newFrame = $frame->newChild( $args, $title );
3000
3001                         if ( $nowiki ) {
3002                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3003                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3004                                 # Expansion is eligible for the empty-frame cache
3005                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3006                                         $text = $this->mTplExpandCache[$titleText];
3007                                 } else {
3008                                         $text = $newFrame->expand( $text );
3009                                         $this->mTplExpandCache[$titleText] = $text;
3010                                 }
3011                         } else {
3012                                 # Uncached expansion
3013                                 $text = $newFrame->expand( $text );
3014                         }
3015                 }
3016                 if ( $isLocalObj && $nowiki ) {
3017                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3018                         $isLocalObj = false;
3019                 }
3020
3021                 # Replace raw HTML by a placeholder
3022                 # Add a blank line preceding, to prevent it from mucking up
3023                 # immediately preceding headings
3024                 if ( $isHTML ) {
3025                         $text = "\n\n" . $this->insertStripItem( $text );
3026                 }
3027                 # Escape nowiki-style return values
3028                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3029                         $text = wfEscapeWikiText( $text );
3030                 }
3031                 # Bug 529: if the template begins with a table or block-level
3032                 # element, it should be treated as beginning a new line.
3033                 # This behaviour is somewhat controversial.
3034                 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3035                         $text = "\n" . $text;
3036                 }
3037
3038                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3039                         # Error, oversize inclusion
3040                         $text = "[[$originalTitle]]" .
3041                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3042                         $this->limitationWarn( 'post-expand-template-inclusion' );
3043                 }
3044
3045                 if ( $isLocalObj ) {
3046                         $ret = array( 'object' => $text );
3047                 } else {
3048                         $ret = array( 'text' => $text );
3049                 }
3050
3051                 wfProfileOut( __METHOD__ );
3052                 return $ret;
3053         }
3054
3055         /**
3056          * Get the semi-parsed DOM representation of a template with a given title,
3057          * and its redirect destination title. Cached.
3058          */
3059         function getTemplateDom( $title ) {
3060                 $cacheTitle = $title;
3061                 $titleText = $title->getPrefixedDBkey();
3062
3063                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3064                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3065                         $title = Title::makeTitle( $ns, $dbk );
3066                         $titleText = $title->getPrefixedDBkey();
3067                 }
3068                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3069                         return array( $this->mTplDomCache[$titleText], $title );
3070                 }
3071
3072                 // Cache miss, go to the database
3073                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3074
3075                 if ( $text === false ) {
3076                         $this->mTplDomCache[$titleText] = false;
3077                         return array( false, $title );
3078                 }
3079
3080                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3081                 $this->mTplDomCache[ $titleText ] = $dom;
3082
3083                 if (! $title->equals($cacheTitle)) {
3084                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3085                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3086                 }
3087
3088                 return array( $dom, $title );
3089         }
3090
3091         /**
3092          * Fetch the unparsed text of a template and register a reference to it.
3093          */
3094         function fetchTemplateAndTitle( $title ) {
3095                 $templateCb = $this->mOptions->getTemplateCallback();
3096                 $stuff = call_user_func( $templateCb, $title, $this );
3097                 $text = $stuff['text'];
3098                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3099                 if ( isset( $stuff['deps'] ) ) {
3100                         foreach ( $stuff['deps'] as $dep ) {
3101                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3102                         }
3103                 }
3104                 return array($text,$finalTitle);
3105         }
3106
3107         function fetchTemplate( $title ) {
3108                 $rv = $this->fetchTemplateAndTitle($title);
3109                 return $rv[0];
3110         }
3111
3112         /**
3113          * Static function to get a template
3114          * Can be overridden via ParserOptions::setTemplateCallback().
3115          */
3116         static function statelessFetchTemplate( $title, $parser=false ) {
3117                 $text = $skip = false;
3118                 $finalTitle = $title;
3119                 $deps = array();
3120
3121                 // Loop to fetch the article, with up to 1 redirect
3122                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3123                         # Give extensions a chance to select the revision instead
3124                         $id = false; // Assume current
3125                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3126
3127                         if( $skip ) {
3128                                 $text = false;
3129                                 $deps[] = array(
3130                                         'title' => $title,
3131                                         'page_id' => $title->getArticleID(),
3132                                         'rev_id' => null );
3133                                 break;
3134                         }
3135                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3136                         $rev_id = $rev ? $rev->getId() : 0;
3137                         // If there is no current revision, there is no page
3138                         if( $id === false && !$rev ) {
3139                                 $linkCache = LinkCache::singleton();
3140                                 $linkCache->addBadLinkObj( $title );
3141                         }
3142
3143                         $deps[] = array(
3144                                 'title' => $title,
3145                                 'page_id' => $title->getArticleID(),
3146                                 'rev_id' => $rev_id );
3147
3148                         if( $rev ) {
3149                                 $text = $rev->getText();
3150                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3151                                 global $wgContLang;
3152                                 $message = $wgContLang->lcfirst( $title->getText() );
3153                                 $text = wfMsgForContentNoTrans( $message );
3154                                 if( wfEmptyMsg( $message, $text ) ) {
3155                                         $text = false;
3156                                         break;
3157                                 }
3158                         } else {
3159                                 break;
3160                         }
3161                         if ( $text === false ) {
3162                                 break;
3163                         }
3164                         // Redirect?
3165                         $finalTitle = $title;
3166                         $title = Title::newFromRedirect( $text );
3167                 }
3168                 return array(
3169                         'text' => $text,
3170                         'finalTitle' => $finalTitle,
3171                         'deps' => $deps );
3172         }
3173
3174         /**
3175          * Transclude an interwiki link.
3176          */
3177         function interwikiTransclude( $title, $action ) {
3178                 global $wgEnableScaryTranscluding;
3179
3180                 if (!$wgEnableScaryTranscluding)
3181                         return wfMsg('scarytranscludedisabled');
3182
3183                 $url = $title->getFullUrl( "action=$action" );
3184
3185                 if (strlen($url) > 255)
3186                         return wfMsg('scarytranscludetoolong');
3187                 return $this->fetchScaryTemplateMaybeFromCache($url);
3188         }
3189
3190         function fetchScaryTemplateMaybeFromCache($url) {
3191                 global $wgTranscludeCacheExpiry;
3192                 $dbr = wfGetDB(DB_SLAVE);
3193                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3194                                 array('tc_url' => $url));
3195                 if ($obj) {
3196                         $time = $obj->tc_time;
3197                         $text = $obj->tc_contents;
3198                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3199                                 return $text;
3200                         }
3201                 }
3202
3203                 $text = Http::get($url);
3204                 if (!$text)
3205                         return wfMsg('scarytranscludefailed', $url);
3206
3207                 $dbw = wfGetDB(DB_MASTER);
3208                 $dbw->replace('transcache', array('tc_url'), array(
3209                         'tc_url' => $url,
3210                         'tc_time' => time(),
3211                         'tc_contents' => $text));
3212                 return $text;
3213         }
3214
3215
3216         /**
3217          * Triple brace replacement -- used for template arguments
3218          * @private
3219          */
3220         function argSubstitution( $piece, $frame ) {
3221                 wfProfileIn( __METHOD__ );
3222
3223                 $error = false;
3224                 $parts = $piece['parts'];
3225                 $nameWithSpaces = $frame->expand( $piece['title'] );
3226                 $argName = trim( $nameWithSpaces );
3227                 $object = false;
3228                 $text = $frame->getArgument( $argName );
3229                 if (  $text === false && $parts->getLength() > 0
3230                   && (
3231                     $this->ot['html']
3232                     || $this->ot['pre']
3233                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3234                   )
3235                 ) {
3236                         # No match in frame, use the supplied default
3237                         $object = $parts->item( 0 )->getChildren();
3238                 }
3239                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3240                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3241                         $this->limitationWarn( 'post-expand-template-argument' );
3242                 }
3243
3244                 if ( $text === false && $object === false ) {
3245                         # No match anywhere
3246                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3247                 }
3248                 if ( $error !== false ) {
3249                         $text .= $error;
3250                 }
3251                 if ( $object !== false ) {
3252                         $ret = array( 'object' => $object );
3253                 } else {
3254                         $ret = array( 'text' => $text );
3255                 }
3256
3257                 wfProfileOut( __METHOD__ );
3258                 return $ret;
3259         }
3260
3261         /**
3262          * Return the text to be used for a given extension tag.
3263          * This is the ghost of strip().
3264          *
3265          * @param array $params Associative array of parameters:
3266          *     name       PPNode for the tag name
3267          *     attr       PPNode for unparsed text where tag attributes are thought to be
3268          *     attributes Optional associative array of parsed attributes
3269          *     inner      Contents of extension element
3270          *     noClose    Original text did not have a close tag
3271          * @param PPFrame $frame
3272          */
3273         function extensionSubstitution( $params, $frame ) {
3274                 global $wgRawHtml, $wgContLang;
3275
3276                 $name = $frame->expand( $params['name'] );
3277                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3278                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3279
3280                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $this->mMarkerIndex++) . self::MARKER_SUFFIX;
3281
3282                 if ( $this->ot['html'] ) {
3283                         $name = strtolower( $name );
3284
3285                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3286                         if ( isset( $params['attributes'] ) ) {
3287                                 $attributes = $attributes + $params['attributes'];
3288                         }
3289                         switch ( $name ) {
3290                                 case 'html':
3291                                         if( $wgRawHtml ) {
3292                                                 $output = $content;
3293                                                 break;
3294                                         } else {
3295                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3296                                         }
3297                                 case 'nowiki':
3298                                         $output = Xml::escapeTagsOnly( $content );
3299                                         break;
3300                                 case 'math':
3301                                         $output = $wgContLang->armourMath(
3302                                                 MathRenderer::renderMath( $content, $attributes ) );
3303                                         break;
3304                                 case 'gallery':
3305                                         $output = $this->renderImageGallery( $content, $attributes );
3306                                         break;
3307                                 case 'poem':
3308                                         $output = $this->renderPoem( $content, $attributes );
3309                                         break;
3310                                 default:
3311                                         if( isset( $this->mTagHooks[$name] ) ) {
3312                                                 # Workaround for PHP bug 35229 and similar
3313                                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3314                                                         throw new MWException( "Tag hook for $name is not callable\n" );
3315                                                 }
3316                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3317                                                         array( $content, $attributes, $this ) );
3318                                         } else {
3319                                                 $output = '<span class="error">Invalid tag extension name: ' .
3320                                                         htmlspecialchars( $name ) . '</span>';
3321                                         }
3322                         }
3323                 } else {
3324                         if ( is_null( $attrText ) ) {
3325                                 $attrText = '';
3326                         }
3327                         if ( isset( $params['attributes'] ) ) {
3328                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3329                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3330                                                 htmlspecialchars( $attrValue ) . '"';
3331                                 }
3332                         }
3333                         if ( $content === null ) {
3334                                 $output = "<$name$attrText/>";
3335                         } else {
3336                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3337                                 $output = "<$name$attrText>$content$close";
3338                         }
3339                 }
3340
3341                 if ( $name === 'html' || $name === 'nowiki' ) {
3342                         $this->mStripState->nowiki->setPair( $marker, $output );
3343                 } else {
3344                         $this->mStripState->general->setPair( $marker, $output );
3345                 }
3346                 return $marker;
3347         }
3348
3349         /**
3350          * Increment an include size counter
3351          *
3352          * @param string $type The type of expansion
3353          * @param integer $size The size of the text
3354          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3355          */
3356         function incrementIncludeSize( $type, $size ) {
3357                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3358                         return false;
3359                 } else {
3360                         $this->mIncludeSizes[$type] += $size;
3361                         return true;
3362                 }
3363         }
3364
3365         /**
3366          * Increment the expensive function count
3367          *
3368          * @return boolean False if the limit has been exceeded
3369          */
3370         function incrementExpensiveFunctionCount() {
3371                 global $wgExpensiveParserFunctionLimit;
3372                 $this->mExpensiveFunctionCount++;
3373                 if($this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit) {
3374                         return true;
3375                 }
3376                 return false;
3377         }
3378
3379         /**
3380          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3381          * Fills $this->mDoubleUnderscores, returns the modified text
3382          */
3383         function doDoubleUnderscore( $text ) {
3384                 // The position of __TOC__ needs to be recorded
3385                 $mw = MagicWord::get( 'toc' );
3386                 if( $mw->match( $text ) ) {
3387                         $this->mShowToc = true;
3388                         $this->mForceTocPosition = true;
3389
3390                         // Set a placeholder. At the end we'll fill it in with the TOC.
3391                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3392
3393                         // Only keep the first one.
3394                         $text = $mw->replace( '', $text );
3395                 }
3396
3397                 // Now match and remove the rest of them
3398                 $mwa = MagicWord::getDoubleUnderscoreArray();
3399                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3400
3401                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3402                         $this->mOutput->mNoGallery = true;
3403                 }
3404                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3405                         $this->mShowToc = false;
3406                 }
3407                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3408                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3409
3410                         $containerCategory = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( 'hidden-category-category' ) );
3411                         if ( $containerCategory ) {
3412                                 $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3413                         } else {
3414                                 wfDebug( __METHOD__.": [[MediaWiki:hidden-category-category]] is not a valid title!\n" );
3415                         }
3416                 }
3417                 # (bug 8068) Allow control over whether robots index a page.
3418                 #
3419                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3420                 # is not desirable, the last one on the page should win.
3421                 if( isset( $this->mDoubleUnderscores['noindex'] ) ) {
3422                         $this->mOutput->setIndexPolicy( 'noindex' );
3423                 } elseif( isset( $this->mDoubleUnderscores['index'] ) ) {
3424                         $this->mOutput->setIndexPolicy( 'index' );
3425                 }
3426
3427                 return $text;
3428         }
3429
3430         /**
3431          * This function accomplishes several tasks:
3432          * 1) Auto-number headings if that option is enabled
3433          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3434          * 3) Add a Table of contents on the top for users who have enabled the option
3435          * 4) Auto-anchor headings
3436          *
3437          * It loops through all headlines, collects the necessary data, then splits up the
3438          * string and re-inserts the newly formatted headlines.
3439          *
3440          * @param string $text
3441          * @param boolean $isMain
3442          * @private
3443          */
3444         function formatHeadings( $text, $isMain=true ) {
3445                 global $wgMaxTocLevel, $wgContLang;
3446
3447                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3448                 $showEditLink = $this->mOptions->getEditSection();
3449
3450                 // Do not call quickUserCan unless necessary
3451                 if( $showEditLink && !$this->mTitle->quickUserCan( 'edit' ) ) {
3452                         $showEditLink = 0;
3453                 }
3454
3455                 # Inhibit editsection links if requested in the page
3456                 if ( isset( $this->mDoubleUnderscores['noeditsection'] ) ) {
3457                         $showEditLink = 0;
3458                 }
3459
3460                 # Get all headlines for numbering them and adding funky stuff like [edit]
3461                 # links - this is for later, but we need the number of headlines right now
3462                 $matches = array();
3463                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3464
3465                 # if there are fewer than 4 headlines in the article, do not show TOC
3466                 # unless it's been explicitly enabled.
3467                 $enoughToc = $this->mShowToc &&
3468                         (($numMatches >= 4) || $this->mForceTocPosition);
3469
3470                 # Allow user to stipulate that a page should have a "new section"
3471                 # link added via __NEWSECTIONLINK__
3472                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3473                         $this->mOutput->setNewSection( true );
3474                 }
3475
3476                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3477                 # override above conditions and always show TOC above first header
3478                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3479                         $this->mShowToc = true;
3480                         $enoughToc = true;
3481                 }
3482
3483                 # We need this to perform operations on the HTML
3484                 $sk = $this->mOptions->getSkin();
3485
3486                 # headline counter
3487                 $headlineCount = 0;
3488                 $numVisible = 0;
3489
3490                 # Ugh .. the TOC should have neat indentation levels which can be
3491                 # passed to the skin functions. These are determined here
3492                 $toc = '';
3493                 $full = '';
3494                 $head = array();
3495                 $sublevelCount = array();
3496                 $levelCount = array();
3497                 $toclevel = 0;
3498                 $level = 0;
3499                 $prevlevel = 0;
3500                 $toclevel = 0;
3501                 $prevtoclevel = 0;
3502                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3503                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3504                 $tocraw = array();
3505
3506                 foreach( $matches[3] as $headline ) {
3507                         $isTemplate = false;
3508                         $titleText = false;
3509                         $sectionIndex = false;
3510                         $numbering = '';
3511                         $markerMatches = array();
3512                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3513                                 $serial = $markerMatches[1];
3514                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3515                                 $isTemplate = ($titleText != $baseTitleText);
3516                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3517                         }
3518
3519                         if( $toclevel ) {
3520                                 $prevlevel = $level;
3521                                 $prevtoclevel = $toclevel;
3522                         }
3523                         $level = $matches[1][$headlineCount];
3524
3525                         if( $doNumberHeadings || $enoughToc ) {
3526
3527                                 if ( $level > $prevlevel ) {
3528                                         # Increase TOC level
3529                                         $toclevel++;
3530                                         $sublevelCount[$toclevel] = 0;
3531                                         if( $toclevel<$wgMaxTocLevel ) {
3532                                                 $prevtoclevel = $toclevel;
3533                                                 $toc .= $sk->tocIndent();
3534                                                 $numVisible++;
3535                                         }
3536                                 }
3537                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3538                                         # Decrease TOC level, find level to jump to
3539
3540                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3541                                                 # Can only go down to level 1
3542                                                 $toclevel = 1;
3543                                         } else {
3544                                                 for ($i = $toclevel; $i > 0; $i--) {
3545                                                         if ( $levelCount[$i] == $level ) {
3546                                                                 # Found last matching level
3547                                                                 $toclevel = $i;
3548                                                                 break;
3549                                                         }
3550                                                         elseif ( $levelCount[$i] < $level ) {
3551                                                                 # Found first matching level below current level
3552                                                                 $toclevel = $i + 1;
3553                                                                 break;
3554                                                         }
3555                                                 }
3556                                         }
3557                                         if( $toclevel<$wgMaxTocLevel ) {
3558                                                 if($prevtoclevel < $wgMaxTocLevel) {
3559                                                         # Unindent only if the previous toc level was shown :p
3560                                                         $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3561                                                         $prevtoclevel = $toclevel;
3562                                                 } else {
3563                                                         $toc .= $sk->tocLineEnd();
3564                                                 }
3565                                         }
3566                                 }
3567                                 else {
3568                                         # No change in level, end TOC line
3569                                         if( $toclevel<$wgMaxTocLevel ) {
3570                                                 $toc .= $sk->tocLineEnd();
3571                                         }
3572                                 }
3573
3574                                 $levelCount[$toclevel] = $level;
3575
3576                                 # count number of headlines for each level
3577                                 @$sublevelCount[$toclevel]++;
3578                                 $dot = 0;
3579                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3580                                         if( !empty( $sublevelCount[$i] ) ) {
3581                                                 if( $dot ) {
3582                                                         $numbering .= '.';
3583                                                 }
3584                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3585                                                 $dot = 1;
3586                                         }
3587                                 }
3588                         }
3589
3590                         # The safe header is a version of the header text safe to use for links
3591                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3592                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3593
3594                         # Remove link placeholders by the link text.
3595                         #     <!--LINK number-->
3596                         # turns into
3597                         #     link text with suffix
3598                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3599
3600                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3601                         $tocline = preg_replace(
3602                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3603                                 array( '',                          '<$1>'),
3604                                 $safeHeadline
3605                         );
3606                         $tocline = trim( $tocline );
3607
3608                         # For the anchor, strip out HTML-y stuff period
3609                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3610                         $safeHeadline = trim( $safeHeadline );
3611
3612                         # Save headline for section edit hint before it's escaped
3613                         $headlineHint = $safeHeadline;
3614                         $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3615                         # HTML names must be case-insensitively unique (bug 10721)
3616                         $arrayKey = strtolower( $safeHeadline );
3617
3618                         # count how many in assoc. array so we can track dupes in anchors
3619                         isset( $refers[$arrayKey] ) ? $refers[$arrayKey]++ : $refers[$arrayKey] = 1;
3620                         $refcount[$headlineCount] = $refers[$arrayKey];
3621
3622                         # Don't number the heading if it is the only one (looks silly)
3623                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3624                                 # the two are different if the line contains a link
3625                                 $headline=$numbering . ' ' . $headline;
3626                         }
3627
3628                         # Create the anchor for linking from the TOC to the section
3629                         $anchor = $safeHeadline;
3630                         if($refcount[$headlineCount] > 1 ) {
3631                                 $anchor .= '_' . $refcount[$headlineCount];
3632                         }
3633                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3634                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3635                                 $tocraw[] = array( 'toclevel' => $toclevel, 'level' => $level, 'line' => $tocline, 'number' => $numbering );
3636                         }
3637                         # give headline the correct <h#> tag
3638                         if( $showEditLink && $sectionIndex !== false ) {
3639                                 if( $isTemplate ) {
3640                                         # Put a T flag in the section identifier, to indicate to extractSections()
3641                                         # that sections inside <includeonly> should be counted.
3642                                         $editlink = $sk->doEditSectionLink(Title::newFromText( $titleText ), "T-$sectionIndex");
3643                                 } else {
3644                                         $editlink = $sk->doEditSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3645                                 }
3646                         } else {
3647                                 $editlink = '';
3648                         }
3649                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
3650
3651                         $headlineCount++;
3652                 }
3653
3654                 $this->mOutput->setSections( $tocraw );
3655
3656                 # Never ever show TOC if no headers
3657                 if( $numVisible < 1 ) {
3658                         $enoughToc = false;
3659                 }
3660
3661                 if( $enoughToc ) {
3662                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3663                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3664                         }
3665                         $toc = $sk->tocList( $toc );
3666                 }
3667
3668                 # split up and insert constructed headlines
3669
3670                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3671                 $i = 0;
3672
3673                 foreach( $blocks as $block ) {
3674                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3675                                 # This is the [edit] link that appears for the top block of text when
3676                                 # section editing is enabled
3677
3678                                 # Disabled because it broke block formatting
3679                                 # For example, a bullet point in the top line
3680                                 # $full .= $sk->editSectionLink(0);
3681                         }
3682                         $full .= $block;
3683                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3684                                 # Top anchor now in skin
3685                                 $full = $full.$toc;
3686                         }
3687
3688                         if( !empty( $head[$i] ) ) {
3689                                 $full .= $head[$i];
3690                         }
3691                         $i++;
3692                 }
3693                 if( $this->mForceTocPosition ) {
3694                         return str_replace( '<!--MWTOC-->', $toc, $full );
3695                 } else {
3696                         return $full;
3697                 }
3698         }
3699
3700         /**
3701          * Transform wiki markup when saving a page by doing \r\n -> \n
3702          * conversion, substitting signatures, {{subst:}} templates, etc.
3703          *
3704          * @param string $text the text to transform
3705          * @param Title &$title the Title object for the current article
3706          * @param User &$user the User object describing the current user
3707          * @param ParserOptions $options parsing options
3708          * @param bool $clearState whether to clear the parser state first
3709          * @return string the altered wiki markup
3710          * @public
3711          */
3712         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3713                 $this->mOptions = $options;
3714                 $this->setTitle( $title );
3715                 $this->setOutputType( self::OT_WIKI );
3716
3717                 if ( $clearState ) {
3718                         $this->clearState();
3719                 }
3720
3721                 $pairs = array(
3722                         "\r\n" => "\n",
3723                 );
3724                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3725                 $text = $this->pstPass2( $text, $user );
3726                 $text = $this->mStripState->unstripBoth( $text );
3727                 return $text;
3728         }
3729
3730         /**
3731          * Pre-save transform helper function
3732          * @private
3733          */
3734         function pstPass2( $text, $user ) {
3735                 global $wgContLang, $wgLocaltimezone;
3736
3737                 /* Note: This is the timestamp saved as hardcoded wikitext to
3738                  * the database, we use $wgContLang here in order to give
3739                  * everyone the same signature and use the default one rather
3740                  * than the one selected in each user's preferences.
3741                  *
3742                  * (see also bug 12815)
3743                  */
3744                 $ts = $this->mOptions->getTimestamp();
3745                 $tz = wfMsgForContent( 'timezone-utc' );
3746                 if ( isset( $wgLocaltimezone ) ) {
3747                         $unixts = wfTimestamp( TS_UNIX, $ts );
3748                         $oldtz = getenv( 'TZ' );
3749                         putenv( 'TZ='.$wgLocaltimezone );
3750                         $ts = date( 'YmdHis', $unixts );
3751                         $tz = date( 'T', $unixts );  # might vary on DST changeover!
3752                         putenv( 'TZ='.$oldtz );
3753                 }
3754
3755                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tz)";
3756
3757                 # Variable replacement
3758                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3759                 $text = $this->replaceVariables( $text );
3760
3761                 # Signatures
3762                 $sigText = $this->getUserSig( $user );
3763                 $text = strtr( $text, array(
3764                         '~~~~~' => $d,
3765                         '~~~~' => "$sigText $d",
3766                         '~~~' => $sigText
3767                 ) );
3768
3769                 # Context links: [[|name]] and [[name (context)|]]
3770                 #
3771                 global $wgLegalTitleChars;
3772                 $tc = "[$wgLegalTitleChars]";
3773                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
3774
3775                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3776                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
3777                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3778                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3779
3780                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3781                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3782                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
3783                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3784
3785                 $t = $this->mTitle->getText();
3786                 $m = array();
3787                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3788                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3789                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3790                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3791                 } else {
3792                         # if there's no context, don't bother duplicating the title
3793                         $text = preg_replace( $p2, '[[\\1]]', $text );
3794                 }
3795
3796                 # Trim trailing whitespace
3797                 $text = rtrim( $text );
3798
3799                 return $text;
3800         }
3801
3802         /**
3803          * Fetch the user's signature text, if any, and normalize to
3804          * validated, ready-to-insert wikitext.
3805          *
3806          * @param User $user
3807          * @return string
3808          * @private
3809          */
3810         function getUserSig( &$user ) {
3811                 global $wgMaxSigChars;
3812
3813                 $username = $user->getName();
3814                 $nickname = $user->getOption( 'nickname' );
3815                 $nickname = $nickname === '' ? $username : $nickname;
3816
3817                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
3818                         $nickname = $username;
3819                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
3820                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
3821                         # Sig. might contain markup; validate this
3822                         if( $this->validateSig( $nickname ) !== false ) {
3823                                 # Validated; clean up (if needed) and return it
3824                                 return $this->cleanSig( $nickname, true );
3825                         } else {
3826                                 # Failed to validate; fall back to the default
3827                                 $nickname = $username;
3828                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
3829                         }
3830                 }
3831
3832                 // Make sure nickname doesnt get a sig in a sig
3833                 $nickname = $this->cleanSigInSig( $nickname );
3834
3835                 # If we're still here, make it a link to the user page
3836                 $userText = wfEscapeWikiText( $username );
3837                 $nickText = wfEscapeWikiText( $nickname );
3838                 if ( $user->isAnon() )  {
3839                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
3840                 } else {
3841                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
3842                 }
3843         }
3844
3845         /**
3846          * Check that the user's signature contains no bad XML
3847          *
3848          * @param string $text
3849          * @return mixed An expanded string, or false if invalid.
3850          */
3851         function validateSig( $text ) {
3852                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
3853         }
3854
3855         /**
3856          * Clean up signature text
3857          *
3858          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3859          * 2) Substitute all transclusions
3860          *
3861          * @param string $text
3862          * @param $parsing Whether we're cleaning (preferences save) or parsing
3863          * @return string Signature text
3864          */
3865         function cleanSig( $text, $parsing = false ) {
3866                 if ( !$parsing ) {
3867                         global $wgTitle;
3868                         $this->clearState();
3869                         $this->setTitle( $wgTitle );
3870                         $this->mOptions = new ParserOptions;
3871                         $this->setOutputType = self::OT_PREPROCESS;
3872                 }
3873
3874                 # Option to disable this feature
3875                 if ( !$this->mOptions->getCleanSignatures() ) {
3876                         return $text;
3877                 }
3878
3879                 # FIXME: regex doesn't respect extension tags or nowiki
3880                 #  => Move this logic to braceSubstitution()
3881                 $substWord = MagicWord::get( 'subst' );
3882                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3883                 $substText = '{{' . $substWord->getSynonym( 0 );
3884
3885                 $text = preg_replace( $substRegex, $substText, $text );
3886                 $text = $this->cleanSigInSig( $text );
3887                 $dom = $this->preprocessToDom( $text );
3888                 $frame = $this->getPreprocessor()->newFrame();
3889                 $text = $frame->expand( $dom );
3890
3891                 if ( !$parsing ) {
3892                         $text = $this->mStripState->unstripBoth( $text );
3893                 }
3894
3895                 return $text;
3896         }
3897
3898         /**
3899          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3900          * @param string $text
3901          * @return string Signature text with /~{3,5}/ removed
3902          */
3903         function cleanSigInSig( $text ) {
3904                 $text = preg_replace( '/~{3,5}/', '', $text );
3905                 return $text;
3906         }
3907
3908         /**
3909          * Set up some variables which are usually set up in parse()
3910          * so that an external function can call some class members with confidence
3911          * @public
3912          */
3913         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3914                 $this->setTitle( $title );
3915                 $this->mOptions = $options;
3916                 $this->setOutputType( $outputType );
3917                 if ( $clearState ) {
3918                         $this->clearState();
3919                 }
3920         }
3921
3922         /**
3923          * Wrapper for preprocess()
3924          *
3925          * @param string $text the text to preprocess
3926          * @param ParserOptions $options  options
3927          * @return string
3928          * @public
3929          */
3930         function transformMsg( $text, $options ) {
3931                 global $wgTitle;
3932                 static $executing = false;
3933
3934                 # Guard against infinite recursion
3935                 if ( $executing ) {
3936                         return $text;
3937                 }
3938                 $executing = true;
3939
3940                 wfProfileIn(__METHOD__);
3941                 $text = $this->preprocess( $text, $wgTitle, $options );
3942
3943                 $executing = false;
3944                 wfProfileOut(__METHOD__);
3945                 return $text;
3946         }
3947
3948         /**
3949          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3950          * The callback should have the following form:
3951          *    function myParserHook( $text, $params, &$parser ) { ... }
3952          *
3953          * Transform and return $text. Use $parser for any required context, e.g. use
3954          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3955          *
3956          * @public
3957          *
3958          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3959          * @param mixed $callback The callback function (and object) to use for the tag
3960          *
3961          * @return The old value of the mTagHooks array associated with the hook
3962          */
3963         function setHook( $tag, $callback ) {
3964                 $tag = strtolower( $tag );
3965                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
3966                 $this->mTagHooks[$tag] = $callback;
3967                 if( !in_array( $tag, $this->mStripList ) ) {
3968                         $this->mStripList[] = $tag;
3969                 }
3970
3971                 return $oldVal;
3972         }
3973
3974         function setTransparentTagHook( $tag, $callback ) {
3975                 $tag = strtolower( $tag );
3976                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
3977                 $this->mTransparentTagHooks[$tag] = $callback;
3978
3979                 return $oldVal;
3980         }
3981
3982         /**
3983          * Remove all tag hooks
3984          */
3985         function clearTagHooks() {
3986                 $this->mTagHooks = array();
3987                 $this->mStripList = $this->mDefaultStripList;
3988         }
3989
3990         /**
3991          * Create a function, e.g. {{sum:1|2|3}}
3992          * The callback function should have the form:
3993          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3994          *
3995          * Or with SFH_OBJECT_ARGS:
3996          *    function myParserFunction( $parser, $frame, $args ) { ... }
3997          *
3998          * The callback may either return the text result of the function, or an array with the text
3999          * in element 0, and a number of flags in the other elements. The names of the flags are
4000          * specified in the keys. Valid flags are:
4001          *   found                     The text returned is valid, stop processing the template. This
4002          *                             is on by default.
4003          *   nowiki                    Wiki markup in the return value should be escaped
4004          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4005          *
4006          * @public
4007          *
4008          * @param string $id The magic word ID
4009          * @param mixed $callback The callback function (and object) to use
4010          * @param integer $flags a combination of the following flags:
4011          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4012          *
4013          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4014          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4015          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4016          *     the arguments, and to control the way they are expanded.
4017          *
4018          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4019          *     arguments, for instance:
4020          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4021          *
4022          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4023          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4024          *     working if/when this is changed.
4025          *
4026          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4027          *     expansion.
4028          *
4029          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4030          *     about the methods available in PPFrame and PPNode.
4031          *
4032          * @return The old callback function for this name, if any
4033          */
4034         function setFunctionHook( $id, $callback, $flags = 0 ) {
4035                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4036                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4037
4038                 # Add to function cache
4039                 $mw = MagicWord::get( $id );
4040                 if( !$mw )
4041                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4042
4043                 $synonyms = $mw->getSynonyms();
4044                 $sensitive = intval( $mw->isCaseSensitive() );
4045
4046                 foreach ( $synonyms as $syn ) {
4047                         # Case
4048                         if ( !$sensitive ) {
4049                                 $syn = strtolower( $syn );
4050                         }
4051                         # Add leading hash
4052                         if ( !( $flags & SFH_NO_HASH ) ) {
4053                                 $syn = '#' . $syn;
4054                         }
4055                         # Remove trailing colon
4056                         if ( substr( $syn, -1, 1 ) === ':' ) {
4057                                 $syn = substr( $syn, 0, -1 );
4058                         }
4059                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4060                 }
4061                 return $oldVal;
4062         }
4063
4064         /**
4065          * Get all registered function hook identifiers
4066          *
4067          * @return array
4068          */
4069         function getFunctionHooks() {
4070                 return array_keys( $this->mFunctionHooks );
4071         }
4072
4073         /**
4074          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4075          * Placeholders created in Skin::makeLinkObj()
4076          * Returns an array of link CSS classes, indexed by PDBK.
4077          */
4078         function replaceLinkHolders( &$text, $options = 0 ) {
4079                 return $this->mLinkHolders->replace( $text );
4080         }
4081
4082         /**
4083          * Replace <!--LINK--> link placeholders with plain text of links
4084          * (not HTML-formatted).
4085          * @param string $text
4086          * @return string
4087          */
4088         function replaceLinkHoldersText( $text ) {
4089                 return $this->mLinkHolders->replaceText( $text );
4090         }
4091
4092         /**
4093          * Tag hook handler for 'pre'.
4094          */
4095         function renderPreTag( $text, $attribs ) {
4096                 // Backwards-compatibility hack
4097                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4098
4099                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4100                 return wfOpenElement( 'pre', $attribs ) .
4101                         Xml::escapeTagsOnly( $content ) .
4102                         '</pre>';
4103         }
4104
4105         /**
4106          * Renders an image gallery from a text with one line per image.
4107          * text labels may be given by using |-style alternative text. E.g.
4108          *   Image:one.jpg|The number "1"
4109          *   Image:tree.jpg|A tree
4110          * given as text will return the HTML of a gallery with two images,
4111          * labeled 'The number "1"' and
4112          * 'A tree'.
4113          */
4114         function renderImageGallery( $text, $params ) {
4115                 $ig = new ImageGallery();
4116                 $ig->setContextTitle( $this->mTitle );
4117                 $ig->setShowBytes( false );
4118                 $ig->setShowFilename( false );
4119                 $ig->setParser( $this );
4120                 $ig->setHideBadImages();
4121                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4122                 $ig->useSkin( $this->mOptions->getSkin() );
4123                 $ig->mRevisionId = $this->mRevisionId;
4124
4125                 if( isset( $params['caption'] ) ) {
4126                         $caption = $params['caption'];
4127                         $caption = htmlspecialchars( $caption );
4128                         $caption = $this->replaceInternalLinks( $caption );
4129                         $ig->setCaptionHtml( $caption );
4130                 }
4131                 if( isset( $params['perrow'] ) ) {
4132                         $ig->setPerRow( $params['perrow'] );
4133                 }
4134                 if( isset( $params['widths'] ) ) {
4135                         $ig->setWidths( $params['widths'] );
4136                 }
4137                 if( isset( $params['heights'] ) ) {
4138                         $ig->setHeights( $params['heights'] );
4139                 }
4140
4141                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4142
4143                 $lines = StringUtils::explode( "\n", $text );
4144                 foreach ( $lines as $line ) {
4145                         # match lines like these:
4146                         # Image:someimage.jpg|This is some image
4147                         $matches = array();
4148                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4149                         # Skip empty lines
4150                         if ( count( $matches ) == 0 ) {
4151                                 continue;
4152                         }
4153
4154                         if ( strpos( $matches[0], '%' ) !== false )
4155                                 $matches[1] = urldecode( $matches[1] );
4156                         $tp = Title::newFromText( $matches[1]/*, NS_IMAGE*/ );
4157                         $nt =& $tp;
4158                         if( is_null( $nt ) ) {
4159                                 # Bogus title. Ignore these so we don't bomb out later.
4160                                 continue;
4161                         }
4162                         if ( isset( $matches[3] ) ) {
4163                                 $label = $matches[3];
4164                         } else {
4165                                 $label = '';
4166                         }
4167
4168                         $html = $this->recursiveTagParse( trim( $label ) );
4169
4170                         $ig->add( $nt, $html );
4171
4172                         # Only add real images (bug #5586)
4173                         if ( $nt->getNamespace() == NS_IMAGE ) {
4174                                 $this->mOutput->addImage( $nt->getDBkey() );
4175                         }
4176                 }
4177                 return $ig->toHTML();
4178         }
4179
4180         /** Renders any text in between <poem></poem> tags
4181          * based on http://www.mediawiki.org/wiki/Extension:Poem
4182          */
4183
4184         function renderPoem( $in, $param=array() ) {
4185
4186                 /* using newlines in the text will cause the parser to add <p> tags,
4187                 * which may not be desired in some cases
4188                 */
4189                 $nl = isset( $param['compact'] ) ? '' : "\n";
4190
4191                 $tag = $this->insertStripItem( "<br />", $this->mStripState );
4192                 $text = preg_replace(
4193                         array( "/^\n/", "/\n$/D", "/\n/", "/^( +)/me" ),
4194                         array( "", "", "$tag\n", "str_replace(' ','&nbsp;','\\1')" ),
4195                         $in );
4196                 $text = $this->recursiveTagParse( $text );
4197
4198                 // Pass HTML attributes through to the output.
4199                 $attribs = Sanitizer::validateTagAttributes( $param, 'div' );
4200
4201                 // Wrap output in a <div> with "poem" class.
4202                 if( isset( $attribs['class'] ) ) {
4203                         $attribs['class'] = 'poem ' . $attribs['class'];
4204                 } else {
4205                         $attribs['class'] = 'poem';
4206                 }
4207
4208                 return wfOpenElement( 'div', $attribs ) .
4209                         $nl .
4210                         trim( $text ) .
4211                         "$nl</div>";
4212         }
4213
4214         function getImageParams( $handler ) {
4215                 if ( $handler ) {
4216                         $handlerClass = get_class( $handler );
4217                 } else {
4218                         $handlerClass = '';
4219                 }
4220                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4221                         // Initialise static lists
4222                         static $internalParamNames = array(
4223                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4224                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4225                                         'bottom', 'text-bottom' ),
4226                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4227                                         'upright', 'border', 'link' ),
4228                         );
4229                         static $internalParamMap;
4230                         if ( !$internalParamMap ) {
4231                                 $internalParamMap = array();
4232                                 foreach ( $internalParamNames as $type => $names ) {
4233                                         foreach ( $names as $name ) {
4234                                                 $magicName = str_replace( '-', '_', "img_$name" );
4235                                                 $internalParamMap[$magicName] = array( $type, $name );
4236                                         }
4237                                 }
4238                         }
4239
4240                         // Add handler params
4241                         $paramMap = $internalParamMap;
4242                         if ( $handler ) {
4243                                 $handlerParamMap = $handler->getParamMap();
4244                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4245                                         $paramMap[$magic] = array( 'handler', $paramName );
4246                                 }
4247                         }
4248                         $this->mImageParams[$handlerClass] = $paramMap;
4249                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4250                 }
4251                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4252         }
4253
4254         /**
4255          * Parse image options text and use it to make an image
4256          * @param Title $title
4257          * @param string $options
4258          * @param LinkHolderArray $holders
4259          */
4260         function makeImage( $title, $options, $holders = false ) {
4261                 # Check if the options text is of the form "options|alt text"
4262                 # Options are:
4263                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4264                 #  * left               no resizing, just left align. label is used for alt= only
4265                 #  * right              same, but right aligned
4266                 #  * none               same, but not aligned
4267                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4268                 #  * center             center the image
4269                 #  * framed             Keep original image size, no magnify-button.
4270                 #  * frameless          like 'thumb' but without a frame. Keeps user preferences for width
4271                 #  * upright            reduce width for upright images, rounded to full __0 px
4272                 #  * border             draw a 1px border around the image
4273                 # vertical-align values (no % or length right now):
4274                 #  * baseline
4275                 #  * sub
4276                 #  * super
4277                 #  * top
4278                 #  * text-top
4279                 #  * middle
4280                 #  * bottom
4281                 #  * text-bottom
4282
4283                 $parts = StringUtils::explode( "|", $options );
4284                 $sk = $this->mOptions->getSkin();
4285
4286                 # Give extensions a chance to select the file revision for us
4287                 $skip = $time = $descQuery = false;
4288                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4289
4290                 if ( $skip ) {
4291                         return $sk->link( $title );
4292                 }
4293
4294                 # Get parameter map
4295                 $file = wfFindFile( $title, $time );
4296                 $handler = $file ? $file->getHandler() : false;
4297
4298                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4299
4300                 # Process the input parameters
4301                 $caption = '';
4302                 $params = array( 'frame' => array(), 'handler' => array(),
4303                         'horizAlign' => array(), 'vertAlign' => array() );
4304                 foreach( $parts as $part ) {
4305                         $part = trim( $part );
4306                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4307                         $validated = false;
4308                         if( isset( $paramMap[$magicName] ) ) {
4309                                 list( $type, $paramName ) = $paramMap[$magicName];
4310
4311                                 // Special case; width and height come in one variable together
4312                                 if( $type === 'handler' && $paramName === 'width' ) {
4313                                         $m = array();
4314                                         # (bug 13500) In both cases (width/height and width only),
4315                                         # permit trailing "px" for backward compatibility.
4316                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4317                                                 $width = intval( $m[1] );
4318                                                 $height = intval( $m[2] );
4319                                                 if ( $handler->validateParam( 'width', $width ) ) {
4320                                                         $params[$type]['width'] = $width;
4321                                                         $validated = true;
4322                                                 }
4323                                                 if ( $handler->validateParam( 'height', $height ) ) {
4324                                                         $params[$type]['height'] = $height;
4325                                                         $validated = true;
4326                                                 }
4327                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4328                                                 $width = intval( $value );
4329                                                 if ( $handler->validateParam( 'width', $width ) ) {
4330                                                         $params[$type]['width'] = $width;
4331                                                         $validated = true;
4332                                                 }
4333                                         } // else no validation -- bug 13436
4334                                 } else {
4335                                         if ( $type === 'handler' ) {
4336                                                 # Validate handler parameter
4337                                                 $validated = $handler->validateParam( $paramName, $value );
4338                                         } else {
4339                                                 # Validate internal parameters
4340                                                 switch( $paramName ) {
4341                                                 case "manualthumb":
4342                                                         /// @fixme - possibly check validity here?
4343                                                         /// downstream behavior seems odd with missing manual thumbs.
4344                                                         $validated = true;
4345                                                         break;
4346                                                 case 'link':
4347                                                         $chars = self::EXT_LINK_URL_CLASS;
4348                                                         $prots = $this->mUrlProtocols;
4349                                                         if ( $value === '' ) {
4350                                                                 $paramName = 'no-link';
4351                                                                 $value = true;
4352                                                                 $validated = true;
4353                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4354                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4355                                                                         $paramName = 'link-url';
4356                                                                         $this->mOutput->addExternalLink( $value );
4357                                                                         $validated = true;
4358                                                                 }
4359                                                         } else {
4360                                                                 $linkTitle = Title::newFromText( $value );
4361                                                                 if ( $linkTitle ) {
4362                                                                         $paramName = 'link-title';
4363                                                                         $value = $linkTitle;
4364                                                                         $this->mOutput->addLink( $linkTitle );
4365                                                                         $validated = true;
4366                                                                 }
4367                                                         }
4368                                                         break;
4369                                                 default:
4370                                                         // Most other things appear to be empty or numeric...
4371                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4372                                                 }
4373                                         }
4374
4375                                         if ( $validated ) {
4376                                                 $params[$type][$paramName] = $value;
4377                                         }
4378                                 }
4379                         }
4380                         if ( !$validated ) {
4381                                 $caption = $part;
4382                         }
4383                 }
4384
4385                 # Process alignment parameters
4386                 if ( $params['horizAlign'] ) {
4387                         $params['frame']['align'] = key( $params['horizAlign'] );
4388                 }
4389                 if ( $params['vertAlign'] ) {
4390                         $params['frame']['valign'] = key( $params['vertAlign'] );
4391                 }
4392
4393                 # Strip bad stuff out of the alt text
4394                 # We can't just use replaceLinkHoldersText() here, because if this function
4395                 # is called from replaceInternalLinks2(), mLinkHolders won't be up to date.
4396                 if ( $holders ) {
4397                         $alt = $holders->replaceText( $caption );
4398                 } else {
4399                         $alt = $this->replaceLinkHoldersText( $caption );
4400                 }
4401
4402                 # make sure there are no placeholders in thumbnail attributes
4403                 # that are later expanded to html- so expand them now and
4404                 # remove the tags
4405                 $alt = $this->mStripState->unstripBoth( $alt );
4406                 $alt = Sanitizer::stripAllTags( $alt );
4407
4408                 $params['frame']['alt'] = $alt;
4409                 $params['frame']['caption'] = $caption;
4410
4411                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4412
4413                 # Linker does the rest
4414                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4415
4416                 # Give the handler a chance to modify the parser object
4417                 if ( $handler ) {
4418                         $handler->parserTransformHook( $this, $file );
4419                 }
4420
4421                 return $ret;
4422         }
4423
4424         /**
4425          * Set a flag in the output object indicating that the content is dynamic and
4426          * shouldn't be cached.
4427          */
4428         function disableCache() {
4429                 wfDebug( "Parser output marked as uncacheable.\n" );
4430                 $this->mOutput->mCacheTime = -1;
4431         }
4432
4433         /**#@+
4434          * Callback from the Sanitizer for expanding items found in HTML attribute
4435          * values, so they can be safely tested and escaped.
4436          * @param string $text
4437          * @param PPFrame $frame
4438          * @return string
4439          * @private
4440          */
4441         function attributeStripCallback( &$text, $frame = false ) {
4442                 $text = $this->replaceVariables( $text, $frame );
4443                 $text = $this->mStripState->unstripBoth( $text );
4444                 return $text;
4445         }
4446
4447         /**#@-*/
4448
4449         /**#@+
4450          * Accessor/mutator
4451          */
4452         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4453         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4454         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4455         /**#@-*/
4456
4457         /**#@+
4458          * Accessor
4459          */
4460         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4461         /**#@-*/
4462
4463
4464         /**
4465          * Break wikitext input into sections, and either pull or replace
4466          * some particular section's text.
4467          *
4468          * External callers should use the getSection and replaceSection methods.
4469          *
4470          * @param string $text Page wikitext
4471          * @param string $section A section identifier string of the form:
4472          *   <flag1> - <flag2> - ... - <section number>
4473          *
4474          * Currently the only recognised flag is "T", which means the target section number
4475          * was derived during a template inclusion parse, in other words this is a template
4476          * section edit link. If no flags are given, it was an ordinary section edit link.
4477          * This flag is required to avoid a section numbering mismatch when a section is
4478          * enclosed by <includeonly> (bug 6563).
4479          *
4480          * The section number 0 pulls the text before the first heading; other numbers will
4481          * pull the given section along with its lower-level subsections. If the section is
4482          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4483          *
4484          * @param string $mode One of "get" or "replace"
4485          * @param string $newText Replacement text for section data.
4486          * @return string for "get", the extracted section text.
4487          *                for "replace", the whole page with the section replaced.
4488          */
4489         private function extractSections( $text, $section, $mode, $newText='' ) {
4490                 global $wgTitle;
4491                 $this->clearState();
4492                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4493                 $this->mOptions = new ParserOptions;
4494                 $this->setOutputType( self::OT_WIKI );
4495                 $outText = '';
4496                 $frame = $this->getPreprocessor()->newFrame();
4497
4498                 // Process section extraction flags
4499                 $flags = 0;
4500                 $sectionParts = explode( '-', $section );
4501                 $sectionIndex = array_pop( $sectionParts );
4502                 foreach ( $sectionParts as $part ) {
4503                         if ( $part === 'T' ) {
4504                                 $flags |= self::PTD_FOR_INCLUSION;
4505                         }
4506                 }
4507                 // Preprocess the text
4508                 $root = $this->preprocessToDom( $text, $flags );
4509
4510                 // <h> nodes indicate section breaks
4511                 // They can only occur at the top level, so we can find them by iterating the root's children
4512                 $node = $root->getFirstChild();
4513
4514                 // Find the target section
4515                 if ( $sectionIndex == 0 ) {
4516                         // Section zero doesn't nest, level=big
4517                         $targetLevel = 1000;
4518                 } else {
4519             while ( $node ) {
4520                 if ( $node->getName() === 'h' ) {
4521                     $bits = $node->splitHeading();
4522                                         if ( $bits['i'] == $sectionIndex ) {
4523                                         $targetLevel = $bits['level'];
4524                                                 break;
4525                                         }
4526                                 }
4527                                 if ( $mode === 'replace' ) {
4528                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4529                                 }
4530                                 $node = $node->getNextSibling();
4531                         }
4532                 }
4533
4534                 if ( !$node ) {
4535                         // Not found
4536                         if ( $mode === 'get' ) {
4537                                 return $newText;
4538                         } else {
4539                                 return $text;
4540                         }
4541                 }
4542
4543                 // Find the end of the section, including nested sections
4544                 do {
4545                         if ( $node->getName() === 'h' ) {
4546                                 $bits = $node->splitHeading();
4547                                 $curLevel = $bits['level'];
4548                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4549                                         break;
4550                                 }
4551                         }
4552                         if ( $mode === 'get' ) {
4553                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4554                         }
4555                         $node = $node->getNextSibling();
4556                 } while ( $node );
4557
4558                 // Write out the remainder (in replace mode only)
4559                 if ( $mode === 'replace' ) {
4560                         // Output the replacement text
4561                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4562                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4563                         $outText .= $newText . "\n\n";
4564                         while ( $node ) {
4565                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4566                                 $node = $node->getNextSibling();
4567                         }
4568                 }
4569
4570                 if ( is_string( $outText ) ) {
4571                         // Re-insert stripped tags
4572                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4573                 }
4574
4575                 return $outText;
4576         }
4577
4578         /**
4579          * This function returns the text of a section, specified by a number ($section).
4580          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4581          * the first section before any such heading (section 0).
4582          *
4583          * If a section contains subsections, these are also returned.
4584          *
4585          * @param string $text text to look in
4586          * @param string $section section identifier
4587          * @param string $deftext default to return if section is not found
4588          * @return string text of the requested section
4589          */
4590         public function getSection( $text, $section, $deftext='' ) {
4591                 return $this->extractSections( $text, $section, "get", $deftext );
4592         }
4593
4594         public function replaceSection( $oldtext, $section, $text ) {
4595                 return $this->extractSections( $oldtext, $section, "replace", $text );
4596         }
4597
4598         /**
4599          * Get the timestamp associated with the current revision, adjusted for
4600          * the default server-local timestamp
4601          */
4602         function getRevisionTimestamp() {
4603                 if ( is_null( $this->mRevisionTimestamp ) ) {
4604                         wfProfileIn( __METHOD__ );
4605                         global $wgContLang;
4606                         $dbr = wfGetDB( DB_SLAVE );
4607                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4608                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4609
4610                         // Normalize timestamp to internal MW format for timezone processing.
4611                         // This has the added side-effect of replacing a null value with
4612                         // the current time, which gives us more sensible behavior for
4613                         // previews.
4614                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4615
4616                         // The cryptic '' timezone parameter tells to use the site-default
4617                         // timezone offset instead of the user settings.
4618                         //
4619                         // Since this value will be saved into the parser cache, served
4620                         // to other users, and potentially even used inside links and such,
4621                         // it needs to be consistent for all visitors.
4622                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4623
4624                         wfProfileOut( __METHOD__ );
4625                 }
4626                 return $this->mRevisionTimestamp;
4627         }
4628
4629         /**
4630          * Mutator for $mDefaultSort
4631          *
4632          * @param $sort New value
4633          */
4634         public function setDefaultSort( $sort ) {
4635                 $this->mDefaultSort = $sort;
4636         }
4637
4638         /**
4639          * Accessor for $mDefaultSort
4640          * Will use the title/prefixed title if none is set
4641          *
4642          * @return string
4643          */
4644         public function getDefaultSort() {
4645                 global $wgCategoryPrefixedDefaultSortkey;
4646                 if( $this->mDefaultSort !== false ) {
4647                         return $this->mDefaultSort;
4648                 } elseif ($this->mTitle->getNamespace() == NS_CATEGORY ||
4649                         !$wgCategoryPrefixedDefaultSortkey) {
4650                         return $this->mTitle->getText();
4651                 } else {
4652                         return $this->mTitle->getPrefixedText();
4653                 }
4654         }
4655
4656         /**
4657          * Try to guess the section anchor name based on a wikitext fragment
4658          * presumably extracted from a heading, for example "Header" from
4659          * "== Header ==".
4660          */
4661         public function guessSectionNameFromWikiText( $text ) {
4662                 # Strip out wikitext links(they break the anchor)
4663                 $text = $this->stripSectionName( $text );
4664                 $headline = Sanitizer::decodeCharReferences( $text );
4665                 # strip out HTML
4666                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
4667                 $headline = trim( $headline );
4668                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
4669                 $replacearray = array(
4670                         '%3A' => ':',
4671                         '%' => '.'
4672                 );
4673                 return str_replace(
4674                         array_keys( $replacearray ),
4675                         array_values( $replacearray ),
4676                         $sectionanchor );
4677         }
4678
4679         /**
4680          * Strips a text string of wikitext for use in a section anchor
4681          *
4682          * Accepts a text string and then removes all wikitext from the
4683          * string and leaves only the resultant text (i.e. the result of
4684          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
4685          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
4686          * to create valid section anchors by mimicing the output of the
4687          * parser when headings are parsed.
4688          *
4689          * @param $text string Text string to be stripped of wikitext
4690          * for use in a Section anchor
4691          * @return Filtered text string
4692          */
4693         public function stripSectionName( $text ) {
4694                 # Strip internal link markup
4695                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
4696                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
4697
4698                 # Strip external link markup (FIXME: Not Tolerant to blank link text
4699                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
4700                 # on how many empty links there are on the page - need to figure that out.
4701                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
4702
4703                 # Parse wikitext quotes (italics & bold)
4704                 $text = $this->doQuotes($text);
4705
4706                 # Strip HTML tags
4707                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
4708                 return $text;
4709         }
4710
4711         function srvus( $text ) {
4712                 return $this->testSrvus( $text, $this->mOutputType );
4713         }
4714
4715         /**
4716          * strip/replaceVariables/unstrip for preprocessor regression testing
4717          */
4718         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
4719                 $this->clearState();
4720                 if ( ! ( $title instanceof Title ) ) {
4721                         $title = Title::newFromText( $title );
4722                 }
4723                 $this->mTitle = $title;
4724                 $this->mOptions = $options;
4725                 $this->setOutputType( $outputType );
4726                 $text = $this->replaceVariables( $text );
4727                 $text = $this->mStripState->unstripBoth( $text );
4728                 $text = Sanitizer::removeHTMLtags( $text );
4729                 return $text;
4730         }
4731
4732         function testPst( $text, $title, $options ) {
4733                 global $wgUser;
4734                 if ( ! ( $title instanceof Title ) ) {
4735                         $title = Title::newFromText( $title );
4736                 }
4737                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
4738         }
4739
4740         function testPreprocess( $text, $title, $options ) {
4741                 if ( ! ( $title instanceof Title ) ) {
4742                         $title = Title::newFromText( $title );
4743                 }
4744                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
4745         }
4746
4747         function markerSkipCallback( $s, $callback ) {
4748                 $i = 0;
4749                 $out = '';
4750                 while ( $i < strlen( $s ) ) {
4751                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
4752                         if ( $markerStart === false ) {
4753                                 $out .= call_user_func( $callback, substr( $s, $i ) );
4754                                 break;
4755                         } else {
4756                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
4757                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
4758                                 if ( $markerEnd === false ) {
4759                                         $out .= substr( $s, $markerStart );
4760                                         break;
4761                                 } else {
4762                                         $markerEnd += strlen( self::MARKER_SUFFIX );
4763                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
4764                                         $i = $markerEnd;
4765                                 }
4766                         }
4767                 }
4768                 return $out;
4769         }
4770 }
4771
4772 /**
4773  * @todo document, briefly.
4774  * @ingroup Parser
4775  */
4776 class StripState {
4777         var $general, $nowiki;
4778
4779         function __construct() {
4780                 $this->general = new ReplacementArray;
4781                 $this->nowiki = new ReplacementArray;
4782         }
4783
4784         function unstripGeneral( $text ) {
4785                 wfProfileIn( __METHOD__ );
4786                 do {
4787                         $oldText = $text;
4788                         $text = $this->general->replace( $text );
4789                 } while ( $text !== $oldText );
4790                 wfProfileOut( __METHOD__ );
4791                 return $text;
4792         }
4793
4794         function unstripNoWiki( $text ) {
4795                 wfProfileIn( __METHOD__ );
4796                 do {
4797                         $oldText = $text;
4798                         $text = $this->nowiki->replace( $text );
4799                 } while ( $text !== $oldText );
4800                 wfProfileOut( __METHOD__ );
4801                 return $text;
4802         }
4803
4804         function unstripBoth( $text ) {
4805                 wfProfileIn( __METHOD__ );
4806                 do {
4807                         $oldText = $text;
4808                         $text = $this->general->replace( $text );
4809                         $text = $this->nowiki->replace( $text );
4810                 } while ( $text !== $oldText );
4811                 wfProfileOut( __METHOD__ );
4812                 return $text;
4813         }
4814 }
4815
4816 /**
4817  * @todo document, briefly.
4818  * @ingroup Parser
4819  */
4820 class OnlyIncludeReplacer {
4821         var $output = '';
4822
4823         function replace( $matches ) {
4824                 if ( substr( $matches[1], -1 ) === "\n" ) {
4825                         $this->output .= substr( $matches[1], 0, -1 );
4826                 } else {
4827                         $this->output .= $matches[1];
4828                 }
4829         }
4830 }