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                 $mFunctionTagHooks;
  97
  98
  99         # Cleared with clearState():
 100         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 101         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 102         var $mLinkHolders, $mLinkID;
 103         var $mIncludeSizes, $mPPNodeCount, $mDefaultSort;
 104         var $mTplExpandCache; // empty-frame expansion cache
 105         var $mTplRedirCache, $mTplDomCache, $mHeadings, $mDoubleUnderscores;
 106         var $mExpensiveFunctionCount; // number of expensive parser function calls
 107         var $mFileCache;
 108
 109         # Temporary
 110         # These are variables reset at least once per parse regardless of $clearState
 111         var $mOptions,      // ParserOptions object
 112                 $mTitle,        // Title context, used for self-link rendering and similar things
 113                 $mOutputType,   // Output type, one of the OT_xxx constants
 114                 $ot,            // Shortcut alias, see setOutputType()
 115                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 116                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 117                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 118
 119         /**#@-*/
 120
 121         /**
 122          * Constructor
 123          *
 124          * @public
 125          */
 126         function __construct( $conf = array() ) {
 127                 $this->mConf = $conf;
 128                 $this->mTagHooks = array();
 129                 $this->mTransparentTagHooks = array();
 130                 $this->mFunctionHooks = array();
 131                 $this->mFunctionTagHooks = array();
 132                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 133                 $this->mDefaultStripList = $this->mStripList = array( 'nowiki', 'gallery' );
 134                 $this->mUrlProtocols = wfUrlProtocols();
 135                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 136                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 137                 $this->mVarCache = array();
 138                 if ( isset( $conf['preprocessorClass'] ) ) {
 139                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 140                 } elseif ( extension_loaded( 'domxml' ) ) {
 141                         // PECL extension that conflicts with the core DOM extension (bug 13770)
 142                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 143                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 144                 } elseif ( extension_loaded( 'dom' ) ) {
 145                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 146                 } else {
 147                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 148                 }
 149                 $this->mMarkerIndex = 0;
 150                 $this->mFirstCall = true;
 151         }
 152
 153         /**
 154          * Reduce memory usage to reduce the impact of circular references
 155          */
 156         function __destruct() {
 157                 if ( isset( $this->mLinkHolders ) ) {
 158                         $this->mLinkHolders->__destruct();
 159                 }
 160                 foreach ( $this as $name => $value ) {
 161                         unset( $this->$name );
 162                 }
 163         }
 164
 165         /**
 166          * Do various kinds of initialisation on the first call of the parser
 167          */
 168         function firstCallInit() {
 169                 if ( !$this->mFirstCall ) {
 170                         return;
 171                 }
 172                 $this->mFirstCall = false;
 173
 174                 wfProfileIn( __METHOD__ );
 175
 176                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 177                 CoreParserFunctions::register( $this );
 178                 $this->initialiseVariables();
 179
 180                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 181                 wfProfileOut( __METHOD__ );
 182         }
 183
 184         /**
 185          * Clear Parser state
 186          *
 187          * @private
 188          */
 189         function clearState() {
 190                 wfProfileIn( __METHOD__ );
 191                 if ( $this->mFirstCall ) {
 192                         $this->firstCallInit();
 193                 }
 194                 $this->mOutput = new ParserOutput;
 195                 $this->mAutonumber = 0;
 196                 $this->mLastSection = '';
 197                 $this->mDTopen = false;
 198                 $this->mIncludeCount = array();
 199                 $this->mStripState = new StripState;
 200                 $this->mArgStack = false;
 201                 $this->mInPre = false;
 202                 $this->mLinkHolders = new LinkHolderArray( $this );
 203                 $this->mLinkID = 0;
 204                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 205
 206                 /**
 207                  * Prefix for temporary replacement strings for the multipass parser.
 208                  * \x07 should never appear in input as it's disallowed in XML.
 209                  * Using it at the front also gives us a little extra robustness
 210                  * since it shouldn't match when butted up against identifier-like
 211                  * string constructs.
 212                  *
 213                  * Must not consist of all title characters, or else it will change
 214                  * the behaviour of <nowiki> in a link.
 215                  */
 216                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 217                 # Changed to \x7f to allow XML double-parsing -- TS
 218                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 219
 220
 221                 # Clear these on every parse, bug 4549
 222                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 223
 224                 $this->mShowToc = true;
 225                 $this->mForceTocPosition = false;
 226                 $this->mIncludeSizes = array(
 227                         'post-expand' => 0,
 228                         'arg' => 0,
 229                 );
 230                 $this->mPPNodeCount = 0;
 231                 $this->mDefaultSort = false;
 232                 $this->mHeadings = array();
 233                 $this->mDoubleUnderscores = array();
 234                 $this->mExpensiveFunctionCount = 0;
 235                 $this->mFileCache = array();
 236
 237                 # Fix cloning
 238                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 239                         $this->mPreprocessor = null;
 240                 }
 241
 242                 wfRunHooks( 'ParserClearState', array( &$this ) );
 243                 wfProfileOut( __METHOD__ );
 244         }
 245
 246         function setOutputType( $ot ) {
 247                 $this->mOutputType = $ot;
 248                 // Shortcut alias
 249                 $this->ot = array(
 250                         'html' => $ot == self::OT_HTML,
 251                         'wiki' => $ot == self::OT_WIKI,
 252                         'pre' => $ot == self::OT_PREPROCESS,
 253                 );
 254         }
 255
 256         /**
 257          * Set the context title
 258          */
 259         function setTitle( $t ) {
 260                 if ( !$t || $t instanceof FakeTitle ) {
 261                         $t = Title::newFromText( 'NO TITLE' );
 262                 }
 263                 if ( strval( $t->getFragment() ) !== '' ) {
 264                         # Strip the fragment to avoid various odd effects
 265                         $this->mTitle = clone $t;
 266                         $this->mTitle->setFragment( '' );
 267                 } else {
 268                         $this->mTitle = $t;
 269                 }
 270         }
 271
 272         /**
 273          * Accessor for mUniqPrefix.
 274          *
 275          * @public
 276          */
 277         function uniqPrefix() {
 278                 if( !isset( $this->mUniqPrefix ) ) {
 279                         // @fixme this is probably *horribly wrong*
 280                         // LanguageConverter seems to want $wgParser's uniqPrefix, however
 281                         // if this is called for a parser cache hit, the parser may not
 282                         // have ever been initialized in the first place.
 283                         // Not really sure what the heck is supposed to be going on here.
 284                         return '';
 285                         //throw new MWException( "Accessing uninitialized mUniqPrefix" );
 286                 }
 287                 return $this->mUniqPrefix;
 288         }
 289
 290         /**
 291          * Convert wikitext to HTML
 292          * Do not call this function recursively.
 293          *
 294          * @param $text String: text we want to parse
 295          * @param $title A title object
 296          * @param $options ParserOptions
 297          * @param $linestart boolean
 298          * @param $clearState boolean
 299          * @param $revid Int: number to pass in {{REVISIONID}}
 300          * @return ParserOutput a ParserOutput
 301          */
 302         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 303                 /**
 304                  * First pass--just handle <nowiki> sections, pass the rest off
 305                  * to internalParse() which does all the real work.
 306                  */
 307
 308                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 309                 $fname = __METHOD__.'-' . wfGetCaller();
 310                 wfProfileIn( __METHOD__ );
 311                 wfProfileIn( $fname );
 312
 313                 if ( $clearState ) {
 314                         $this->clearState();
 315                 }
 316
 317                 $this->mOptions = $options;
 318                 $this->setTitle( $title );
 319                 $oldRevisionId = $this->mRevisionId;
 320                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 321                 if( $revid !== null ) {
 322                         $this->mRevisionId = $revid;
 323                         $this->mRevisionTimestamp = null;
 324                 }
 325                 $this->setOutputType( self::OT_HTML );
 326                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 327                 # No more strip!
 328                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 329                 $text = $this->internalParse( $text );
 330                 $text = $this->mStripState->unstripGeneral( $text );
 331
 332                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 333                 $fixtags = array(
 334                         # french spaces, last one Guillemet-left
 335                         # only if there is something before the space
 336                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 337                         # french spaces, Guillemet-right
 338                         '/(\\302\\253) /' => '\\1&nbsp;',
 339                         '/&nbsp;(!\s*important)/' => ' \\1', #Beware of CSS magic word !important, bug #11874.
 340                 );
 341                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 342
 343                 $text = $this->doBlockLevels( $text, $linestart );
 344
 345                 $this->replaceLinkHolders( $text );
 346
 347                 # the position of the parserConvert() call should not be changed. it
 348                 # assumes that the links are all replaced and the only thing left
 349                 # is the <nowiki> mark.
 350                 # Side-effects: this calls $this->mOutput->setTitleText()
 351                 $text = $wgContLang->parserConvert( $text, $this );
 352
 353                 $text = $this->mStripState->unstripNoWiki( $text );
 354
 355                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 356
 357 //!JF Move to its own function
 358
 359                 $uniq_prefix = $this->mUniqPrefix;
 360                 $matches = array();
 361                 $elements = array_keys( $this->mTransparentTagHooks );
 362                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 363
 364                 foreach( $matches as $marker => $data ) {
 365                         list( $element, $content, $params, $tag ) = $data;
 366                         $tagName = strtolower( $element );
 367                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 368                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 369                                         array( $content, $params, $this ) );
 370                         } else {
 371                                 $output = $tag;
 372                         }
 373                         $this->mStripState->general->setPair( $marker, $output );
 374                 }
 375                 $text = $this->mStripState->unstripGeneral( $text );
 376
 377                 $text = Sanitizer::normalizeCharReferences( $text );
 378
 379                 if ( ( $wgUseTidy && $this->mOptions->mTidy ) || $wgAlwaysUseTidy ) {
 380                         $text = MWTidy::tidy( $text );
 381                 } else {
 382                         # attempt to sanitize at least some nesting problems
 383                         # (bug #2702 and quite a few others)
 384                         $tidyregs = array(
 385                                 # ''Something [http://www.cool.com cool''] -->
 386                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 387                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 388                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 389                                 # fix up an anchor inside another anchor, only
 390                                 # at least for a single single nested link (bug 3695)
 391                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 392                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 393                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 394                                 # contains a div, so fix it up here; replace
 395                                 # div with escaped text
 396                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 397                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 398                                 # remove empty italic or bold tag pairs, some
 399                                 # introduced by rules above
 400                                 '/<([bi])><\/\\1>/' => '',
 401                         );
 402
 403                         $text = preg_replace(
 404                                 array_keys( $tidyregs ),
 405                                 array_values( $tidyregs ),
 406                                 $text );
 407                 }
 408                 global $wgExpensiveParserFunctionLimit;
 409                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 410                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 411                 }
 412
 413                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 414
 415                 # Information on include size limits, for the benefit of users who try to skirt them
 416                 if ( $this->mOptions->getEnableLimitReport() ) {
 417                         global $wgExpensiveParserFunctionLimit;
 418                         $max = $this->mOptions->getMaxIncludeSize();
 419                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 420                         $limitReport =
 421                                 "NewPP limit report\n" .
 422                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 423                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 424                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 425                                 $PFreport;
 426                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 427                         $text .= "\n<!-- \n$limitReport-->\n";
 428                 }
 429                 $this->mOutput->setText( $text );
 430                 $this->mRevisionId = $oldRevisionId;
 431                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 432                 wfProfileOut( $fname );
 433                 wfProfileOut( __METHOD__ );
 434
 435                 return $this->mOutput;
 436         }
 437
 438         /**
 439          * Recursive parser entry point that can be called from an extension tag
 440          * hook.
 441          */
 442         function recursiveTagParse( $text ) {
 443                 wfProfileIn( __METHOD__ );
 444                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 445                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 446                 $text = $this->internalParse( $text, false );
 447                 wfProfileOut( __METHOD__ );
 448                 return $text;
 449         }
 450
 451         /**
 452          * Expand templates and variables in the text, producing valid, static wikitext.
 453          * Also removes comments.
 454          */
 455         function preprocess( $text, $title, $options, $revid = null ) {
 456                 wfProfileIn( __METHOD__ );
 457                 $this->clearState();
 458                 $this->setOutputType( self::OT_PREPROCESS );
 459                 $this->mOptions = $options;
 460                 $this->setTitle( $title );
 461                 if( $revid !== null ) {
 462                         $this->mRevisionId = $revid;
 463                 }
 464                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 465                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 466                 $text = $this->replaceVariables( $text );
 467                 $text = $this->mStripState->unstripBoth( $text );
 468                 wfProfileOut( __METHOD__ );
 469                 return $text;
 470         }
 471
 472         /**
 473          * Get a random string
 474          *
 475          * @private
 476          * @static
 477          */
 478         function getRandomString() {
 479                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 480         }
 481
 482         function &getTitle() { return $this->mTitle; }
 483         function getOptions() { return $this->mOptions; }
 484         function getRevisionId() { return $this->mRevisionId; }
 485         function getOutput() { return $this->mOutput; }
 486         function nextLinkID() { return $this->mLinkID++; }
 487
 488         function getFunctionLang() {
 489                 global $wgLang, $wgContLang;
 490
 491                 $target = $this->mOptions->getTargetLanguage();
 492                 if ( $target !== null ) {
 493                         return $target;
 494                 } else {
 495                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 496                 }
 497         }
 498
 499         /**
 500          * Get a preprocessor object
 501          */
 502         function getPreprocessor() {
 503                 if ( !isset( $this->mPreprocessor ) ) {
 504                         $class = $this->mPreprocessorClass;
 505                         $this->mPreprocessor = new $class( $this );
 506                 }
 507                 return $this->mPreprocessor;
 508         }
 509
 510         /**
 511          * Replaces all occurrences of HTML-style comments and the given tags
 512          * in the text with a random marker and returns the next text. The output
 513          * parameter $matches will be an associative array filled with data in
 514          * the form:
 515          *   'UNIQ-xxxxx' => array(
 516          *     'element',
 517          *     'tag content',
 518          *     array( 'param' => 'x' ),
 519          *     '<element param="x">tag content</element>' ) )
 520          *
 521          * @param $elements list of element names. Comments are always extracted.
 522          * @param $text Source text string.
 523          * @param $uniq_prefix
 524          *
 525          * @public
 526          * @static
 527          */
 528         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 529                 static $n = 1;
 530                 $stripped = '';
 531                 $matches = array();
 532
 533                 $taglist = implode( '|', $elements );
 534                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 535
 536                 while ( '' != $text ) {
 537                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 538                         $stripped .= $p[0];
 539                         if( count( $p ) < 5 ) {
 540                                 break;
 541                         }
 542                         if( count( $p ) > 5 ) {
 543                                 // comment
 544                                 $element    = $p[4];
 545                                 $attributes = '';
 546                                 $close      = '';
 547                                 $inside     = $p[5];
 548                         } else {
 549                                 // tag
 550                                 $element    = $p[1];
 551                                 $attributes = $p[2];
 552                                 $close      = $p[3];
 553                                 $inside     = $p[4];
 554                         }
 555
 556                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . self::MARKER_SUFFIX;
 557                         $stripped .= $marker;
 558
 559                         if ( $close === '/>' ) {
 560                                 // Empty element tag, <tag />
 561                                 $content = null;
 562                                 $text = $inside;
 563                                 $tail = null;
 564                         } else {
 565                                 if( $element === '!--' ) {
 566                                         $end = '/(-->)/';
 567                                 } else {
 568                                         $end = "/(<\\/$element\\s*>)/i";
 569                                 }
 570                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 571                                 $content = $q[0];
 572                                 if( count( $q ) < 3 ) {
 573                                         # No end tag -- let it run out to the end of the text.
 574                                         $tail = '';
 575                                         $text = '';
 576                                 } else {
 577                                         $tail = $q[1];
 578                                         $text = $q[2];
 579                                 }
 580                         }
 581
 582                         $matches[$marker] = array( $element,
 583                                 $content,
 584                                 Sanitizer::decodeTagAttributes( $attributes ),
 585                                 "<$element$attributes$close$content$tail" );
 586                 }
 587                 return $stripped;
 588         }
 589
 590         /**
 591          * Get a list of strippable XML-like elements
 592          */
 593         function getStripList() {
 594                 global $wgRawHtml;
 595                 $elements = $this->mStripList;
 596                 if( $wgRawHtml ) {
 597                         $elements[] = 'html';
 598                 }
 599                 if( $this->mOptions->getUseTeX() ) {
 600                         $elements[] = 'math';
 601                 }
 602                 return $elements;
 603         }
 604
 605         /**
 606          * @deprecated use replaceVariables
 607          */
 608         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 609                 return $text;
 610         }
 611
 612         /**
 613          * Restores pre, math, and other extensions removed by strip()
 614          *
 615          * always call unstripNoWiki() after this one
 616          * @private
 617          * @deprecated use $this->mStripState->unstrip()
 618          */
 619         function unstrip( $text, $state ) {
 620                 return $state->unstripGeneral( $text );
 621         }
 622
 623         /**
 624          * Always call this after unstrip() to preserve the order
 625          *
 626          * @private
 627          * @deprecated use $this->mStripState->unstrip()
 628          */
 629         function unstripNoWiki( $text, $state ) {
 630                 return $state->unstripNoWiki( $text );
 631         }
 632
 633         /**
 634          * @deprecated use $this->mStripState->unstripBoth()
 635          */
 636         function unstripForHTML( $text ) {
 637                 return $this->mStripState->unstripBoth( $text );
 638         }
 639
 640         /**
 641          * Add an item to the strip state
 642          * Returns the unique tag which must be inserted into the stripped text
 643          * The tag will be replaced with the original text in unstrip()
 644          *
 645          * @private
 646          */
 647         function insertStripItem( $text ) {
 648                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 649                 $this->mMarkerIndex++;
 650                 $this->mStripState->general->setPair( $rnd, $text );
 651                 return $rnd;
 652         }
 653
 654         /**
 655          * Interface with html tidy
 656          * @deprecated Use MWTidy::tidy()
 657          */
 658         public static function tidy( $text ) {
 659                 wfDeprecated( __METHOD__ );
 660                 return MWTidy::tidy( $text );
 661         }
 662
 663         /**
 664          * parse the wiki syntax used to render tables
 665          *
 666          * @private
 667          */
 668         function doTableStuff ( $text ) {
 669                 wfProfileIn( __METHOD__ );
 670
 671                 $lines = StringUtils::explode( "\n", $text );
 672                 $out = '';
 673                 $td_history = array (); // Is currently a td tag open?
 674                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 675                 $tr_history = array (); // Is currently a tr tag open?
 676                 $tr_attributes = array (); // history of tr attributes
 677                 $has_opened_tr = array(); // Did this table open a <tr> element?
 678                 $indent_level = 0; // indent level of the table
 679
 680                 foreach ( $lines as $outLine ) {
 681                         $line = trim( $outLine );
 682
 683                         if( $line == '' ) { // empty line, go to next line
 684                                 $out .= $outLine."\n";
 685                                 continue;
 686                         }
 687                         $first_character = $line[0];
 688                         $matches = array();
 689
 690                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 691                                 // First check if we are starting a new table
 692                                 $indent_level = strlen( $matches[1] );
 693
 694                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 695                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 696
 697                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 698                                 array_push ( $td_history , false );
 699                                 array_push ( $last_tag_history , '' );
 700                                 array_push ( $tr_history , false );
 701                                 array_push ( $tr_attributes , '' );
 702                                 array_push ( $has_opened_tr , false );
 703                         } else if ( count ( $td_history ) == 0 ) {
 704                                 // Don't do any of the following
 705                                 $out .= $outLine."\n";
 706                                 continue;
 707                         } else if ( substr ( $line , 0 , 2 ) === '|}' ) {
 708                                 // We are ending a table
 709                                 $line = '</table>' . substr ( $line , 2 );
 710                                 $last_tag = array_pop ( $last_tag_history );
 711
 712                                 if ( !array_pop ( $has_opened_tr ) ) {
 713                                         $line = "<tr><td></td></tr>{$line}";
 714                                 }
 715
 716                                 if ( array_pop ( $tr_history ) ) {
 717                                         $line = "</tr>{$line}";
 718                                 }
 719
 720                                 if ( array_pop ( $td_history ) ) {
 721                                         $line = "</{$last_tag}>{$line}";
 722                                 }
 723                                 array_pop ( $tr_attributes );
 724                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 725                         } else if ( substr ( $line , 0 , 2 ) === '|-' ) {
 726                                 // Now we have a table row
 727                                 $line = preg_replace( '#^\|-+#', '', $line );
 728
 729                                 // Whats after the tag is now only attributes
 730                                 $attributes = $this->mStripState->unstripBoth( $line );
 731                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 732                                 array_pop ( $tr_attributes );
 733                                 array_push ( $tr_attributes , $attributes );
 734
 735                                 $line = '';
 736                                 $last_tag = array_pop ( $last_tag_history );
 737                                 array_pop ( $has_opened_tr );
 738                                 array_push ( $has_opened_tr , true );
 739
 740                                 if ( array_pop ( $tr_history ) ) {
 741                                         $line = '</tr>';
 742                                 }
 743
 744                                 if ( array_pop ( $td_history ) ) {
 745                                         $line = "</{$last_tag}>{$line}";
 746                                 }
 747
 748                                 $outLine = $line;
 749                                 array_push ( $tr_history , false );
 750                                 array_push ( $td_history , false );
 751                                 array_push ( $last_tag_history , '' );
 752                         }
 753                         else if ( $first_character === '|' || $first_character === '!' || substr ( $line , 0 , 2 )  === '|+' ) {
 754                                 // This might be cell elements, td, th or captions
 755                                 if ( substr ( $line , 0 , 2 ) === '|+' ) {
 756                                         $first_character = '+';
 757                                         $line = substr ( $line , 1 );
 758                                 }
 759
 760                                 $line = substr ( $line , 1 );
 761
 762                                 if ( $first_character === '!' ) {
 763                                         $line = str_replace ( '!!' , '||' , $line );
 764                                 }
 765
 766                                 // Split up multiple cells on the same line.
 767                                 // FIXME : This can result in improper nesting of tags processed
 768                                 // by earlier parser steps, but should avoid splitting up eg
 769                                 // attribute values containing literal "||".
 770                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 771
 772                                 $outLine = '';
 773
 774                                 // Loop through each table cell
 775                                 foreach ( $cells as $cell )
 776                                 {
 777                                         $previous = '';
 778                                         if ( $first_character !== '+' )
 779                                         {
 780                                                 $tr_after = array_pop ( $tr_attributes );
 781                                                 if ( !array_pop ( $tr_history ) ) {
 782                                                         $previous = "<tr{$tr_after}>\n";
 783                                                 }
 784                                                 array_push ( $tr_history , true );
 785                                                 array_push ( $tr_attributes , '' );
 786                                                 array_pop ( $has_opened_tr );
 787                                                 array_push ( $has_opened_tr , true );
 788                                         }
 789
 790                                         $last_tag = array_pop ( $last_tag_history );
 791
 792                                         if ( array_pop ( $td_history ) ) {
 793                                                 $previous = "</{$last_tag}>{$previous}";
 794                                         }
 795
 796                                         if ( $first_character === '|' ) {
 797                                                 $last_tag = 'td';
 798                                         } else if ( $first_character === '!' ) {
 799                                                 $last_tag = 'th';
 800                                         } else if ( $first_character === '+' ) {
 801                                                 $last_tag = 'caption';
 802                                         } else {
 803                                                 $last_tag = '';
 804                                         }
 805
 806                                         array_push ( $last_tag_history , $last_tag );
 807
 808                                         // A cell could contain both parameters and data
 809                                         $cell_data = explode ( '|' , $cell , 2 );
 810
 811                                         // Bug 553: Note that a '|' inside an invalid link should not
 812                                         // be mistaken as delimiting cell parameters
 813                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 814                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 815                                         } else if ( count ( $cell_data ) == 1 )
 816                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 817                                         else {
 818                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 819                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 820                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 821                                         }
 822
 823                                         $outLine .= $cell;
 824                                         array_push ( $td_history , true );
 825                                 }
 826                         }
 827                         $out .= $outLine . "\n";
 828                 }
 829
 830                 // Closing open td, tr && table
 831                 while ( count ( $td_history ) > 0 )
 832                 {
 833                         if ( array_pop ( $td_history ) ) {
 834                                 $out .= "</td>\n";
 835                         }
 836                         if ( array_pop ( $tr_history ) ) {
 837                                 $out .= "</tr>\n";
 838                         }
 839                         if ( !array_pop ( $has_opened_tr ) ) {
 840                                 $out .= "<tr><td></td></tr>\n" ;
 841                         }
 842
 843                         $out .= "</table>\n";
 844                 }
 845
 846                 // Remove trailing line-ending (b/c)
 847                 if ( substr( $out, -1 ) === "\n" ) {
 848                         $out = substr( $out, 0, -1 );
 849                 }
 850
 851                 // special case: don't return empty table
 852                 if( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 853                         $out = '';
 854                 }
 855
 856                 wfProfileOut( __METHOD__ );
 857
 858                 return $out;
 859         }
 860
 861         /**
 862          * Helper function for parse() that transforms wiki markup into
 863          * HTML. Only called for $mOutputType == self::OT_HTML.
 864          *
 865          * @private
 866          */
 867         function internalParse( $text, $isMain = true ) {
 868                 wfProfileIn( __METHOD__ );
 869
 870                 $origText = $text;
 871
 872                 # Hook to suspend the parser in this state
 873                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 874                         wfProfileOut( __METHOD__ );
 875                         return $text ;
 876                 }
 877
 878                 $text = $this->replaceVariables( $text );
 879                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 880                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 881
 882                 // Tables need to come after variable replacement for things to work
 883                 // properly; putting them before other transformations should keep
 884                 // exciting things like link expansions from showing up in surprising
 885                 // places.
 886                 $text = $this->doTableStuff( $text );
 887
 888                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 889
 890                 $text = $this->doDoubleUnderscore( $text );
 891                 $text = $this->doHeadings( $text );
 892                 if( $this->mOptions->getUseDynamicDates() ) {
 893                         $df = DateFormatter::getInstance();
 894                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 895                 }
 896                 $text = $this->doAllQuotes( $text );
 897                 $text = $this->replaceInternalLinks( $text );
 898                 $text = $this->replaceExternalLinks( $text );
 899
 900                 # replaceInternalLinks may sometimes leave behind
 901                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 902                 $text = str_replace($this->mUniqPrefix.'NOPARSE', '', $text);
 903
 904                 $text = $this->doMagicLinks( $text );
 905                 $text = $this->formatHeadings( $text, $origText, $isMain );
 906
 907                 wfProfileOut( __METHOD__ );
 908                 return $text;
 909         }
 910
 911         /**
 912          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 913          * magic external links.
 914          *
 915          * DML
 916          * @private
 917          */
 918         function doMagicLinks( $text ) {
 919                 wfProfileIn( __METHOD__ );
 920                 $prots = $this->mUrlProtocols;
 921                 $urlChar = self::EXT_LINK_URL_CLASS;
 922                 $text = preg_replace_callback(
 923                         '!(?:                           # Start cases
 924                                 (<a.*?</a>) |               # m[1]: Skip link text
 925                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
 926                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
 927                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
 928                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
 929                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 930                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 931                                     [0-9Xx]                 # check digit
 932                                     \b)
 933                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 934                 wfProfileOut( __METHOD__ );
 935                 return $text;
 936         }
 937
 938         function magicLinkCallback( $m ) {
 939                 if ( isset( $m[1] ) && $m[1] !== '' ) {
 940                         # Skip anchor
 941                         return $m[0];
 942                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
 943                         # Skip HTML element
 944                         return $m[0];
 945                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
 946                         # Free external link
 947                         return $this->makeFreeExternalLink( $m[0] );
 948                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
 949                         # RFC or PMID
 950                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
 951                                 $keyword = 'RFC';
 952                                 $urlmsg = 'rfcurl';
 953                                 $id = $m[4];
 954                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
 955                                 $keyword = 'PMID';
 956                                 $urlmsg = 'pubmedurl';
 957                                 $id = $m[4];
 958                         } else {
 959                                 throw new MWException( __METHOD__.': unrecognised match type "' .
 960                                         substr($m[0], 0, 20 ) . '"' );
 961                         }
 962                         $url = wfMsg( $urlmsg, $id);
 963                         $sk = $this->mOptions->getSkin();
 964                         $la = $sk->getExternalLinkAttributes();
 965                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
 966                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
 967                         # ISBN
 968                         $isbn = $m[5];
 969                         $num = strtr( $isbn, array(
 970                                 '-' => '',
 971                                 ' ' => '',
 972                                 'x' => 'X',
 973                         ));
 974                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
 975                         return'<a href="' .
 976                                 $titleObj->escapeLocalUrl() .
 977                                 "\" class=\"internal\">ISBN $isbn</a>";
 978                 } else {
 979                         return $m[0];
 980                 }
 981         }
 982
 983         /**
 984          * Make a free external link, given a user-supplied URL
 985          * @return HTML
 986          * @private
 987          */
 988         function makeFreeExternalLink( $url ) {
 989                 global $wgContLang;
 990                 wfProfileIn( __METHOD__ );
 991
 992                 $sk = $this->mOptions->getSkin();
 993                 $trail = '';
 994
 995                 # The characters '<' and '>' (which were escaped by
 996                 # removeHTMLtags()) should not be included in
 997                 # URLs, per RFC 2396.
 998                 $m2 = array();
 999                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1000                         $trail = substr($url, $m2[0][1]) . $trail;
1001                         $url = substr($url, 0, $m2[0][1]);
1002                 }
1003
1004                 # Move trailing punctuation to $trail
1005                 $sep = ',;\.:!?';
1006                 # If there is no left bracket, then consider right brackets fair game too
1007                 if ( strpos( $url, '(' ) === false ) {
1008                         $sep .= ')';
1009                 }
1010
1011                 $numSepChars = strspn( strrev( $url ), $sep );
1012                 if ( $numSepChars ) {
1013                         $trail = substr( $url, -$numSepChars ) . $trail;
1014                         $url = substr( $url, 0, -$numSepChars );
1015                 }
1016
1017                 $url = Sanitizer::cleanUrl( $url );
1018
1019                 # Is this an external image?
1020                 $text = $this->maybeMakeExternalImage( $url );
1021                 if ( $text === false ) {
1022                         # Not an image, make a link
1023                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1024                                 $this->getExternalLinkAttribs( $url ) );
1025                         # Register it in the output object...
1026                         # Replace unnecessary URL escape codes with their equivalent characters
1027                         $pasteurized = self::replaceUnusualEscapes( $url );
1028                         $this->mOutput->addExternalLink( $pasteurized );
1029                 }
1030                 wfProfileOut( __METHOD__ );
1031                 return $text . $trail;
1032         }
1033
1034
1035         /**
1036          * Parse headers and return html
1037          *
1038          * @private
1039          */
1040         function doHeadings( $text ) {
1041                 wfProfileIn( __METHOD__ );
1042                 for ( $i = 6; $i >= 1; --$i ) {
1043                         $h = str_repeat( '=', $i );
1044                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1045                           "<h$i>\\1</h$i>", $text );
1046                 }
1047                 wfProfileOut( __METHOD__ );
1048                 return $text;
1049         }
1050
1051         /**
1052          * Replace single quotes with HTML markup
1053          * @private
1054          * @return string the altered text
1055          */
1056         function doAllQuotes( $text ) {
1057                 wfProfileIn( __METHOD__ );
1058                 $outtext = '';
1059                 $lines = StringUtils::explode( "\n", $text );
1060                 foreach ( $lines as $line ) {
1061                         $outtext .= $this->doQuotes( $line ) . "\n";
1062                 }
1063                 $outtext = substr($outtext, 0,-1);
1064                 wfProfileOut( __METHOD__ );
1065                 return $outtext;
1066         }
1067
1068         /**
1069          * Helper function for doAllQuotes()
1070          */
1071         public function doQuotes( $text ) {
1072                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1073                 if ( count( $arr ) == 1 )
1074                         return $text;
1075                 else
1076                 {
1077                         # First, do some preliminary work. This may shift some apostrophes from
1078                         # being mark-up to being text. It also counts the number of occurrences
1079                         # of bold and italics mark-ups.
1080                         $i = 0;
1081                         $numbold = 0;
1082                         $numitalics = 0;
1083                         foreach ( $arr as $r )
1084                         {
1085                                 if ( ( $i % 2 ) == 1 )
1086                                 {
1087                                         # If there are ever four apostrophes, assume the first is supposed to
1088                                         # be text, and the remaining three constitute mark-up for bold text.
1089                                         if ( strlen( $arr[$i] ) == 4 )
1090                                         {
1091                                                 $arr[$i-1] .= "'";
1092                                                 $arr[$i] = "'''";
1093                                         }
1094                                         # If there are more than 5 apostrophes in a row, assume they're all
1095                                         # text except for the last 5.
1096                                         else if ( strlen( $arr[$i] ) > 5 )
1097                                         {
1098                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1099                                                 $arr[$i] = "'''''";
1100                                         }
1101                                         # Count the number of occurrences of bold and italics mark-ups.
1102                                         # We are not counting sequences of five apostrophes.
1103                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1104                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1105                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1106                                 }
1107                                 $i++;
1108                         }
1109
1110                         # If there is an odd number of both bold and italics, it is likely
1111                         # that one of the bold ones was meant to be an apostrophe followed
1112                         # by italics. Which one we cannot know for certain, but it is more
1113                         # likely to be one that has a single-letter word before it.
1114                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1115                         {
1116                                 $i = 0;
1117                                 $firstsingleletterword = -1;
1118                                 $firstmultiletterword = -1;
1119                                 $firstspace = -1;
1120                                 foreach ( $arr as $r )
1121                                 {
1122                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1123                                         {
1124                                                 $x1 = substr ($arr[$i-1], -1);
1125                                                 $x2 = substr ($arr[$i-1], -2, 1);
1126                                                 if ($x1 === ' ') {
1127                                                         if ($firstspace == -1) $firstspace = $i;
1128                                                 } else if ($x2 === ' ') {
1129                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1130                                                 } else {
1131                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1132                                                 }
1133                                         }
1134                                         $i++;
1135                                 }
1136
1137                                 # If there is a single-letter word, use it!
1138                                 if ($firstsingleletterword > -1)
1139                                 {
1140                                         $arr [ $firstsingleletterword ] = "''";
1141                                         $arr [ $firstsingleletterword-1 ] .= "'";
1142                                 }
1143                                 # If not, but there's a multi-letter word, use that one.
1144                                 else if ($firstmultiletterword > -1)
1145                                 {
1146                                         $arr [ $firstmultiletterword ] = "''";
1147                                         $arr [ $firstmultiletterword-1 ] .= "'";
1148                                 }
1149                                 # ... otherwise use the first one that has neither.
1150                                 # (notice that it is possible for all three to be -1 if, for example,
1151                                 # there is only one pentuple-apostrophe in the line)
1152                                 else if ($firstspace > -1)
1153                                 {
1154                                         $arr [ $firstspace ] = "''";
1155                                         $arr [ $firstspace-1 ] .= "'";
1156                                 }
1157                         }
1158
1159                         # Now let's actually convert our apostrophic mush to HTML!
1160                         $output = '';
1161                         $buffer = '';
1162                         $state = '';
1163                         $i = 0;
1164                         foreach ($arr as $r)
1165                         {
1166                                 if (($i % 2) == 0)
1167                                 {
1168                                         if ($state === 'both')
1169                                                 $buffer .= $r;
1170                                         else
1171                                                 $output .= $r;
1172                                 }
1173                                 else
1174                                 {
1175                                         if (strlen ($r) == 2)
1176                                         {
1177                                                 if ($state === 'i')
1178                                                 { $output .= '</i>'; $state = ''; }
1179                                                 else if ($state === 'bi')
1180                                                 { $output .= '</i>'; $state = 'b'; }
1181                                                 else if ($state === 'ib')
1182                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1183                                                 else if ($state === 'both')
1184                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1185                                                 else # $state can be 'b' or ''
1186                                                 { $output .= '<i>'; $state .= 'i'; }
1187                                         }
1188                                         else if (strlen ($r) == 3)
1189                                         {
1190                                                 if ($state === 'b')
1191                                                 { $output .= '</b>'; $state = ''; }
1192                                                 else if ($state === 'bi')
1193                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1194                                                 else if ($state === 'ib')
1195                                                 { $output .= '</b>'; $state = 'i'; }
1196                                                 else if ($state === 'both')
1197                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1198                                                 else # $state can be 'i' or ''
1199                                                 { $output .= '<b>'; $state .= 'b'; }
1200                                         }
1201                                         else if (strlen ($r) == 5)
1202                                         {
1203                                                 if ($state === 'b')
1204                                                 { $output .= '</b><i>'; $state = 'i'; }
1205                                                 else if ($state === 'i')
1206                                                 { $output .= '</i><b>'; $state = 'b'; }
1207                                                 else if ($state === 'bi')
1208                                                 { $output .= '</i></b>'; $state = ''; }
1209                                                 else if ($state === 'ib')
1210                                                 { $output .= '</b></i>'; $state = ''; }
1211                                                 else if ($state === 'both')
1212                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1213                                                 else # ($state == '')
1214                                                 { $buffer = ''; $state = 'both'; }
1215                                         }
1216                                 }
1217                                 $i++;
1218                         }
1219                         # Now close all remaining tags.  Notice that the order is important.
1220                         if ($state === 'b' || $state === 'ib')
1221                                 $output .= '</b>';
1222                         if ($state === 'i' || $state === 'bi' || $state === 'ib')
1223                                 $output .= '</i>';
1224                         if ($state === 'bi')
1225                                 $output .= '</b>';
1226                         # There might be lonely ''''', so make sure we have a buffer
1227                         if ($state === 'both' && $buffer)
1228                                 $output .= '<b><i>'.$buffer.'</i></b>';
1229                         return $output;
1230                 }
1231         }
1232
1233         /**
1234          * Replace external links (REL)
1235          *
1236          * Note: this is all very hackish and the order of execution matters a lot.
1237          * Make sure to run maintenance/parserTests.php if you change this code.
1238          *
1239          * @private
1240          */
1241         function replaceExternalLinks( $text ) {
1242                 global $wgContLang;
1243                 wfProfileIn( __METHOD__ );
1244
1245                 $sk = $this->mOptions->getSkin();
1246
1247                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1248                 $s = array_shift( $bits );
1249
1250                 $i = 0;
1251                 while ( $i<count( $bits ) ) {
1252                         $url = $bits[$i++];
1253                         $protocol = $bits[$i++];
1254                         $text = $bits[$i++];
1255                         $trail = $bits[$i++];
1256
1257                         # The characters '<' and '>' (which were escaped by
1258                         # removeHTMLtags()) should not be included in
1259                         # URLs, per RFC 2396.
1260                         $m2 = array();
1261                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1262                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1263                                 $url = substr($url, 0, $m2[0][1]);
1264                         }
1265
1266                         # If the link text is an image URL, replace it with an <img> tag
1267                         # This happened by accident in the original parser, but some people used it extensively
1268                         $img = $this->maybeMakeExternalImage( $text );
1269                         if ( $img !== false ) {
1270                                 $text = $img;
1271                         }
1272
1273                         $dtrail = '';
1274
1275                         # Set linktype for CSS - if URL==text, link is essentially free
1276                         $linktype = ($text === $url) ? 'free' : 'text';
1277
1278                         # No link text, e.g. [http://domain.tld/some.link]
1279                         if ( $text == '' ) {
1280                                 # Autonumber if allowed. See bug #5918
1281                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1282                                         $langObj = $this->getFunctionLang();
1283                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1284                                         $linktype = 'autonumber';
1285                                 } else {
1286                                         # Otherwise just use the URL
1287                                         $text = htmlspecialchars( $url );
1288                                         $linktype = 'free';
1289                                 }
1290                         } else {
1291                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1292                                 # Check for trail
1293                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1294                         }
1295
1296                         $text = $wgContLang->markNoConversion($text);
1297
1298                         $url = Sanitizer::cleanUrl( $url );
1299
1300                         # Use the encoded URL
1301                         # This means that users can paste URLs directly into the text
1302                         # Funny characters like &ouml; aren't valid in URLs anyway
1303                         # This was changed in August 2004
1304                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1305                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1306
1307                         # Register link in the output object.
1308                         # Replace unnecessary URL escape codes with the referenced character
1309                         # This prevents spammers from hiding links from the filters
1310                         $pasteurized = self::replaceUnusualEscapes( $url );
1311                         $this->mOutput->addExternalLink( $pasteurized );
1312                 }
1313
1314                 wfProfileOut( __METHOD__ );
1315                 return $s;
1316         }
1317
1318         /**
1319          * Get an associative array of additional HTML attributes appropriate for a
1320          * particular external link.  This currently may include rel => nofollow
1321          * (depending on configuration, namespace, and the URL's domain) and/or a
1322          * target attribute (depending on configuration).
1323          *
1324          * @param string $url Optional URL, to extract the domain from for rel =>
1325          *   nofollow if appropriate
1326          * @return array Associative array of HTML attributes
1327          */
1328         function getExternalLinkAttribs( $url = false ) {
1329                 $attribs = array();
1330                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1331                 $ns = $this->mTitle->getNamespace();
1332                 if( $wgNoFollowLinks && !in_array($ns, $wgNoFollowNsExceptions) ) {
1333                         $attribs['rel'] = 'nofollow';
1334
1335                         global $wgNoFollowDomainExceptions;
1336                         if ( $wgNoFollowDomainExceptions ) {
1337                                 $bits = wfParseUrl( $url );
1338                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1339                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1340                                                 if( substr( $bits['host'], -strlen( $domain ) )
1341                                                 == $domain ) {
1342                                                         unset( $attribs['rel'] );
1343                                                         break;
1344                                                 }
1345                                         }
1346                                 }
1347                         }
1348                 }
1349                 if ( $this->mOptions->getExternalLinkTarget() ) {
1350                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1351                 }
1352                 return $attribs;
1353         }
1354
1355
1356         /**
1357          * Replace unusual URL escape codes with their equivalent characters
1358          * @param string
1359          * @return string
1360          * @static
1361          * @todo  This can merge genuinely required bits in the path or query string,
1362          *        breaking legit URLs. A proper fix would treat the various parts of
1363          *        the URL differently; as a workaround, just use the output for
1364          *        statistical records, not for actual linking/output.
1365          */
1366         static function replaceUnusualEscapes( $url ) {
1367                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1368                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1369         }
1370
1371         /**
1372          * Callback function used in replaceUnusualEscapes().
1373          * Replaces unusual URL escape codes with their equivalent character
1374          * @static
1375          * @private
1376          */
1377         private static function replaceUnusualEscapesCallback( $matches ) {
1378                 $char = urldecode( $matches[0] );
1379                 $ord = ord( $char );
1380                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1381                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1382                         // No, shouldn't be escaped
1383                         return $char;
1384                 } else {
1385                         // Yes, leave it escaped
1386                         return $matches[0];
1387                 }
1388         }
1389
1390         /**
1391          * make an image if it's allowed, either through the global
1392          * option, through the exception, or through the on-wiki whitelist
1393          * @private
1394          */
1395         function maybeMakeExternalImage( $url ) {
1396                 $sk = $this->mOptions->getSkin();
1397                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1398                 $imagesexception = !empty($imagesfrom);
1399                 $text = false;
1400                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1401                 if( $imagesexception && is_array( $imagesfrom ) ) {
1402                         $imagematch = false;
1403                         foreach( $imagesfrom as $match ) {
1404                                 if( strpos( $url, $match ) === 0 ) {
1405                                         $imagematch = true;
1406                                         break;
1407                                 }
1408                         }
1409                 } elseif( $imagesexception ) {
1410                         $imagematch = (strpos( $url, $imagesfrom ) === 0);
1411                 } else {
1412                         $imagematch = false;
1413                 }
1414                 if ( $this->mOptions->getAllowExternalImages()
1415                      || ( $imagesexception && $imagematch ) ) {
1416                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1417                                 # Image found
1418                                 $text = $sk->makeExternalImage( $url );
1419                         }
1420                 }
1421                 if( !$text && $this->mOptions->getEnableImageWhitelist()
1422                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1423                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1424                         foreach( $whitelist as $entry ) {
1425                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1426                                 if( strpos( $entry, '#' ) === 0 || $entry === '' )
1427                                         continue;
1428                                 if( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1429                                         # Image matches a whitelist entry
1430                                         $text = $sk->makeExternalImage( $url );
1431                                         break;
1432                                 }
1433                         }
1434                 }
1435                 return $text;
1436         }
1437
1438         /**
1439          * Process [[ ]] wikilinks
1440          * @return processed text
1441          *
1442          * @private
1443          */
1444         function replaceInternalLinks( $s ) {
1445                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1446                 return $s;
1447         }
1448
1449         /**
1450          * Process [[ ]] wikilinks (RIL)
1451          * @return LinkHolderArray
1452          *
1453          * @private
1454          */
1455         function replaceInternalLinks2( &$s ) {
1456                 global $wgContLang;
1457
1458                 wfProfileIn( __METHOD__ );
1459
1460                 wfProfileIn( __METHOD__.'-setup' );
1461                 static $tc = FALSE, $e1, $e1_img;
1462                 # the % is needed to support urlencoded titles as well
1463                 if ( !$tc ) {
1464                         $tc = Title::legalChars() . '#%';
1465                         # Match a link having the form [[namespace:link|alternate]]trail
1466                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1467                         # Match cases where there is no "]]", which might still be images
1468                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1469                 }
1470
1471                 $sk = $this->mOptions->getSkin();
1472                 $holders = new LinkHolderArray( $this );
1473
1474                 #split the entire text string on occurences of [[
1475                 $a = StringUtils::explode( '[[', ' ' . $s );
1476                 #get the first element (all text up to first [[), and remove the space we added
1477                 $s = $a->current();
1478                 $a->next();
1479                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1480                 $s = substr( $s, 1 );
1481
1482                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1483                 $e2 = null;
1484                 if ( $useLinkPrefixExtension ) {
1485                         # Match the end of a line for a word that's not followed by whitespace,
1486                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1487                         $e2 = wfMsgForContent( 'linkprefix' );
1488                 }
1489
1490                 if( is_null( $this->mTitle ) ) {
1491                         wfProfileOut( __METHOD__.'-setup' );
1492                         wfProfileOut( __METHOD__ );
1493                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1494                 }
1495                 $nottalk = !$this->mTitle->isTalkPage();
1496
1497                 if ( $useLinkPrefixExtension ) {
1498                         $m = array();
1499                         if ( preg_match( $e2, $s, $m ) ) {
1500                                 $first_prefix = $m[2];
1501                         } else {
1502                                 $first_prefix = false;
1503                         }
1504                 } else {
1505                         $prefix = '';
1506                 }
1507
1508                 if($wgContLang->hasVariants()) {
1509                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1510                 } else {
1511                         $selflink = array($this->mTitle->getPrefixedText());
1512                 }
1513                 $useSubpages = $this->areSubpagesAllowed();
1514                 wfProfileOut( __METHOD__.'-setup' );
1515
1516                 # Loop for each link
1517                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1518                         # Check for excessive memory usage
1519                         if ( $holders->isBig() ) {
1520                                 # Too big
1521                                 # Do the existence check, replace the link holders and clear the array
1522                                 $holders->replace( $s );
1523                                 $holders->clear();
1524                         }
1525
1526                         if ( $useLinkPrefixExtension ) {
1527                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1528                                 if ( preg_match( $e2, $s, $m ) ) {
1529                                         $prefix = $m[2];
1530                                         $s = $m[1];
1531                                 } else {
1532                                         $prefix='';
1533                                 }
1534                                 # first link
1535                                 if($first_prefix) {
1536                                         $prefix = $first_prefix;
1537                                         $first_prefix = false;
1538                                 }
1539                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1540                         }
1541
1542                         $might_be_img = false;
1543
1544                         wfProfileIn( __METHOD__."-e1" );
1545                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1546                                 $text = $m[2];
1547                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1548                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1549                                 # the real problem is with the $e1 regex
1550                                 # See bug 1300.
1551                                 #
1552                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1553                                 # and no image is in sight. See bug 2095.
1554                                 #
1555                                 if( $text !== '' &&
1556                                         substr( $m[3], 0, 1 ) === ']' &&
1557                                         strpos($text, '[') !== false
1558                                 )
1559                                 {
1560                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1561                                         $m[3] = substr( $m[3], 1 );
1562                                 }
1563                                 # fix up urlencoded title texts
1564                                 if( strpos( $m[1], '%' ) !== false ) {
1565                                         # Should anchors '#' also be rejected?
1566                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1567                                 }
1568                                 $trail = $m[3];
1569                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1570                                 $might_be_img = true;
1571                                 $text = $m[2];
1572                                 if ( strpos( $m[1], '%' ) !== false ) {
1573                                         $m[1] = urldecode($m[1]);
1574                                 }
1575                                 $trail = "";
1576                         } else { # Invalid form; output directly
1577                                 $s .= $prefix . '[[' . $line ;
1578                                 wfProfileOut( __METHOD__."-e1" );
1579                                 continue;
1580                         }
1581                         wfProfileOut( __METHOD__."-e1" );
1582                         wfProfileIn( __METHOD__."-misc" );
1583
1584                         # Don't allow internal links to pages containing
1585                         # PROTO: where PROTO is a valid URL protocol; these
1586                         # should be external links.
1587                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1588                                 $s .= $prefix . '[[' . $line ;
1589                                 wfProfileOut( __METHOD__."-misc" );
1590                                 continue;
1591                         }
1592
1593                         # Make subpage if necessary
1594                         if ( $useSubpages ) {
1595                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1596                         } else {
1597                                 $link = $m[1];
1598                         }
1599
1600                         $noforce = (substr( $m[1], 0, 1 ) !== ':');
1601                         if (!$noforce) {
1602                                 # Strip off leading ':'
1603                                 $link = substr( $link, 1 );
1604                         }
1605
1606                         wfProfileOut( __METHOD__."-misc" );
1607                         wfProfileIn( __METHOD__."-title" );
1608                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1609                         if ( $nt === NULL ) {
1610                                 $s .= $prefix . '[[' . $line;
1611                                 wfProfileOut( __METHOD__."-title" );
1612                                 continue;
1613                         }
1614
1615                         $ns = $nt->getNamespace();
1616                         $iw = $nt->getInterWiki();
1617                         wfProfileOut( __METHOD__."-title" );
1618
1619                         if ( $might_be_img ) { # if this is actually an invalid link
1620                                 wfProfileIn( __METHOD__."-might_be_img" );
1621                                 if ( $ns == NS_FILE && $noforce ) { #but might be an image
1622                                         $found = false;
1623                                         while ( true ) {
1624                                                 #look at the next 'line' to see if we can close it there
1625                                                 $a->next();
1626                                                 $next_line = $a->current();
1627                                                 if ( $next_line === false || $next_line === null ) {
1628                                                         break;
1629                                                 }
1630                                                 $m = explode( ']]', $next_line, 3 );
1631                                                 if ( count( $m ) == 3 ) {
1632                                                         # the first ]] closes the inner link, the second the image
1633                                                         $found = true;
1634                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1635                                                         $trail = $m[2];
1636                                                         break;
1637                                                 } elseif ( count( $m ) == 2 ) {
1638                                                         #if there's exactly one ]] that's fine, we'll keep looking
1639                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1640                                                 } else {
1641                                                         #if $next_line is invalid too, we need look no further
1642                                                         $text .= '[[' . $next_line;
1643                                                         break;
1644                                                 }
1645                                         }
1646                                         if ( !$found ) {
1647                                                 # we couldn't find the end of this imageLink, so output it raw
1648                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1649                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1650                                                 $s .= "{$prefix}[[$link|$text";
1651                                                 # note: no $trail, because without an end, there *is* no trail
1652                                                 wfProfileOut( __METHOD__."-might_be_img" );
1653                                                 continue;
1654                                         }
1655                                 } else { #it's not an image, so output it raw
1656                                         $s .= "{$prefix}[[$link|$text";
1657                                         # note: no $trail, because without an end, there *is* no trail
1658                                         wfProfileOut( __METHOD__."-might_be_img" );
1659                                         continue;
1660                                 }
1661                                 wfProfileOut( __METHOD__."-might_be_img" );
1662                         }
1663
1664                         $wasblank = ( '' == $text );
1665                         if ( $wasblank ) $text = $link;
1666
1667                         # Link not escaped by : , create the various objects
1668                         if ( $noforce ) {
1669
1670                                 # Interwikis
1671                                 wfProfileIn( __METHOD__."-interwiki" );
1672                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1673                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1674                                         $s = rtrim($s . $prefix);
1675                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1676                                         wfProfileOut( __METHOD__."-interwiki" );
1677                                         continue;
1678                                 }
1679                                 wfProfileOut( __METHOD__."-interwiki" );
1680
1681                                 if ( $ns == NS_FILE ) {
1682                                         wfProfileIn( __METHOD__."-image" );
1683                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1684                                                 if ( $wasblank ) {
1685                                                         # if no parameters were passed, $text
1686                                                         # becomes something like "File:Foo.png",
1687                                                         # which we don't want to pass on to the
1688                                                         # image generator
1689                                                         $text = '';
1690                                                 } else {
1691                                                         # recursively parse links inside the image caption
1692                                                         # actually, this will parse them in any other parameters, too,
1693                                                         # but it might be hard to fix that, and it doesn't matter ATM
1694                                                         $text = $this->replaceExternalLinks($text);
1695                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1696                                                 }
1697                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1698                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1699                                         }
1700                                         $this->mOutput->addImage( $nt->getDBkey() );
1701                                         wfProfileOut( __METHOD__."-image" );
1702                                         continue;
1703
1704                                 }
1705
1706                                 if ( $ns == NS_CATEGORY ) {
1707                                         wfProfileIn( __METHOD__."-category" );
1708                                         $s = rtrim($s . "\n"); # bug 87
1709
1710                                         if ( $wasblank ) {
1711                                                 $sortkey = $this->getDefaultSort();
1712                                         } else {
1713                                                 $sortkey = $text;
1714                                         }
1715                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1716                                         $sortkey = str_replace( "\n", '', $sortkey );
1717                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1718                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1719
1720                                         /**
1721                                          * Strip the whitespace Category links produce, see bug 87
1722                                          * @todo We might want to use trim($tmp, "\n") here.
1723                                          */
1724                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1725
1726                                         wfProfileOut( __METHOD__."-category" );
1727                                         continue;
1728                                 }
1729                         }
1730
1731                         # Self-link checking
1732                         if( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1733                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1734                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1735                                         continue;
1736                                 }
1737                         }
1738
1739                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1740                         # FIXME: Should do batch file existence checks, see comment below
1741                         if( $ns == NS_MEDIA ) {
1742                                 wfProfileIn( __METHOD__."-media" );
1743                                 # Give extensions a chance to select the file revision for us
1744                                 $skip = $time = false;
1745                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1746                                 if ( $skip ) {
1747                                         $link = $sk->link( $nt );
1748                                 } else {
1749                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1750                                 }
1751                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1752                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1753                                 $this->mOutput->addImage( $nt->getDBkey() );
1754                                 wfProfileOut( __METHOD__."-media" );
1755                                 continue;
1756                         }
1757
1758                         wfProfileIn( __METHOD__."-always_known" );
1759                         # Some titles, such as valid special pages or files in foreign repos, should
1760                         # be shown as bluelinks even though they're not included in the page table
1761                         #
1762                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1763                         # batch file existence checks for NS_FILE and NS_MEDIA
1764                         if( $iw == '' && $nt->isAlwaysKnown() ) {
1765                                 $this->mOutput->addLink( $nt );
1766                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1767                         } else {
1768                                 # Links will be added to the output link list after checking
1769                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1770                         }
1771                         wfProfileOut( __METHOD__."-always_known" );
1772                 }
1773                 wfProfileOut( __METHOD__ );
1774                 return $holders;
1775         }
1776
1777         /**
1778          * Make a link placeholder. The text returned can be later resolved to a real link with
1779          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1780          * parsing of interwiki links, and secondly to allow all existence checks and
1781          * article length checks (for stub links) to be bundled into a single query.
1782          *
1783          * @deprecated
1784          */
1785         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1786                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1787         }
1788
1789         /**
1790          * Render a forced-blue link inline; protect against double expansion of
1791          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1792          * Since this little disaster has to split off the trail text to avoid
1793          * breaking URLs in the following text without breaking trails on the
1794          * wiki links, it's been made into a horrible function.
1795          *
1796          * @param Title $nt
1797          * @param string $text
1798          * @param string $query
1799          * @param string $trail
1800          * @param string $prefix
1801          * @return string HTML-wikitext mix oh yuck
1802          */
1803         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1804                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1805                 $sk = $this->mOptions->getSkin();
1806                 // FIXME: use link() instead of deprecated makeKnownLinkObj()
1807                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1808                 return $this->armorLinks( $link ) . $trail;
1809         }
1810
1811         /**
1812          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1813          * going to go through further parsing steps before inline URL expansion.
1814          *
1815          * Not needed quite as much as it used to be since free links are a bit
1816          * more sensible these days. But bracketed links are still an issue.
1817          *
1818          * @param string more-or-less HTML
1819          * @return string less-or-more HTML with NOPARSE bits
1820          */
1821         function armorLinks( $text ) {
1822                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1823                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1824         }
1825
1826         /**
1827          * Return true if subpage links should be expanded on this page.
1828          * @return bool
1829          */
1830         function areSubpagesAllowed() {
1831                 # Some namespaces don't allow subpages
1832                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1833         }
1834
1835         /**
1836          * Handle link to subpage if necessary
1837          * @param string $target the source of the link
1838          * @param string &$text the link text, modified as necessary
1839          * @return string the full name of the link
1840          * @private
1841          */
1842         function maybeDoSubpageLink($target, &$text) {
1843                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
1844         }
1845
1846         /**#@+
1847          * Used by doBlockLevels()
1848          * @private
1849          */
1850         /* private */ function closeParagraph() {
1851                 $result = '';
1852                 if ( '' != $this->mLastSection ) {
1853                         $result = '</' . $this->mLastSection  . ">\n";
1854                 }
1855                 $this->mInPre = false;
1856                 $this->mLastSection = '';
1857                 return $result;
1858         }
1859         # getCommon() returns the length of the longest common substring
1860         # of both arguments, starting at the beginning of both.
1861         #
1862         /* private */ function getCommon( $st1, $st2 ) {
1863                 $fl = strlen( $st1 );
1864                 $shorter = strlen( $st2 );
1865                 if ( $fl < $shorter ) { $shorter = $fl; }
1866
1867                 for ( $i = 0; $i < $shorter; ++$i ) {
1868                         if ( $st1{$i} != $st2{$i} ) { break; }
1869                 }
1870                 return $i;
1871         }
1872         # These next three functions open, continue, and close the list
1873         # element appropriate to the prefix character passed into them.
1874         #
1875         /* private */ function openList( $char ) {
1876                 $result = $this->closeParagraph();
1877
1878                 if ( '*' === $char ) { $result .= '<ul><li>'; }
1879                 elseif ( '#' === $char ) { $result .= '<ol><li>'; }
1880                 elseif ( ':' === $char ) { $result .= '<dl><dd>'; }
1881                 elseif ( ';' === $char ) {
1882                         $result .= '<dl><dt>';
1883                         $this->mDTopen = true;
1884                 } elseif ( '>' === $char ) { $result .= "<blockquote><p>"; }
1885                 else { $result = '<!-- ERR 1 -->'; }
1886
1887                 return $result;
1888         }
1889
1890         /* private */ function nextItem( $char ) {
1891                 if ( '*' === $char || '#' === $char ) { return '</li><li>'; }
1892                 elseif ( '>' === $char ) { return "</p><p>"; }
1893                 elseif ( ':' === $char || ';' === $char ) {
1894                         $close = '</dd>';
1895                         if ( $this->mDTopen ) { $close = '</dt>'; }
1896                         if ( ';' === $char ) {
1897                                 $this->mDTopen = true;
1898                                 return $close . '<dt>';
1899                         } else {
1900                                 $this->mDTopen = false;
1901                                 return $close . '<dd>';
1902                         }
1903                 }
1904                 return '<!-- ERR 2 -->';
1905         }
1906
1907         /* private */ function closeList( $char ) {
1908                 if ( '*' === $char ) { $text = '</li></ul>'; }
1909                 elseif ( '#' === $char ) { $text = '</li></ol>'; }
1910                 elseif ( '>' === $char ) { $text = "</p></blockquote>"; }
1911                 elseif ( ':' === $char ) {
1912                         if ( $this->mDTopen ) {
1913                                 $this->mDTopen = false;
1914                                 $text = '</dt></dl>';
1915                         } else {
1916                                 $text = '</dd></dl>';
1917                         }
1918                 }
1919                 else {  return '<!-- ERR 3 -->'; }
1920                 return $text."\n";
1921         }
1922         /**#@-*/
1923
1924         /**
1925          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
1926          *
1927          * @param $linestart bool whether or not this is at the start of a line.
1928          * @private
1929          * @return string the lists rendered as HTML
1930          */
1931         function doBlockLevels( $text, $linestart ) {
1932                 wfProfileIn( __METHOD__ );
1933
1934                 # Parsing through the text line by line.  The main thing
1935                 # happening here is handling of block-level elements p, pre,
1936                 # and making lists from lines starting with * # : etc.
1937                 #
1938                 $textLines = StringUtils::explode( "\n", $text );
1939
1940                 $lastPrefix = $output = '';
1941                 $this->mDTopen = $inBlockElem = false;
1942                 $prefixLength = 0;
1943                 $paragraphStack = false;
1944
1945                 foreach ( $textLines as $oLine ) {
1946                         # Fix up $linestart
1947                         if ( !$linestart ) {
1948                                 $output .= $oLine;
1949                                 $linestart = true;
1950                                 continue;
1951                         }
1952                         // * = ul
1953                         // # = ol
1954                         // ; = dt
1955                         // : = dd
1956                         // > = blockquote
1957
1958                         $lastPrefixLength = strlen( $lastPrefix );
1959                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1960                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1961
1962                         // Need to decode &gt; --> > for blockquote syntax. Re-encode later.
1963                         // To avoid collision with real >s, we temporarily convert them to &gt;
1964                         // This is a weird choice of armouring, but it's totally resistant to any
1965                         //  collision.
1966                         $orig = $oLine;
1967                         $oLine = strtr( $oLine, array( '&gt;' => '>', '>' => '&gt;' ) );
1968
1969                         // If not in a <pre> element, scan for and figure out what prefixes are there.
1970                         if ( !$this->mInPre ) {
1971                                 # Multiple prefixes may abut each other for nested lists.
1972                                 $prefixLength = strspn( $oLine, '*#:;>' );
1973                                 $prefix = substr( $oLine, 0, $prefixLength );
1974
1975                                 # eh?
1976                                 // ; and : are both from definition-lists, so they're equivalent
1977                                 //  for the purposes of determining whether or not we need to open/close
1978                                 //  elements.
1979                                 $prefix2 = str_replace( ';', ':', $prefix );
1980                                 $t = substr( $oLine, $prefixLength );
1981                                 $this->mInPre = (bool)$preOpenMatch;
1982                         } else {
1983                                 # Don't interpret any other prefixes in preformatted text
1984                                 $prefixLength = 0;
1985                                 $prefix = $prefix2 = '';
1986                                 $t = $oLine;
1987                         }
1988
1989                         // Re-encode >s now
1990                         $t = strtr( $t, array( '&gt;' => '>', '>' => '&gt;' ) );
1991
1992                         # List generation
1993                         if( $prefixLength && $lastPrefix === $prefix2 ) {
1994                                 # Same as the last item, so no need to deal with nesting or opening stuff
1995                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
1996                                 $paragraphStack = false;
1997
1998                                 if ( substr( $prefix, -1 ) === ';') {
1999                                         # The one nasty exception: definition lists work like this:
2000                                         # ; title : definition text
2001                                         # So we check for : in the remainder text to split up the
2002                                         # title and definition, without b0rking links.
2003                                         $term = $t2 = '';
2004                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2005                                                 $t = $t2;
2006                                                 $output .= $term . $this->nextItem( ':' );
2007                                         }
2008                                 }
2009                         } elseif( $prefixLength || $lastPrefixLength ) {
2010                                 // We need to open or close prefixes, or both.
2011
2012                                 # Either open or close a level...
2013                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2014                                 $paragraphStack = false;
2015
2016                                 // Close all the prefixes which aren't shared.
2017                                 while( $commonPrefixLength < $lastPrefixLength ) {
2018                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2019                                         --$lastPrefixLength;
2020                                 }
2021
2022                                 // Continue the current prefix if appropriate.
2023                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2024                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2025                                 }
2026
2027                                 // Open prefixes where appropriate.
2028                                 while ( $prefixLength > $commonPrefixLength ) {
2029                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2030                                         $output .= $this->openList( $char );
2031
2032                                         if ( ';' === $char ) {
2033                                                 # FIXME: This is dupe of code above
2034                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2035                                                         $t = $t2;
2036                                                         $output .= $term . $this->nextItem( ':' );
2037                                                 }
2038                                         }
2039                                         ++$commonPrefixLength;
2040                                 }
2041                                 $lastPrefix = $prefix2;
2042                         }
2043
2044                         // If we have no prefixes, go to paragraph mode.
2045                         if( 0 == $prefixLength ) {
2046                                 wfProfileIn( __METHOD__."-paragraph" );
2047                                 # No prefix (not in list)--go to paragraph mode
2048                                 // XXX: use a stack for nestable elements like span, table and div
2049                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2050                                 $closematch = preg_match(
2051                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2052                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2053                                 if ( $openmatch or $closematch ) {
2054                                         $paragraphStack = false;
2055                                         # TODO bug 5718: paragraph closed
2056                                         $output .= $this->closeParagraph();
2057                                         if ( $preOpenMatch and !$preCloseMatch ) {
2058                                                 $this->mInPre = true;
2059                                         }
2060                                         if ( $closematch ) {
2061                                                 $inBlockElem = false;
2062                                         } else {
2063                                                 $inBlockElem = true;
2064                                         }
2065                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2066                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' or trim($t) != '' ) ) {
2067                                                 // pre
2068                                                 if ($this->mLastSection !== 'pre') {
2069                                                         $paragraphStack = false;
2070                                                         $output .= $this->closeParagraph().'<pre>';
2071                                                         $this->mLastSection = 'pre';
2072                                                 }
2073                                                 $t = substr( $t, 1 );
2074                                         } else {
2075                                                 // paragraph
2076                                                 if ( '' == trim($t) ) {
2077                                                         if ( $paragraphStack ) {
2078                                                                 $output .= $paragraphStack.'<br />';
2079                                                                 $paragraphStack = false;
2080                                                                 $this->mLastSection = 'p';
2081                                                         } else {
2082                                                                 if ($this->mLastSection !== 'p' ) {
2083                                                                         $output .= $this->closeParagraph();
2084                                                                         $this->mLastSection = '';
2085                                                                         $paragraphStack = '<p>';
2086                                                                 } else {
2087                                                                         $paragraphStack = '</p><p>';
2088                                                                 }
2089                                                         }
2090                                                 } else {
2091                                                         if ( $paragraphStack ) {
2092                                                                 $output .= $paragraphStack;
2093                                                                 $paragraphStack = false;
2094                                                                 $this->mLastSection = 'p';
2095                                                         } else if ($this->mLastSection !== 'p') {
2096                                                                 $output .= $this->closeParagraph().'<p>';
2097                                                                 $this->mLastSection = 'p';
2098                                                         }
2099                                                 }
2100                                         }
2101                                 }
2102                                 wfProfileOut( __METHOD__."-paragraph" );
2103                         }
2104                         // somewhere above we forget to get out of pre block (bug 785)
2105                         if($preCloseMatch && $this->mInPre) {
2106                                 $this->mInPre = false;
2107                         }
2108                         if ($paragraphStack === false) {
2109                                 $output .= $t."\n";
2110                         }
2111                 }
2112                 while ( $prefixLength ) {
2113                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2114                         --$prefixLength;
2115                 }
2116                 if ( '' != $this->mLastSection ) {
2117                         $output .= '</' . $this->mLastSection . '>';
2118                         $this->mLastSection = '';
2119                 }
2120
2121                 wfProfileOut( __METHOD__ );
2122                 return $output;
2123         }
2124
2125         /**
2126          * Split up a string on ':', ignoring any occurences inside tags
2127          * to prevent illegal overlapping.
2128          * @param string $str the string to split
2129          * @param string &$before set to everything before the ':'
2130          * @param string &$after set to everything after the ':'
2131          * return string the position of the ':', or false if none found
2132          */
2133         function findColonNoLinks($str, &$before, &$after) {
2134                 wfProfileIn( __METHOD__ );
2135
2136                 $pos = strpos( $str, ':' );
2137                 if( $pos === false ) {
2138                         // Nothing to find!
2139                         wfProfileOut( __METHOD__ );
2140                         return false;
2141                 }
2142
2143                 $lt = strpos( $str, '<' );
2144                 if( $lt === false || $lt > $pos ) {
2145                         // Easy; no tag nesting to worry about
2146                         $before = substr( $str, 0, $pos );
2147                         $after = substr( $str, $pos+1 );
2148                         wfProfileOut( __METHOD__ );
2149                         return $pos;
2150                 }
2151
2152                 // Ugly state machine to walk through avoiding tags.
2153                 $state = self::COLON_STATE_TEXT;
2154                 $stack = 0;
2155                 $len = strlen( $str );
2156                 for( $i = 0; $i < $len; $i++ ) {
2157                         $c = $str{$i};
2158
2159                         switch( $state ) {
2160                         // (Using the number is a performance hack for common cases)
2161                         case 0: // self::COLON_STATE_TEXT:
2162                                 switch( $c ) {
2163                                 case "<":
2164                                         // Could be either a <start> tag or an </end> tag
2165                                         $state = self::COLON_STATE_TAGSTART;
2166                                         break;
2167                                 case ":":
2168                                         if( $stack == 0 ) {
2169                                                 // We found it!
2170                                                 $before = substr( $str, 0, $i );
2171                                                 $after = substr( $str, $i + 1 );
2172                                                 wfProfileOut( __METHOD__ );
2173                                                 return $i;
2174                                         }
2175                                         // Embedded in a tag; don't break it.
2176                                         break;
2177                                 default:
2178                                         // Skip ahead looking for something interesting
2179                                         $colon = strpos( $str, ':', $i );
2180                                         if( $colon === false ) {
2181                                                 // Nothing else interesting
2182                                                 wfProfileOut( __METHOD__ );
2183                                                 return false;
2184                                         }
2185                                         $lt = strpos( $str, '<', $i );
2186                                         if( $stack === 0 ) {
2187                                                 if( $lt === false || $colon < $lt ) {
2188                                                         // We found it!
2189                                                         $before = substr( $str, 0, $colon );
2190                                                         $after = substr( $str, $colon + 1 );
2191                                                         wfProfileOut( __METHOD__ );
2192                                                         return $i;
2193                                                 }
2194                                         }
2195                                         if( $lt === false ) {
2196                                                 // Nothing else interesting to find; abort!
2197                                                 // We're nested, but there's no close tags left. Abort!
2198                                                 break 2;
2199                                         }
2200                                         // Skip ahead to next tag start
2201                                         $i = $lt;
2202                                         $state = self::COLON_STATE_TAGSTART;
2203                                 }
2204                                 break;
2205                         case 1: // self::COLON_STATE_TAG:
2206                                 // In a <tag>
2207                                 switch( $c ) {
2208                                 case ">":
2209                                         $stack++;
2210                                         $state = self::COLON_STATE_TEXT;
2211                                         break;
2212                                 case "/":
2213                                         // Slash may be followed by >?
2214                                         $state = self::COLON_STATE_TAGSLASH;
2215                                         break;
2216                                 default:
2217                                         // ignore
2218                                 }
2219                                 break;
2220                         case 2: // self::COLON_STATE_TAGSTART:
2221                                 switch( $c ) {
2222                                 case "/":
2223                                         $state = self::COLON_STATE_CLOSETAG;
2224                                         break;
2225                                 case "!":
2226                                         $state = self::COLON_STATE_COMMENT;
2227                                         break;
2228                                 case ">":
2229                                         // Illegal early close? This shouldn't happen D:
2230                                         $state = self::COLON_STATE_TEXT;
2231                                         break;
2232                                 default:
2233                                         $state = self::COLON_STATE_TAG;
2234                                 }
2235                                 break;
2236                         case 3: // self::COLON_STATE_CLOSETAG:
2237                                 // In a </tag>
2238                                 if( $c === ">" ) {
2239                                         $stack--;
2240                                         if( $stack < 0 ) {
2241                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2242                                                 wfProfileOut( __METHOD__ );
2243                                                 return false;
2244                                         }
2245                                         $state = self::COLON_STATE_TEXT;
2246                                 }
2247                                 break;
2248                         case self::COLON_STATE_TAGSLASH:
2249                                 if( $c === ">" ) {
2250                                         // Yes, a self-closed tag <blah/>
2251                                         $state = self::COLON_STATE_TEXT;
2252                                 } else {
2253                                         // Probably we're jumping the gun, and this is an attribute
2254                                         $state = self::COLON_STATE_TAG;
2255                                 }
2256                                 break;
2257                         case 5: // self::COLON_STATE_COMMENT:
2258                                 if( $c === "-" ) {
2259                                         $state = self::COLON_STATE_COMMENTDASH;
2260                                 }
2261                                 break;
2262                         case self::COLON_STATE_COMMENTDASH:
2263                                 if( $c === "-" ) {
2264                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2265                                 } else {
2266                                         $state = self::COLON_STATE_COMMENT;
2267                                 }
2268                                 break;
2269                         case self::COLON_STATE_COMMENTDASHDASH:
2270                                 if( $c === ">" ) {
2271                                         $state = self::COLON_STATE_TEXT;
2272                                 } else {
2273                                         $state = self::COLON_STATE_COMMENT;
2274                                 }
2275                                 break;
2276                         default:
2277                                 throw new MWException( "State machine error in " . __METHOD__ );
2278                         }
2279                 }
2280                 if( $stack > 0 ) {
2281                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2282                         return false;
2283                 }
2284                 wfProfileOut( __METHOD__ );
2285                 return false;
2286         }
2287
2288         /**
2289          * Return value of a magic variable (like PAGENAME)
2290          *
2291          * @private
2292          */
2293         function getVariableValue( $index ) {
2294                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2295
2296                 /**
2297                  * Some of these require message or data lookups and can be
2298                  * expensive to check many times.
2299                  */
2300                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2301                         if ( isset( $this->mVarCache[$index] ) ) {
2302                                 return $this->mVarCache[$index];
2303                         }
2304                 }
2305
2306                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2307                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2308
2309                 # Use the time zone
2310                 global $wgLocaltimezone;
2311                 if ( isset( $wgLocaltimezone ) ) {
2312                         $oldtz = getenv( 'TZ' );
2313                         putenv( 'TZ='.$wgLocaltimezone );
2314                 }
2315
2316                 wfSuppressWarnings(); // E_STRICT system time bitching
2317                 $localTimestamp = date( 'YmdHis', $ts );
2318                 $localMonth = date( 'm', $ts );
2319                 $localMonth1 = date( 'n', $ts );
2320                 $localMonthName = date( 'n', $ts );
2321                 $localDay = date( 'j', $ts );
2322                 $localDay2 = date( 'd', $ts );
2323                 $localDayOfWeek = date( 'w', $ts );
2324                 $localWeek = date( 'W', $ts );
2325                 $localYear = date( 'Y', $ts );
2326                 $localHour = date( 'H', $ts );
2327                 if ( isset( $wgLocaltimezone ) ) {
2328                         putenv( 'TZ='.$oldtz );
2329                 }
2330                 wfRestoreWarnings();
2331
2332                 switch ( $index ) {
2333                         case 'currentmonth':
2334                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2335                         case 'currentmonth1':
2336                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2337                         case 'currentmonthname':
2338                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2339                         case 'currentmonthnamegen':
2340                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2341                         case 'currentmonthabbrev':
2342                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2343                         case 'currentday':
2344                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2345                         case 'currentday2':
2346                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2347                         case 'localmonth':
2348                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localMonth );
2349                         case 'localmonth1':
2350                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localMonth1 );
2351                         case 'localmonthname':
2352                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( $localMonthName );
2353                         case 'localmonthnamegen':
2354                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2355                         case 'localmonthabbrev':
2356                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2357                         case 'localday':
2358                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay );
2359                         case 'localday2':
2360                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay2 );
2361                         case 'pagename':
2362                                 return wfEscapeWikiText( $this->mTitle->getText() );
2363                         case 'pagenamee':
2364                                 return $this->mTitle->getPartialURL();
2365                         case 'fullpagename':
2366                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2367                         case 'fullpagenamee':
2368                                 return $this->mTitle->getPrefixedURL();
2369                         case 'subpagename':
2370                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2371                         case 'subpagenamee':
2372                                 return $this->mTitle->getSubpageUrlForm();
2373                         case 'basepagename':
2374                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2375                         case 'basepagenamee':
2376                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2377                         case 'talkpagename':
2378                                 if( $this->mTitle->canTalk() ) {
2379                                         $talkPage = $this->mTitle->getTalkPage();
2380                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2381                                 } else {
2382                                         return '';
2383                                 }
2384                         case 'talkpagenamee':
2385                                 if( $this->mTitle->canTalk() ) {
2386                                         $talkPage = $this->mTitle->getTalkPage();
2387                                         return $talkPage->getPrefixedUrl();
2388                                 } else {
2389                                         return '';
2390                                 }
2391                         case 'subjectpagename':
2392                                 $subjPage = $this->mTitle->getSubjectPage();
2393                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2394                         case 'subjectpagenamee':
2395                                 $subjPage = $this->mTitle->getSubjectPage();
2396                                 return $subjPage->getPrefixedUrl();
2397                         case 'revisionid':
2398                                 // Let the edit saving system know we should parse the page
2399                                 // *after* a revision ID has been assigned.
2400                                 $this->mOutput->setFlag( 'vary-revision' );
2401                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2402                                 return $this->mRevisionId;
2403                         case 'revisionday':
2404                                 // Let the edit saving system know we should parse the page
2405                                 // *after* a revision ID has been assigned. This is for null edits.
2406                                 $this->mOutput->setFlag( 'vary-revision' );
2407                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2408                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2409                         case 'revisionday2':
2410                                 // Let the edit saving system know we should parse the page
2411                                 // *after* a revision ID has been assigned. This is for null edits.
2412                                 $this->mOutput->setFlag( 'vary-revision' );
2413                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2414                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2415                         case 'revisionmonth':
2416                                 // Let the edit saving system know we should parse the page
2417                                 // *after* a revision ID has been assigned. This is for null edits.
2418                                 $this->mOutput->setFlag( 'vary-revision' );
2419                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2420                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2421                         case 'revisionyear':
2422                                 // Let the edit saving system know we should parse the page
2423                                 // *after* a revision ID has been assigned. This is for null edits.
2424                                 $this->mOutput->setFlag( 'vary-revision' );
2425                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2426                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2427                         case 'revisiontimestamp':
2428                                 // Let the edit saving system know we should parse the page
2429                                 // *after* a revision ID has been assigned. This is for null edits.
2430                                 $this->mOutput->setFlag( 'vary-revision' );
2431                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2432                                 return $this->getRevisionTimestamp();
2433                         case 'revisionuser':
2434                                 // Let the edit saving system know we should parse the page
2435                                 // *after* a revision ID has been assigned. This is for null edits.
2436                                 $this->mOutput->setFlag( 'vary-revision' );
2437                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2438                                 return $this->getRevisionUser();
2439                         case 'namespace':
2440                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2441                         case 'namespacee':
2442                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2443                         case 'talkspace':
2444                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2445                         case 'talkspacee':
2446                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2447                         case 'subjectspace':
2448                                 return $this->mTitle->getSubjectNsText();
2449                         case 'subjectspacee':
2450                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2451                         case 'currentdayname':
2452                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2453                         case 'currentyear':
2454                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2455                         case 'currenttime':
2456                                 return $this->mVarCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2457                         case 'currenthour':
2458                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2459                         case 'currentweek':
2460                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2461                                 // int to remove the padding
2462                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2463                         case 'currentdow':
2464                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2465                         case 'localdayname':
2466                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2467                         case 'localyear':
2468                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localYear, true );
2469                         case 'localtime':
2470                                 return $this->mVarCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2471                         case 'localhour':
2472                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localHour, true );
2473                         case 'localweek':
2474                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2475                                 // int to remove the padding
2476                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2477                         case 'localdow':
2478                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2479                         case 'numberofarticles':
2480                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2481                         case 'numberoffiles':
2482                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2483                         case 'numberofusers':
2484                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2485                         case 'numberofactiveusers':
2486                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::activeUsers() );
2487                         case 'numberofpages':
2488                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2489                         case 'numberofadmins':
2490                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::numberingroup('sysop') );
2491                         case 'numberofedits':
2492                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
2493                         case 'numberofviews':
2494                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::views() );
2495                         case 'currenttimestamp':
2496                                 return $this->mVarCache[$index] = wfTimestamp( TS_MW, $ts );
2497                         case 'localtimestamp':
2498                                 return $this->mVarCache[$index] = $localTimestamp;
2499                         case 'currentversion':
2500                                 return $this->mVarCache[$index] = SpecialVersion::getVersion();
2501                         case 'sitename':
2502                                 return $wgSitename;
2503                         case 'server':
2504                                 return $wgServer;
2505                         case 'servername':
2506                                 return $wgServerName;
2507                         case 'scriptpath':
2508                                 return $wgScriptPath;
2509                         case 'directionmark':
2510                                 return $wgContLang->getDirMark();
2511                         case 'contentlanguage':
2512                                 global $wgContLanguageCode;
2513                                 return $wgContLanguageCode;
2514                         default:
2515                                 $ret = null;
2516                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret ) ) )
2517                                         return $ret;
2518                                 else
2519                                         return null;
2520                 }
2521         }
2522
2523         /**
2524          * initialise the magic variables (like CURRENTMONTHNAME)
2525          *
2526          * @private
2527          */
2528         function initialiseVariables() {
2529                 wfProfileIn( __METHOD__ );
2530                 $variableIDs = MagicWord::getVariableIDs();
2531
2532                 $this->mVariables = new MagicWordArray( $variableIDs );
2533                 wfProfileOut( __METHOD__ );
2534         }
2535
2536         /**
2537          * Preprocess some wikitext and return the document tree.
2538          * This is the ghost of replace_variables().
2539          *
2540          * @param string $text The text to parse
2541          * @param integer flags Bitwise combination of:
2542          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2543          *                                     included. Default is to assume a direct page view.
2544          *
2545          * The generated DOM tree must depend only on the input text and the flags.
2546          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2547          *
2548          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2549          * change in the DOM tree for a given text, must be passed through the section identifier
2550          * in the section edit link and thus back to extractSections().
2551          *
2552          * The output of this function is currently only cached in process memory, but a persistent
2553          * cache may be implemented at a later date which takes further advantage of these strict
2554          * dependency requirements.
2555          *
2556          * @private
2557          */
2558         function preprocessToDom ( $text, $flags = 0 ) {
2559                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2560                 return $dom;
2561         }
2562
2563         /*
2564          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2565          */
2566         public static function splitWhitespace( $s ) {
2567                 $ltrimmed = ltrim( $s );
2568                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2569                 $trimmed = rtrim( $ltrimmed );
2570                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2571                 if ( $diff > 0 ) {
2572                         $w2 = substr( $ltrimmed, -$diff );
2573                 } else {
2574                         $w2 = '';
2575                 }
2576                 return array( $w1, $trimmed, $w2 );
2577         }
2578
2579         /**
2580          * Replace magic variables, templates, and template arguments
2581          * with the appropriate text. Templates are substituted recursively,
2582          * taking care to avoid infinite loops.
2583          *
2584          * Note that the substitution depends on value of $mOutputType:
2585          *  self::OT_WIKI: only {{subst:}} templates
2586          *  self::OT_PREPROCESS: templates but not extension tags
2587          *  self::OT_HTML: all templates and extension tags
2588          *
2589          * @param string $tex The text to transform
2590          * @param PPFrame $frame Object describing the arguments passed to the template.
2591          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2592          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2593          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2594          * @private
2595          */
2596         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2597                 # Is there any text? Also, Prevent too big inclusions!
2598                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2599                         return $text;
2600                 }
2601                 wfProfileIn( __METHOD__ );
2602
2603                 if ( $frame === false ) {
2604                         $frame = $this->getPreprocessor()->newFrame();
2605                 } elseif ( !( $frame instanceof PPFrame ) ) {
2606                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2607                         $frame = $this->getPreprocessor()->newCustomFrame($frame);
2608                 }
2609
2610                 $dom = $this->preprocessToDom( $text );
2611                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2612                 $text = $frame->expand( $dom, $flags );
2613
2614                 wfProfileOut( __METHOD__ );
2615                 return $text;
2616         }
2617
2618         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2619         static function createAssocArgs( $args ) {
2620                 $assocArgs = array();
2621                 $index = 1;
2622                 foreach( $args as $arg ) {
2623                         $eqpos = strpos( $arg, '=' );
2624                         if ( $eqpos === false ) {
2625                                 $assocArgs[$index++] = $arg;
2626                         } else {
2627                                 $name = trim( substr( $arg, 0, $eqpos ) );
2628                                 $value = trim( substr( $arg, $eqpos+1 ) );
2629                                 if ( $value === false ) {
2630                                         $value = '';
2631                                 }
2632                                 if ( $name !== false ) {
2633                                         $assocArgs[$name] = $value;
2634                                 }
2635                         }
2636                 }
2637
2638                 return $assocArgs;
2639         }
2640
2641         /**
2642          * Warn the user when a parser limitation is reached
2643          * Will warn at most once the user per limitation type
2644          *
2645          * @param string $limitationType, should be one of:
2646          *   'expensive-parserfunction' (corresponding messages: 'expensive-parserfunction-warning', 'expensive-parserfunction-category')
2647          *   'post-expand-template-argument' (corresponding messages: 'post-expand-template-argument-warning', 'post-expand-template-argument-category')
2648          *   'post-expand-template-inclusion' (corresponding messages: 'post-expand-template-inclusion-warning', 'post-expand-template-inclusion-category')
2649          * @params int $current, $max When an explicit limit has been
2650          *       exceeded, provide the values (optional)
2651          */
2652         function limitationWarn( $limitationType, $current=null, $max=null) {
2653                 $msgName = $limitationType . '-warning';
2654                 //does no harm if $current and $max are present but are unnecessary for the message
2655                 $warning = wfMsgExt( $msgName, array( 'parsemag', 'escape' ), $current, $max );
2656                 $this->mOutput->addWarning( $warning );
2657                 $cat = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( $limitationType . '-category' ) );
2658                 if ( $cat ) {
2659                         $this->mOutput->addCategory( $cat->getDBkey(), $this->getDefaultSort() );
2660                 }
2661         }
2662
2663         /**
2664          * Return the text of a template, after recursively
2665          * replacing any variables or templates within the template.
2666          *
2667          * @param array $piece The parts of the template
2668          *  $piece['title']: the title, i.e. the part before the |
2669          *  $piece['parts']: the parameter array
2670          *  $piece['lineStart']: whether the brace was at the start of a line
2671          * @param PPFrame The current frame, contains template arguments
2672          * @return string the text of the template
2673          * @private
2674          */
2675         function braceSubstitution( $piece, $frame ) {
2676                 global $wgContLang, $wgNonincludableNamespaces;
2677                 wfProfileIn( __METHOD__ );
2678                 wfProfileIn( __METHOD__.'-setup' );
2679
2680                 # Flags
2681                 $found = false;             # $text has been filled
2682                 $nowiki = false;            # wiki markup in $text should be escaped
2683                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2684                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2685                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2686                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2687
2688                 # Title object, where $text came from
2689                 $title = NULL;
2690
2691                 # $part1 is the bit before the first |, and must contain only title characters.
2692                 # Various prefixes will be stripped from it later.
2693                 $titleWithSpaces = $frame->expand( $piece['title'] );
2694                 $part1 = trim( $titleWithSpaces );
2695                 $titleText = false;
2696
2697                 # Original title text preserved for various purposes
2698                 $originalTitle = $part1;
2699
2700                 # $args is a list of argument nodes, starting from index 0, not including $part1
2701                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2702                 wfProfileOut( __METHOD__.'-setup' );
2703
2704                 # SUBST
2705                 wfProfileIn( __METHOD__.'-modifiers' );
2706                 if ( !$found ) {
2707                         $mwSubst = MagicWord::get( 'subst' );
2708                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2709                                 # One of two possibilities is true:
2710                                 # 1) Found SUBST but not in the PST phase
2711                                 # 2) Didn't find SUBST and in the PST phase
2712                                 # In either case, return without further processing
2713                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2714                                 $isLocalObj = true;
2715                                 $found = true;
2716                         }
2717                 }
2718
2719                 # Variables
2720                 if ( !$found && $args->getLength() == 0 ) {
2721                         $id = $this->mVariables->matchStartToEnd( $part1 );
2722                         if ( $id !== false ) {
2723                                 $text = $this->getVariableValue( $id );
2724                                 if (MagicWord::getCacheTTL($id)>-1)
2725                                         $this->mOutput->mContainsOldMagic = true;
2726                                 $found = true;
2727                         }
2728                 }
2729
2730                 # MSG, MSGNW and RAW
2731                 if ( !$found ) {
2732                         # Check for MSGNW:
2733                         $mwMsgnw = MagicWord::get( 'msgnw' );
2734                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2735                                 $nowiki = true;
2736                         } else {
2737                                 # Remove obsolete MSG:
2738                                 $mwMsg = MagicWord::get( 'msg' );
2739                                 $mwMsg->matchStartAndRemove( $part1 );
2740                         }
2741
2742                         # Check for RAW:
2743                         $mwRaw = MagicWord::get( 'raw' );
2744                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2745                                 $forceRawInterwiki = true;
2746                         }
2747                 }
2748                 wfProfileOut( __METHOD__.'-modifiers' );
2749
2750                 # Parser functions
2751                 if ( !$found ) {
2752                         wfProfileIn( __METHOD__ . '-pfunc' );
2753
2754                         $colonPos = strpos( $part1, ':' );
2755                         if ( $colonPos !== false ) {
2756                                 # Case sensitive functions
2757                                 $function = substr( $part1, 0, $colonPos );
2758                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2759                                         $function = $this->mFunctionSynonyms[1][$function];
2760                                 } else {
2761                                         # Case insensitive functions
2762                                         $function = strtolower( $function );
2763                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2764                                                 $function = $this->mFunctionSynonyms[0][$function];
2765                                         } else {
2766                                                 $function = false;
2767                                         }
2768                                 }
2769                                 if ( $function ) {
2770                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2771                                         $initialArgs = array( &$this );
2772                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2773                                         if ( $flags & SFH_OBJECT_ARGS ) {
2774                                                 # Add a frame parameter, and pass the arguments as an array
2775                                                 $allArgs = $initialArgs;
2776                                                 $allArgs[] = $frame;
2777                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2778                                                         $funcArgs[] = $args->item( $i );
2779                                                 }
2780                                                 $allArgs[] = $funcArgs;
2781                                         } else {
2782                                                 # Convert arguments to plain text
2783                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2784                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2785                                                 }
2786                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2787                                         }
2788
2789                                         # Workaround for PHP bug 35229 and similar
2790                                         if ( !is_callable( $callback ) ) {
2791                                                 wfProfileOut( __METHOD__ . '-pfunc' );
2792                                                 wfProfileOut( __METHOD__ );
2793                                                 throw new MWException( "Tag hook for $function is not callable\n" );
2794                                         }
2795                                         $result = call_user_func_array( $callback, $allArgs );
2796                                         $found = true;
2797                                         $noparse = true;
2798                                         $preprocessFlags = 0;
2799
2800                                         if ( is_array( $result ) ) {
2801                                                 if ( isset( $result[0] ) ) {
2802                                                         $text = $result[0];
2803                                                         unset( $result[0] );
2804                                                 }
2805
2806                                                 // Extract flags into the local scope
2807                                                 // This allows callers to set flags such as nowiki, found, etc.
2808                                                 extract( $result );
2809                                         } else {
2810                                                 $text = $result;
2811                                         }
2812                                         if ( !$noparse ) {
2813                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
2814                                                 $isChildObj = true;
2815                                         }
2816                                 }
2817                         }
2818                         wfProfileOut( __METHOD__ . '-pfunc' );
2819                 }
2820
2821                 # Finish mangling title and then check for loops.
2822                 # Set $title to a Title object and $titleText to the PDBK
2823                 if ( !$found ) {
2824                         $ns = NS_TEMPLATE;
2825                         # Split the title into page and subpage
2826                         $subpage = '';
2827                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2828                         if ($subpage !== '') {
2829                                 $ns = $this->mTitle->getNamespace();
2830                         }
2831                         $title = Title::newFromText( $part1, $ns );
2832                         if ( $title ) {
2833                                 $titleText = $title->getPrefixedText();
2834                                 # Check for language variants if the template is not found
2835                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
2836                                         $wgContLang->findVariantLink( $part1, $title, true );
2837                                 }
2838                                 # Do recursion depth check
2839                                 $limit = $this->mOptions->getMaxTemplateDepth();
2840                                 if ( $frame->depth >= $limit ) {
2841                                         $found = true;
2842                                         $text = '<span class="error">' . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit ) . '</span>';
2843                                 }
2844                         }
2845                 }
2846
2847                 # Load from database
2848                 if ( !$found && $title ) {
2849                         wfProfileIn( __METHOD__ . '-loadtpl' );
2850                         if ( !$title->isExternal() ) {
2851                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
2852                                         $text = SpecialPage::capturePath( $title );
2853                                         if ( is_string( $text ) ) {
2854                                                 $found = true;
2855                                                 $isHTML = true;
2856                                                 $this->disableCache();
2857                                         }
2858                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
2859                                         $found = false; //access denied
2860                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
2861                                 } else {
2862                                         list( $text, $title ) = $this->getTemplateDom( $title );
2863                                         if ( $text !== false ) {
2864                                                 $found = true;
2865                                                 $isChildObj = true;
2866                                         }
2867                                 }
2868
2869                                 # If the title is valid but undisplayable, make a link to it
2870                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2871                                         $text = "[[:$titleText]]";
2872                                         $found = true;
2873                                 }
2874                         } elseif ( $title->isTrans() ) {
2875                                 // Interwiki transclusion
2876                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
2877                                         $text = $this->interwikiTransclude( $title, 'render' );
2878                                         $isHTML = true;
2879                                 } else {
2880                                         $text = $this->interwikiTransclude( $title, 'raw' );
2881                                         // Preprocess it like a template
2882                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
2883                                         $isChildObj = true;
2884                                 }
2885                                 $found = true;
2886                         }
2887
2888                         # Do infinite loop check
2889                         # This has to be done after redirect resolution to avoid infinite loops via redirects
2890                         if ( !$frame->loopCheck( $title ) ) {
2891                                 $found = true;
2892                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
2893                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
2894                         }
2895                         wfProfileOut( __METHOD__ . '-loadtpl' );
2896                 }
2897
2898                 # If we haven't found text to substitute by now, we're done
2899                 # Recover the source wikitext and return it
2900                 if ( !$found ) {
2901                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2902                         wfProfileOut( __METHOD__ );
2903                         return array( 'object' => $text );
2904                 }
2905
2906                 # Expand DOM-style return values in a child frame
2907                 if ( $isChildObj ) {
2908                         # Clean up argument array
2909                         $newFrame = $frame->newChild( $args, $title );
2910
2911                         if ( $nowiki ) {
2912                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
2913                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
2914                                 # Expansion is eligible for the empty-frame cache
2915                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
2916                                         $text = $this->mTplExpandCache[$titleText];
2917                                 } else {
2918                                         $text = $newFrame->expand( $text );
2919                                         $this->mTplExpandCache[$titleText] = $text;
2920                                 }
2921                         } else {
2922                                 # Uncached expansion
2923                                 $text = $newFrame->expand( $text );
2924                         }
2925                 }
2926                 if ( $isLocalObj && $nowiki ) {
2927                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
2928                         $isLocalObj = false;
2929                 }
2930
2931                 # Replace raw HTML by a placeholder
2932                 # Add a blank line preceding, to prevent it from mucking up
2933                 # immediately preceding headings
2934                 if ( $isHTML ) {
2935                         $text = "\n\n" . $this->insertStripItem( $text );
2936                 }
2937                 # Escape nowiki-style return values
2938                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2939                         $text = wfEscapeWikiText( $text );
2940                 }
2941                 # Bug 529: if the template begins with a table or block-level
2942                 # element, it should be treated as beginning a new line.
2943                 # This behaviour is somewhat controversial.
2944                 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
2945                         $text = "\n" . $text;
2946                 }
2947
2948                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
2949                         # Error, oversize inclusion
2950                         $text = "[[$originalTitle]]" .
2951                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
2952                         $this->limitationWarn( 'post-expand-template-inclusion' );
2953                 }
2954
2955                 if ( $isLocalObj ) {
2956                         $ret = array( 'object' => $text );
2957                 } else {
2958                         $ret = array( 'text' => $text );
2959                 }
2960
2961                 wfProfileOut( __METHOD__ );
2962                 return $ret;
2963         }
2964
2965         /**
2966          * Get the semi-parsed DOM representation of a template with a given title,
2967          * and its redirect destination title. Cached.
2968          */
2969         function getTemplateDom( $title ) {
2970                 $cacheTitle = $title;
2971                 $titleText = $title->getPrefixedDBkey();
2972
2973                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
2974                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
2975                         $title = Title::makeTitle( $ns, $dbk );
2976                         $titleText = $title->getPrefixedDBkey();
2977                 }
2978                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
2979                         return array( $this->mTplDomCache[$titleText], $title );
2980                 }
2981
2982                 // Cache miss, go to the database
2983                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
2984
2985                 if ( $text === false ) {
2986                         $this->mTplDomCache[$titleText] = false;
2987                         return array( false, $title );
2988                 }
2989
2990                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
2991                 $this->mTplDomCache[ $titleText ] = $dom;
2992
2993                 if (! $title->equals($cacheTitle)) {
2994                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
2995                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
2996                 }
2997
2998                 return array( $dom, $title );
2999         }
3000
3001         /**
3002          * Fetch the unparsed text of a template and register a reference to it.
3003          */
3004         function fetchTemplateAndTitle( $title ) {
3005                 $templateCb = $this->mOptions->getTemplateCallback();
3006                 $stuff = call_user_func( $templateCb, $title, $this );
3007                 $text = $stuff['text'];
3008                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3009                 if ( isset( $stuff['deps'] ) ) {
3010                         foreach ( $stuff['deps'] as $dep ) {
3011                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3012                         }
3013                 }
3014                 return array($text,$finalTitle);
3015         }
3016
3017         function fetchTemplate( $title ) {
3018                 $rv = $this->fetchTemplateAndTitle($title);
3019                 return $rv[0];
3020         }
3021
3022         /**
3023          * Static function to get a template
3024          * Can be overridden via ParserOptions::setTemplateCallback().
3025          */
3026         static function statelessFetchTemplate( $title, $parser=false ) {
3027                 $text = $skip = false;
3028                 $finalTitle = $title;
3029                 $deps = array();
3030
3031                 // Loop to fetch the article, with up to 1 redirect
3032                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3033                         # Give extensions a chance to select the revision instead
3034                         $id = false; // Assume current
3035                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3036
3037                         if( $skip ) {
3038                                 $text = false;
3039                                 $deps[] = array(
3040                                         'title' => $title,
3041                                         'page_id' => $title->getArticleID(),
3042                                         'rev_id' => null );
3043                                 break;
3044                         }
3045                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3046                         $rev_id = $rev ? $rev->getId() : 0;
3047                         // If there is no current revision, there is no page
3048                         if( $id === false && !$rev ) {
3049                                 $linkCache = LinkCache::singleton();
3050                                 $linkCache->addBadLinkObj( $title );
3051                         }
3052
3053                         $deps[] = array(
3054                                 'title' => $title,
3055                                 'page_id' => $title->getArticleID(),
3056                                 'rev_id' => $rev_id );
3057
3058                         if( $rev ) {
3059                                 $text = $rev->getText();
3060                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3061                                 global $wgContLang;
3062                                 $message = $wgContLang->lcfirst( $title->getText() );
3063                                 $text = wfMsgForContentNoTrans( $message );
3064                                 if( wfEmptyMsg( $message, $text ) ) {
3065                                         $text = false;
3066                                         break;
3067                                 }
3068                         } else {
3069                                 break;
3070                         }
3071                         if ( $text === false ) {
3072                                 break;
3073                         }
3074                         // Redirect?
3075                         $finalTitle = $title;
3076                         $title = Title::newFromRedirect( $text );
3077                 }
3078                 return array(
3079                         'text' => $text,
3080                         'finalTitle' => $finalTitle,
3081                         'deps' => $deps );
3082         }
3083
3084         /**
3085          * Transclude an interwiki link.
3086          */
3087         function interwikiTransclude( $title, $action ) {
3088                 global $wgEnableScaryTranscluding;
3089
3090                 if (!$wgEnableScaryTranscluding)
3091                         return wfMsg('scarytranscludedisabled');
3092
3093                 $url = $title->getFullUrl( "action=$action" );
3094
3095                 if (strlen($url) > 255)
3096                         return wfMsg('scarytranscludetoolong');
3097                 return $this->fetchScaryTemplateMaybeFromCache($url);
3098         }
3099
3100         function fetchScaryTemplateMaybeFromCache($url) {
3101                 global $wgTranscludeCacheExpiry;
3102                 $dbr = wfGetDB(DB_SLAVE);
3103                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3104                                 array('tc_url' => $url));
3105                 if ($obj) {
3106                         $time = $obj->tc_time;
3107                         $text = $obj->tc_contents;
3108                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3109                                 return $text;
3110                         }
3111                 }
3112
3113                 $text = Http::get($url);
3114                 if (!$text)
3115                         return wfMsg('scarytranscludefailed', $url);
3116
3117                 $dbw = wfGetDB(DB_MASTER);
3118                 $dbw->replace('transcache', array('tc_url'), array(
3119                         'tc_url' => $url,
3120                         'tc_time' => time(),
3121                         'tc_contents' => $text));
3122                 return $text;
3123         }
3124
3125
3126         /**
3127          * Triple brace replacement -- used for template arguments
3128          * @private
3129          */
3130         function argSubstitution( $piece, $frame ) {
3131                 wfProfileIn( __METHOD__ );
3132
3133                 $error = false;
3134                 $parts = $piece['parts'];
3135                 $nameWithSpaces = $frame->expand( $piece['title'] );
3136                 $argName = trim( $nameWithSpaces );
3137                 $object = false;
3138                 $text = $frame->getArgument( $argName );
3139                 if (  $text === false && $parts->getLength() > 0
3140                   && (
3141                     $this->ot['html']
3142                     || $this->ot['pre']
3143                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3144                   )
3145                 ) {
3146                         # No match in frame, use the supplied default
3147                         $object = $parts->item( 0 )->getChildren();
3148                 }
3149                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3150                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3151                         $this->limitationWarn( 'post-expand-template-argument' );
3152                 }
3153
3154                 if ( $text === false && $object === false ) {
3155                         # No match anywhere
3156                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3157                 }
3158                 if ( $error !== false ) {
3159                         $text .= $error;
3160                 }
3161                 if ( $object !== false ) {
3162                         $ret = array( 'object' => $object );
3163                 } else {
3164                         $ret = array( 'text' => $text );
3165                 }
3166
3167                 wfProfileOut( __METHOD__ );
3168                 return $ret;
3169         }
3170
3171         /**
3172          * Return the text to be used for a given extension tag.
3173          * This is the ghost of strip().
3174          *
3175          * @param array $params Associative array of parameters:
3176          *     name       PPNode for the tag name
3177          *     attr       PPNode for unparsed text where tag attributes are thought to be
3178          *     attributes Optional associative array of parsed attributes
3179          *     inner      Contents of extension element
3180          *     noClose    Original text did not have a close tag
3181          * @param PPFrame $frame
3182          */
3183         function extensionSubstitution( $params, $frame ) {
3184                 global $wgRawHtml, $wgContLang;
3185
3186                 $name = $frame->expand( $params['name'] );
3187                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3188                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3189
3190                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $this->mMarkerIndex++) . self::MARKER_SUFFIX;
3191
3192                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3193                         ( $this->ot['html'] || $this->ot['pre'] );
3194                 if ( $this->ot['html'] || $isFunctionTag ) {
3195                         $name = strtolower( $name );
3196                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3197                         if ( isset( $params['attributes'] ) ) {
3198                                 $attributes = $attributes + $params['attributes'];
3199                         }
3200                         switch ( $name ) {
3201                                 case 'html':
3202                                         if( $wgRawHtml ) {
3203                                                 $output = $content;
3204                                                 break;
3205                                         } else {
3206                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3207                                         }
3208                                 case 'nowiki':
3209                                         $content = strtr($content, array('-{' => '-&#123;', '}-' => '&#125;-'));
3210                                         $output = Xml::escapeTagsOnly( $content );
3211                                         break;
3212                                 case 'math':
3213                                         $output = $wgContLang->armourMath(
3214                                                 MathRenderer::renderMath( $content, $attributes ) );
3215                                         break;
3216                                 case 'gallery':
3217                                         $output = $this->renderImageGallery( $content, $attributes );
3218                                         break;
3219                                 default:
3220                                         if( isset( $this->mTagHooks[$name] ) ) {
3221                                                 # Workaround for PHP bug 35229 and similar
3222                                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3223                                                         throw new MWException( "Tag hook for $name is not callable\n" );
3224                                                 }
3225                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3226                                                         array( $content, $attributes, $this ) );
3227                                         } elseif( isset( $this->mFunctionTagHooks[$name] ) ) {
3228                                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3229                                                 if( !is_callable( $callback ) )
3230                                                         throw new MWException( "Tag hook for $name is not callable\n" );
3231
3232                                                 $output = call_user_func_array( $callback,
3233                                                         array( &$this, $frame, $content, $attributes ) );
3234                                         } else {
3235                                                 $output = '<span class="error">Invalid tag extension name: ' .
3236                                                         htmlspecialchars( $name ) . '</span>';
3237                                         }
3238                         }
3239                 } else {
3240                         if ( is_null( $attrText ) ) {
3241                                 $attrText = '';
3242                         }
3243                         if ( isset( $params['attributes'] ) ) {
3244                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3245                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3246                                                 htmlspecialchars( $attrValue ) . '"';
3247                                 }
3248                         }
3249                         if ( $content === null ) {
3250                                 $output = "<$name$attrText/>";
3251                         } else {
3252                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3253                                 $output = "<$name$attrText>$content$close";
3254                         }
3255                 }
3256
3257                 if( $isFunctionTag ) {
3258                         return $output;
3259                 } elseif ( $name === 'html' || $name === 'nowiki' ) {
3260                         $this->mStripState->nowiki->setPair( $marker, $output );
3261                 } else {
3262                         $this->mStripState->general->setPair( $marker, $output );
3263                 }
3264                 return $marker;
3265         }
3266
3267         /**
3268          * Increment an include size counter
3269          *
3270          * @param string $type The type of expansion
3271          * @param integer $size The size of the text
3272          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3273          */
3274         function incrementIncludeSize( $type, $size ) {
3275                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3276                         return false;
3277                 } else {
3278                         $this->mIncludeSizes[$type] += $size;
3279                         return true;
3280                 }
3281         }
3282
3283         /**
3284          * Increment the expensive function count
3285          *
3286          * @return boolean False if the limit has been exceeded
3287          */
3288         function incrementExpensiveFunctionCount() {
3289                 global $wgExpensiveParserFunctionLimit;
3290                 $this->mExpensiveFunctionCount++;
3291                 if($this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit) {
3292                         return true;
3293                 }
3294                 return false;
3295         }
3296
3297         /**
3298          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3299          * Fills $this->mDoubleUnderscores, returns the modified text
3300          */
3301         function doDoubleUnderscore( $text ) {
3302                 wfProfileIn( __METHOD__ );
3303                 // The position of __TOC__ needs to be recorded
3304                 $mw = MagicWord::get( 'toc' );
3305                 if( $mw->match( $text ) ) {
3306                         $this->mShowToc = true;
3307                         $this->mForceTocPosition = true;
3308
3309                         // Set a placeholder. At the end we'll fill it in with the TOC.
3310                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3311
3312                         // Only keep the first one.
3313                         $text = $mw->replace( '', $text );
3314                 }
3315
3316                 // Now match and remove the rest of them
3317                 $mwa = MagicWord::getDoubleUnderscoreArray();
3318                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3319
3320                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3321                         $this->mOutput->mNoGallery = true;
3322                 }
3323                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3324                         $this->mShowToc = false;
3325                 }
3326                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3327                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3328
3329                         $containerCategory = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( 'hidden-category-category' ) );
3330                         if ( $containerCategory ) {
3331                                 $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3332                         } else {
3333                                 wfDebug( __METHOD__.": [[MediaWiki:hidden-category-category]] is not a valid title!\n" );
3334                         }
3335                 }
3336                 # (bug 8068) Allow control over whether robots index a page.
3337                 #
3338                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3339                 # is not desirable, the last one on the page should win.
3340                 if( isset( $this->mDoubleUnderscores['noindex'] ) ) {
3341                         $this->mOutput->setIndexPolicy( 'noindex' );
3342                 } elseif( isset( $this->mDoubleUnderscores['index'] ) ) {
3343                         $this->mOutput->setIndexPolicy( 'index' );
3344                 }
3345                 wfProfileOut( __METHOD__ );
3346                 return $text;
3347         }
3348
3349         /**
3350          * This function accomplishes several tasks:
3351          * 1) Auto-number headings if that option is enabled
3352          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3353          * 3) Add a Table of contents on the top for users who have enabled the option
3354          * 4) Auto-anchor headings
3355          *
3356          * It loops through all headlines, collects the necessary data, then splits up the
3357          * string and re-inserts the newly formatted headlines.
3358          *
3359          * @param string $text
3360          * @param string $origText Original, untouched wikitext
3361          * @param boolean $isMain
3362          * @private
3363          */
3364         function formatHeadings( $text, $origText, $isMain=true ) {
3365                 global $wgMaxTocLevel, $wgContLang, $wgEnforceHtmlIds;
3366
3367                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3368                 $showEditLink = $this->mOptions->getEditSection();
3369
3370                 // Do not call quickUserCan unless necessary
3371                 if( $showEditLink && !$this->mTitle->quickUserCan( 'edit' ) ) {
3372                         $showEditLink = 0;
3373                 }
3374
3375                 # Inhibit editsection links if requested in the page
3376                 if ( isset( $this->mDoubleUnderscores['noeditsection'] )  || $this->mOptions->getIsPrintable() ) {
3377                         $showEditLink = 0;
3378                 }
3379
3380                 # Get all headlines for numbering them and adding funky stuff like [edit]
3381                 # links - this is for later, but we need the number of headlines right now
3382                 $matches = array();
3383                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3384
3385                 # if there are fewer than 4 headlines in the article, do not show TOC
3386                 # unless it's been explicitly enabled.
3387                 $enoughToc = $this->mShowToc &&
3388                         (($numMatches >= 4) || $this->mForceTocPosition);
3389
3390                 # Allow user to stipulate that a page should have a "new section"
3391                 # link added via __NEWSECTIONLINK__
3392                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3393                         $this->mOutput->setNewSection( true );
3394                 }
3395
3396                 # Allow user to remove the "new section"
3397                 # link via __NONEWSECTIONLINK__
3398                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3399                         $this->mOutput->hideNewSection( true );
3400                 }
3401
3402                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3403                 # override above conditions and always show TOC above first header
3404                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3405                         $this->mShowToc = true;
3406                         $enoughToc = true;
3407                 }
3408
3409                 # We need this to perform operations on the HTML
3410                 $sk = $this->mOptions->getSkin();
3411
3412                 # headline counter
3413                 $headlineCount = 0;
3414                 $numVisible = 0;
3415
3416                 # Ugh .. the TOC should have neat indentation levels which can be
3417                 # passed to the skin functions. These are determined here
3418                 $toc = '';
3419                 $full = '';
3420                 $head = array();
3421                 $sublevelCount = array();
3422                 $levelCount = array();
3423                 $toclevel = 0;
3424                 $level = 0;
3425                 $prevlevel = 0;
3426                 $toclevel = 0;
3427                 $prevtoclevel = 0;
3428                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3429                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3430                 $oldType = $this->mOutputType;
3431                 $this->setOutputType( self::OT_WIKI );
3432                 $frame = $this->getPreprocessor()->newFrame();
3433                 $root = $this->preprocessToDom( $origText );
3434                 $node = $root->getFirstChild();
3435                 $byteOffset = 0;
3436                 $tocraw = array();
3437
3438                 foreach( $matches[3] as $headline ) {
3439                         $isTemplate = false;
3440                         $titleText = false;
3441                         $sectionIndex = false;
3442                         $numbering = '';
3443                         $markerMatches = array();
3444                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3445                                 $serial = $markerMatches[1];
3446                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3447                                 $isTemplate = ($titleText != $baseTitleText);
3448                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3449                         }
3450
3451                         if( $toclevel ) {
3452                                 $prevlevel = $level;
3453                                 $prevtoclevel = $toclevel;
3454                         }
3455                         $level = $matches[1][$headlineCount];
3456
3457                         if ( $level > $prevlevel ) {
3458                                 # Increase TOC level
3459                                 $toclevel++;
3460                                 $sublevelCount[$toclevel] = 0;
3461                                 if( $toclevel<$wgMaxTocLevel ) {
3462                                         $prevtoclevel = $toclevel;
3463                                         $toc .= $sk->tocIndent();
3464                                         $numVisible++;
3465                                 }
3466                         }
3467                         elseif ( $level < $prevlevel && $toclevel > 1 ) {
3468                                 # Decrease TOC level, find level to jump to
3469
3470                                 for ($i = $toclevel; $i > 0; $i--) {
3471                                         if ( $levelCount[$i] == $level ) {
3472                                                 # Found last matching level
3473                                                 $toclevel = $i;
3474                                                 break;
3475                                         }
3476                                         elseif ( $levelCount[$i] < $level ) {
3477                                                 # Found first matching level below current level
3478                                                 $toclevel = $i + 1;
3479                                                 break;
3480                                         }
3481                                 }
3482                                 if( $i == 0 ) $toclevel = 1;
3483                                 if( $toclevel<$wgMaxTocLevel ) {
3484                                         if($prevtoclevel < $wgMaxTocLevel) {
3485                                                 # Unindent only if the previous toc level was shown :p
3486                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3487                                                 $prevtoclevel = $toclevel;
3488                                         } else {
3489                                                 $toc .= $sk->tocLineEnd();
3490                                         }
3491                                 }
3492                         }
3493                         else {
3494                                 # No change in level, end TOC line
3495                                 if( $toclevel<$wgMaxTocLevel ) {
3496                                         $toc .= $sk->tocLineEnd();
3497                                 }
3498                         }
3499
3500                         $levelCount[$toclevel] = $level;
3501
3502                         # count number of headlines for each level
3503                         @$sublevelCount[$toclevel]++;
3504                         $dot = 0;
3505                         for( $i = 1; $i <= $toclevel; $i++ ) {
3506                                 if( !empty( $sublevelCount[$i] ) ) {
3507                                         if( $dot ) {
3508                                                 $numbering .= '.';
3509                                         }
3510                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3511                                         $dot = 1;
3512                                 }
3513                         }
3514
3515                         # The safe header is a version of the header text safe to use for links
3516                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3517                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3518
3519                         # Remove link placeholders by the link text.
3520                         #     <!--LINK number-->
3521                         # turns into
3522                         #     link text with suffix
3523                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3524
3525                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3526                         $tocline = preg_replace(
3527                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3528                                 array( '',                          '<$1>'),
3529                                 $safeHeadline
3530                         );
3531                         $tocline = trim( $tocline );
3532
3533                         # For the anchor, strip out HTML-y stuff period
3534                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3535                         $safeHeadline = preg_replace( '/[ _]+/', ' ', $safeHeadline );
3536                         $safeHeadline = trim( $safeHeadline );
3537
3538                         # Save headline for section edit hint before it's escaped
3539                         $headlineHint = $safeHeadline;
3540
3541                         if ( $wgEnforceHtmlIds ) {
3542                                 $legacyHeadline = false;
3543                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3544                                         'noninitial' );
3545                         } else {
3546                                 # For reverse compatibility, provide an id that's
3547                                 # HTML4-compatible, like we used to.
3548                                 #
3549                                 # It may be worth noting, academically, that it's possible for
3550                                 # the legacy anchor to conflict with a non-legacy headline
3551                                 # anchor on the page.  In this case likely the "correct" thing
3552                                 # would be to either drop the legacy anchors or make sure
3553                                 # they're numbered first.  However, this would require people
3554                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3555                                 # manually, so let's not bother worrying about it.
3556                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3557                                         'noninitial' );
3558                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline, 'xml' );
3559
3560                                 if ( $legacyHeadline == $safeHeadline ) {
3561                                         # No reason to have both (in fact, we can't)
3562                                         $legacyHeadline = false;
3563                                 } elseif ( $legacyHeadline != Sanitizer::escapeId(
3564                                 $legacyHeadline, 'xml' ) ) {
3565                                         # The legacy id is invalid XML.  We used to allow this, but
3566                                         # there's no reason to do so anymore.  Backward
3567                                         # compatibility will fail slightly in this case, but it's
3568                                         # no big deal.
3569                                         $legacyHeadline = false;
3570                                 }
3571                         }
3572
3573                         # HTML names must be case-insensitively unique (bug 10721).  FIXME:
3574                         # Does this apply to Unicode characters?  Because we aren't
3575                         # handling those here.
3576                         $arrayKey = strtolower( $safeHeadline );
3577                         if ( $legacyHeadline === false ) {
3578                                 $legacyArrayKey = false;
3579                         } else {
3580                                 $legacyArrayKey = strtolower( $legacyHeadline );
3581                         }
3582
3583                         # count how many in assoc. array so we can track dupes in anchors
3584                         if ( isset( $refers[$arrayKey] ) ) {
3585                                 $refers[$arrayKey]++;
3586                         } else {
3587                                 $refers[$arrayKey] = 1;
3588                         }
3589                         if ( isset( $refers[$legacyArrayKey] ) ) {
3590                                 $refers[$legacyArrayKey]++;
3591                         } else {
3592                                 $refers[$legacyArrayKey] = 1;
3593                         }
3594
3595                         # Don't number the heading if it is the only one (looks silly)
3596                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3597                                 # the two are different if the line contains a link
3598                                 $headline=$numbering . ' ' . $headline;
3599                         }
3600
3601                         # Create the anchor for linking from the TOC to the section
3602                         $anchor = $safeHeadline;
3603                         $legacyAnchor = $legacyHeadline;
3604                         if ( $refers[$arrayKey] > 1 ) {
3605                                 $anchor .= '_' . $refers[$arrayKey];
3606                         }
3607                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3608                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3609                         }
3610                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3611                                 $toc .= $sk->tocLine($anchor, $tocline,
3612                                         $numbering, $toclevel, ($isTemplate ? false : $sectionIndex));
3613                         }
3614
3615                         # Add the section to the section tree
3616                         # Find the DOM node for this header
3617                         while ( $node && !$isTemplate ) {
3618                                 if ( $node->getName() === 'h' ) {
3619                                         $bits = $node->splitHeading();
3620                                         if ( $bits['i'] == $sectionIndex )
3621                                                 break;
3622                                 }
3623                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
3624                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
3625                                 $node = $node->getNextSibling();
3626                         }
3627                         $tocraw[] = array(
3628                                 'toclevel' => $toclevel,
3629                                 'level' => $level,
3630                                 'line' => $tocline,
3631                                 'number' => $numbering,
3632                                 'index' => ($isTemplate ? 'T-' : '' ) . $sectionIndex,
3633                                 'fromtitle' => $titleText,
3634                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
3635                                 'anchor' => $anchor,
3636                         );
3637
3638                         # give headline the correct <h#> tag
3639                         if( $showEditLink && $sectionIndex !== false ) {
3640                                 if( $isTemplate ) {
3641                                         # Put a T flag in the section identifier, to indicate to extractSections()
3642                                         # that sections inside <includeonly> should be counted.
3643                                         $editlink = $sk->doEditSectionLink(Title::newFromText( $titleText ), "T-$sectionIndex");
3644                                 } else {
3645                                         $editlink = $sk->doEditSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3646                                 }
3647                         } else {
3648                                 $editlink = '';
3649                         }
3650                         $head[$headlineCount] = $sk->makeHeadline( $level,
3651                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3652                                 $editlink, $legacyAnchor );
3653
3654                         $headlineCount++;
3655                 }
3656
3657                 $this->setOutputType( $oldType );
3658
3659                 # Never ever show TOC if no headers
3660                 if( $numVisible < 1 ) {
3661                         $enoughToc = false;
3662                 }
3663
3664                 if( $enoughToc ) {
3665                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3666                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3667                         }
3668                         $toc = $sk->tocList( $toc );
3669                         $this->mOutput->setTOCHTML( $toc );
3670                 }
3671
3672                 if ( $isMain ) {
3673                         $this->mOutput->setSections( $tocraw );
3674                 }
3675
3676                 # split up and insert constructed headlines
3677
3678                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3679                 $i = 0;
3680
3681                 foreach( $blocks as $block ) {
3682                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3683                                 # This is the [edit] link that appears for the top block of text when
3684                                 # section editing is enabled
3685
3686                                 # Disabled because it broke block formatting
3687                                 # For example, a bullet point in the top line
3688                                 # $full .= $sk->editSectionLink(0);
3689                         }
3690                         $full .= $block;
3691                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3692                                 # Top anchor now in skin
3693                                 $full = $full.$toc;
3694                         }
3695
3696                         if( !empty( $head[$i] ) ) {
3697                                 $full .= $head[$i];
3698                         }
3699                         $i++;
3700                 }
3701                 if( $this->mForceTocPosition ) {
3702                         return str_replace( '<!--MWTOC-->', $toc, $full );
3703                 } else {
3704                         return $full;
3705                 }
3706         }
3707
3708         /**
3709          * Merge $tree2 into $tree1 by replacing the section with index
3710          * $section in $tree1 and its descendants with the sections in $tree2.
3711          * Note that in the returned section tree, only the 'index' and
3712          * 'byteoffset' fields are guaranteed to be correct.
3713          * @param $tree1 array Section tree from ParserOutput::getSectons()
3714         *  @param $tree2 array Section tree
3715          * @param $section int Section index
3716          * @param $title Title Title both section trees come from
3717          * @param $len2 int Length of the original wikitext for $tree2
3718          * @return array Merged section tree
3719          */
3720         public static function mergeSectionTrees( $tree1, $tree2, $section, $title, $len2 ) {
3721                 global $wgContLang;
3722                 $newTree = array();
3723                 $targetLevel = false;
3724                 $merged = false;
3725                 $lastLevel = 1;
3726                 $nextIndex = 1;
3727                 $numbering = array( 0 );
3728                 $titletext = $title->getPrefixedDBkey();
3729                 foreach ( $tree1 as $s ) {
3730                         if ( $targetLevel !== false ) {
3731                                 if ( $s['level'] <= $targetLevel )
3732                                         // We've skipped enough
3733                                         $targetLevel = false;
3734                                 else
3735                                         continue;
3736                         }
3737                         if ( $s['index'] != $section ||
3738                                         $s['fromtitle'] != $titletext ) {
3739                                 self::incrementNumbering( $numbering,
3740                                         $s['toclevel'], $lastLevel );
3741
3742                                 // Rewrite index, byteoffset and number
3743                                 if ( $s['fromtitle'] == $titletext ) {
3744                                         $s['index'] = $nextIndex++;
3745                                         if ( $merged )
3746                                                 $s['byteoffset'] += $len2;
3747                                 }
3748                                 $s['number']  = implode( '.', array_map(
3749                                         array( $wgContLang, 'formatnum' ),
3750                                         $numbering ) );
3751                                 $lastLevel = $s['toclevel'];
3752                                 $newTree[] = $s;
3753                         } else {
3754                                 // We're at $section
3755                                 // Insert sections from $tree2 here
3756                                 foreach ( $tree2 as $s2 ) {
3757                                         // Rewrite the fields in $s2
3758                                         // before inserting it
3759                                         $s2['toclevel'] += $s['toclevel'] - 1;
3760                                         $s2['level'] += $s['level'] - 1;
3761                                         $s2['index'] = $nextIndex++;
3762                                         $s2['byteoffset'] += $s['byteoffset'];
3763
3764                                         self::incrementNumbering( $numbering,
3765                                                 $s2['toclevel'], $lastLevel );
3766                                         $s2['number']  = implode( '.', array_map(
3767                                                 array( $wgContLang, 'formatnum' ),
3768                                                 $numbering ) );
3769                                         $lastLevel = $s2['toclevel'];
3770                                         $newTree[] = $s2;
3771                                 }
3772                                 // Skip all descendants of $section in $tree1
3773                                 $targetLevel = $s['level'];
3774                                 $merged = true;
3775                         }
3776                 }
3777                 return $newTree;
3778         }
3779
3780         /**
3781          * Increment a section number. Helper function for mergeSectionTrees()
3782          * @param $number array Array representing a section number
3783          * @param $level int Current TOC level (depth)
3784          * @param $lastLevel int Level of previous TOC entry
3785          */
3786         private static function incrementNumbering( &$number, $level, $lastLevel ) {
3787                 if ( $level > $lastLevel )
3788                         $number[$level - 1] = 1;
3789                 else if ( $level < $lastLevel ) {
3790                         foreach ( $number as $key => $unused )
3791                                 if ( $key >= $level )
3792                                         unset( $number[$key] );
3793                         $number[$level - 1]++;
3794                 } else
3795                         $number[$level - 1]++;
3796         }
3797
3798         /**
3799          * Transform wiki markup when saving a page by doing \r\n -> \n
3800          * conversion, substitting signatures, {{subst:}} templates, etc.
3801          *
3802          * @param string $text the text to transform
3803          * @param Title &$title the Title object for the current article
3804          * @param User $user the User object describing the current user
3805          * @param ParserOptions $options parsing options
3806          * @param bool $clearState whether to clear the parser state first
3807          * @return string the altered wiki markup
3808          * @public
3809          */
3810         function preSaveTransform( $text, Title $title, $user, $options, $clearState = true ) {
3811                 $this->mOptions = $options;
3812                 $this->setTitle( $title );
3813                 $this->setOutputType( self::OT_WIKI );
3814
3815                 if ( $clearState ) {
3816                         $this->clearState();
3817                 }
3818
3819                 $pairs = array(
3820                         "\r\n" => "\n",
3821                 );
3822                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3823                 $text = $this->pstPass2( $text, $user );
3824                 $text = $this->mStripState->unstripBoth( $text );
3825                 return $text;
3826         }
3827
3828         /**
3829          * Pre-save transform helper function
3830          * @private
3831          */
3832         function pstPass2( $text, $user ) {
3833                 global $wgContLang, $wgLocaltimezone;
3834
3835                 /* Note: This is the timestamp saved as hardcoded wikitext to
3836                  * the database, we use $wgContLang here in order to give
3837                  * everyone the same signature and use the default one rather
3838                  * than the one selected in each user's preferences.
3839                  *
3840                  * (see also bug 12815)
3841                  */
3842                 $ts = $this->mOptions->getTimestamp();
3843                 $tz = wfMsgForContent( 'timezone-utc' );
3844                 if ( isset( $wgLocaltimezone ) ) {
3845                         $unixts = wfTimestamp( TS_UNIX, $ts );
3846                         $oldtz = getenv( 'TZ' );
3847                         putenv( 'TZ='.$wgLocaltimezone );
3848                         $ts = date( 'YmdHis', $unixts );
3849                         $tz = date( 'T', $unixts );  # might vary on DST changeover!
3850
3851                         /* Allow translation of timezones trough wiki. date() can return
3852                          * whatever crap the system uses, localised or not, so we cannot
3853                          * ship premade translations.
3854                          */
3855                         $key = 'timezone-' . strtolower( trim( $tz ) );
3856                         $value = wfMsgForContent( $key );
3857                         if ( !wfEmptyMsg( $key, $value ) ) $tz = $value;
3858
3859                         putenv( 'TZ='.$oldtz );
3860                 }
3861
3862                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tz)";
3863
3864                 # Variable replacement
3865                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3866                 $text = $this->replaceVariables( $text );
3867
3868                 # Signatures
3869                 $sigText = $this->getUserSig( $user );
3870                 $text = strtr( $text, array(
3871                         '~~~~~' => $d,
3872                         '~~~~' => "$sigText $d",
3873                         '~~~' => $sigText
3874                 ) );
3875
3876                 # Context links: [[|name]] and [[name (context)|]]
3877                 #
3878                 global $wgLegalTitleChars;
3879                 $tc = "[$wgLegalTitleChars]";
3880                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
3881
3882                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3883                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
3884                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3885                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3886
3887                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3888                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3889                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
3890                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3891
3892                 $t = $this->mTitle->getText();
3893                 $m = array();
3894                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3895                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3896                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3897                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3898                 } else {
3899                         # if there's no context, don't bother duplicating the title
3900                         $text = preg_replace( $p2, '[[\\1]]', $text );
3901                 }
3902
3903                 # Trim trailing whitespace
3904                 $text = rtrim( $text );
3905
3906                 return $text;
3907         }
3908
3909         /**
3910          * Fetch the user's signature text, if any, and normalize to
3911          * validated, ready-to-insert wikitext.
3912          *
3913          * @param User $user
3914          * @return string
3915          * @private
3916          */
3917         function getUserSig( &$user ) {
3918                 global $wgMaxSigChars;
3919
3920                 $username = $user->getName();
3921                 $nickname = $user->getOption( 'nickname' );
3922                 $nickname = $nickname === null ? $username : $nickname;
3923
3924                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
3925                         $nickname = $username;
3926                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
3927                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
3928                         # Sig. might contain markup; validate this
3929                         if( $this->validateSig( $nickname ) !== false ) {
3930                                 # Validated; clean up (if needed) and return it
3931                                 return $this->cleanSig( $nickname, true );
3932                         } else {
3933                                 # Failed to validate; fall back to the default
3934                                 $nickname = $username;
3935                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
3936                         }
3937                 }
3938
3939                 // Make sure nickname doesnt get a sig in a sig
3940                 $nickname = $this->cleanSigInSig( $nickname );
3941
3942                 # If we're still here, make it a link to the user page
3943                 $userText = wfEscapeWikiText( $username );
3944                 $nickText = wfEscapeWikiText( $nickname );
3945                 if ( $user->isAnon() )  {
3946                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
3947                 } else {
3948                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
3949                 }
3950         }
3951
3952         /**
3953          * Check that the user's signature contains no bad XML
3954          *
3955          * @param string $text
3956          * @return mixed An expanded string, or false if invalid.
3957          */
3958         function validateSig( $text ) {
3959                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
3960         }
3961
3962         /**
3963          * Clean up signature text
3964          *
3965          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3966          * 2) Substitute all transclusions
3967          *
3968          * @param string $text
3969          * @param $parsing Whether we're cleaning (preferences save) or parsing
3970          * @return string Signature text
3971          */
3972         function cleanSig( $text, $parsing = false ) {
3973                 if ( !$parsing ) {
3974                         global $wgTitle;
3975                         $this->clearState();
3976                         $this->setTitle( $wgTitle );
3977                         $this->mOptions = new ParserOptions;
3978                         $this->setOutputType = self::OT_PREPROCESS;
3979                 }
3980
3981                 # Option to disable this feature
3982                 if ( !$this->mOptions->getCleanSignatures() ) {
3983                         return $text;
3984                 }
3985
3986                 # FIXME: regex doesn't respect extension tags or nowiki
3987                 #  => Move this logic to braceSubstitution()
3988                 $substWord = MagicWord::get( 'subst' );
3989                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3990                 $substText = '{{' . $substWord->getSynonym( 0 );
3991
3992                 $text = preg_replace( $substRegex, $substText, $text );
3993                 $text = $this->cleanSigInSig( $text );
3994                 $dom = $this->preprocessToDom( $text );
3995                 $frame = $this->getPreprocessor()->newFrame();
3996                 $text = $frame->expand( $dom );
3997
3998                 if ( !$parsing ) {
3999                         $text = $this->mStripState->unstripBoth( $text );
4000                 }
4001
4002                 return $text;
4003         }
4004
4005         /**
4006          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4007          * @param string $text
4008          * @return string Signature text with /~{3,5}/ removed
4009          */
4010         function cleanSigInSig( $text ) {
4011                 $text = preg_replace( '/~{3,5}/', '', $text );
4012                 return $text;
4013         }
4014
4015         /**
4016          * Set up some variables which are usually set up in parse()
4017          * so that an external function can call some class members with confidence
4018          * @public
4019          */
4020         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4021                 $this->setTitle( $title );
4022                 $this->mOptions = $options;
4023                 $this->setOutputType( $outputType );
4024                 if ( $clearState ) {
4025                         $this->clearState();
4026                 }
4027         }
4028
4029         /**
4030          * Wrapper for preprocess()
4031          *
4032          * @param string $text the text to preprocess
4033          * @param ParserOptions $options  options
4034          * @return string
4035          * @public
4036          */
4037         function transformMsg( $text, $options ) {
4038                 global $wgTitle;
4039                 static $executing = false;
4040
4041                 # Guard against infinite recursion
4042                 if ( $executing ) {
4043                         return $text;
4044                 }
4045                 $executing = true;
4046
4047                 wfProfileIn(__METHOD__);
4048                 $text = $this->preprocess( $text, $wgTitle, $options );
4049
4050                 $executing = false;
4051                 wfProfileOut(__METHOD__);
4052                 return $text;
4053         }
4054
4055         /**
4056          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4057          * The callback should have the following form:
4058          *    function myParserHook( $text, $params, &$parser ) { ... }
4059          *
4060          * Transform and return $text. Use $parser for any required context, e.g. use
4061          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4062          *
4063          * @public
4064          *
4065          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4066          * @param mixed $callback The callback function (and object) to use for the tag
4067          *
4068          * @return The old value of the mTagHooks array associated with the hook
4069          */
4070         function setHook( $tag, $callback ) {
4071                 $tag = strtolower( $tag );
4072                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4073                 $this->mTagHooks[$tag] = $callback;
4074                 if( !in_array( $tag, $this->mStripList ) ) {
4075                         $this->mStripList[] = $tag;
4076                 }
4077
4078                 return $oldVal;
4079         }
4080
4081         function setTransparentTagHook( $tag, $callback ) {
4082                 $tag = strtolower( $tag );
4083                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4084                 $this->mTransparentTagHooks[$tag] = $callback;
4085
4086                 return $oldVal;
4087         }
4088
4089         /**
4090          * Remove all tag hooks
4091          */
4092         function clearTagHooks() {
4093                 $this->mTagHooks = array();
4094                 $this->mStripList = $this->mDefaultStripList;
4095         }
4096
4097         /**
4098          * Create a function, e.g. {{sum:1|2|3}}
4099          * The callback function should have the form:
4100          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4101          *
4102          * Or with SFH_OBJECT_ARGS:
4103          *    function myParserFunction( $parser, $frame, $args ) { ... }
4104          *
4105          * The callback may either return the text result of the function, or an array with the text
4106          * in element 0, and a number of flags in the other elements. The names of the flags are
4107          * specified in the keys. Valid flags are:
4108          *   found                     The text returned is valid, stop processing the template. This
4109          *                             is on by default.
4110          *   nowiki                    Wiki markup in the return value should be escaped
4111          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4112          *
4113          * @public
4114          *
4115          * @param string $id The magic word ID
4116          * @param mixed $callback The callback function (and object) to use
4117          * @param integer $flags a combination of the following flags:
4118          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4119          *
4120          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4121          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4122          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4123          *     the arguments, and to control the way they are expanded.
4124          *
4125          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4126          *     arguments, for instance:
4127          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4128          *
4129          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4130          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4131          *     working if/when this is changed.
4132          *
4133          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4134          *     expansion.
4135          *
4136          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4137          *     about the methods available in PPFrame and PPNode.
4138          *
4139          * @return The old callback function for this name, if any
4140          */
4141         function setFunctionHook( $id, $callback, $flags = 0 ) {
4142                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4143                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4144
4145                 # Add to function cache
4146                 $mw = MagicWord::get( $id );
4147                 if( !$mw )
4148                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4149
4150                 $synonyms = $mw->getSynonyms();
4151                 $sensitive = intval( $mw->isCaseSensitive() );
4152
4153                 foreach ( $synonyms as $syn ) {
4154                         # Case
4155                         if ( !$sensitive ) {
4156                                 $syn = strtolower( $syn );
4157                         }
4158                         # Add leading hash
4159                         if ( !( $flags & SFH_NO_HASH ) ) {
4160                                 $syn = '#' . $syn;
4161                         }
4162                         # Remove trailing colon
4163                         if ( substr( $syn, -1, 1 ) === ':' ) {
4164                                 $syn = substr( $syn, 0, -1 );
4165                         }
4166                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4167                 }
4168                 return $oldVal;
4169         }
4170
4171         /**
4172          * Get all registered function hook identifiers
4173          *
4174          * @return array
4175          */
4176         function getFunctionHooks() {
4177                 return array_keys( $this->mFunctionHooks );
4178         }
4179
4180         /**
4181          * Create a tag function, e.g. <test>some stuff</test>.
4182          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4183          * Unlike parser functions, their content is not preprocessed.
4184          */
4185         function setFunctionTagHook( $tag, $callback, $flags ) {
4186                 $tag = strtolower( $tag );
4187                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4188                         $this->mFunctionTagHooks[$tag] : null;
4189                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4190
4191                 if( !in_array( $tag, $this->mStripList ) ) {
4192                         $this->mStripList[] = $tag;
4193                 }
4194
4195                 return $old;
4196         }
4197
4198         /**
4199          * FIXME: update documentation. makeLinkObj() is deprecated.
4200          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4201          * Placeholders created in Skin::makeLinkObj()
4202          * Returns an array of link CSS classes, indexed by PDBK.
4203          */
4204         function replaceLinkHolders( &$text, $options = 0 ) {
4205                 return $this->mLinkHolders->replace( $text );
4206         }
4207
4208         /**
4209          * Replace <!--LINK--> link placeholders with plain text of links
4210          * (not HTML-formatted).
4211          * @param string $text
4212          * @return string
4213          */
4214         function replaceLinkHoldersText( $text ) {
4215                 return $this->mLinkHolders->replaceText( $text );
4216         }
4217
4218         /**
4219          * Tag hook handler for 'pre'.
4220          */
4221         function renderPreTag( $text, $attribs ) {
4222                 // Backwards-compatibility hack
4223                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4224
4225                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4226                 return Xml::openElement( 'pre', $attribs ) .
4227                         Xml::escapeTagsOnly( $content ) .
4228                         '</pre>';
4229         }
4230
4231         /**
4232          * Renders an image gallery from a text with one line per image.
4233          * text labels may be given by using |-style alternative text. E.g.
4234          *   Image:one.jpg|The number "1"
4235          *   Image:tree.jpg|A tree
4236          * given as text will return the HTML of a gallery with two images,
4237          * labeled 'The number "1"' and
4238          * 'A tree'.
4239          */
4240         function renderImageGallery( $text, $params ) {
4241                 $ig = new ImageGallery();
4242                 $ig->setContextTitle( $this->mTitle );
4243                 $ig->setShowBytes( false );
4244                 $ig->setShowFilename( false );
4245                 $ig->setParser( $this );
4246                 $ig->setHideBadImages();
4247                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4248                 $ig->useSkin( $this->mOptions->getSkin() );
4249                 $ig->mRevisionId = $this->mRevisionId;
4250
4251                 if( isset( $params['caption'] ) ) {
4252                         $caption = $params['caption'];
4253                         $caption = htmlspecialchars( $caption );
4254                         $caption = $this->replaceInternalLinks( $caption );
4255                         $ig->setCaptionHtml( $caption );
4256                 }
4257                 if( isset( $params['perrow'] ) ) {
4258                         $ig->setPerRow( $params['perrow'] );
4259                 }
4260                 if( isset( $params['widths'] ) ) {
4261                         $ig->setWidths( $params['widths'] );
4262                 }
4263                 if( isset( $params['heights'] ) ) {
4264                         $ig->setHeights( $params['heights'] );
4265                 }
4266
4267                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4268
4269                 $lines = StringUtils::explode( "\n", $text );
4270                 foreach ( $lines as $line ) {
4271                         # match lines like these:
4272                         # Image:someimage.jpg|This is some image
4273                         $matches = array();
4274                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4275                         # Skip empty lines
4276                         if ( count( $matches ) == 0 ) {
4277                                 continue;
4278                         }
4279
4280                         if ( strpos( $matches[0], '%' ) !== false )
4281                                 $matches[1] = urldecode( $matches[1] );
4282                         $tp = Title::newFromText( $matches[1]/*, NS_FILE*/ );
4283                         $nt =& $tp;
4284                         if( is_null( $nt ) ) {
4285                                 # Bogus title. Ignore these so we don't bomb out later.
4286                                 continue;
4287                         }
4288                         if ( isset( $matches[3] ) ) {
4289                                 $label = $matches[3];
4290                         } else {
4291                                 $label = '';
4292                         }
4293
4294                         $html = $this->recursiveTagParse( trim( $label ) );
4295
4296                         $ig->add( $nt, $html );
4297
4298                         # Only add real images (bug #5586)
4299                         if ( $nt->getNamespace() == NS_FILE ) {
4300                                 $this->mOutput->addImage( $nt->getDBkey() );
4301                         }
4302                 }
4303                 return $ig->toHTML();
4304         }
4305
4306         function getImageParams( $handler ) {
4307                 if ( $handler ) {
4308                         $handlerClass = get_class( $handler );
4309                 } else {
4310                         $handlerClass = '';
4311                 }
4312                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4313                         // Initialise static lists
4314                         static $internalParamNames = array(
4315                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4316                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4317                                         'bottom', 'text-bottom' ),
4318                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4319                                         'upright', 'border', 'link', 'alt' ),
4320                         );
4321                         static $internalParamMap;
4322                         if ( !$internalParamMap ) {
4323                                 $internalParamMap = array();
4324                                 foreach ( $internalParamNames as $type => $names ) {
4325                                         foreach ( $names as $name ) {
4326                                                 $magicName = str_replace( '-', '_', "img_$name" );
4327                                                 $internalParamMap[$magicName] = array( $type, $name );
4328                                         }
4329                                 }
4330                         }
4331
4332                         // Add handler params
4333                         $paramMap = $internalParamMap;
4334                         if ( $handler ) {
4335                                 $handlerParamMap = $handler->getParamMap();
4336                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4337                                         $paramMap[$magic] = array( 'handler', $paramName );
4338                                 }
4339                         }
4340                         $this->mImageParams[$handlerClass] = $paramMap;
4341                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4342                 }
4343                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4344         }
4345
4346         /**
4347          * Parse image options text and use it to make an image
4348          * @param Title $title
4349          * @param string $options
4350          * @param LinkHolderArray $holders
4351          */
4352         function makeImage( $title, $options, $holders = false ) {
4353                 # Check if the options text is of the form "options|alt text"
4354                 # Options are:
4355                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4356                 #  * left       no resizing, just left align. label is used for alt= only
4357                 #  * right      same, but right aligned
4358                 #  * none       same, but not aligned
4359                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4360                 #  * center     center the image
4361                 #  * frame      Keep original image size, no magnify-button.
4362                 #  * framed     Same as "frame"
4363                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4364                 #  * upright    reduce width for upright images, rounded to full __0 px
4365                 #  * border     draw a 1px border around the image
4366                 #  * alt        Text for HTML alt attribute (defaults to empty)
4367                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4368                 # vertical-align values (no % or length right now):
4369                 #  * baseline
4370                 #  * sub
4371                 #  * super
4372                 #  * top
4373                 #  * text-top
4374                 #  * middle
4375                 #  * bottom
4376                 #  * text-bottom
4377
4378                 $parts = StringUtils::explode( "|", $options );
4379                 $sk = $this->mOptions->getSkin();
4380
4381                 # Give extensions a chance to select the file revision for us
4382                 $skip = $time = $descQuery = false;
4383                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4384
4385                 if ( $skip ) {
4386                         return $sk->link( $title );
4387                 }
4388
4389                 # Get the file
4390                 $imagename = $title->getDBkey();
4391                 if ( isset( $this->mFileCache[$imagename][$time] ) ) {
4392                         $file = $this->mFileCache[$imagename][$time];
4393                 } else {
4394                         $file = wfFindFile( $title, $time );
4395                         if ( count( $this->mFileCache ) > 1000 ) {
4396                                 $this->mFileCache = array();
4397                         }
4398                         $this->mFileCache[$imagename][$time] = $file;
4399                 }
4400                 # Get parameter map
4401                 $handler = $file ? $file->getHandler() : false;
4402
4403                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4404
4405                 # Process the input parameters
4406                 $caption = '';
4407                 $params = array( 'frame' => array(), 'handler' => array(),
4408                         'horizAlign' => array(), 'vertAlign' => array() );
4409                 foreach( $parts as $part ) {
4410                         $part = trim( $part );
4411                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4412                         $validated = false;
4413                         if( isset( $paramMap[$magicName] ) ) {
4414                                 list( $type, $paramName ) = $paramMap[$magicName];
4415
4416                                 // Special case; width and height come in one variable together
4417                                 if( $type === 'handler' && $paramName === 'width' ) {
4418                                         $m = array();
4419                                         # (bug 13500) In both cases (width/height and width only),
4420                                         # permit trailing "px" for backward compatibility.
4421                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4422                                                 $width = intval( $m[1] );
4423                                                 $height = intval( $m[2] );
4424                                                 if ( $handler->validateParam( 'width', $width ) ) {
4425                                                         $params[$type]['width'] = $width;
4426                                                         $validated = true;
4427                                                 }
4428                                                 if ( $handler->validateParam( 'height', $height ) ) {
4429                                                         $params[$type]['height'] = $height;
4430                                                         $validated = true;
4431                                                 }
4432                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4433                                                 $width = intval( $value );
4434                                                 if ( $handler->validateParam( 'width', $width ) ) {
4435                                                         $params[$type]['width'] = $width;
4436                                                         $validated = true;
4437                                                 }
4438                                         } // else no validation -- bug 13436
4439                                 } else {
4440                                         if ( $type === 'handler' ) {
4441                                                 # Validate handler parameter
4442                                                 $validated = $handler->validateParam( $paramName, $value );
4443                                         } else {
4444                                                 # Validate internal parameters
4445                                                 switch( $paramName ) {
4446                                                 case 'manualthumb':
4447                                                 case 'alt':
4448                                                         // @fixme - possibly check validity here for
4449                                                         // manualthumb? downstream behavior seems odd with
4450                                                         // missing manual thumbs.
4451                                                         $validated = true;
4452                                                         $value = $this->stripAltText( $value, $holders );
4453                                                         break;
4454                                                 case 'link':
4455                                                         $chars = self::EXT_LINK_URL_CLASS;
4456                                                         $prots = $this->mUrlProtocols;
4457                                                         if ( $value === '' ) {
4458                                                                 $paramName = 'no-link';
4459                                                                 $value = true;
4460                                                                 $validated = true;
4461                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4462                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4463                                                                         $paramName = 'link-url';
4464                                                                         $this->mOutput->addExternalLink( $value );
4465                                                                         $validated = true;
4466                                                                 }
4467                                                         } else {
4468                                                                 $linkTitle = Title::newFromText( $value );
4469                                                                 if ( $linkTitle ) {
4470                                                                         $paramName = 'link-title';
4471                                                                         $value = $linkTitle;
4472                                                                         $this->mOutput->addLink( $linkTitle );
4473                                                                         $validated = true;
4474                                                                 }
4475                                                         }
4476                                                         break;
4477                                                 default:
4478                                                         // Most other things appear to be empty or numeric...
4479                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4480                                                 }
4481                                         }
4482
4483                                         if ( $validated ) {
4484                                                 $params[$type][$paramName] = $value;
4485                                         }
4486                                 }
4487                         }
4488                         if ( !$validated ) {
4489                                 $caption = $part;
4490                         }
4491                 }
4492
4493                 # Process alignment parameters
4494                 if ( $params['horizAlign'] ) {
4495                         $params['frame']['align'] = key( $params['horizAlign'] );
4496                 }
4497                 if ( $params['vertAlign'] ) {
4498                         $params['frame']['valign'] = key( $params['vertAlign'] );
4499                 }
4500
4501                 $params['frame']['caption'] = $caption;
4502
4503                 # Will the image be presented in a frame, with the caption below?
4504                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4505                                  isset( $params['frame']['framed'] ) ||
4506                                  isset( $params['frame']['thumbnail'] ) ||
4507                                  isset( $params['frame']['manualthumb'] );
4508
4509                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4510                 # came to also set the caption, ordinary text after the image -- which
4511                 # makes no sense, because that just repeats the text multiple times in
4512                 # screen readers.  It *also* came to set the title attribute.
4513                 #
4514                 # Now that we have an alt attribute, we should not set the alt text to
4515                 # equal the caption: that's worse than useless, it just repeats the
4516                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4517                 # use the unnamed parameter for alt text as well, just for the time be-
4518                 # ing, if the unnamed param is set and the alt param is not.
4519                 #
4520                 # For the future, we need to figure out if we want to tweak this more,
4521                 # e.g., introducing a title= parameter for the title; ignoring the un-
4522                 # named parameter entirely for images without a caption; adding an ex-
4523                 # plicit caption= parameter and preserving the old magic unnamed para-
4524                 # meter for BC; ...
4525                 if ( $imageIsFramed ) { # Framed image
4526                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4527                                 # No caption or alt text, add the filename as the alt text so
4528                                 # that screen readers at least get some description of the image
4529                                 $params['frame']['alt'] = $title->getText();
4530                         }
4531                         # Do not set $params['frame']['title'] because tooltips don't make sense
4532                         # for framed images
4533                 } else { # Inline image
4534                         if ( !isset( $params['frame']['alt'] ) ) {
4535                                 # No alt text, use the "caption" for the alt text
4536                                 if ( $caption !== '') {
4537                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4538                                 } else {
4539                                         # No caption, fall back to using the filename for the
4540                                         # alt text
4541                                         $params['frame']['alt'] = $title->getText();
4542                                 }
4543                         }
4544                         # Use the "caption" for the tooltip text
4545                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4546                 }
4547
4548                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4549
4550                 # Linker does the rest
4551                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4552
4553                 # Give the handler a chance to modify the parser object
4554                 if ( $handler ) {
4555                         $handler->parserTransformHook( $this, $file );
4556                 }
4557
4558                 return $ret;
4559         }
4560
4561         protected function stripAltText( $caption, $holders ) {
4562                 # Strip bad stuff out of the title (tooltip).  We can't just use
4563                 # replaceLinkHoldersText() here, because if this function is called
4564                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4565                 if ( $holders ) {
4566                         $tooltip = $holders->replaceText( $caption );
4567                 } else {
4568                         $tooltip = $this->replaceLinkHoldersText( $caption );
4569                 }
4570
4571                 # make sure there are no placeholders in thumbnail attributes
4572                 # that are later expanded to html- so expand them now and
4573                 # remove the tags
4574                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4575                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4576
4577                 return $tooltip;
4578         }
4579
4580         /**
4581          * Set a flag in the output object indicating that the content is dynamic and
4582          * shouldn't be cached.
4583          */
4584         function disableCache() {
4585                 wfDebug( "Parser output marked as uncacheable.\n" );
4586                 $this->mOutput->mCacheTime = -1;
4587         }
4588
4589         /**#@+
4590          * Callback from the Sanitizer for expanding items found in HTML attribute
4591          * values, so they can be safely tested and escaped.
4592          * @param string $text
4593          * @param PPFrame $frame
4594          * @return string
4595          * @private
4596          */
4597         function attributeStripCallback( &$text, $frame = false ) {
4598                 $text = $this->replaceVariables( $text, $frame );
4599                 $text = $this->mStripState->unstripBoth( $text );
4600                 return $text;
4601         }
4602
4603         /**#@-*/
4604
4605         /**#@+
4606          * Accessor/mutator
4607          */
4608         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4609         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4610         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4611         /**#@-*/
4612
4613         /**#@+
4614          * Accessor
4615          */
4616         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4617         /**#@-*/
4618
4619
4620         /**
4621          * Break wikitext input into sections, and either pull or replace
4622          * some particular section's text.
4623          *
4624          * External callers should use the getSection and replaceSection methods.
4625          *
4626          * @param string $text Page wikitext
4627          * @param string $section A section identifier string of the form:
4628          *   <flag1> - <flag2> - ... - <section number>
4629          *
4630          * Currently the only recognised flag is "T", which means the target section number
4631          * was derived during a template inclusion parse, in other words this is a template
4632          * section edit link. If no flags are given, it was an ordinary section edit link.
4633          * This flag is required to avoid a section numbering mismatch when a section is
4634          * enclosed by <includeonly> (bug 6563).
4635          *
4636          * The section number 0 pulls the text before the first heading; other numbers will
4637          * pull the given section along with its lower-level subsections. If the section is
4638          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4639          *
4640          * @param string $mode One of "get" or "replace"
4641          * @param string $newText Replacement text for section data.
4642          * @return string for "get", the extracted section text.
4643          *                for "replace", the whole page with the section replaced.
4644          */
4645         private function extractSections( $text, $section, $mode, $newText='' ) {
4646                 global $wgTitle;
4647                 $this->clearState();
4648                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4649                 $this->mOptions = new ParserOptions;
4650                 $this->setOutputType( self::OT_WIKI );
4651                 $outText = '';
4652                 $frame = $this->getPreprocessor()->newFrame();
4653
4654                 // Process section extraction flags
4655                 $flags = 0;
4656                 $sectionParts = explode( '-', $section );
4657                 $sectionIndex = array_pop( $sectionParts );
4658                 foreach ( $sectionParts as $part ) {
4659                         if ( $part === 'T' ) {
4660                                 $flags |= self::PTD_FOR_INCLUSION;
4661                         }
4662                 }
4663                 // Preprocess the text
4664                 $root = $this->preprocessToDom( $text, $flags );
4665
4666                 // <h> nodes indicate section breaks
4667                 // They can only occur at the top level, so we can find them by iterating the root's children
4668                 $node = $root->getFirstChild();
4669
4670                 // Find the target section
4671                 if ( $sectionIndex == 0 ) {
4672                         // Section zero doesn't nest, level=big
4673                         $targetLevel = 1000;
4674                 } else {
4675             while ( $node ) {
4676                 if ( $node->getName() === 'h' ) {
4677                     $bits = $node->splitHeading();
4678                                         if ( $bits['i'] == $sectionIndex ) {
4679                                         $targetLevel = $bits['level'];
4680                                                 break;
4681                                         }
4682                                 }
4683                                 if ( $mode === 'replace' ) {
4684                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4685                                 }
4686                                 $node = $node->getNextSibling();
4687                         }
4688                 }
4689
4690                 if ( !$node ) {
4691                         // Not found
4692                         if ( $mode === 'get' ) {
4693                                 return $newText;
4694                         } else {
4695                                 return $text;
4696                         }
4697                 }
4698
4699                 // Find the end of the section, including nested sections
4700                 do {
4701                         if ( $node->getName() === 'h' ) {
4702                                 $bits = $node->splitHeading();
4703                                 $curLevel = $bits['level'];
4704                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4705                                         break;
4706                                 }
4707                         }
4708                         if ( $mode === 'get' ) {
4709                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4710                         }
4711                         $node = $node->getNextSibling();
4712                 } while ( $node );
4713
4714                 // Write out the remainder (in replace mode only)
4715                 if ( $mode === 'replace' ) {
4716                         // Output the replacement text
4717                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4718                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4719                         // Only add trailing whitespace if there is newText
4720                         if($newText != "") {
4721                                 $outText .= $newText . "\n\n";
4722                         }
4723
4724                         while ( $node ) {
4725                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4726                                 $node = $node->getNextSibling();
4727                         }
4728                 }
4729
4730                 if ( is_string( $outText ) ) {
4731                         // Re-insert stripped tags
4732                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4733                 }
4734
4735                 return $outText;
4736         }
4737
4738         /**
4739          * This function returns the text of a section, specified by a number ($section).
4740          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4741          * the first section before any such heading (section 0).
4742          *
4743          * If a section contains subsections, these are also returned.
4744          *
4745          * @param string $text text to look in
4746          * @param string $section section identifier
4747          * @param string $deftext default to return if section is not found
4748          * @return string text of the requested section
4749          */
4750         public function getSection( $text, $section, $deftext='' ) {
4751                 return $this->extractSections( $text, $section, "get", $deftext );
4752         }
4753
4754         public function replaceSection( $oldtext, $section, $text ) {
4755                 return $this->extractSections( $oldtext, $section, "replace", $text );
4756         }
4757
4758         /**
4759          * Get the timestamp associated with the current revision, adjusted for
4760          * the default server-local timestamp
4761          */
4762         function getRevisionTimestamp() {
4763                 if ( is_null( $this->mRevisionTimestamp ) ) {
4764                         wfProfileIn( __METHOD__ );
4765                         global $wgContLang;
4766                         $dbr = wfGetDB( DB_SLAVE );
4767                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4768                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4769
4770                         // Normalize timestamp to internal MW format for timezone processing.
4771                         // This has the added side-effect of replacing a null value with
4772                         // the current time, which gives us more sensible behavior for
4773                         // previews.
4774                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4775
4776                         // The cryptic '' timezone parameter tells to use the site-default
4777                         // timezone offset instead of the user settings.
4778                         //
4779                         // Since this value will be saved into the parser cache, served
4780                         // to other users, and potentially even used inside links and such,
4781                         // it needs to be consistent for all visitors.
4782                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4783
4784                         wfProfileOut( __METHOD__ );
4785                 }
4786                 return $this->mRevisionTimestamp;
4787         }
4788
4789         /**
4790          * Get the name of the user that edited the last revision
4791          */
4792         function getRevisionUser() {
4793                 // if this template is subst: the revision id will be blank,
4794                 // so just use the current user's name
4795                 if( $this->mRevisionId ) {
4796                         $revision = Revision::newFromId( $this->mRevisionId );
4797                         $revuser = $revision->getUserText();
4798                 } else {
4799                         global $wgUser;
4800                         $revuser = $wgUser->getName();
4801                 }
4802                 return $revuser;
4803         }
4804
4805         /**
4806          * Mutator for $mDefaultSort
4807          *
4808          * @param $sort New value
4809          */
4810         public function setDefaultSort( $sort ) {
4811                 $this->mDefaultSort = $sort;
4812         }
4813
4814         /**
4815          * Accessor for $mDefaultSort
4816          * Will use the title/prefixed title if none is set
4817          *
4818          * @return string
4819          */
4820         public function getDefaultSort() {
4821                 global $wgCategoryPrefixedDefaultSortkey;
4822                 if( $this->mDefaultSort !== false ) {
4823                         return $this->mDefaultSort;
4824                 } elseif ($this->mTitle->getNamespace() == NS_CATEGORY ||
4825                         !$wgCategoryPrefixedDefaultSortkey) {
4826                         return $this->mTitle->getText();
4827                 } else {
4828                         return $this->mTitle->getPrefixedText();
4829                 }
4830         }
4831
4832         /**
4833          * Accessor for $mDefaultSort
4834          * Unlike getDefaultSort(), will return false if none is set
4835          *
4836          * @return string or false
4837          */
4838         public function getCustomDefaultSort() {
4839                 return $this->mDefaultSort;
4840         }
4841
4842         /**
4843          * Try to guess the section anchor name based on a wikitext fragment
4844          * presumably extracted from a heading, for example "Header" from
4845          * "== Header ==".
4846          */
4847         public function guessSectionNameFromWikiText( $text ) {
4848                 # Strip out wikitext links(they break the anchor)
4849                 $text = $this->stripSectionName( $text );
4850                 $headline = Sanitizer::decodeCharReferences( $text );
4851                 # strip out HTML
4852                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
4853                 $headline = trim( $headline );
4854                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
4855                 $replacearray = array(
4856                         '%3A' => ':',
4857                         '%' => '.'
4858                 );
4859                 return str_replace(
4860                         array_keys( $replacearray ),
4861                         array_values( $replacearray ),
4862                         $sectionanchor );
4863         }
4864
4865         /**
4866          * Strips a text string of wikitext for use in a section anchor
4867          *
4868          * Accepts a text string and then removes all wikitext from the
4869          * string and leaves only the resultant text (i.e. the result of
4870          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
4871          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
4872          * to create valid section anchors by mimicing the output of the
4873          * parser when headings are parsed.
4874          *
4875          * @param $text string Text string to be stripped of wikitext
4876          * for use in a Section anchor
4877          * @return Filtered text string
4878          */
4879         public function stripSectionName( $text ) {
4880                 # Strip internal link markup
4881                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
4882                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
4883
4884                 # Strip external link markup (FIXME: Not Tolerant to blank link text
4885                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
4886                 # on how many empty links there are on the page - need to figure that out.
4887                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
4888
4889                 # Parse wikitext quotes (italics & bold)
4890                 $text = $this->doQuotes($text);
4891
4892                 # Strip HTML tags
4893                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
4894                 return $text;
4895         }
4896
4897         function srvus( $text ) {
4898                 return $this->testSrvus( $text, $this->mOutputType );
4899         }
4900
4901         /**
4902          * strip/replaceVariables/unstrip for preprocessor regression testing
4903          */
4904         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
4905                 $this->clearState();
4906                 if ( ! ( $title instanceof Title ) ) {
4907                         $title = Title::newFromText( $title );
4908                 }
4909                 $this->mTitle = $title;
4910                 $this->mOptions = $options;
4911                 $this->setOutputType( $outputType );
4912                 $text = $this->replaceVariables( $text );
4913                 $text = $this->mStripState->unstripBoth( $text );
4914                 $text = Sanitizer::removeHTMLtags( $text );
4915                 return $text;
4916         }
4917
4918         function testPst( $text, $title, $options ) {
4919                 global $wgUser;
4920                 if ( ! ( $title instanceof Title ) ) {
4921                         $title = Title::newFromText( $title );
4922                 }
4923                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
4924         }
4925
4926         function testPreprocess( $text, $title, $options ) {
4927                 if ( ! ( $title instanceof Title ) ) {
4928                         $title = Title::newFromText( $title );
4929                 }
4930                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
4931         }
4932
4933         function markerSkipCallback( $s, $callback ) {
4934                 $i = 0;
4935                 $out = '';
4936                 while ( $i < strlen( $s ) ) {
4937                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
4938                         if ( $markerStart === false ) {
4939                                 $out .= call_user_func( $callback, substr( $s, $i ) );
4940                                 break;
4941                         } else {
4942                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
4943                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
4944                                 if ( $markerEnd === false ) {
4945                                         $out .= substr( $s, $markerStart );
4946                                         break;
4947                                 } else {
4948                                         $markerEnd += strlen( self::MARKER_SUFFIX );
4949                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
4950                                         $i = $markerEnd;
4951                                 }
4952                         }
4953                 }
4954                 return $out;
4955         }
4956
4957         function serialiseHalfParsedText( $text ) {
4958                 $data = array();
4959                 $data['text'] = $text;
4960
4961                 // First, find all strip markers, and store their
4962                 //  data in an array.
4963                 $stripState = new StripState;
4964                 $pos = 0;
4965                 while( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) ) && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) ) {
4966                         $end_pos += strlen( self::MARKER_SUFFIX );
4967                         $marker = substr( $text, $start_pos, $end_pos-$start_pos );
4968
4969                         if ( !empty( $this->mStripState->general->data[$marker] ) ) {
4970                                 $replaceArray = $stripState->general;
4971                                 $stripText = $this->mStripState->general->data[$marker];
4972                         } elseif ( !empty( $this->mStripState->nowiki->data[$marker] ) ) {
4973                                 $replaceArray = $stripState->nowiki;
4974                                 $stripText = $this->mStripState->nowiki->data[$marker];
4975                         } else {
4976                                 throw new MWException( "Hanging strip marker: '$marker'." );
4977                         }
4978
4979                         $replaceArray->setPair( $marker, $stripText );
4980                         $pos = $end_pos;
4981                 }
4982                 $data['stripstate'] = $stripState;
4983
4984                 // Now, find all of our links, and store THEIR
4985                 //  data in an array! :)
4986                 $links = array( 'internal' => array(), 'interwiki' => array() );
4987                 $pos = 0;
4988
4989                 // Internal links
4990                 while( ( $start_pos = strpos( $text, '<!--LINK ', $pos ) ) ) {
4991                         list( $ns, $trail ) = explode( ':', substr( $text, $start_pos + strlen( '<!--LINK ' ) ), 2 );
4992
4993                         $ns = trim($ns);
4994                         if (empty( $links['internal'][$ns] )) {
4995                                 $links['internal'][$ns] = array();
4996                         }
4997
4998                         $key = trim( substr( $trail, 0, strpos( $trail, '-->' ) ) );
4999                         $links['internal'][$ns][] = $this->mLinkHolders->internals[$ns][$key];
5000                         $pos = $start_pos + strlen( "<!--LINK $ns:$key-->" );
5001                 }
5002
5003                 $pos = 0;
5004
5005                 // Interwiki links
5006                 while( ( $start_pos = strpos( $text, '<!--IWLINK ', $pos ) ) ) {
5007                         $data = substr( $text, $start_pos );
5008                         $key = trim( substr( $data, 0, strpos( $data, '-->' ) ) );
5009                         $links['interwiki'][] = $this->mLinkHolders->interwiki[$key];
5010                         $pos = $start_pos + strlen( "<!--IWLINK $key-->" );
5011                 }
5012
5013                 $data['linkholder'] = $links;
5014
5015                 return $data;
5016         }
5017
5018         function unserialiseHalfParsedText( $data, $intPrefix = null /* Unique identifying prefix */ ) {
5019                 if (!$intPrefix)
5020                         $intPrefix = $this->getRandomString();
5021
5022                 // First, extract the strip state.
5023                 $stripState = $data['stripstate'];
5024                 $this->mStripState->general->merge( $stripState->general );
5025                 $this->mStripState->nowiki->merge( $stripState->nowiki );
5026
5027                 // Now, extract the text, and renumber links
5028                 $text = $data['text'];
5029                 $links = $data['linkholder'];
5030
5031                 // Internal...
5032                 foreach( $links['internal'] as $ns => $nsLinks ) {
5033                         foreach( $nsLinks as $key => $entry ) {
5034                                 $newKey = $intPrefix . '-' . $key;
5035                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
5036
5037                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
5038                         }
5039                 }
5040
5041                 // Interwiki...
5042                 foreach( $links['interwiki'] as $key => $entry ) {
5043                         $newKey = "$intPrefix-$key";
5044                         $this->mLinkHolders->interwikis[$newKey] = $entry;
5045
5046                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
5047                 }
5048
5049                 // Should be good to go.
5050                 return $text;
5051         }
5052 }
5053
5054 /**
5055  * @todo document, briefly.
5056  * @ingroup Parser
5057  */
5058 class StripState {
5059         var $general, $nowiki;
5060
5061         function __construct() {
5062                 $this->general = new ReplacementArray;
5063                 $this->nowiki = new ReplacementArray;
5064         }
5065
5066         function unstripGeneral( $text ) {
5067                 wfProfileIn( __METHOD__ );
5068                 do {
5069                         $oldText = $text;
5070                         $text = $this->general->replace( $text );
5071                 } while ( $text !== $oldText );
5072                 wfProfileOut( __METHOD__ );
5073                 return $text;
5074         }
5075
5076         function unstripNoWiki( $text ) {
5077                 wfProfileIn( __METHOD__ );
5078                 do {
5079                         $oldText = $text;
5080                         $text = $this->nowiki->replace( $text );
5081                 } while ( $text !== $oldText );
5082                 wfProfileOut( __METHOD__ );
5083                 return $text;
5084         }
5085
5086         function unstripBoth( $text ) {
5087                 wfProfileIn( __METHOD__ );
5088                 do {
5089                         $oldText = $text;
5090                         $text = $this->general->replace( $text );
5091                         $text = $this->nowiki->replace( $text );
5092                 } while ( $text !== $oldText );
5093                 wfProfileOut( __METHOD__ );
5094                 return $text;
5095         }
5096 }
5097
5098 /**
5099  * @todo document, briefly.
5100  * @ingroup Parser
5101  */
5102 class OnlyIncludeReplacer {
5103         var $output = '';
5104
5105         function replace( $matches ) {
5106                 if ( substr( $matches[1], -1 ) === "\n" ) {
5107                         $this->output .= substr( $matches[1], 0, -1 );
5108                 } else {
5109                         $this->output .= $matches[1];
5110                 }
5111         }
5112 }