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
 108         # Temporary
 109         # These are variables reset at least once per parse regardless of $clearState
 110         var $mOptions,      // ParserOptions object
 111                 $mTitle,        // Title context, used for self-link rendering and similar things
 112                 $mOutputType,   // Output type, one of the OT_xxx constants
 113                 $ot,            // Shortcut alias, see setOutputType()
 114                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 115                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 116                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 117
 118         /**#@-*/
 119
 120         /**
 121          * Constructor
 122          *
 123          * @public
 124          */
 125         function __construct( $conf = array() ) {
 126                 $this->mConf = $conf;
 127                 $this->mTagHooks = array();
 128                 $this->mTransparentTagHooks = array();
 129                 $this->mFunctionHooks = array();
 130                 $this->mFunctionTagHooks = array();
 131                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 132                 $this->mDefaultStripList = $this->mStripList = array( 'nowiki', 'gallery' );
 133                 $this->mUrlProtocols = wfUrlProtocols();
 134                 $this->mExtLinkBracketedRegex = '/\[(\b(' . wfUrlProtocols() . ')'.
 135                         '[^][<>"\\x00-\\x20\\x7F]+) *([^\]\\x0a\\x0d]*?)\]/S';
 136                 $this->mVarCache = array();
 137                 if ( isset( $conf['preprocessorClass'] ) ) {
 138                         $this->mPreprocessorClass = $conf['preprocessorClass'];
 139                 } elseif ( extension_loaded( 'domxml' ) ) {
 140                         // PECL extension that conflicts with the core DOM extension (bug 13770)
 141                         wfDebug( "Warning: you have the obsolete domxml extension for PHP. Please remove it!\n" );
 142                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 143                 } elseif ( extension_loaded( 'dom' ) ) {
 144                         $this->mPreprocessorClass = 'Preprocessor_DOM';
 145                 } else {
 146                         $this->mPreprocessorClass = 'Preprocessor_Hash';
 147                 }
 148                 $this->mMarkerIndex = 0;
 149                 $this->mFirstCall = true;
 150         }
 151
 152         /**
 153          * Reduce memory usage to reduce the impact of circular references
 154          */
 155         function __destruct() {
 156                 if ( isset( $this->mLinkHolders ) ) {
 157                         $this->mLinkHolders->__destruct();
 158                 }
 159                 foreach ( $this as $name => $value ) {
 160                         unset( $this->$name );
 161                 }
 162         }
 163
 164         /**
 165          * Do various kinds of initialisation on the first call of the parser
 166          */
 167         function firstCallInit() {
 168                 if ( !$this->mFirstCall ) {
 169                         return;
 170                 }
 171                 $this->mFirstCall = false;
 172
 173                 wfProfileIn( __METHOD__ );
 174
 175                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 176                 CoreParserFunctions::register( $this );
 177                 $this->initialiseVariables();
 178
 179                 wfRunHooks( 'ParserFirstCallInit', array( &$this ) );
 180                 wfProfileOut( __METHOD__ );
 181         }
 182
 183         /**
 184          * Clear Parser state
 185          *
 186          * @private
 187          */
 188         function clearState() {
 189                 wfProfileIn( __METHOD__ );
 190                 if ( $this->mFirstCall ) {
 191                         $this->firstCallInit();
 192                 }
 193                 $this->mOutput = new ParserOutput;
 194                 $this->mAutonumber = 0;
 195                 $this->mLastSection = '';
 196                 $this->mDTopen = false;
 197                 $this->mIncludeCount = array();
 198                 $this->mStripState = new StripState;
 199                 $this->mArgStack = false;
 200                 $this->mInPre = false;
 201                 $this->mLinkHolders = new LinkHolderArray( $this );
 202                 $this->mLinkID = 0;
 203                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 204
 205                 /**
 206                  * Prefix for temporary replacement strings for the multipass parser.
 207                  * \x07 should never appear in input as it's disallowed in XML.
 208                  * Using it at the front also gives us a little extra robustness
 209                  * since it shouldn't match when butted up against identifier-like
 210                  * string constructs.
 211                  *
 212                  * Must not consist of all title characters, or else it will change
 213                  * the behaviour of <nowiki> in a link.
 214                  */
 215                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 216                 # Changed to \x7f to allow XML double-parsing -- TS
 217                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 218
 219
 220                 # Clear these on every parse, bug 4549
 221                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 222
 223                 $this->mShowToc = true;
 224                 $this->mForceTocPosition = false;
 225                 $this->mIncludeSizes = array(
 226                         'post-expand' => 0,
 227                         'arg' => 0,
 228                 );
 229                 $this->mPPNodeCount = 0;
 230                 $this->mDefaultSort = false;
 231                 $this->mHeadings = array();
 232                 $this->mDoubleUnderscores = array();
 233                 $this->mExpensiveFunctionCount = 0;
 234
 235                 # Fix cloning
 236                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 237                         $this->mPreprocessor = null;
 238                 }
 239
 240                 wfRunHooks( 'ParserClearState', array( &$this ) );
 241                 wfProfileOut( __METHOD__ );
 242         }
 243
 244         function setOutputType( $ot ) {
 245                 $this->mOutputType = $ot;
 246                 // Shortcut alias
 247                 $this->ot = array(
 248                         'html' => $ot == self::OT_HTML,
 249                         'wiki' => $ot == self::OT_WIKI,
 250                         'pre' => $ot == self::OT_PREPROCESS,
 251                 );
 252         }
 253
 254         /**
 255          * Set the context title
 256          */
 257         function setTitle( $t ) {
 258                 if ( !$t || $t instanceof FakeTitle ) {
 259                         $t = Title::newFromText( 'NO TITLE' );
 260                 }
 261                 if ( strval( $t->getFragment() ) !== '' ) {
 262                         # Strip the fragment to avoid various odd effects
 263                         $this->mTitle = clone $t;
 264                         $this->mTitle->setFragment( '' );
 265                 } else {
 266                         $this->mTitle = $t;
 267                 }
 268         }
 269
 270         /**
 271          * Accessor for mUniqPrefix.
 272          *
 273          * @public
 274          */
 275         function uniqPrefix() {
 276                 if( !isset( $this->mUniqPrefix ) ) {
 277                         // @fixme this is probably *horribly wrong*
 278                         // LanguageConverter seems to want $wgParser's uniqPrefix, however
 279                         // if this is called for a parser cache hit, the parser may not
 280                         // have ever been initialized in the first place.
 281                         // Not really sure what the heck is supposed to be going on here.
 282                         return '';
 283                         //throw new MWException( "Accessing uninitialized mUniqPrefix" );
 284                 }
 285                 return $this->mUniqPrefix;
 286         }
 287
 288         /**
 289          * Convert wikitext to HTML
 290          * Do not call this function recursively.
 291          *
 292          * @param $text String: text we want to parse
 293          * @param $title A title object
 294          * @param $options ParserOptions
 295          * @param $linestart boolean
 296          * @param $clearState boolean
 297          * @param $revid Int: number to pass in {{REVISIONID}}
 298          * @return ParserOutput a ParserOutput
 299          */
 300         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 301                 /**
 302                  * First pass--just handle <nowiki> sections, pass the rest off
 303                  * to internalParse() which does all the real work.
 304                  */
 305
 306                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 307                 $fname = __METHOD__.'-' . wfGetCaller();
 308                 wfProfileIn( __METHOD__ );
 309                 wfProfileIn( $fname );
 310
 311                 if ( $clearState ) {
 312                         $this->clearState();
 313                 }
 314
 315                 $this->mOptions = $options;
 316                 $this->setTitle( $title );
 317                 $oldRevisionId = $this->mRevisionId;
 318                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 319                 if( $revid !== null ) {
 320                         $this->mRevisionId = $revid;
 321                         $this->mRevisionTimestamp = null;
 322                 }
 323                 $this->setOutputType( self::OT_HTML );
 324                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 325                 # No more strip!
 326                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 327                 $text = $this->internalParse( $text );
 328                 $text = $this->mStripState->unstripGeneral( $text );
 329
 330                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 331                 $fixtags = array(
 332                         # french spaces, last one Guillemet-left
 333                         # only if there is something before the space
 334                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 335                         # french spaces, Guillemet-right
 336                         '/(\\302\\253) /' => '\\1&nbsp;',
 337                         '/&nbsp;(!\s*important)/' => ' \\1', #Beware of CSS magic word !important, bug #11874.
 338                 );
 339                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 340
 341                 $text = $this->doBlockLevels( $text, $linestart );
 342
 343                 $this->replaceLinkHolders( $text );
 344
 345                 # the position of the parserConvert() call should not be changed. it
 346                 # assumes that the links are all replaced and the only thing left
 347                 # is the <nowiki> mark.
 348                 # Side-effects: this calls $this->mOutput->setTitleText()
 349                 $text = $wgContLang->parserConvert( $text, $this );
 350
 351                 $text = $this->mStripState->unstripNoWiki( $text );
 352
 353                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 354
 355 //!JF Move to its own function
 356
 357                 $uniq_prefix = $this->mUniqPrefix;
 358                 $matches = array();
 359                 $elements = array_keys( $this->mTransparentTagHooks );
 360                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 361
 362                 foreach( $matches as $marker => $data ) {
 363                         list( $element, $content, $params, $tag ) = $data;
 364                         $tagName = strtolower( $element );
 365                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 366                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 367                                         array( $content, $params, $this ) );
 368                         } else {
 369                                 $output = $tag;
 370                         }
 371                         $this->mStripState->general->setPair( $marker, $output );
 372                 }
 373                 $text = $this->mStripState->unstripGeneral( $text );
 374
 375                 $text = Sanitizer::normalizeCharReferences( $text );
 376
 377                 if ( ( $wgUseTidy && $this->mOptions->mTidy ) || $wgAlwaysUseTidy ) {
 378                         $text = MWTidy::tidy( $text );
 379                 } else {
 380                         # attempt to sanitize at least some nesting problems
 381                         # (bug #2702 and quite a few others)
 382                         $tidyregs = array(
 383                                 # ''Something [http://www.cool.com cool''] -->
 384                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 385                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 386                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 387                                 # fix up an anchor inside another anchor, only
 388                                 # at least for a single single nested link (bug 3695)
 389                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 390                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 391                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 392                                 # contains a div, so fix it up here; replace
 393                                 # div with escaped text
 394                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 395                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 396                                 # remove empty italic or bold tag pairs, some
 397                                 # introduced by rules above
 398                                 '/<([bi])><\/\\1>/' => '',
 399                         );
 400
 401                         $text = preg_replace(
 402                                 array_keys( $tidyregs ),
 403                                 array_values( $tidyregs ),
 404                                 $text );
 405                 }
 406                 global $wgExpensiveParserFunctionLimit;
 407                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 408                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 409                 }
 410
 411                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 412
 413                 # Information on include size limits, for the benefit of users who try to skirt them
 414                 if ( $this->mOptions->getEnableLimitReport() ) {
 415                         global $wgExpensiveParserFunctionLimit;
 416                         $max = $this->mOptions->getMaxIncludeSize();
 417                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 418                         $limitReport =
 419                                 "NewPP limit report\n" .
 420                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 421                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 422                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 423                                 $PFreport;
 424                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 425                         $text .= "\n<!-- \n$limitReport-->\n";
 426                 }
 427                 $this->mOutput->setText( $text );
 428                 $this->mRevisionId = $oldRevisionId;
 429                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 430                 wfProfileOut( $fname );
 431                 wfProfileOut( __METHOD__ );
 432
 433                 return $this->mOutput;
 434         }
 435
 436         /**
 437          * Recursive parser entry point that can be called from an extension tag
 438          * hook.
 439          *
 440          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 441          *
 442          * @param $text String: text extension wants to have parsed
 443          * @param PPFrame $frame: The frame to use for expanding any template variables
 444          */
 445         function recursiveTagParse( $text, $frame=false ) {
 446                 wfProfileIn( __METHOD__ );
 447                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 448                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 449                 $text = $this->internalParse( $text, false, $frame );
 450                 wfProfileOut( __METHOD__ );
 451                 return $text;
 452         }
 453
 454         /**
 455          * Expand templates and variables in the text, producing valid, static wikitext.
 456          * Also removes comments.
 457          */
 458         function preprocess( $text, $title, $options, $revid = null ) {
 459                 wfProfileIn( __METHOD__ );
 460                 $this->clearState();
 461                 $this->setOutputType( self::OT_PREPROCESS );
 462                 $this->mOptions = $options;
 463                 $this->setTitle( $title );
 464                 if( $revid !== null ) {
 465                         $this->mRevisionId = $revid;
 466                 }
 467                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 468                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 469                 $text = $this->replaceVariables( $text );
 470                 $text = $this->mStripState->unstripBoth( $text );
 471                 wfProfileOut( __METHOD__ );
 472                 return $text;
 473         }
 474
 475         /**
 476          * Get a random string
 477          *
 478          * @private
 479          * @static
 480          */
 481         function getRandomString() {
 482                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 483         }
 484
 485         function &getTitle() { return $this->mTitle; }
 486         function getOptions() { return $this->mOptions; }
 487         function getRevisionId() { return $this->mRevisionId; }
 488         function getOutput() { return $this->mOutput; }
 489         function nextLinkID() { return $this->mLinkID++; }
 490
 491         function getFunctionLang() {
 492                 global $wgLang, $wgContLang;
 493
 494                 $target = $this->mOptions->getTargetLanguage();
 495                 if ( $target !== null ) {
 496                         return $target;
 497                 } else {
 498                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 499                 }
 500         }
 501
 502         /**
 503          * Get a preprocessor object
 504          */
 505         function getPreprocessor() {
 506                 if ( !isset( $this->mPreprocessor ) ) {
 507                         $class = $this->mPreprocessorClass;
 508                         $this->mPreprocessor = new $class( $this );
 509                 }
 510                 return $this->mPreprocessor;
 511         }
 512
 513         /**
 514          * Replaces all occurrences of HTML-style comments and the given tags
 515          * in the text with a random marker and returns the next text. The output
 516          * parameter $matches will be an associative array filled with data in
 517          * the form:
 518          *   'UNIQ-xxxxx' => array(
 519          *     'element',
 520          *     'tag content',
 521          *     array( 'param' => 'x' ),
 522          *     '<element param="x">tag content</element>' ) )
 523          *
 524          * @param $elements list of element names. Comments are always extracted.
 525          * @param $text Source text string.
 526          * @param $uniq_prefix
 527          *
 528          * @public
 529          * @static
 530          */
 531         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 532                 static $n = 1;
 533                 $stripped = '';
 534                 $matches = array();
 535
 536                 $taglist = implode( '|', $elements );
 537                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 538
 539                 while ( '' != $text ) {
 540                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 541                         $stripped .= $p[0];
 542                         if( count( $p ) < 5 ) {
 543                                 break;
 544                         }
 545                         if( count( $p ) > 5 ) {
 546                                 // comment
 547                                 $element    = $p[4];
 548                                 $attributes = '';
 549                                 $close      = '';
 550                                 $inside     = $p[5];
 551                         } else {
 552                                 // tag
 553                                 $element    = $p[1];
 554                                 $attributes = $p[2];
 555                                 $close      = $p[3];
 556                                 $inside     = $p[4];
 557                         }
 558
 559                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . self::MARKER_SUFFIX;
 560                         $stripped .= $marker;
 561
 562                         if ( $close === '/>' ) {
 563                                 // Empty element tag, <tag />
 564                                 $content = null;
 565                                 $text = $inside;
 566                                 $tail = null;
 567                         } else {
 568                                 if( $element === '!--' ) {
 569                                         $end = '/(-->)/';
 570                                 } else {
 571                                         $end = "/(<\\/$element\\s*>)/i";
 572                                 }
 573                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 574                                 $content = $q[0];
 575                                 if( count( $q ) < 3 ) {
 576                                         # No end tag -- let it run out to the end of the text.
 577                                         $tail = '';
 578                                         $text = '';
 579                                 } else {
 580                                         $tail = $q[1];
 581                                         $text = $q[2];
 582                                 }
 583                         }
 584
 585                         $matches[$marker] = array( $element,
 586                                 $content,
 587                                 Sanitizer::decodeTagAttributes( $attributes ),
 588                                 "<$element$attributes$close$content$tail" );
 589                 }
 590                 return $stripped;
 591         }
 592
 593         /**
 594          * Get a list of strippable XML-like elements
 595          */
 596         function getStripList() {
 597                 global $wgRawHtml;
 598                 $elements = $this->mStripList;
 599                 if( $wgRawHtml ) {
 600                         $elements[] = 'html';
 601                 }
 602                 if( $this->mOptions->getUseTeX() ) {
 603                         $elements[] = 'math';
 604                 }
 605                 return $elements;
 606         }
 607
 608         /**
 609          * @deprecated use replaceVariables
 610          */
 611         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 612                 return $text;
 613         }
 614
 615         /**
 616          * Restores pre, math, and other extensions removed by strip()
 617          *
 618          * always call unstripNoWiki() after this one
 619          * @private
 620          * @deprecated use $this->mStripState->unstrip()
 621          */
 622         function unstrip( $text, $state ) {
 623                 return $state->unstripGeneral( $text );
 624         }
 625
 626         /**
 627          * Always call this after unstrip() to preserve the order
 628          *
 629          * @private
 630          * @deprecated use $this->mStripState->unstrip()
 631          */
 632         function unstripNoWiki( $text, $state ) {
 633                 return $state->unstripNoWiki( $text );
 634         }
 635
 636         /**
 637          * @deprecated use $this->mStripState->unstripBoth()
 638          */
 639         function unstripForHTML( $text ) {
 640                 return $this->mStripState->unstripBoth( $text );
 641         }
 642
 643         /**
 644          * Add an item to the strip state
 645          * Returns the unique tag which must be inserted into the stripped text
 646          * The tag will be replaced with the original text in unstrip()
 647          *
 648          * @private
 649          */
 650         function insertStripItem( $text ) {
 651                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 652                 $this->mMarkerIndex++;
 653                 $this->mStripState->general->setPair( $rnd, $text );
 654                 return $rnd;
 655         }
 656
 657         /**
 658          * Interface with html tidy
 659          * @deprecated Use MWTidy::tidy()
 660          */
 661         public static function tidy( $text ) {
 662                 wfDeprecated( __METHOD__ );
 663                 return MWTidy::tidy( $text );
 664         }
 665
 666         /**
 667          * parse the wiki syntax used to render tables
 668          *
 669          * @private
 670          */
 671         function doTableStuff ( $text ) {
 672                 wfProfileIn( __METHOD__ );
 673
 674                 $lines = StringUtils::explode( "\n", $text );
 675                 $out = '';
 676                 $td_history = array (); // Is currently a td tag open?
 677                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 678                 $tr_history = array (); // Is currently a tr tag open?
 679                 $tr_attributes = array (); // history of tr attributes
 680                 $has_opened_tr = array(); // Did this table open a <tr> element?
 681                 $indent_level = 0; // indent level of the table
 682
 683                 foreach ( $lines as $outLine ) {
 684                         $line = trim( $outLine );
 685
 686                         if( $line == '' ) { // empty line, go to next line
 687                                 $out .= $outLine."\n";
 688                                 continue;
 689                         }
 690                         $first_character = $line[0];
 691                         $matches = array();
 692
 693                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 694                                 // First check if we are starting a new table
 695                                 $indent_level = strlen( $matches[1] );
 696
 697                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 698                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 699
 700                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 701                                 array_push ( $td_history , false );
 702                                 array_push ( $last_tag_history , '' );
 703                                 array_push ( $tr_history , false );
 704                                 array_push ( $tr_attributes , '' );
 705                                 array_push ( $has_opened_tr , false );
 706                         } else if ( count ( $td_history ) == 0 ) {
 707                                 // Don't do any of the following
 708                                 $out .= $outLine."\n";
 709                                 continue;
 710                         } else if ( substr ( $line , 0 , 2 ) === '|}' ) {
 711                                 // We are ending a table
 712                                 $line = '</table>' . substr ( $line , 2 );
 713                                 $last_tag = array_pop ( $last_tag_history );
 714
 715                                 if ( !array_pop ( $has_opened_tr ) ) {
 716                                         $line = "<tr><td></td></tr>{$line}";
 717                                 }
 718
 719                                 if ( array_pop ( $tr_history ) ) {
 720                                         $line = "</tr>{$line}";
 721                                 }
 722
 723                                 if ( array_pop ( $td_history ) ) {
 724                                         $line = "</{$last_tag}>{$line}";
 725                                 }
 726                                 array_pop ( $tr_attributes );
 727                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 728                         } else if ( substr ( $line , 0 , 2 ) === '|-' ) {
 729                                 // Now we have a table row
 730                                 $line = preg_replace( '#^\|-+#', '', $line );
 731
 732                                 // Whats after the tag is now only attributes
 733                                 $attributes = $this->mStripState->unstripBoth( $line );
 734                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 735                                 array_pop ( $tr_attributes );
 736                                 array_push ( $tr_attributes , $attributes );
 737
 738                                 $line = '';
 739                                 $last_tag = array_pop ( $last_tag_history );
 740                                 array_pop ( $has_opened_tr );
 741                                 array_push ( $has_opened_tr , true );
 742
 743                                 if ( array_pop ( $tr_history ) ) {
 744                                         $line = '</tr>';
 745                                 }
 746
 747                                 if ( array_pop ( $td_history ) ) {
 748                                         $line = "</{$last_tag}>{$line}";
 749                                 }
 750
 751                                 $outLine = $line;
 752                                 array_push ( $tr_history , false );
 753                                 array_push ( $td_history , false );
 754                                 array_push ( $last_tag_history , '' );
 755                         }
 756                         else if ( $first_character === '|' || $first_character === '!' || substr ( $line , 0 , 2 )  === '|+' ) {
 757                                 // This might be cell elements, td, th or captions
 758                                 if ( substr ( $line , 0 , 2 ) === '|+' ) {
 759                                         $first_character = '+';
 760                                         $line = substr ( $line , 1 );
 761                                 }
 762
 763                                 $line = substr ( $line , 1 );
 764
 765                                 if ( $first_character === '!' ) {
 766                                         $line = str_replace ( '!!' , '||' , $line );
 767                                 }
 768
 769                                 // Split up multiple cells on the same line.
 770                                 // FIXME : This can result in improper nesting of tags processed
 771                                 // by earlier parser steps, but should avoid splitting up eg
 772                                 // attribute values containing literal "||".
 773                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 774
 775                                 $outLine = '';
 776
 777                                 // Loop through each table cell
 778                                 foreach ( $cells as $cell )
 779                                 {
 780                                         $previous = '';
 781                                         if ( $first_character !== '+' )
 782                                         {
 783                                                 $tr_after = array_pop ( $tr_attributes );
 784                                                 if ( !array_pop ( $tr_history ) ) {
 785                                                         $previous = "<tr{$tr_after}>\n";
 786                                                 }
 787                                                 array_push ( $tr_history , true );
 788                                                 array_push ( $tr_attributes , '' );
 789                                                 array_pop ( $has_opened_tr );
 790                                                 array_push ( $has_opened_tr , true );
 791                                         }
 792
 793                                         $last_tag = array_pop ( $last_tag_history );
 794
 795                                         if ( array_pop ( $td_history ) ) {
 796                                                 $previous = "</{$last_tag}>{$previous}";
 797                                         }
 798
 799                                         if ( $first_character === '|' ) {
 800                                                 $last_tag = 'td';
 801                                         } else if ( $first_character === '!' ) {
 802                                                 $last_tag = 'th';
 803                                         } else if ( $first_character === '+' ) {
 804                                                 $last_tag = 'caption';
 805                                         } else {
 806                                                 $last_tag = '';
 807                                         }
 808
 809                                         array_push ( $last_tag_history , $last_tag );
 810
 811                                         // A cell could contain both parameters and data
 812                                         $cell_data = explode ( '|' , $cell , 2 );
 813
 814                                         // Bug 553: Note that a '|' inside an invalid link should not
 815                                         // be mistaken as delimiting cell parameters
 816                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 817                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 818                                         } else if ( count ( $cell_data ) == 1 )
 819                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 820                                         else {
 821                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 822                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 823                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 824                                         }
 825
 826                                         $outLine .= $cell;
 827                                         array_push ( $td_history , true );
 828                                 }
 829                         }
 830                         $out .= $outLine . "\n";
 831                 }
 832
 833                 // Closing open td, tr && table
 834                 while ( count ( $td_history ) > 0 )
 835                 {
 836                         if ( array_pop ( $td_history ) ) {
 837                                 $out .= "</td>\n";
 838                         }
 839                         if ( array_pop ( $tr_history ) ) {
 840                                 $out .= "</tr>\n";
 841                         }
 842                         if ( !array_pop ( $has_opened_tr ) ) {
 843                                 $out .= "<tr><td></td></tr>\n" ;
 844                         }
 845
 846                         $out .= "</table>\n";
 847                 }
 848
 849                 // Remove trailing line-ending (b/c)
 850                 if ( substr( $out, -1 ) === "\n" ) {
 851                         $out = substr( $out, 0, -1 );
 852                 }
 853
 854                 // special case: don't return empty table
 855                 if( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 856                         $out = '';
 857                 }
 858
 859                 wfProfileOut( __METHOD__ );
 860
 861                 return $out;
 862         }
 863
 864         /**
 865          * Helper function for parse() that transforms wiki markup into
 866          * HTML. Only called for $mOutputType == self::OT_HTML.
 867          *
 868          * @private
 869          */
 870         function internalParse( $text, $isMain = true, $frame=false ) {
 871                 wfProfileIn( __METHOD__ );
 872
 873                 $origText = $text;
 874
 875                 # Hook to suspend the parser in this state
 876                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 877                         wfProfileOut( __METHOD__ );
 878                         return $text ;
 879                 }
 880
 881                 // if $frame is provided, then use $frame for replacing any variables
 882                 if ($frame) {
 883                         // use frame depth to infer how include/noinclude tags should be handled
 884                         // depth=0 means this is the top-level document; otherwise it's an included document
 885                         if( !$frame->depth )
 886                                 $flag = 0;
 887                         else
 888                                 $flag = Parser::PTD_FOR_INCLUSION;
 889                         $dom = $this->preprocessToDom( $text, $flag );
 890                         $text = $frame->expand( $dom );
 891                 }
 892                 // if $frame is not provided, then use old-style replaceVariables
 893                 else {
 894                         $text = $this->replaceVariables( $text );
 895                 }
 896
 897                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 898                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 899
 900                 // Tables need to come after variable replacement for things to work
 901                 // properly; putting them before other transformations should keep
 902                 // exciting things like link expansions from showing up in surprising
 903                 // places.
 904                 $text = $this->doTableStuff( $text );
 905
 906                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 907
 908                 $text = $this->doDoubleUnderscore( $text );
 909                 $text = $this->doHeadings( $text );
 910                 if( $this->mOptions->getUseDynamicDates() ) {
 911                         $df = DateFormatter::getInstance();
 912                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 913                 }
 914                 $text = $this->doAllQuotes( $text );
 915                 $text = $this->replaceInternalLinks( $text );
 916                 $text = $this->replaceExternalLinks( $text );
 917
 918                 # replaceInternalLinks may sometimes leave behind
 919                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 920                 $text = str_replace($this->mUniqPrefix.'NOPARSE', '', $text);
 921
 922                 $text = $this->doMagicLinks( $text );
 923                 $text = $this->formatHeadings( $text, $origText, $isMain );
 924
 925                 wfProfileOut( __METHOD__ );
 926                 return $text;
 927         }
 928
 929         /**
 930          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 931          * magic external links.
 932          *
 933          * DML
 934          * @private
 935          */
 936         function doMagicLinks( $text ) {
 937                 wfProfileIn( __METHOD__ );
 938                 $prots = $this->mUrlProtocols;
 939                 $urlChar = self::EXT_LINK_URL_CLASS;
 940                 $text = preg_replace_callback(
 941                         '!(?:                           # Start cases
 942                                 (<a.*?</a>) |               # m[1]: Skip link text
 943                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
 944                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
 945                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
 946                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
 947                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 948                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 949                                     [0-9Xx]                 # check digit
 950                                     \b)
 951                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 952                 wfProfileOut( __METHOD__ );
 953                 return $text;
 954         }
 955
 956         function magicLinkCallback( $m ) {
 957                 if ( isset( $m[1] ) && $m[1] !== '' ) {
 958                         # Skip anchor
 959                         return $m[0];
 960                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
 961                         # Skip HTML element
 962                         return $m[0];
 963                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
 964                         # Free external link
 965                         return $this->makeFreeExternalLink( $m[0] );
 966                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
 967                         # RFC or PMID
 968                         $CssClass = '';
 969                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
 970                                 $keyword = 'RFC';
 971                                 $urlmsg = 'rfcurl';
 972                                 $CssClass = 'mw-magiclink-rfc';
 973                                 $id = $m[4];
 974                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
 975                                 $keyword = 'PMID';
 976                                 $urlmsg = 'pubmedurl';
 977                                 $CssClass = 'mw-magiclink-pmid';
 978                                 $id = $m[4];
 979                         } else {
 980                                 throw new MWException( __METHOD__.': unrecognised match type "' .
 981                                         substr($m[0], 0, 20 ) . '"' );
 982                         }
 983                         $url = wfMsg( $urlmsg, $id);
 984                         $sk = $this->mOptions->getSkin();
 985                         $la = $sk->getExternalLinkAttributes( "external $CssClass" );
 986                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
 987                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
 988                         # ISBN
 989                         $isbn = $m[5];
 990                         $num = strtr( $isbn, array(
 991                                 '-' => '',
 992                                 ' ' => '',
 993                                 'x' => 'X',
 994                         ));
 995                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
 996                         return'<a href="' .
 997                                 $titleObj->escapeLocalUrl() .
 998                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
 999                 } else {
1000                         return $m[0];
1001                 }
1002         }
1003
1004         /**
1005          * Make a free external link, given a user-supplied URL
1006          * @return HTML
1007          * @private
1008          */
1009         function makeFreeExternalLink( $url ) {
1010                 global $wgContLang;
1011                 wfProfileIn( __METHOD__ );
1012
1013                 $sk = $this->mOptions->getSkin();
1014                 $trail = '';
1015
1016                 # The characters '<' and '>' (which were escaped by
1017                 # removeHTMLtags()) should not be included in
1018                 # URLs, per RFC 2396.
1019                 $m2 = array();
1020                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1021                         $trail = substr($url, $m2[0][1]) . $trail;
1022                         $url = substr($url, 0, $m2[0][1]);
1023                 }
1024
1025                 # Move trailing punctuation to $trail
1026                 $sep = ',;\.:!?';
1027                 # If there is no left bracket, then consider right brackets fair game too
1028                 if ( strpos( $url, '(' ) === false ) {
1029                         $sep .= ')';
1030                 }
1031
1032                 $numSepChars = strspn( strrev( $url ), $sep );
1033                 if ( $numSepChars ) {
1034                         $trail = substr( $url, -$numSepChars ) . $trail;
1035                         $url = substr( $url, 0, -$numSepChars );
1036                 }
1037
1038                 $url = Sanitizer::cleanUrl( $url );
1039
1040                 # Is this an external image?
1041                 $text = $this->maybeMakeExternalImage( $url );
1042                 if ( $text === false ) {
1043                         # Not an image, make a link
1044                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1045                                 $this->getExternalLinkAttribs( $url ) );
1046                         # Register it in the output object...
1047                         # Replace unnecessary URL escape codes with their equivalent characters
1048                         $pasteurized = self::replaceUnusualEscapes( $url );
1049                         $this->mOutput->addExternalLink( $pasteurized );
1050                 }
1051                 wfProfileOut( __METHOD__ );
1052                 return $text . $trail;
1053         }
1054
1055
1056         /**
1057          * Parse headers and return html
1058          *
1059          * @private
1060          */
1061         function doHeadings( $text ) {
1062                 wfProfileIn( __METHOD__ );
1063                 for ( $i = 6; $i >= 1; --$i ) {
1064                         $h = str_repeat( '=', $i );
1065                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1066                           "<h$i>\\1</h$i>", $text );
1067                 }
1068                 wfProfileOut( __METHOD__ );
1069                 return $text;
1070         }
1071
1072         /**
1073          * Replace single quotes with HTML markup
1074          * @private
1075          * @return string the altered text
1076          */
1077         function doAllQuotes( $text ) {
1078                 wfProfileIn( __METHOD__ );
1079                 $outtext = '';
1080                 $lines = StringUtils::explode( "\n", $text );
1081                 foreach ( $lines as $line ) {
1082                         $outtext .= $this->doQuotes( $line ) . "\n";
1083                 }
1084                 $outtext = substr($outtext, 0,-1);
1085                 wfProfileOut( __METHOD__ );
1086                 return $outtext;
1087         }
1088
1089         /**
1090          * Helper function for doAllQuotes()
1091          */
1092         public function doQuotes( $text ) {
1093                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1094                 if ( count( $arr ) == 1 )
1095                         return $text;
1096                 else
1097                 {
1098                         # First, do some preliminary work. This may shift some apostrophes from
1099                         # being mark-up to being text. It also counts the number of occurrences
1100                         # of bold and italics mark-ups.
1101                         $i = 0;
1102                         $numbold = 0;
1103                         $numitalics = 0;
1104                         foreach ( $arr as $r )
1105                         {
1106                                 if ( ( $i % 2 ) == 1 )
1107                                 {
1108                                         # If there are ever four apostrophes, assume the first is supposed to
1109                                         # be text, and the remaining three constitute mark-up for bold text.
1110                                         if ( strlen( $arr[$i] ) == 4 )
1111                                         {
1112                                                 $arr[$i-1] .= "'";
1113                                                 $arr[$i] = "'''";
1114                                         }
1115                                         # If there are more than 5 apostrophes in a row, assume they're all
1116                                         # text except for the last 5.
1117                                         else if ( strlen( $arr[$i] ) > 5 )
1118                                         {
1119                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1120                                                 $arr[$i] = "'''''";
1121                                         }
1122                                         # Count the number of occurrences of bold and italics mark-ups.
1123                                         # We are not counting sequences of five apostrophes.
1124                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1125                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1126                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1127                                 }
1128                                 $i++;
1129                         }
1130
1131                         # If there is an odd number of both bold and italics, it is likely
1132                         # that one of the bold ones was meant to be an apostrophe followed
1133                         # by italics. Which one we cannot know for certain, but it is more
1134                         # likely to be one that has a single-letter word before it.
1135                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1136                         {
1137                                 $i = 0;
1138                                 $firstsingleletterword = -1;
1139                                 $firstmultiletterword = -1;
1140                                 $firstspace = -1;
1141                                 foreach ( $arr as $r )
1142                                 {
1143                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1144                                         {
1145                                                 $x1 = substr ($arr[$i-1], -1);
1146                                                 $x2 = substr ($arr[$i-1], -2, 1);
1147                                                 if ($x1 === ' ') {
1148                                                         if ($firstspace == -1) $firstspace = $i;
1149                                                 } else if ($x2 === ' ') {
1150                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1151                                                 } else {
1152                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1153                                                 }
1154                                         }
1155                                         $i++;
1156                                 }
1157
1158                                 # If there is a single-letter word, use it!
1159                                 if ($firstsingleletterword > -1)
1160                                 {
1161                                         $arr [ $firstsingleletterword ] = "''";
1162                                         $arr [ $firstsingleletterword-1 ] .= "'";
1163                                 }
1164                                 # If not, but there's a multi-letter word, use that one.
1165                                 else if ($firstmultiletterword > -1)
1166                                 {
1167                                         $arr [ $firstmultiletterword ] = "''";
1168                                         $arr [ $firstmultiletterword-1 ] .= "'";
1169                                 }
1170                                 # ... otherwise use the first one that has neither.
1171                                 # (notice that it is possible for all three to be -1 if, for example,
1172                                 # there is only one pentuple-apostrophe in the line)
1173                                 else if ($firstspace > -1)
1174                                 {
1175                                         $arr [ $firstspace ] = "''";
1176                                         $arr [ $firstspace-1 ] .= "'";
1177                                 }
1178                         }
1179
1180                         # Now let's actually convert our apostrophic mush to HTML!
1181                         $output = '';
1182                         $buffer = '';
1183                         $state = '';
1184                         $i = 0;
1185                         foreach ($arr as $r)
1186                         {
1187                                 if (($i % 2) == 0)
1188                                 {
1189                                         if ($state === 'both')
1190                                                 $buffer .= $r;
1191                                         else
1192                                                 $output .= $r;
1193                                 }
1194                                 else
1195                                 {
1196                                         if (strlen ($r) == 2)
1197                                         {
1198                                                 if ($state === 'i')
1199                                                 { $output .= '</i>'; $state = ''; }
1200                                                 else if ($state === 'bi')
1201                                                 { $output .= '</i>'; $state = 'b'; }
1202                                                 else if ($state === 'ib')
1203                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1204                                                 else if ($state === 'both')
1205                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1206                                                 else # $state can be 'b' or ''
1207                                                 { $output .= '<i>'; $state .= 'i'; }
1208                                         }
1209                                         else if (strlen ($r) == 3)
1210                                         {
1211                                                 if ($state === 'b')
1212                                                 { $output .= '</b>'; $state = ''; }
1213                                                 else if ($state === 'bi')
1214                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1215                                                 else if ($state === 'ib')
1216                                                 { $output .= '</b>'; $state = 'i'; }
1217                                                 else if ($state === 'both')
1218                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1219                                                 else # $state can be 'i' or ''
1220                                                 { $output .= '<b>'; $state .= 'b'; }
1221                                         }
1222                                         else if (strlen ($r) == 5)
1223                                         {
1224                                                 if ($state === 'b')
1225                                                 { $output .= '</b><i>'; $state = 'i'; }
1226                                                 else if ($state === 'i')
1227                                                 { $output .= '</i><b>'; $state = 'b'; }
1228                                                 else if ($state === 'bi')
1229                                                 { $output .= '</i></b>'; $state = ''; }
1230                                                 else if ($state === 'ib')
1231                                                 { $output .= '</b></i>'; $state = ''; }
1232                                                 else if ($state === 'both')
1233                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1234                                                 else # ($state == '')
1235                                                 { $buffer = ''; $state = 'both'; }
1236                                         }
1237                                 }
1238                                 $i++;
1239                         }
1240                         # Now close all remaining tags.  Notice that the order is important.
1241                         if ($state === 'b' || $state === 'ib')
1242                                 $output .= '</b>';
1243                         if ($state === 'i' || $state === 'bi' || $state === 'ib')
1244                                 $output .= '</i>';
1245                         if ($state === 'bi')
1246                                 $output .= '</b>';
1247                         # There might be lonely ''''', so make sure we have a buffer
1248                         if ($state === 'both' && $buffer)
1249                                 $output .= '<b><i>'.$buffer.'</i></b>';
1250                         return $output;
1251                 }
1252         }
1253
1254         /**
1255          * Replace external links (REL)
1256          *
1257          * Note: this is all very hackish and the order of execution matters a lot.
1258          * Make sure to run maintenance/parserTests.php if you change this code.
1259          *
1260          * @private
1261          */
1262         function replaceExternalLinks( $text ) {
1263                 global $wgContLang;
1264                 wfProfileIn( __METHOD__ );
1265
1266                 $sk = $this->mOptions->getSkin();
1267
1268                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1269                 $s = array_shift( $bits );
1270
1271                 $i = 0;
1272                 while ( $i<count( $bits ) ) {
1273                         $url = $bits[$i++];
1274                         $protocol = $bits[$i++];
1275                         $text = $bits[$i++];
1276                         $trail = $bits[$i++];
1277
1278                         # The characters '<' and '>' (which were escaped by
1279                         # removeHTMLtags()) should not be included in
1280                         # URLs, per RFC 2396.
1281                         $m2 = array();
1282                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1283                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1284                                 $url = substr($url, 0, $m2[0][1]);
1285                         }
1286
1287                         # If the link text is an image URL, replace it with an <img> tag
1288                         # This happened by accident in the original parser, but some people used it extensively
1289                         $img = $this->maybeMakeExternalImage( $text );
1290                         if ( $img !== false ) {
1291                                 $text = $img;
1292                         }
1293
1294                         $dtrail = '';
1295
1296                         # Set linktype for CSS - if URL==text, link is essentially free
1297                         $linktype = ($text === $url) ? 'free' : 'text';
1298
1299                         # No link text, e.g. [http://domain.tld/some.link]
1300                         if ( $text == '' ) {
1301                                 # Autonumber if allowed. See bug #5918
1302                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1303                                         $langObj = $this->getFunctionLang();
1304                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1305                                         $linktype = 'autonumber';
1306                                 } else {
1307                                         # Otherwise just use the URL
1308                                         $text = htmlspecialchars( $url );
1309                                         $linktype = 'free';
1310                                 }
1311                         } else {
1312                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1313                                 # Check for trail
1314                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1315                         }
1316
1317                         $text = $wgContLang->markNoConversion($text);
1318
1319                         $url = Sanitizer::cleanUrl( $url );
1320
1321                         # Use the encoded URL
1322                         # This means that users can paste URLs directly into the text
1323                         # Funny characters like &ouml; aren't valid in URLs anyway
1324                         # This was changed in August 2004
1325                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1326                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1327
1328                         # Register link in the output object.
1329                         # Replace unnecessary URL escape codes with the referenced character
1330                         # This prevents spammers from hiding links from the filters
1331                         $pasteurized = self::replaceUnusualEscapes( $url );
1332                         $this->mOutput->addExternalLink( $pasteurized );
1333                 }
1334
1335                 wfProfileOut( __METHOD__ );
1336                 return $s;
1337         }
1338
1339         /**
1340          * Get an associative array of additional HTML attributes appropriate for a
1341          * particular external link.  This currently may include rel => nofollow
1342          * (depending on configuration, namespace, and the URL's domain) and/or a
1343          * target attribute (depending on configuration).
1344          *
1345          * @param string $url Optional URL, to extract the domain from for rel =>
1346          *   nofollow if appropriate
1347          * @return array Associative array of HTML attributes
1348          */
1349         function getExternalLinkAttribs( $url = false ) {
1350                 $attribs = array();
1351                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1352                 $ns = $this->mTitle->getNamespace();
1353                 if( $wgNoFollowLinks && !in_array($ns, $wgNoFollowNsExceptions) ) {
1354                         $attribs['rel'] = 'nofollow';
1355
1356                         global $wgNoFollowDomainExceptions;
1357                         if ( $wgNoFollowDomainExceptions ) {
1358                                 $bits = wfParseUrl( $url );
1359                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1360                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1361                                                 if( substr( $bits['host'], -strlen( $domain ) )
1362                                                 == $domain ) {
1363                                                         unset( $attribs['rel'] );
1364                                                         break;
1365                                                 }
1366                                         }
1367                                 }
1368                         }
1369                 }
1370                 if ( $this->mOptions->getExternalLinkTarget() ) {
1371                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1372                 }
1373                 return $attribs;
1374         }
1375
1376
1377         /**
1378          * Replace unusual URL escape codes with their equivalent characters
1379          * @param string
1380          * @return string
1381          * @static
1382          * @todo  This can merge genuinely required bits in the path or query string,
1383          *        breaking legit URLs. A proper fix would treat the various parts of
1384          *        the URL differently; as a workaround, just use the output for
1385          *        statistical records, not for actual linking/output.
1386          */
1387         static function replaceUnusualEscapes( $url ) {
1388                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1389                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1390         }
1391
1392         /**
1393          * Callback function used in replaceUnusualEscapes().
1394          * Replaces unusual URL escape codes with their equivalent character
1395          * @static
1396          * @private
1397          */
1398         private static function replaceUnusualEscapesCallback( $matches ) {
1399                 $char = urldecode( $matches[0] );
1400                 $ord = ord( $char );
1401                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1402                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1403                         // No, shouldn't be escaped
1404                         return $char;
1405                 } else {
1406                         // Yes, leave it escaped
1407                         return $matches[0];
1408                 }
1409         }
1410
1411         /**
1412          * make an image if it's allowed, either through the global
1413          * option, through the exception, or through the on-wiki whitelist
1414          * @private
1415          */
1416         function maybeMakeExternalImage( $url ) {
1417                 $sk = $this->mOptions->getSkin();
1418                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1419                 $imagesexception = !empty($imagesfrom);
1420                 $text = false;
1421                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1422                 if( $imagesexception && is_array( $imagesfrom ) ) {
1423                         $imagematch = false;
1424                         foreach( $imagesfrom as $match ) {
1425                                 if( strpos( $url, $match ) === 0 ) {
1426                                         $imagematch = true;
1427                                         break;
1428                                 }
1429                         }
1430                 } elseif( $imagesexception ) {
1431                         $imagematch = (strpos( $url, $imagesfrom ) === 0);
1432                 } else {
1433                         $imagematch = false;
1434                 }
1435                 if ( $this->mOptions->getAllowExternalImages()
1436                      || ( $imagesexception && $imagematch ) ) {
1437                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1438                                 # Image found
1439                                 $text = $sk->makeExternalImage( $url );
1440                         }
1441                 }
1442                 if( !$text && $this->mOptions->getEnableImageWhitelist()
1443                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1444                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1445                         foreach( $whitelist as $entry ) {
1446                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1447                                 if( strpos( $entry, '#' ) === 0 || $entry === '' )
1448                                         continue;
1449                                 if( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1450                                         # Image matches a whitelist entry
1451                                         $text = $sk->makeExternalImage( $url );
1452                                         break;
1453                                 }
1454                         }
1455                 }
1456                 return $text;
1457         }
1458
1459         /**
1460          * Process [[ ]] wikilinks
1461          * @return processed text
1462          *
1463          * @private
1464          */
1465         function replaceInternalLinks( $s ) {
1466                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1467                 return $s;
1468         }
1469
1470         /**
1471          * Process [[ ]] wikilinks (RIL)
1472          * @return LinkHolderArray
1473          *
1474          * @private
1475          */
1476         function replaceInternalLinks2( &$s ) {
1477                 global $wgContLang;
1478
1479                 wfProfileIn( __METHOD__ );
1480
1481                 wfProfileIn( __METHOD__.'-setup' );
1482                 static $tc = FALSE, $e1, $e1_img;
1483                 # the % is needed to support urlencoded titles as well
1484                 if ( !$tc ) {
1485                         $tc = Title::legalChars() . '#%';
1486                         # Match a link having the form [[namespace:link|alternate]]trail
1487                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1488                         # Match cases where there is no "]]", which might still be images
1489                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1490                 }
1491
1492                 $sk = $this->mOptions->getSkin();
1493                 $holders = new LinkHolderArray( $this );
1494
1495                 #split the entire text string on occurences of [[
1496                 $a = StringUtils::explode( '[[', ' ' . $s );
1497                 #get the first element (all text up to first [[), and remove the space we added
1498                 $s = $a->current();
1499                 $a->next();
1500                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1501                 $s = substr( $s, 1 );
1502
1503                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1504                 $e2 = null;
1505                 if ( $useLinkPrefixExtension ) {
1506                         # Match the end of a line for a word that's not followed by whitespace,
1507                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1508                         $e2 = wfMsgForContent( 'linkprefix' );
1509                 }
1510
1511                 if( is_null( $this->mTitle ) ) {
1512                         wfProfileOut( __METHOD__.'-setup' );
1513                         wfProfileOut( __METHOD__ );
1514                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1515                 }
1516                 $nottalk = !$this->mTitle->isTalkPage();
1517
1518                 if ( $useLinkPrefixExtension ) {
1519                         $m = array();
1520                         if ( preg_match( $e2, $s, $m ) ) {
1521                                 $first_prefix = $m[2];
1522                         } else {
1523                                 $first_prefix = false;
1524                         }
1525                 } else {
1526                         $prefix = '';
1527                 }
1528
1529                 if($wgContLang->hasVariants()) {
1530                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1531                 } else {
1532                         $selflink = array($this->mTitle->getPrefixedText());
1533                 }
1534                 $useSubpages = $this->areSubpagesAllowed();
1535                 wfProfileOut( __METHOD__.'-setup' );
1536
1537                 # Loop for each link
1538                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1539                         # Check for excessive memory usage
1540                         if ( $holders->isBig() ) {
1541                                 # Too big
1542                                 # Do the existence check, replace the link holders and clear the array
1543                                 $holders->replace( $s );
1544                                 $holders->clear();
1545                         }
1546
1547                         if ( $useLinkPrefixExtension ) {
1548                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1549                                 if ( preg_match( $e2, $s, $m ) ) {
1550                                         $prefix = $m[2];
1551                                         $s = $m[1];
1552                                 } else {
1553                                         $prefix='';
1554                                 }
1555                                 # first link
1556                                 if($first_prefix) {
1557                                         $prefix = $first_prefix;
1558                                         $first_prefix = false;
1559                                 }
1560                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1561                         }
1562
1563                         $might_be_img = false;
1564
1565                         wfProfileIn( __METHOD__."-e1" );
1566                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1567                                 $text = $m[2];
1568                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1569                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1570                                 # the real problem is with the $e1 regex
1571                                 # See bug 1300.
1572                                 #
1573                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1574                                 # and no image is in sight. See bug 2095.
1575                                 #
1576                                 if( $text !== '' &&
1577                                         substr( $m[3], 0, 1 ) === ']' &&
1578                                         strpos($text, '[') !== false
1579                                 )
1580                                 {
1581                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1582                                         $m[3] = substr( $m[3], 1 );
1583                                 }
1584                                 # fix up urlencoded title texts
1585                                 if( strpos( $m[1], '%' ) !== false ) {
1586                                         # Should anchors '#' also be rejected?
1587                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1588                                 }
1589                                 $trail = $m[3];
1590                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1591                                 $might_be_img = true;
1592                                 $text = $m[2];
1593                                 if ( strpos( $m[1], '%' ) !== false ) {
1594                                         $m[1] = urldecode($m[1]);
1595                                 }
1596                                 $trail = "";
1597                         } else { # Invalid form; output directly
1598                                 $s .= $prefix . '[[' . $line ;
1599                                 wfProfileOut( __METHOD__."-e1" );
1600                                 continue;
1601                         }
1602                         wfProfileOut( __METHOD__."-e1" );
1603                         wfProfileIn( __METHOD__."-misc" );
1604
1605                         # Don't allow internal links to pages containing
1606                         # PROTO: where PROTO is a valid URL protocol; these
1607                         # should be external links.
1608                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1609                                 $s .= $prefix . '[[' . $line ;
1610                                 wfProfileOut( __METHOD__."-misc" );
1611                                 continue;
1612                         }
1613
1614                         # Make subpage if necessary
1615                         if ( $useSubpages ) {
1616                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1617                         } else {
1618                                 $link = $m[1];
1619                         }
1620
1621                         $noforce = (substr( $m[1], 0, 1 ) !== ':');
1622                         if (!$noforce) {
1623                                 # Strip off leading ':'
1624                                 $link = substr( $link, 1 );
1625                         }
1626
1627                         wfProfileOut( __METHOD__."-misc" );
1628                         wfProfileIn( __METHOD__."-title" );
1629                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1630                         if ( $nt === NULL ) {
1631                                 $s .= $prefix . '[[' . $line;
1632                                 wfProfileOut( __METHOD__."-title" );
1633                                 continue;
1634                         }
1635
1636                         $ns = $nt->getNamespace();
1637                         $iw = $nt->getInterWiki();
1638                         wfProfileOut( __METHOD__."-title" );
1639
1640                         if ( $might_be_img ) { # if this is actually an invalid link
1641                                 wfProfileIn( __METHOD__."-might_be_img" );
1642                                 if ( $ns == NS_FILE && $noforce ) { #but might be an image
1643                                         $found = false;
1644                                         while ( true ) {
1645                                                 #look at the next 'line' to see if we can close it there
1646                                                 $a->next();
1647                                                 $next_line = $a->current();
1648                                                 if ( $next_line === false || $next_line === null ) {
1649                                                         break;
1650                                                 }
1651                                                 $m = explode( ']]', $next_line, 3 );
1652                                                 if ( count( $m ) == 3 ) {
1653                                                         # the first ]] closes the inner link, the second the image
1654                                                         $found = true;
1655                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1656                                                         $trail = $m[2];
1657                                                         break;
1658                                                 } elseif ( count( $m ) == 2 ) {
1659                                                         #if there's exactly one ]] that's fine, we'll keep looking
1660                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1661                                                 } else {
1662                                                         #if $next_line is invalid too, we need look no further
1663                                                         $text .= '[[' . $next_line;
1664                                                         break;
1665                                                 }
1666                                         }
1667                                         if ( !$found ) {
1668                                                 # we couldn't find the end of this imageLink, so output it raw
1669                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1670                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1671                                                 $s .= "{$prefix}[[$link|$text";
1672                                                 # note: no $trail, because without an end, there *is* no trail
1673                                                 wfProfileOut( __METHOD__."-might_be_img" );
1674                                                 continue;
1675                                         }
1676                                 } else { #it's not an image, so output it raw
1677                                         $s .= "{$prefix}[[$link|$text";
1678                                         # note: no $trail, because without an end, there *is* no trail
1679                                         wfProfileOut( __METHOD__."-might_be_img" );
1680                                         continue;
1681                                 }
1682                                 wfProfileOut( __METHOD__."-might_be_img" );
1683                         }
1684
1685                         $wasblank = ( '' == $text );
1686                         if ( $wasblank ) $text = $link;
1687
1688                         # Link not escaped by : , create the various objects
1689                         if ( $noforce ) {
1690
1691                                 # Interwikis
1692                                 wfProfileIn( __METHOD__."-interwiki" );
1693                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1694                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1695                                         $s = rtrim($s . $prefix);
1696                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1697                                         wfProfileOut( __METHOD__."-interwiki" );
1698                                         continue;
1699                                 }
1700                                 wfProfileOut( __METHOD__."-interwiki" );
1701
1702                                 if ( $ns == NS_FILE ) {
1703                                         wfProfileIn( __METHOD__."-image" );
1704                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1705                                                 if ( $wasblank ) {
1706                                                         # if no parameters were passed, $text
1707                                                         # becomes something like "File:Foo.png",
1708                                                         # which we don't want to pass on to the
1709                                                         # image generator
1710                                                         $text = '';
1711                                                 } else {
1712                                                         # recursively parse links inside the image caption
1713                                                         # actually, this will parse them in any other parameters, too,
1714                                                         # but it might be hard to fix that, and it doesn't matter ATM
1715                                                         $text = $this->replaceExternalLinks($text);
1716                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1717                                                 }
1718                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1719                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1720                                         }
1721                                         $this->mOutput->addImage( $nt->getDBkey() );
1722                                         wfProfileOut( __METHOD__."-image" );
1723                                         continue;
1724
1725                                 }
1726
1727                                 if ( $ns == NS_CATEGORY ) {
1728                                         wfProfileIn( __METHOD__."-category" );
1729                                         $s = preg_replace( "/(\s*\n)+\s*$/D", '', $s ); # bug 87
1730
1731                                         if ( $wasblank ) {
1732                                                 $sortkey = $this->getDefaultSort();
1733                                         } else {
1734                                                 $sortkey = $text;
1735                                         }
1736                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1737                                         $sortkey = str_replace( "\n", '', $sortkey );
1738                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1739                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1740
1741                                         /**
1742                                          * Strip the whitespace Category links produce, see bug 87
1743                                          * @todo We might want to use trim($tmp, "\n") here.
1744                                          */
1745                                         $s .= trim( $prefix . $trail, "\n" ) == '' ? '' : $prefix . $trail;
1746
1747                                         wfProfileOut( __METHOD__."-category" );
1748                                         continue;
1749                                 }
1750                         }
1751
1752                         # Self-link checking
1753                         if( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1754                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1755                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1756                                         continue;
1757                                 }
1758                         }
1759
1760                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1761                         # FIXME: Should do batch file existence checks, see comment below
1762                         if( $ns == NS_MEDIA ) {
1763                                 wfProfileIn( __METHOD__."-media" );
1764                                 # Give extensions a chance to select the file revision for us
1765                                 $skip = $time = false;
1766                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1767                                 if ( $skip ) {
1768                                         $link = $sk->link( $nt );
1769                                 } else {
1770                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1771                                 }
1772                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1773                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1774                                 $this->mOutput->addImage( $nt->getDBkey() );
1775                                 wfProfileOut( __METHOD__."-media" );
1776                                 continue;
1777                         }
1778
1779                         wfProfileIn( __METHOD__."-always_known" );
1780                         # Some titles, such as valid special pages or files in foreign repos, should
1781                         # be shown as bluelinks even though they're not included in the page table
1782                         #
1783                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1784                         # batch file existence checks for NS_FILE and NS_MEDIA
1785                         if( $iw == '' && $nt->isAlwaysKnown() ) {
1786                                 $this->mOutput->addLink( $nt );
1787                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1788                         } else {
1789                                 # Links will be added to the output link list after checking
1790                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1791                         }
1792                         wfProfileOut( __METHOD__."-always_known" );
1793                 }
1794                 wfProfileOut( __METHOD__ );
1795                 return $holders;
1796         }
1797
1798         /**
1799          * Make a link placeholder. The text returned can be later resolved to a real link with
1800          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1801          * parsing of interwiki links, and secondly to allow all existence checks and
1802          * article length checks (for stub links) to be bundled into a single query.
1803          *
1804          * @deprecated
1805          */
1806         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1807                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1808         }
1809
1810         /**
1811          * Render a forced-blue link inline; protect against double expansion of
1812          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1813          * Since this little disaster has to split off the trail text to avoid
1814          * breaking URLs in the following text without breaking trails on the
1815          * wiki links, it's been made into a horrible function.
1816          *
1817          * @param Title $nt
1818          * @param string $text
1819          * @param string $query
1820          * @param string $trail
1821          * @param string $prefix
1822          * @return string HTML-wikitext mix oh yuck
1823          */
1824         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1825                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1826                 $sk = $this->mOptions->getSkin();
1827                 // FIXME: use link() instead of deprecated makeKnownLinkObj()
1828                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1829                 return $this->armorLinks( $link ) . $trail;
1830         }
1831
1832         /**
1833          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1834          * going to go through further parsing steps before inline URL expansion.
1835          *
1836          * Not needed quite as much as it used to be since free links are a bit
1837          * more sensible these days. But bracketed links are still an issue.
1838          *
1839          * @param string more-or-less HTML
1840          * @return string less-or-more HTML with NOPARSE bits
1841          */
1842         function armorLinks( $text ) {
1843                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1844                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1845         }
1846
1847         /**
1848          * Return true if subpage links should be expanded on this page.
1849          * @return bool
1850          */
1851         function areSubpagesAllowed() {
1852                 # Some namespaces don't allow subpages
1853                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1854         }
1855
1856         /**
1857          * Handle link to subpage if necessary
1858          * @param string $target the source of the link
1859          * @param string &$text the link text, modified as necessary
1860          * @return string the full name of the link
1861          * @private
1862          */
1863         function maybeDoSubpageLink($target, &$text) {
1864                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
1865         }
1866
1867         /**#@+
1868          * Used by doBlockLevels()
1869          * @private
1870          */
1871         /* private */ function closeParagraph() {
1872                 $result = '';
1873                 if ( '' != $this->mLastSection ) {
1874                         $result = '</' . $this->mLastSection  . ">\n";
1875                 }
1876                 $this->mInPre = false;
1877                 $this->mLastSection = '';
1878                 return $result;
1879         }
1880         # getCommon() returns the length of the longest common substring
1881         # of both arguments, starting at the beginning of both.
1882         #
1883         /* private */ function getCommon( $st1, $st2 ) {
1884                 $fl = strlen( $st1 );
1885                 $shorter = strlen( $st2 );
1886                 if ( $fl < $shorter ) { $shorter = $fl; }
1887
1888                 for ( $i = 0; $i < $shorter; ++$i ) {
1889                         if ( $st1{$i} != $st2{$i} ) { break; }
1890                 }
1891                 return $i;
1892         }
1893         # These next three functions open, continue, and close the list
1894         # element appropriate to the prefix character passed into them.
1895         #
1896         /* private */ function openList( $char ) {
1897                 $result = $this->closeParagraph();
1898
1899                 if ( '*' === $char ) { $result .= '<ul><li>'; }
1900                 elseif ( '#' === $char ) { $result .= '<ol><li>'; }
1901                 elseif ( ':' === $char ) { $result .= '<dl><dd>'; }
1902                 elseif ( ';' === $char ) {
1903                         $result .= '<dl><dt>';
1904                         $this->mDTopen = true;
1905                 }
1906                 else { $result = '<!-- ERR 1 -->'; }
1907
1908                 return $result;
1909         }
1910
1911         /* private */ function nextItem( $char ) {
1912                 if ( '*' === $char || '#' === $char ) { return '</li><li>'; }
1913                 elseif ( ':' === $char || ';' === $char ) {
1914                         $close = '</dd>';
1915                         if ( $this->mDTopen ) { $close = '</dt>'; }
1916                         if ( ';' === $char ) {
1917                                 $this->mDTopen = true;
1918                                 return $close . '<dt>';
1919                         } else {
1920                                 $this->mDTopen = false;
1921                                 return $close . '<dd>';
1922                         }
1923                 }
1924                 return '<!-- ERR 2 -->';
1925         }
1926
1927         /* private */ function closeList( $char ) {
1928                 if ( '*' === $char ) { $text = '</li></ul>'; }
1929                 elseif ( '#' === $char ) { $text = '</li></ol>'; }
1930                 elseif ( ':' === $char ) {
1931                         if ( $this->mDTopen ) {
1932                                 $this->mDTopen = false;
1933                                 $text = '</dt></dl>';
1934                         } else {
1935                                 $text = '</dd></dl>';
1936                         }
1937                 }
1938                 else {  return '<!-- ERR 3 -->'; }
1939                 return $text."\n";
1940         }
1941         /**#@-*/
1942
1943         /**
1944          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
1945          *
1946          * @param $linestart bool whether or not this is at the start of a line.
1947          * @private
1948          * @return string the lists rendered as HTML
1949          */
1950         function doBlockLevels( $text, $linestart ) {
1951                 wfProfileIn( __METHOD__ );
1952
1953                 # Parsing through the text line by line.  The main thing
1954                 # happening here is handling of block-level elements p, pre,
1955                 # and making lists from lines starting with * # : etc.
1956                 #
1957                 $textLines = StringUtils::explode( "\n", $text );
1958
1959                 $lastPrefix = $output = '';
1960                 $this->mDTopen = $inBlockElem = false;
1961                 $prefixLength = 0;
1962                 $paragraphStack = false;
1963
1964                 foreach ( $textLines as $oLine ) {
1965                         # Fix up $linestart
1966                         if ( !$linestart ) {
1967                                 $output .= $oLine;
1968                                 $linestart = true;
1969                                 continue;
1970                         }
1971                         // * = ul
1972                         // # = ol
1973                         // ; = dt
1974                         // : = dd
1975
1976                         $lastPrefixLength = strlen( $lastPrefix );
1977                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1978                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1979                         // If not in a <pre> element, scan for and figure out what prefixes are there.
1980                         if ( !$this->mInPre ) {
1981                                 # Multiple prefixes may abut each other for nested lists.
1982                                 $prefixLength = strspn( $oLine, '*#:;' );
1983                                 $prefix = substr( $oLine, 0, $prefixLength );
1984
1985                                 # eh?
1986                                 // ; and : are both from definition-lists, so they're equivalent
1987                                 //  for the purposes of determining whether or not we need to open/close
1988                                 //  elements.
1989                                 $prefix2 = str_replace( ';', ':', $prefix );
1990                                 $t = substr( $oLine, $prefixLength );
1991                                 $this->mInPre = (bool)$preOpenMatch;
1992                         } else {
1993                                 # Don't interpret any other prefixes in preformatted text
1994                                 $prefixLength = 0;
1995                                 $prefix = $prefix2 = '';
1996                                 $t = $oLine;
1997                         }
1998
1999                         # List generation
2000                         if( $prefixLength && $lastPrefix === $prefix2 ) {
2001                                 # Same as the last item, so no need to deal with nesting or opening stuff
2002                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2003                                 $paragraphStack = false;
2004
2005                                 if ( substr( $prefix, -1 ) === ';') {
2006                                         # The one nasty exception: definition lists work like this:
2007                                         # ; title : definition text
2008                                         # So we check for : in the remainder text to split up the
2009                                         # title and definition, without b0rking links.
2010                                         $term = $t2 = '';
2011                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2012                                                 $t = $t2;
2013                                                 $output .= $term . $this->nextItem( ':' );
2014                                         }
2015                                 }
2016                         } elseif( $prefixLength || $lastPrefixLength ) {
2017                                 // We need to open or close prefixes, or both.
2018
2019                                 # Either open or close a level...
2020                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2021                                 $paragraphStack = false;
2022
2023                                 // Close all the prefixes which aren't shared.
2024                                 while( $commonPrefixLength < $lastPrefixLength ) {
2025                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2026                                         --$lastPrefixLength;
2027                                 }
2028
2029                                 // Continue the current prefix if appropriate.
2030                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2031                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2032                                 }
2033
2034                                 // Open prefixes where appropriate.
2035                                 while ( $prefixLength > $commonPrefixLength ) {
2036                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2037                                         $output .= $this->openList( $char );
2038
2039                                         if ( ';' === $char ) {
2040                                                 # FIXME: This is dupe of code above
2041                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2042                                                         $t = $t2;
2043                                                         $output .= $term . $this->nextItem( ':' );
2044                                                 }
2045                                         }
2046                                         ++$commonPrefixLength;
2047                                 }
2048                                 $lastPrefix = $prefix2;
2049                         }
2050
2051                         // If we have no prefixes, go to paragraph mode.
2052                         if( 0 == $prefixLength ) {
2053                                 wfProfileIn( __METHOD__."-paragraph" );
2054                                 # No prefix (not in list)--go to paragraph mode
2055                                 // XXX: use a stack for nestable elements like span, table and div
2056                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2057                                 $closematch = preg_match(
2058                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2059                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2060                                 if ( $openmatch or $closematch ) {
2061                                         $paragraphStack = false;
2062                                         # TODO bug 5718: paragraph closed
2063                                         $output .= $this->closeParagraph();
2064                                         if ( $preOpenMatch and !$preCloseMatch ) {
2065                                                 $this->mInPre = true;
2066                                         }
2067                                         if ( $closematch ) {
2068                                                 $inBlockElem = false;
2069                                         } else {
2070                                                 $inBlockElem = true;
2071                                         }
2072                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2073                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' or trim($t) != '' ) ) {
2074                                                 // pre
2075                                                 if ($this->mLastSection !== 'pre') {
2076                                                         $paragraphStack = false;
2077                                                         $output .= $this->closeParagraph().'<pre>';
2078                                                         $this->mLastSection = 'pre';
2079                                                 }
2080                                                 $t = substr( $t, 1 );
2081                                         } else {
2082                                                 // paragraph
2083                                                 if ( '' == trim($t) ) {
2084                                                         if ( $paragraphStack ) {
2085                                                                 $output .= $paragraphStack.'<br />';
2086                                                                 $paragraphStack = false;
2087                                                                 $this->mLastSection = 'p';
2088                                                         } else {
2089                                                                 if ($this->mLastSection !== 'p' ) {
2090                                                                         $output .= $this->closeParagraph();
2091                                                                         $this->mLastSection = '';
2092                                                                         $paragraphStack = '<p>';
2093                                                                 } else {
2094                                                                         $paragraphStack = '</p><p>';
2095                                                                 }
2096                                                         }
2097                                                 } else {
2098                                                         if ( $paragraphStack ) {
2099                                                                 $output .= $paragraphStack;
2100                                                                 $paragraphStack = false;
2101                                                                 $this->mLastSection = 'p';
2102                                                         } else if ($this->mLastSection !== 'p') {
2103                                                                 $output .= $this->closeParagraph().'<p>';
2104                                                                 $this->mLastSection = 'p';
2105                                                         }
2106                                                 }
2107                                         }
2108                                 }
2109                                 wfProfileOut( __METHOD__."-paragraph" );
2110                         }
2111                         // somewhere above we forget to get out of pre block (bug 785)
2112                         if($preCloseMatch && $this->mInPre) {
2113                                 $this->mInPre = false;
2114                         }
2115                         if ($paragraphStack === false) {
2116                                 $output .= $t."\n";
2117                         }
2118                 }
2119                 while ( $prefixLength ) {
2120                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2121                         --$prefixLength;
2122                 }
2123                 if ( '' != $this->mLastSection ) {
2124                         $output .= '</' . $this->mLastSection . '>';
2125                         $this->mLastSection = '';
2126                 }
2127
2128                 wfProfileOut( __METHOD__ );
2129                 return $output;
2130         }
2131
2132         /**
2133          * Split up a string on ':', ignoring any occurences inside tags
2134          * to prevent illegal overlapping.
2135          * @param string $str the string to split
2136          * @param string &$before set to everything before the ':'
2137          * @param string &$after set to everything after the ':'
2138          * return string the position of the ':', or false if none found
2139          */
2140         function findColonNoLinks($str, &$before, &$after) {
2141                 wfProfileIn( __METHOD__ );
2142
2143                 $pos = strpos( $str, ':' );
2144                 if( $pos === false ) {
2145                         // Nothing to find!
2146                         wfProfileOut( __METHOD__ );
2147                         return false;
2148                 }
2149
2150                 $lt = strpos( $str, '<' );
2151                 if( $lt === false || $lt > $pos ) {
2152                         // Easy; no tag nesting to worry about
2153                         $before = substr( $str, 0, $pos );
2154                         $after = substr( $str, $pos+1 );
2155                         wfProfileOut( __METHOD__ );
2156                         return $pos;
2157                 }
2158
2159                 // Ugly state machine to walk through avoiding tags.
2160                 $state = self::COLON_STATE_TEXT;
2161                 $stack = 0;
2162                 $len = strlen( $str );
2163                 for( $i = 0; $i < $len; $i++ ) {
2164                         $c = $str{$i};
2165
2166                         switch( $state ) {
2167                         // (Using the number is a performance hack for common cases)
2168                         case 0: // self::COLON_STATE_TEXT:
2169                                 switch( $c ) {
2170                                 case "<":
2171                                         // Could be either a <start> tag or an </end> tag
2172                                         $state = self::COLON_STATE_TAGSTART;
2173                                         break;
2174                                 case ":":
2175                                         if( $stack == 0 ) {
2176                                                 // We found it!
2177                                                 $before = substr( $str, 0, $i );
2178                                                 $after = substr( $str, $i + 1 );
2179                                                 wfProfileOut( __METHOD__ );
2180                                                 return $i;
2181                                         }
2182                                         // Embedded in a tag; don't break it.
2183                                         break;
2184                                 default:
2185                                         // Skip ahead looking for something interesting
2186                                         $colon = strpos( $str, ':', $i );
2187                                         if( $colon === false ) {
2188                                                 // Nothing else interesting
2189                                                 wfProfileOut( __METHOD__ );
2190                                                 return false;
2191                                         }
2192                                         $lt = strpos( $str, '<', $i );
2193                                         if( $stack === 0 ) {
2194                                                 if( $lt === false || $colon < $lt ) {
2195                                                         // We found it!
2196                                                         $before = substr( $str, 0, $colon );
2197                                                         $after = substr( $str, $colon + 1 );
2198                                                         wfProfileOut( __METHOD__ );
2199                                                         return $i;
2200                                                 }
2201                                         }
2202                                         if( $lt === false ) {
2203                                                 // Nothing else interesting to find; abort!
2204                                                 // We're nested, but there's no close tags left. Abort!
2205                                                 break 2;
2206                                         }
2207                                         // Skip ahead to next tag start
2208                                         $i = $lt;
2209                                         $state = self::COLON_STATE_TAGSTART;
2210                                 }
2211                                 break;
2212                         case 1: // self::COLON_STATE_TAG:
2213                                 // In a <tag>
2214                                 switch( $c ) {
2215                                 case ">":
2216                                         $stack++;
2217                                         $state = self::COLON_STATE_TEXT;
2218                                         break;
2219                                 case "/":
2220                                         // Slash may be followed by >?
2221                                         $state = self::COLON_STATE_TAGSLASH;
2222                                         break;
2223                                 default:
2224                                         // ignore
2225                                 }
2226                                 break;
2227                         case 2: // self::COLON_STATE_TAGSTART:
2228                                 switch( $c ) {
2229                                 case "/":
2230                                         $state = self::COLON_STATE_CLOSETAG;
2231                                         break;
2232                                 case "!":
2233                                         $state = self::COLON_STATE_COMMENT;
2234                                         break;
2235                                 case ">":
2236                                         // Illegal early close? This shouldn't happen D:
2237                                         $state = self::COLON_STATE_TEXT;
2238                                         break;
2239                                 default:
2240                                         $state = self::COLON_STATE_TAG;
2241                                 }
2242                                 break;
2243                         case 3: // self::COLON_STATE_CLOSETAG:
2244                                 // In a </tag>
2245                                 if( $c === ">" ) {
2246                                         $stack--;
2247                                         if( $stack < 0 ) {
2248                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2249                                                 wfProfileOut( __METHOD__ );
2250                                                 return false;
2251                                         }
2252                                         $state = self::COLON_STATE_TEXT;
2253                                 }
2254                                 break;
2255                         case self::COLON_STATE_TAGSLASH:
2256                                 if( $c === ">" ) {
2257                                         // Yes, a self-closed tag <blah/>
2258                                         $state = self::COLON_STATE_TEXT;
2259                                 } else {
2260                                         // Probably we're jumping the gun, and this is an attribute
2261                                         $state = self::COLON_STATE_TAG;
2262                                 }
2263                                 break;
2264                         case 5: // self::COLON_STATE_COMMENT:
2265                                 if( $c === "-" ) {
2266                                         $state = self::COLON_STATE_COMMENTDASH;
2267                                 }
2268                                 break;
2269                         case self::COLON_STATE_COMMENTDASH:
2270                                 if( $c === "-" ) {
2271                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2272                                 } else {
2273                                         $state = self::COLON_STATE_COMMENT;
2274                                 }
2275                                 break;
2276                         case self::COLON_STATE_COMMENTDASHDASH:
2277                                 if( $c === ">" ) {
2278                                         $state = self::COLON_STATE_TEXT;
2279                                 } else {
2280                                         $state = self::COLON_STATE_COMMENT;
2281                                 }
2282                                 break;
2283                         default:
2284                                 throw new MWException( "State machine error in " . __METHOD__ );
2285                         }
2286                 }
2287                 if( $stack > 0 ) {
2288                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2289                         return false;
2290                 }
2291                 wfProfileOut( __METHOD__ );
2292                 return false;
2293         }
2294
2295         /**
2296          * Return value of a magic variable (like PAGENAME)
2297          *
2298          * @private
2299          */
2300         function getVariableValue( $index, $frame=false ) {
2301                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2302
2303                 /**
2304                  * Some of these require message or data lookups and can be
2305                  * expensive to check many times.
2306                  */
2307                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2308                         if ( isset( $this->mVarCache[$index] ) ) {
2309                                 return $this->mVarCache[$index];
2310                         }
2311                 }
2312
2313                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2314                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2315
2316                 # Use the time zone
2317                 global $wgLocaltimezone;
2318                 if ( isset( $wgLocaltimezone ) ) {
2319                         $oldtz = getenv( 'TZ' );
2320                         putenv( 'TZ='.$wgLocaltimezone );
2321                 }
2322
2323                 wfSuppressWarnings(); // E_STRICT system time bitching
2324                 $localTimestamp = date( 'YmdHis', $ts );
2325                 $localMonth = date( 'm', $ts );
2326                 $localMonth1 = date( 'n', $ts );
2327                 $localMonthName = date( 'n', $ts );
2328                 $localDay = date( 'j', $ts );
2329                 $localDay2 = date( 'd', $ts );
2330                 $localDayOfWeek = date( 'w', $ts );
2331                 $localWeek = date( 'W', $ts );
2332                 $localYear = date( 'Y', $ts );
2333                 $localHour = date( 'H', $ts );
2334                 if ( isset( $wgLocaltimezone ) ) {
2335                         putenv( 'TZ='.$oldtz );
2336                 }
2337                 wfRestoreWarnings();
2338
2339                 switch ( $index ) {
2340                         case 'currentmonth':
2341                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2342                         case 'currentmonth1':
2343                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2344                         case 'currentmonthname':
2345                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2346                         case 'currentmonthnamegen':
2347                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2348                         case 'currentmonthabbrev':
2349                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2350                         case 'currentday':
2351                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2352                         case 'currentday2':
2353                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2354                         case 'localmonth':
2355                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localMonth );
2356                         case 'localmonth1':
2357                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localMonth1 );
2358                         case 'localmonthname':
2359                                 return $this->mVarCache[$index] = $wgContLang->getMonthName( $localMonthName );
2360                         case 'localmonthnamegen':
2361                                 return $this->mVarCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2362                         case 'localmonthabbrev':
2363                                 return $this->mVarCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2364                         case 'localday':
2365                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay );
2366                         case 'localday2':
2367                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDay2 );
2368                         case 'pagename':
2369                                 return wfEscapeWikiText( $this->mTitle->getText() );
2370                         case 'pagenamee':
2371                                 return $this->mTitle->getPartialURL();
2372                         case 'fullpagename':
2373                                 return wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2374                         case 'fullpagenamee':
2375                                 return $this->mTitle->getPrefixedURL();
2376                         case 'subpagename':
2377                                 return wfEscapeWikiText( $this->mTitle->getSubpageText() );
2378                         case 'subpagenamee':
2379                                 return $this->mTitle->getSubpageUrlForm();
2380                         case 'basepagename':
2381                                 return wfEscapeWikiText( $this->mTitle->getBaseText() );
2382                         case 'basepagenamee':
2383                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2384                         case 'talkpagename':
2385                                 if( $this->mTitle->canTalk() ) {
2386                                         $talkPage = $this->mTitle->getTalkPage();
2387                                         return wfEscapeWikiText( $talkPage->getPrefixedText() );
2388                                 } else {
2389                                         return '';
2390                                 }
2391                         case 'talkpagenamee':
2392                                 if( $this->mTitle->canTalk() ) {
2393                                         $talkPage = $this->mTitle->getTalkPage();
2394                                         return $talkPage->getPrefixedUrl();
2395                                 } else {
2396                                         return '';
2397                                 }
2398                         case 'subjectpagename':
2399                                 $subjPage = $this->mTitle->getSubjectPage();
2400                                 return wfEscapeWikiText( $subjPage->getPrefixedText() );
2401                         case 'subjectpagenamee':
2402                                 $subjPage = $this->mTitle->getSubjectPage();
2403                                 return $subjPage->getPrefixedUrl();
2404                         case 'revisionid':
2405                                 // Let the edit saving system know we should parse the page
2406                                 // *after* a revision ID has been assigned.
2407                                 $this->mOutput->setFlag( 'vary-revision' );
2408                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2409                                 return $this->mRevisionId;
2410                         case 'revisionday':
2411                                 // Let the edit saving system know we should parse the page
2412                                 // *after* a revision ID has been assigned. This is for null edits.
2413                                 $this->mOutput->setFlag( 'vary-revision' );
2414                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2415                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2416                         case 'revisionday2':
2417                                 // Let the edit saving system know we should parse the page
2418                                 // *after* a revision ID has been assigned. This is for null edits.
2419                                 $this->mOutput->setFlag( 'vary-revision' );
2420                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2421                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2422                         case 'revisionmonth':
2423                                 // Let the edit saving system know we should parse the page
2424                                 // *after* a revision ID has been assigned. This is for null edits.
2425                                 $this->mOutput->setFlag( 'vary-revision' );
2426                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2427                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2428                         case 'revisionyear':
2429                                 // Let the edit saving system know we should parse the page
2430                                 // *after* a revision ID has been assigned. This is for null edits.
2431                                 $this->mOutput->setFlag( 'vary-revision' );
2432                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2433                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2434                         case 'revisiontimestamp':
2435                                 // Let the edit saving system know we should parse the page
2436                                 // *after* a revision ID has been assigned. This is for null edits.
2437                                 $this->mOutput->setFlag( 'vary-revision' );
2438                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2439                                 return $this->getRevisionTimestamp();
2440                         case 'revisionuser':
2441                                 // Let the edit saving system know we should parse the page
2442                                 // *after* a revision ID has been assigned. This is for null edits.
2443                                 $this->mOutput->setFlag( 'vary-revision' );
2444                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2445                                 return $this->getRevisionUser();
2446                         case 'namespace':
2447                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2448                         case 'namespacee':
2449                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2450                         case 'talkspace':
2451                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2452                         case 'talkspacee':
2453                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2454                         case 'subjectspace':
2455                                 return $this->mTitle->getSubjectNsText();
2456                         case 'subjectspacee':
2457                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2458                         case 'currentdayname':
2459                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2460                         case 'currentyear':
2461                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2462                         case 'currenttime':
2463                                 return $this->mVarCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2464                         case 'currenthour':
2465                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2466                         case 'currentweek':
2467                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2468                                 // int to remove the padding
2469                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2470                         case 'currentdow':
2471                                 return $this->mVarCache[$index] = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2472                         case 'localdayname':
2473                                 return $this->mVarCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2474                         case 'localyear':
2475                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localYear, true );
2476                         case 'localtime':
2477                                 return $this->mVarCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2478                         case 'localhour':
2479                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localHour, true );
2480                         case 'localweek':
2481                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2482                                 // int to remove the padding
2483                                 return $this->mVarCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2484                         case 'localdow':
2485                                 return $this->mVarCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2486                         case 'numberofarticles':
2487                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2488                         case 'numberoffiles':
2489                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2490                         case 'numberofusers':
2491                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2492                         case 'numberofactiveusers':
2493                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::activeUsers() );
2494                         case 'numberofpages':
2495                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2496                         case 'numberofadmins':
2497                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::numberingroup('sysop') );
2498                         case 'numberofedits':
2499                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::edits() );
2500                         case 'numberofviews':
2501                                 return $this->mVarCache[$index] = $wgContLang->formatNum( SiteStats::views() );
2502                         case 'currenttimestamp':
2503                                 return $this->mVarCache[$index] = wfTimestamp( TS_MW, $ts );
2504                         case 'localtimestamp':
2505                                 return $this->mVarCache[$index] = $localTimestamp;
2506                         case 'currentversion':
2507                                 return $this->mVarCache[$index] = SpecialVersion::getVersion();
2508                         case 'sitename':
2509                                 return $wgSitename;
2510                         case 'server':
2511                                 return $wgServer;
2512                         case 'servername':
2513                                 return $wgServerName;
2514                         case 'scriptpath':
2515                                 return $wgScriptPath;
2516                         case 'directionmark':
2517                                 return $wgContLang->getDirMark();
2518                         case 'contentlanguage':
2519                                 global $wgContLanguageCode;
2520                                 return $wgContLanguageCode;
2521                         default:
2522                                 $ret = null;
2523                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) )
2524                                         return $ret;
2525                                 else
2526                                         return null;
2527                 }
2528         }
2529
2530         /**
2531          * initialise the magic variables (like CURRENTMONTHNAME)
2532          *
2533          * @private
2534          */
2535         function initialiseVariables() {
2536                 wfProfileIn( __METHOD__ );
2537                 $variableIDs = MagicWord::getVariableIDs();
2538
2539                 $this->mVariables = new MagicWordArray( $variableIDs );
2540                 wfProfileOut( __METHOD__ );
2541         }
2542
2543         /**
2544          * Preprocess some wikitext and return the document tree.
2545          * This is the ghost of replace_variables().
2546          *
2547          * @param string $text The text to parse
2548          * @param integer flags Bitwise combination of:
2549          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2550          *                                     included. Default is to assume a direct page view.
2551          *
2552          * The generated DOM tree must depend only on the input text and the flags.
2553          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2554          *
2555          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2556          * change in the DOM tree for a given text, must be passed through the section identifier
2557          * in the section edit link and thus back to extractSections().
2558          *
2559          * The output of this function is currently only cached in process memory, but a persistent
2560          * cache may be implemented at a later date which takes further advantage of these strict
2561          * dependency requirements.
2562          *
2563          * @private
2564          */
2565         function preprocessToDom ( $text, $flags = 0 ) {
2566                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2567                 return $dom;
2568         }
2569
2570         /*
2571          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2572          */
2573         public static function splitWhitespace( $s ) {
2574                 $ltrimmed = ltrim( $s );
2575                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2576                 $trimmed = rtrim( $ltrimmed );
2577                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2578                 if ( $diff > 0 ) {
2579                         $w2 = substr( $ltrimmed, -$diff );
2580                 } else {
2581                         $w2 = '';
2582                 }
2583                 return array( $w1, $trimmed, $w2 );
2584         }
2585
2586         /**
2587          * Replace magic variables, templates, and template arguments
2588          * with the appropriate text. Templates are substituted recursively,
2589          * taking care to avoid infinite loops.
2590          *
2591          * Note that the substitution depends on value of $mOutputType:
2592          *  self::OT_WIKI: only {{subst:}} templates
2593          *  self::OT_PREPROCESS: templates but not extension tags
2594          *  self::OT_HTML: all templates and extension tags
2595          *
2596          * @param string $tex The text to transform
2597          * @param PPFrame $frame Object describing the arguments passed to the template.
2598          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2599          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2600          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2601          * @private
2602          */
2603         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2604                 # Is there any text? Also, Prevent too big inclusions!
2605                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2606                         return $text;
2607                 }
2608                 wfProfileIn( __METHOD__ );
2609
2610                 if ( $frame === false ) {
2611                         $frame = $this->getPreprocessor()->newFrame();
2612                 } elseif ( !( $frame instanceof PPFrame ) ) {
2613                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2614                         $frame = $this->getPreprocessor()->newCustomFrame($frame);
2615                 }
2616
2617                 $dom = $this->preprocessToDom( $text );
2618                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2619                 $text = $frame->expand( $dom, $flags );
2620
2621                 wfProfileOut( __METHOD__ );
2622                 return $text;
2623         }
2624
2625         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2626         static function createAssocArgs( $args ) {
2627                 $assocArgs = array();
2628                 $index = 1;
2629                 foreach( $args as $arg ) {
2630                         $eqpos = strpos( $arg, '=' );
2631                         if ( $eqpos === false ) {
2632                                 $assocArgs[$index++] = $arg;
2633                         } else {
2634                                 $name = trim( substr( $arg, 0, $eqpos ) );
2635                                 $value = trim( substr( $arg, $eqpos+1 ) );
2636                                 if ( $value === false ) {
2637                                         $value = '';
2638                                 }
2639                                 if ( $name !== false ) {
2640                                         $assocArgs[$name] = $value;
2641                                 }
2642                         }
2643                 }
2644
2645                 return $assocArgs;
2646         }
2647
2648         /**
2649          * Warn the user when a parser limitation is reached
2650          * Will warn at most once the user per limitation type
2651          *
2652          * @param string $limitationType, should be one of:
2653          *   'expensive-parserfunction' (corresponding messages: 'expensive-parserfunction-warning', 'expensive-parserfunction-category')
2654          *   'post-expand-template-argument' (corresponding messages: 'post-expand-template-argument-warning', 'post-expand-template-argument-category')
2655          *   'post-expand-template-inclusion' (corresponding messages: 'post-expand-template-inclusion-warning', 'post-expand-template-inclusion-category')
2656          * @params int $current, $max When an explicit limit has been
2657          *       exceeded, provide the values (optional)
2658          */
2659         function limitationWarn( $limitationType, $current=null, $max=null) {
2660                 $msgName = $limitationType . '-warning';
2661                 //does no harm if $current and $max are present but are unnecessary for the message
2662                 $warning = wfMsgExt( $msgName, array( 'parsemag', 'escape' ), $current, $max );
2663                 $this->mOutput->addWarning( $warning );
2664                 $cat = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( $limitationType . '-category' ) );
2665                 if ( $cat ) {
2666                         $this->mOutput->addCategory( $cat->getDBkey(), $this->getDefaultSort() );
2667                 }
2668         }
2669
2670         /**
2671          * Return the text of a template, after recursively
2672          * replacing any variables or templates within the template.
2673          *
2674          * @param array $piece The parts of the template
2675          *  $piece['title']: the title, i.e. the part before the |
2676          *  $piece['parts']: the parameter array
2677          *  $piece['lineStart']: whether the brace was at the start of a line
2678          * @param PPFrame The current frame, contains template arguments
2679          * @return string the text of the template
2680          * @private
2681          */
2682         function braceSubstitution( $piece, $frame ) {
2683                 global $wgContLang, $wgNonincludableNamespaces;
2684                 wfProfileIn( __METHOD__ );
2685                 wfProfileIn( __METHOD__.'-setup' );
2686
2687                 # Flags
2688                 $found = false;             # $text has been filled
2689                 $nowiki = false;            # wiki markup in $text should be escaped
2690                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2691                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2692                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2693                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2694
2695                 # Title object, where $text came from
2696                 $title = NULL;
2697
2698                 # $part1 is the bit before the first |, and must contain only title characters.
2699                 # Various prefixes will be stripped from it later.
2700                 $titleWithSpaces = $frame->expand( $piece['title'] );
2701                 $part1 = trim( $titleWithSpaces );
2702                 $titleText = false;
2703
2704                 # Original title text preserved for various purposes
2705                 $originalTitle = $part1;
2706
2707                 # $args is a list of argument nodes, starting from index 0, not including $part1
2708                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2709                 wfProfileOut( __METHOD__.'-setup' );
2710
2711                 # SUBST
2712                 wfProfileIn( __METHOD__.'-modifiers' );
2713                 if ( !$found ) {
2714                         $mwSubst = MagicWord::get( 'subst' );
2715                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2716                                 # One of two possibilities is true:
2717                                 # 1) Found SUBST but not in the PST phase
2718                                 # 2) Didn't find SUBST and in the PST phase
2719                                 # In either case, return without further processing
2720                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2721                                 $isLocalObj = true;
2722                                 $found = true;
2723                         }
2724                 }
2725
2726                 # Variables
2727                 if ( !$found && $args->getLength() == 0 ) {
2728                         $id = $this->mVariables->matchStartToEnd( $part1 );
2729                         if ( $id !== false ) {
2730                                 $text = $this->getVariableValue( $id, $frame );
2731                                 if (MagicWord::getCacheTTL($id)>-1)
2732                                         $this->mOutput->mContainsOldMagic = true;
2733                                 $found = true;
2734                         }
2735                 }
2736
2737                 # MSG, MSGNW and RAW
2738                 if ( !$found ) {
2739                         # Check for MSGNW:
2740                         $mwMsgnw = MagicWord::get( 'msgnw' );
2741                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2742                                 $nowiki = true;
2743                         } else {
2744                                 # Remove obsolete MSG:
2745                                 $mwMsg = MagicWord::get( 'msg' );
2746                                 $mwMsg->matchStartAndRemove( $part1 );
2747                         }
2748
2749                         # Check for RAW:
2750                         $mwRaw = MagicWord::get( 'raw' );
2751                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2752                                 $forceRawInterwiki = true;
2753                         }
2754                 }
2755                 wfProfileOut( __METHOD__.'-modifiers' );
2756
2757                 # Parser functions
2758                 if ( !$found ) {
2759                         wfProfileIn( __METHOD__ . '-pfunc' );
2760
2761                         $colonPos = strpos( $part1, ':' );
2762                         if ( $colonPos !== false ) {
2763                                 # Case sensitive functions
2764                                 $function = substr( $part1, 0, $colonPos );
2765                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2766                                         $function = $this->mFunctionSynonyms[1][$function];
2767                                 } else {
2768                                         # Case insensitive functions
2769                                         $function = $wgContLang->lc( $function );
2770                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2771                                                 $function = $this->mFunctionSynonyms[0][$function];
2772                                         } else {
2773                                                 $function = false;
2774                                         }
2775                                 }
2776                                 if ( $function ) {
2777                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2778                                         $initialArgs = array( &$this );
2779                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2780                                         if ( $flags & SFH_OBJECT_ARGS ) {
2781                                                 # Add a frame parameter, and pass the arguments as an array
2782                                                 $allArgs = $initialArgs;
2783                                                 $allArgs[] = $frame;
2784                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2785                                                         $funcArgs[] = $args->item( $i );
2786                                                 }
2787                                                 $allArgs[] = $funcArgs;
2788                                         } else {
2789                                                 # Convert arguments to plain text
2790                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2791                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2792                                                 }
2793                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2794                                         }
2795
2796                                         # Workaround for PHP bug 35229 and similar
2797                                         if ( !is_callable( $callback ) ) {
2798                                                 wfProfileOut( __METHOD__ . '-pfunc' );
2799                                                 wfProfileOut( __METHOD__ );
2800                                                 throw new MWException( "Tag hook for $function is not callable\n" );
2801                                         }
2802                                         $result = call_user_func_array( $callback, $allArgs );
2803                                         $found = true;
2804                                         $noparse = true;
2805                                         $preprocessFlags = 0;
2806
2807                                         if ( is_array( $result ) ) {
2808                                                 if ( isset( $result[0] ) ) {
2809                                                         $text = $result[0];
2810                                                         unset( $result[0] );
2811                                                 }
2812
2813                                                 // Extract flags into the local scope
2814                                                 // This allows callers to set flags such as nowiki, found, etc.
2815                                                 extract( $result );
2816                                         } else {
2817                                                 $text = $result;
2818                                         }
2819                                         if ( !$noparse ) {
2820                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
2821                                                 $isChildObj = true;
2822                                         }
2823                                 }
2824                         }
2825                         wfProfileOut( __METHOD__ . '-pfunc' );
2826                 }
2827
2828                 # Finish mangling title and then check for loops.
2829                 # Set $title to a Title object and $titleText to the PDBK
2830                 if ( !$found ) {
2831                         $ns = NS_TEMPLATE;
2832                         # Split the title into page and subpage
2833                         $subpage = '';
2834                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2835                         if ($subpage !== '') {
2836                                 $ns = $this->mTitle->getNamespace();
2837                         }
2838                         $title = Title::newFromText( $part1, $ns );
2839                         if ( $title ) {
2840                                 $titleText = $title->getPrefixedText();
2841                                 # Check for language variants if the template is not found
2842                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
2843                                         $wgContLang->findVariantLink( $part1, $title, true );
2844                                 }
2845                                 # Do recursion depth check
2846                                 $limit = $this->mOptions->getMaxTemplateDepth();
2847                                 if ( $frame->depth >= $limit ) {
2848                                         $found = true;
2849                                         $text = '<span class="error">' . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit ) . '</span>';
2850                                 }
2851                         }
2852                 }
2853
2854                 # Load from database
2855                 if ( !$found && $title ) {
2856                         wfProfileIn( __METHOD__ . '-loadtpl' );
2857                         if ( !$title->isExternal() ) {
2858                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
2859                                         $text = SpecialPage::capturePath( $title );
2860                                         if ( is_string( $text ) ) {
2861                                                 $found = true;
2862                                                 $isHTML = true;
2863                                                 $this->disableCache();
2864                                         }
2865                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
2866                                         $found = false; //access denied
2867                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
2868                                 } else {
2869                                         list( $text, $title ) = $this->getTemplateDom( $title );
2870                                         if ( $text !== false ) {
2871                                                 $found = true;
2872                                                 $isChildObj = true;
2873                                         }
2874                                 }
2875
2876                                 # If the title is valid but undisplayable, make a link to it
2877                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2878                                         $text = "[[:$titleText]]";
2879                                         $found = true;
2880                                 }
2881                         } elseif ( $title->isTrans() ) {
2882                                 // Interwiki transclusion
2883                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
2884                                         $text = $this->interwikiTransclude( $title, 'render' );
2885                                         $isHTML = true;
2886                                 } else {
2887                                         $text = $this->interwikiTransclude( $title, 'raw' );
2888                                         // Preprocess it like a template
2889                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
2890                                         $isChildObj = true;
2891                                 }
2892                                 $found = true;
2893                         }
2894
2895                         # Do infinite loop check
2896                         # This has to be done after redirect resolution to avoid infinite loops via redirects
2897                         if ( !$frame->loopCheck( $title ) ) {
2898                                 $found = true;
2899                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
2900                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
2901                         }
2902                         wfProfileOut( __METHOD__ . '-loadtpl' );
2903                 }
2904
2905                 # If we haven't found text to substitute by now, we're done
2906                 # Recover the source wikitext and return it
2907                 if ( !$found ) {
2908                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2909                         wfProfileOut( __METHOD__ );
2910                         return array( 'object' => $text );
2911                 }
2912
2913                 # Expand DOM-style return values in a child frame
2914                 if ( $isChildObj ) {
2915                         # Clean up argument array
2916                         $newFrame = $frame->newChild( $args, $title );
2917
2918                         if ( $nowiki ) {
2919                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
2920                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
2921                                 # Expansion is eligible for the empty-frame cache
2922                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
2923                                         $text = $this->mTplExpandCache[$titleText];
2924                                 } else {
2925                                         $text = $newFrame->expand( $text );
2926                                         $this->mTplExpandCache[$titleText] = $text;
2927                                 }
2928                         } else {
2929                                 # Uncached expansion
2930                                 $text = $newFrame->expand( $text );
2931                         }
2932                 }
2933                 if ( $isLocalObj && $nowiki ) {
2934                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
2935                         $isLocalObj = false;
2936                 }
2937
2938                 # Replace raw HTML by a placeholder
2939                 # Add a blank line preceding, to prevent it from mucking up
2940                 # immediately preceding headings
2941                 if ( $isHTML ) {
2942                         $text = "\n\n" . $this->insertStripItem( $text );
2943                 }
2944                 # Escape nowiki-style return values
2945                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2946                         $text = wfEscapeWikiText( $text );
2947                 }
2948                 # Bug 529: if the template begins with a table or block-level
2949                 # element, it should be treated as beginning a new line.
2950                 # This behaviour is somewhat controversial.
2951                 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
2952                         $text = "\n" . $text;
2953                 }
2954
2955                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
2956                         # Error, oversize inclusion
2957                         $text = "[[$originalTitle]]" .
2958                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
2959                         $this->limitationWarn( 'post-expand-template-inclusion' );
2960                 }
2961
2962                 if ( $isLocalObj ) {
2963                         $ret = array( 'object' => $text );
2964                 } else {
2965                         $ret = array( 'text' => $text );
2966                 }
2967
2968                 wfProfileOut( __METHOD__ );
2969                 return $ret;
2970         }
2971
2972         /**
2973          * Get the semi-parsed DOM representation of a template with a given title,
2974          * and its redirect destination title. Cached.
2975          */
2976         function getTemplateDom( $title ) {
2977                 $cacheTitle = $title;
2978                 $titleText = $title->getPrefixedDBkey();
2979
2980                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
2981                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
2982                         $title = Title::makeTitle( $ns, $dbk );
2983                         $titleText = $title->getPrefixedDBkey();
2984                 }
2985                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
2986                         return array( $this->mTplDomCache[$titleText], $title );
2987                 }
2988
2989                 // Cache miss, go to the database
2990                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
2991
2992                 if ( $text === false ) {
2993                         $this->mTplDomCache[$titleText] = false;
2994                         return array( false, $title );
2995                 }
2996
2997                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
2998                 $this->mTplDomCache[ $titleText ] = $dom;
2999
3000                 if (! $title->equals($cacheTitle)) {
3001                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3002                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3003                 }
3004
3005                 return array( $dom, $title );
3006         }
3007
3008         /**
3009          * Fetch the unparsed text of a template and register a reference to it.
3010          */
3011         function fetchTemplateAndTitle( $title ) {
3012                 $templateCb = $this->mOptions->getTemplateCallback();
3013                 $stuff = call_user_func( $templateCb, $title, $this );
3014                 $text = $stuff['text'];
3015                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3016                 if ( isset( $stuff['deps'] ) ) {
3017                         foreach ( $stuff['deps'] as $dep ) {
3018                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3019                         }
3020                 }
3021                 return array($text,$finalTitle);
3022         }
3023
3024         function fetchTemplate( $title ) {
3025                 $rv = $this->fetchTemplateAndTitle($title);
3026                 return $rv[0];
3027         }
3028
3029         /**
3030          * Static function to get a template
3031          * Can be overridden via ParserOptions::setTemplateCallback().
3032          */
3033         static function statelessFetchTemplate( $title, $parser=false ) {
3034                 $text = $skip = false;
3035                 $finalTitle = $title;
3036                 $deps = array();
3037
3038                 // Loop to fetch the article, with up to 1 redirect
3039                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3040                         # Give extensions a chance to select the revision instead
3041                         $id = false; // Assume current
3042                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3043
3044                         if( $skip ) {
3045                                 $text = false;
3046                                 $deps[] = array(
3047                                         'title' => $title,
3048                                         'page_id' => $title->getArticleID(),
3049                                         'rev_id' => null );
3050                                 break;
3051                         }
3052                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3053                         $rev_id = $rev ? $rev->getId() : 0;
3054                         // If there is no current revision, there is no page
3055                         if( $id === false && !$rev ) {
3056                                 $linkCache = LinkCache::singleton();
3057                                 $linkCache->addBadLinkObj( $title );
3058                         }
3059
3060                         $deps[] = array(
3061                                 'title' => $title,
3062                                 'page_id' => $title->getArticleID(),
3063                                 'rev_id' => $rev_id );
3064
3065                         if( $rev ) {
3066                                 $text = $rev->getText();
3067                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3068                                 global $wgContLang;
3069                                 $message = $wgContLang->lcfirst( $title->getText() );
3070                                 $text = wfMsgForContentNoTrans( $message );
3071                                 if( wfEmptyMsg( $message, $text ) ) {
3072                                         $text = false;
3073                                         break;
3074                                 }
3075                         } else {
3076                                 break;
3077                         }
3078                         if ( $text === false ) {
3079                                 break;
3080                         }
3081                         // Redirect?
3082                         $finalTitle = $title;
3083                         $title = Title::newFromRedirect( $text );
3084                 }
3085                 return array(
3086                         'text' => $text,
3087                         'finalTitle' => $finalTitle,
3088                         'deps' => $deps );
3089         }
3090
3091         /**
3092          * Transclude an interwiki link.
3093          */
3094         function interwikiTransclude( $title, $action ) {
3095                 global $wgEnableScaryTranscluding;
3096
3097                 if (!$wgEnableScaryTranscluding)
3098                         return wfMsg('scarytranscludedisabled');
3099
3100                 $url = $title->getFullUrl( "action=$action" );
3101
3102                 if (strlen($url) > 255)
3103                         return wfMsg('scarytranscludetoolong');
3104                 return $this->fetchScaryTemplateMaybeFromCache($url);
3105         }
3106
3107         function fetchScaryTemplateMaybeFromCache($url) {
3108                 global $wgTranscludeCacheExpiry;
3109                 $dbr = wfGetDB(DB_SLAVE);
3110                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3111                                 array('tc_url' => $url));
3112                 if ($obj) {
3113                         $time = $obj->tc_time;
3114                         $text = $obj->tc_contents;
3115                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3116                                 return $text;
3117                         }
3118                 }
3119
3120                 $text = Http::get($url);
3121                 if (!$text)
3122                         return wfMsg('scarytranscludefailed', $url);
3123
3124                 $dbw = wfGetDB(DB_MASTER);
3125                 $dbw->replace('transcache', array('tc_url'), array(
3126                         'tc_url' => $url,
3127                         'tc_time' => time(),
3128                         'tc_contents' => $text));
3129                 return $text;
3130         }
3131
3132
3133         /**
3134          * Triple brace replacement -- used for template arguments
3135          * @private
3136          */
3137         function argSubstitution( $piece, $frame ) {
3138                 wfProfileIn( __METHOD__ );
3139
3140                 $error = false;
3141                 $parts = $piece['parts'];
3142                 $nameWithSpaces = $frame->expand( $piece['title'] );
3143                 $argName = trim( $nameWithSpaces );
3144                 $object = false;
3145                 $text = $frame->getArgument( $argName );
3146                 if (  $text === false && $parts->getLength() > 0
3147                   && (
3148                     $this->ot['html']
3149                     || $this->ot['pre']
3150                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3151                   )
3152                 ) {
3153                         # No match in frame, use the supplied default
3154                         $object = $parts->item( 0 )->getChildren();
3155                 }
3156                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3157                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3158                         $this->limitationWarn( 'post-expand-template-argument' );
3159                 }
3160
3161                 if ( $text === false && $object === false ) {
3162                         # No match anywhere
3163                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3164                 }
3165                 if ( $error !== false ) {
3166                         $text .= $error;
3167                 }
3168                 if ( $object !== false ) {
3169                         $ret = array( 'object' => $object );
3170                 } else {
3171                         $ret = array( 'text' => $text );
3172                 }
3173
3174                 wfProfileOut( __METHOD__ );
3175                 return $ret;
3176         }
3177
3178         /**
3179          * Return the text to be used for a given extension tag.
3180          * This is the ghost of strip().
3181          *
3182          * @param array $params Associative array of parameters:
3183          *     name       PPNode for the tag name
3184          *     attr       PPNode for unparsed text where tag attributes are thought to be
3185          *     attributes Optional associative array of parsed attributes
3186          *     inner      Contents of extension element
3187          *     noClose    Original text did not have a close tag
3188          * @param PPFrame $frame
3189          */
3190         function extensionSubstitution( $params, $frame ) {
3191                 global $wgRawHtml, $wgContLang;
3192
3193                 $name = $frame->expand( $params['name'] );
3194                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3195                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3196
3197                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $this->mMarkerIndex++) . self::MARKER_SUFFIX;
3198
3199                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3200                         ( $this->ot['html'] || $this->ot['pre'] );
3201                 if ( $this->ot['html'] || $isFunctionTag ) {
3202                         $name = strtolower( $name );
3203                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3204                         if ( isset( $params['attributes'] ) ) {
3205                                 $attributes = $attributes + $params['attributes'];
3206                         }
3207                         switch ( $name ) {
3208                                 case 'html':
3209                                         if( $wgRawHtml ) {
3210                                                 $output = $content;
3211                                                 break;
3212                                         } else {
3213                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3214                                         }
3215                                 case 'nowiki':
3216                                         $content = strtr($content, array('-{' => '-&#123;', '}-' => '&#125;-'));
3217                                         $output = Xml::escapeTagsOnly( $content );
3218                                         break;
3219                                 case 'math':
3220                                         $output = $wgContLang->armourMath(
3221                                                 MathRenderer::renderMath( $content, $attributes ) );
3222                                         break;
3223                                 case 'gallery':
3224                                         $output = $this->renderImageGallery( $content, $attributes );
3225                                         break;
3226                                 default:
3227                                         if( isset( $this->mTagHooks[$name] ) ) {
3228                                                 # Workaround for PHP bug 35229 and similar
3229                                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3230                                                         throw new MWException( "Tag hook for $name is not callable\n" );
3231                                                 }
3232                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3233                                                         array( $content, $attributes, $this, $frame ) );
3234                                         } elseif( isset( $this->mFunctionTagHooks[$name] ) ) {
3235                                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3236                                                 if( !is_callable( $callback ) )
3237                                                         throw new MWException( "Tag hook for $name is not callable\n" );
3238
3239                                                 $output = call_user_func_array( $callback,
3240                                                         array( &$this, $frame, $content, $attributes ) );
3241                                         } else {
3242                                                 $output = '<span class="error">Invalid tag extension name: ' .
3243                                                         htmlspecialchars( $name ) . '</span>';
3244                                         }
3245                         }
3246                 } else {
3247                         if ( is_null( $attrText ) ) {
3248                                 $attrText = '';
3249                         }
3250                         if ( isset( $params['attributes'] ) ) {
3251                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3252                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3253                                                 htmlspecialchars( $attrValue ) . '"';
3254                                 }
3255                         }
3256                         if ( $content === null ) {
3257                                 $output = "<$name$attrText/>";
3258                         } else {
3259                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3260                                 $output = "<$name$attrText>$content$close";
3261                         }
3262                 }
3263
3264                 if( $isFunctionTag ) {
3265                         return $output;
3266                 } elseif ( $name === 'html' || $name === 'nowiki' ) {
3267                         $this->mStripState->nowiki->setPair( $marker, $output );
3268                 } else {
3269                         $this->mStripState->general->setPair( $marker, $output );
3270                 }
3271                 return $marker;
3272         }
3273
3274         /**
3275          * Increment an include size counter
3276          *
3277          * @param string $type The type of expansion
3278          * @param integer $size The size of the text
3279          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3280          */
3281         function incrementIncludeSize( $type, $size ) {
3282                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3283                         return false;
3284                 } else {
3285                         $this->mIncludeSizes[$type] += $size;
3286                         return true;
3287                 }
3288         }
3289
3290         /**
3291          * Increment the expensive function count
3292          *
3293          * @return boolean False if the limit has been exceeded
3294          */
3295         function incrementExpensiveFunctionCount() {
3296                 global $wgExpensiveParserFunctionLimit;
3297                 $this->mExpensiveFunctionCount++;
3298                 if($this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit) {
3299                         return true;
3300                 }
3301                 return false;
3302         }
3303
3304         /**
3305          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3306          * Fills $this->mDoubleUnderscores, returns the modified text
3307          */
3308         function doDoubleUnderscore( $text ) {
3309                 wfProfileIn( __METHOD__ );
3310                 // The position of __TOC__ needs to be recorded
3311                 $mw = MagicWord::get( 'toc' );
3312                 if( $mw->match( $text ) ) {
3313                         $this->mShowToc = true;
3314                         $this->mForceTocPosition = true;
3315
3316                         // Set a placeholder. At the end we'll fill it in with the TOC.
3317                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3318
3319                         // Only keep the first one.
3320                         $text = $mw->replace( '', $text );
3321                 }
3322
3323                 // Now match and remove the rest of them
3324                 $mwa = MagicWord::getDoubleUnderscoreArray();
3325                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3326
3327                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3328                         $this->mOutput->mNoGallery = true;
3329                 }
3330                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3331                         $this->mShowToc = false;
3332                 }
3333                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3334                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3335
3336                         $containerCategory = Title::makeTitleSafe( NS_CATEGORY, wfMsgForContent( 'hidden-category-category' ) );
3337                         if ( $containerCategory ) {
3338                                 $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3339                         } else {
3340                                 wfDebug( __METHOD__.": [[MediaWiki:hidden-category-category]] is not a valid title!\n" );
3341                         }
3342                 }
3343                 # (bug 8068) Allow control over whether robots index a page.
3344                 #
3345                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3346                 # is not desirable, the last one on the page should win.
3347                 if( isset( $this->mDoubleUnderscores['noindex'] ) ) {
3348                         $this->mOutput->setIndexPolicy( 'noindex' );
3349                 } elseif( isset( $this->mDoubleUnderscores['index'] ) ) {
3350                         $this->mOutput->setIndexPolicy( 'index' );
3351                 }
3352                 wfProfileOut( __METHOD__ );
3353                 return $text;
3354         }
3355
3356         /**
3357          * This function accomplishes several tasks:
3358          * 1) Auto-number headings if that option is enabled
3359          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3360          * 3) Add a Table of contents on the top for users who have enabled the option
3361          * 4) Auto-anchor headings
3362          *
3363          * It loops through all headlines, collects the necessary data, then splits up the
3364          * string and re-inserts the newly formatted headlines.
3365          *
3366          * @param string $text
3367          * @param string $origText Original, untouched wikitext
3368          * @param boolean $isMain
3369          * @private
3370          */
3371         function formatHeadings( $text, $origText, $isMain=true ) {
3372                 global $wgMaxTocLevel, $wgContLang, $wgEnforceHtmlIds;
3373
3374                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3375                 $showEditLink = $this->mOptions->getEditSection();
3376
3377                 // Do not call quickUserCan unless necessary
3378                 if( $showEditLink && !$this->mTitle->quickUserCan( 'edit' ) ) {
3379                         $showEditLink = 0;
3380                 }
3381
3382                 # Inhibit editsection links if requested in the page
3383                 if ( isset( $this->mDoubleUnderscores['noeditsection'] )  || $this->mOptions->getIsPrintable() ) {
3384                         $showEditLink = 0;
3385                 }
3386
3387                 # Get all headlines for numbering them and adding funky stuff like [edit]
3388                 # links - this is for later, but we need the number of headlines right now
3389                 $matches = array();
3390                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3391
3392                 # if there are fewer than 4 headlines in the article, do not show TOC
3393                 # unless it's been explicitly enabled.
3394                 $enoughToc = $this->mShowToc &&
3395                         (($numMatches >= 4) || $this->mForceTocPosition);
3396
3397                 # Allow user to stipulate that a page should have a "new section"
3398                 # link added via __NEWSECTIONLINK__
3399                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3400                         $this->mOutput->setNewSection( true );
3401                 }
3402
3403                 # Allow user to remove the "new section"
3404                 # link via __NONEWSECTIONLINK__
3405                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3406                         $this->mOutput->hideNewSection( true );
3407                 }
3408
3409                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3410                 # override above conditions and always show TOC above first header
3411                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3412                         $this->mShowToc = true;
3413                         $enoughToc = true;
3414                 }
3415
3416                 # We need this to perform operations on the HTML
3417                 $sk = $this->mOptions->getSkin();
3418
3419                 # headline counter
3420                 $headlineCount = 0;
3421                 $numVisible = 0;
3422
3423                 # Ugh .. the TOC should have neat indentation levels which can be
3424                 # passed to the skin functions. These are determined here
3425                 $toc = '';
3426                 $full = '';
3427                 $head = array();
3428                 $sublevelCount = array();
3429                 $levelCount = array();
3430                 $toclevel = 0;
3431                 $level = 0;
3432                 $prevlevel = 0;
3433                 $toclevel = 0;
3434                 $prevtoclevel = 0;
3435                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3436                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3437                 $oldType = $this->mOutputType;
3438                 $this->setOutputType( self::OT_WIKI );
3439                 $frame = $this->getPreprocessor()->newFrame();
3440                 $root = $this->preprocessToDom( $origText );
3441                 $node = $root->getFirstChild();
3442                 $byteOffset = 0;
3443                 $tocraw = array();
3444
3445                 foreach( $matches[3] as $headline ) {
3446                         $isTemplate = false;
3447                         $titleText = false;
3448                         $sectionIndex = false;
3449                         $numbering = '';
3450                         $markerMatches = array();
3451                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3452                                 $serial = $markerMatches[1];
3453                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3454                                 $isTemplate = ($titleText != $baseTitleText);
3455                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3456                         }
3457
3458                         if( $toclevel ) {
3459                                 $prevlevel = $level;
3460                                 $prevtoclevel = $toclevel;
3461                         }
3462                         $level = $matches[1][$headlineCount];
3463
3464                         if ( $level > $prevlevel ) {
3465                                 # Increase TOC level
3466                                 $toclevel++;
3467                                 $sublevelCount[$toclevel] = 0;
3468                                 if( $toclevel<$wgMaxTocLevel ) {
3469                                         $prevtoclevel = $toclevel;
3470                                         $toc .= $sk->tocIndent();
3471                                         $numVisible++;
3472                                 }
3473                         }
3474                         elseif ( $level < $prevlevel && $toclevel > 1 ) {
3475                                 # Decrease TOC level, find level to jump to
3476
3477                                 for ($i = $toclevel; $i > 0; $i--) {
3478                                         if ( $levelCount[$i] == $level ) {
3479                                                 # Found last matching level
3480                                                 $toclevel = $i;
3481                                                 break;
3482                                         }
3483                                         elseif ( $levelCount[$i] < $level ) {
3484                                                 # Found first matching level below current level
3485                                                 $toclevel = $i + 1;
3486                                                 break;
3487                                         }
3488                                 }
3489                                 if( $i == 0 ) $toclevel = 1;
3490                                 if( $toclevel<$wgMaxTocLevel ) {
3491                                         if($prevtoclevel < $wgMaxTocLevel) {
3492                                                 # Unindent only if the previous toc level was shown :p
3493                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3494                                                 $prevtoclevel = $toclevel;
3495                                         } else {
3496                                                 $toc .= $sk->tocLineEnd();
3497                                         }
3498                                 }
3499                         }
3500                         else {
3501                                 # No change in level, end TOC line
3502                                 if( $toclevel<$wgMaxTocLevel ) {
3503                                         $toc .= $sk->tocLineEnd();
3504                                 }
3505                         }
3506
3507                         $levelCount[$toclevel] = $level;
3508
3509                         # count number of headlines for each level
3510                         @$sublevelCount[$toclevel]++;
3511                         $dot = 0;
3512                         for( $i = 1; $i <= $toclevel; $i++ ) {
3513                                 if( !empty( $sublevelCount[$i] ) ) {
3514                                         if( $dot ) {
3515                                                 $numbering .= '.';
3516                                         }
3517                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3518                                         $dot = 1;
3519                                 }
3520                         }
3521
3522                         # The safe header is a version of the header text safe to use for links
3523                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3524                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3525
3526                         # Remove link placeholders by the link text.
3527                         #     <!--LINK number-->
3528                         # turns into
3529                         #     link text with suffix
3530                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3531
3532                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3533                         $tocline = preg_replace(
3534                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3535                                 array( '',                          '<$1>'),
3536                                 $safeHeadline
3537                         );
3538                         $tocline = trim( $tocline );
3539
3540                         # For the anchor, strip out HTML-y stuff period
3541                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3542                         $safeHeadline = preg_replace( '/[ _]+/', ' ', $safeHeadline );
3543                         $safeHeadline = trim( $safeHeadline );
3544
3545                         # Save headline for section edit hint before it's escaped
3546                         $headlineHint = $safeHeadline;
3547
3548                         if ( $wgEnforceHtmlIds ) {
3549                                 $legacyHeadline = false;
3550                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3551                                         'noninitial' );
3552                         } else {
3553                                 # For reverse compatibility, provide an id that's
3554                                 # HTML4-compatible, like we used to.
3555                                 #
3556                                 # It may be worth noting, academically, that it's possible for
3557                                 # the legacy anchor to conflict with a non-legacy headline
3558                                 # anchor on the page.  In this case likely the "correct" thing
3559                                 # would be to either drop the legacy anchors or make sure
3560                                 # they're numbered first.  However, this would require people
3561                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3562                                 # manually, so let's not bother worrying about it.
3563                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3564                                         'noninitial' );
3565                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline, 'xml' );
3566
3567                                 if ( $legacyHeadline == $safeHeadline ) {
3568                                         # No reason to have both (in fact, we can't)
3569                                         $legacyHeadline = false;
3570                                 } elseif ( $legacyHeadline != Sanitizer::escapeId(
3571                                 $legacyHeadline, 'xml' ) ) {
3572                                         # The legacy id is invalid XML.  We used to allow this, but
3573                                         # there's no reason to do so anymore.  Backward
3574                                         # compatibility will fail slightly in this case, but it's
3575                                         # no big deal.
3576                                         $legacyHeadline = false;
3577                                 }
3578                         }
3579
3580                         # HTML names must be case-insensitively unique (bug 10721).  FIXME:
3581                         # Does this apply to Unicode characters?  Because we aren't
3582                         # handling those here.
3583                         $arrayKey = strtolower( $safeHeadline );
3584                         if ( $legacyHeadline === false ) {
3585                                 $legacyArrayKey = false;
3586                         } else {
3587                                 $legacyArrayKey = strtolower( $legacyHeadline );
3588                         }
3589
3590                         # count how many in assoc. array so we can track dupes in anchors
3591                         if ( isset( $refers[$arrayKey] ) ) {
3592                                 $refers[$arrayKey]++;
3593                         } else {
3594                                 $refers[$arrayKey] = 1;
3595                         }
3596                         if ( isset( $refers[$legacyArrayKey] ) ) {
3597                                 $refers[$legacyArrayKey]++;
3598                         } else {
3599                                 $refers[$legacyArrayKey] = 1;
3600                         }
3601
3602                         # Don't number the heading if it is the only one (looks silly)
3603                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3604                                 # the two are different if the line contains a link
3605                                 $headline=$numbering . ' ' . $headline;
3606                         }
3607
3608                         # Create the anchor for linking from the TOC to the section
3609                         $anchor = $safeHeadline;
3610                         $legacyAnchor = $legacyHeadline;
3611                         if ( $refers[$arrayKey] > 1 ) {
3612                                 $anchor .= '_' . $refers[$arrayKey];
3613                         }
3614                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3615                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3616                         }
3617                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3618                                 $toc .= $sk->tocLine($anchor, $tocline,
3619                                         $numbering, $toclevel, ($isTemplate ? false : $sectionIndex));
3620                         }
3621
3622                         # Add the section to the section tree
3623                         # Find the DOM node for this header
3624                         while ( $node && !$isTemplate ) {
3625                                 if ( $node->getName() === 'h' ) {
3626                                         $bits = $node->splitHeading();
3627                                         if ( $bits['i'] == $sectionIndex )
3628                                                 break;
3629                                 }
3630                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
3631                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
3632                                 $node = $node->getNextSibling();
3633                         }
3634                         $tocraw[] = array(
3635                                 'toclevel' => $toclevel,
3636                                 'level' => $level,
3637                                 'line' => $tocline,
3638                                 'number' => $numbering,
3639                                 'index' => ($isTemplate ? 'T-' : '' ) . $sectionIndex,
3640                                 'fromtitle' => $titleText,
3641                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
3642                                 'anchor' => $anchor,
3643                         );
3644
3645                         # give headline the correct <h#> tag
3646                         if( $showEditLink && $sectionIndex !== false ) {
3647                                 if( $isTemplate ) {
3648                                         # Put a T flag in the section identifier, to indicate to extractSections()
3649                                         # that sections inside <includeonly> should be counted.
3650                                         $editlink = $sk->doEditSectionLink(Title::newFromText( $titleText ), "T-$sectionIndex");
3651                                 } else {
3652                                         $editlink = $sk->doEditSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3653                                 }
3654                         } else {
3655                                 $editlink = '';
3656                         }
3657                         $head[$headlineCount] = $sk->makeHeadline( $level,
3658                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3659                                 $editlink, $legacyAnchor );
3660
3661                         $headlineCount++;
3662                 }
3663
3664                 $this->setOutputType( $oldType );
3665
3666                 # Never ever show TOC if no headers
3667                 if( $numVisible < 1 ) {
3668                         $enoughToc = false;
3669                 }
3670
3671                 if( $enoughToc ) {
3672                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3673                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3674                         }
3675                         $toc = $sk->tocList( $toc );
3676                         $this->mOutput->setTOCHTML( $toc );
3677                 }
3678
3679                 if ( $isMain ) {
3680                         $this->mOutput->setSections( $tocraw );
3681                 }
3682
3683                 # split up and insert constructed headlines
3684
3685                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3686                 $i = 0;
3687
3688                 foreach( $blocks as $block ) {
3689                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3690                                 # This is the [edit] link that appears for the top block of text when
3691                                 # section editing is enabled
3692
3693                                 # Disabled because it broke block formatting
3694                                 # For example, a bullet point in the top line
3695                                 # $full .= $sk->editSectionLink(0);
3696                         }
3697                         $full .= $block;
3698                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3699                                 # Top anchor now in skin
3700                                 $full = $full.$toc;
3701                         }
3702
3703                         if( !empty( $head[$i] ) ) {
3704                                 $full .= $head[$i];
3705                         }
3706                         $i++;
3707                 }
3708                 if( $this->mForceTocPosition ) {
3709                         return str_replace( '<!--MWTOC-->', $toc, $full );
3710                 } else {
3711                         return $full;
3712                 }
3713         }
3714
3715         /**
3716          * Merge $tree2 into $tree1 by replacing the section with index
3717          * $section in $tree1 and its descendants with the sections in $tree2.
3718          * Note that in the returned section tree, only the 'index' and
3719          * 'byteoffset' fields are guaranteed to be correct.
3720          * @param $tree1 array Section tree from ParserOutput::getSectons()
3721         *  @param $tree2 array Section tree
3722          * @param $section int Section index
3723          * @param $title Title Title both section trees come from
3724          * @param $len2 int Length of the original wikitext for $tree2
3725          * @return array Merged section tree
3726          */
3727         public static function mergeSectionTrees( $tree1, $tree2, $section, $title, $len2 ) {
3728                 global $wgContLang;
3729                 $newTree = array();
3730                 $targetLevel = false;
3731                 $merged = false;
3732                 $lastLevel = 1;
3733                 $nextIndex = 1;
3734                 $numbering = array( 0 );
3735                 $titletext = $title->getPrefixedDBkey();
3736                 foreach ( $tree1 as $s ) {
3737                         if ( $targetLevel !== false ) {
3738                                 if ( $s['level'] <= $targetLevel )
3739                                         // We've skipped enough
3740                                         $targetLevel = false;
3741                                 else
3742                                         continue;
3743                         }
3744                         if ( $s['index'] != $section ||
3745                                         $s['fromtitle'] != $titletext ) {
3746                                 self::incrementNumbering( $numbering,
3747                                         $s['toclevel'], $lastLevel );
3748
3749                                 // Rewrite index, byteoffset and number
3750                                 if ( $s['fromtitle'] == $titletext ) {
3751                                         $s['index'] = $nextIndex++;
3752                                         if ( $merged )
3753                                                 $s['byteoffset'] += $len2;
3754                                 }
3755                                 $s['number']  = implode( '.', array_map(
3756                                         array( $wgContLang, 'formatnum' ),
3757                                         $numbering ) );
3758                                 $lastLevel = $s['toclevel'];
3759                                 $newTree[] = $s;
3760                         } else {
3761                                 // We're at $section
3762                                 // Insert sections from $tree2 here
3763                                 foreach ( $tree2 as $s2 ) {
3764                                         // Rewrite the fields in $s2
3765                                         // before inserting it
3766                                         $s2['toclevel'] += $s['toclevel'] - 1;
3767                                         $s2['level'] += $s['level'] - 1;
3768                                         $s2['index'] = $nextIndex++;
3769                                         $s2['byteoffset'] += $s['byteoffset'];
3770
3771                                         self::incrementNumbering( $numbering,
3772                                                 $s2['toclevel'], $lastLevel );
3773                                         $s2['number']  = implode( '.', array_map(
3774                                                 array( $wgContLang, 'formatnum' ),
3775                                                 $numbering ) );
3776                                         $lastLevel = $s2['toclevel'];
3777                                         $newTree[] = $s2;
3778                                 }
3779                                 // Skip all descendants of $section in $tree1
3780                                 $targetLevel = $s['level'];
3781                                 $merged = true;
3782                         }
3783                 }
3784                 return $newTree;
3785         }
3786
3787         /**
3788          * Increment a section number. Helper function for mergeSectionTrees()
3789          * @param $number array Array representing a section number
3790          * @param $level int Current TOC level (depth)
3791          * @param $lastLevel int Level of previous TOC entry
3792          */
3793         private static function incrementNumbering( &$number, $level, $lastLevel ) {
3794                 if ( $level > $lastLevel )
3795                         $number[$level - 1] = 1;
3796                 else if ( $level < $lastLevel ) {
3797                         foreach ( $number as $key => $unused )
3798                                 if ( $key >= $level )
3799                                         unset( $number[$key] );
3800                         $number[$level - 1]++;
3801                 } else
3802                         $number[$level - 1]++;
3803         }
3804
3805         /**
3806          * Transform wiki markup when saving a page by doing \r\n -> \n
3807          * conversion, substitting signatures, {{subst:}} templates, etc.
3808          *
3809          * @param string $text the text to transform
3810          * @param Title &$title the Title object for the current article
3811          * @param User $user the User object describing the current user
3812          * @param ParserOptions $options parsing options
3813          * @param bool $clearState whether to clear the parser state first
3814          * @return string the altered wiki markup
3815          * @public
3816          */
3817         function preSaveTransform( $text, Title $title, $user, $options, $clearState = true ) {
3818                 $this->mOptions = $options;
3819                 $this->setTitle( $title );
3820                 $this->setOutputType( self::OT_WIKI );
3821
3822                 if ( $clearState ) {
3823                         $this->clearState();
3824                 }
3825
3826                 $pairs = array(
3827                         "\r\n" => "\n",
3828                 );
3829                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3830                 $text = $this->pstPass2( $text, $user );
3831                 $text = $this->mStripState->unstripBoth( $text );
3832                 return $text;
3833         }
3834
3835         /**
3836          * Pre-save transform helper function
3837          * @private
3838          */
3839         function pstPass2( $text, $user ) {
3840                 global $wgContLang, $wgLocaltimezone;
3841
3842                 /* Note: This is the timestamp saved as hardcoded wikitext to
3843                  * the database, we use $wgContLang here in order to give
3844                  * everyone the same signature and use the default one rather
3845                  * than the one selected in each user's preferences.
3846                  *
3847                  * (see also bug 12815)
3848                  */
3849                 $ts = $this->mOptions->getTimestamp();
3850                 $tz = wfMsgForContent( 'timezone-utc' );
3851                 if ( isset( $wgLocaltimezone ) ) {
3852                         $unixts = wfTimestamp( TS_UNIX, $ts );
3853                         $oldtz = getenv( 'TZ' );
3854                         putenv( 'TZ='.$wgLocaltimezone );
3855                         $ts = date( 'YmdHis', $unixts );
3856                         $tz = date( 'T', $unixts );  # might vary on DST changeover!
3857
3858                         /* Allow translation of timezones trough wiki. date() can return
3859                          * whatever crap the system uses, localised or not, so we cannot
3860                          * ship premade translations.
3861                          */
3862                         $key = 'timezone-' . strtolower( trim( $tz ) );
3863                         $value = wfMsgForContent( $key );
3864                         if ( !wfEmptyMsg( $key, $value ) ) $tz = $value;
3865
3866                         putenv( 'TZ='.$oldtz );
3867                 }
3868
3869                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tz)";
3870
3871                 # Variable replacement
3872                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3873                 $text = $this->replaceVariables( $text );
3874
3875                 # Signatures
3876                 $sigText = $this->getUserSig( $user );
3877                 $text = strtr( $text, array(
3878                         '~~~~~' => $d,
3879                         '~~~~' => "$sigText $d",
3880                         '~~~' => $sigText
3881                 ) );
3882
3883                 # Context links: [[|name]] and [[name (context)|]]
3884                 #
3885                 global $wgLegalTitleChars;
3886                 $tc = "[$wgLegalTitleChars]";
3887                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
3888
3889                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3890                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
3891                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3892                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3893
3894                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3895                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3896                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
3897                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3898
3899                 $t = $this->mTitle->getText();
3900                 $m = array();
3901                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3902                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3903                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3904                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3905                 } else {
3906                         # if there's no context, don't bother duplicating the title
3907                         $text = preg_replace( $p2, '[[\\1]]', $text );
3908                 }
3909
3910                 # Trim trailing whitespace
3911                 $text = rtrim( $text );
3912
3913                 return $text;
3914         }
3915
3916         /**
3917          * Fetch the user's signature text, if any, and normalize to
3918          * validated, ready-to-insert wikitext.
3919          *
3920          * @param User $user
3921          * @return string
3922          * @private
3923          */
3924         function getUserSig( &$user ) {
3925                 global $wgMaxSigChars;
3926
3927                 $username = $user->getName();
3928                 $nickname = $user->getOption( 'nickname' );
3929                 $nickname = $nickname === null ? $username : $nickname;
3930
3931                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
3932                         $nickname = $username;
3933                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
3934                 } elseif( $user->getBoolOption( 'fancysig' ) !== false ) {
3935                         # Sig. might contain markup; validate this
3936                         if( $this->validateSig( $nickname ) !== false ) {
3937                                 # Validated; clean up (if needed) and return it
3938                                 return $this->cleanSig( $nickname, true );
3939                         } else {
3940                                 # Failed to validate; fall back to the default
3941                                 $nickname = $username;
3942                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
3943                         }
3944                 }
3945
3946                 // Make sure nickname doesnt get a sig in a sig
3947                 $nickname = $this->cleanSigInSig( $nickname );
3948
3949                 # If we're still here, make it a link to the user page
3950                 $userText = wfEscapeWikiText( $username );
3951                 $nickText = wfEscapeWikiText( $nickname );
3952                 if ( $user->isAnon() )  {
3953                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
3954                 } else {
3955                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
3956                 }
3957         }
3958
3959         /**
3960          * Check that the user's signature contains no bad XML
3961          *
3962          * @param string $text
3963          * @return mixed An expanded string, or false if invalid.
3964          */
3965         function validateSig( $text ) {
3966                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
3967         }
3968
3969         /**
3970          * Clean up signature text
3971          *
3972          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3973          * 2) Substitute all transclusions
3974          *
3975          * @param string $text
3976          * @param $parsing Whether we're cleaning (preferences save) or parsing
3977          * @return string Signature text
3978          */
3979         function cleanSig( $text, $parsing = false ) {
3980                 if ( !$parsing ) {
3981                         global $wgTitle;
3982                         $this->clearState();
3983                         $this->setTitle( $wgTitle );
3984                         $this->mOptions = new ParserOptions;
3985                         $this->setOutputType = self::OT_PREPROCESS;
3986                 }
3987
3988                 # Option to disable this feature
3989                 if ( !$this->mOptions->getCleanSignatures() ) {
3990                         return $text;
3991                 }
3992
3993                 # FIXME: regex doesn't respect extension tags or nowiki
3994                 #  => Move this logic to braceSubstitution()
3995                 $substWord = MagicWord::get( 'subst' );
3996                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3997                 $substText = '{{' . $substWord->getSynonym( 0 );
3998
3999                 $text = preg_replace( $substRegex, $substText, $text );
4000                 $text = $this->cleanSigInSig( $text );
4001                 $dom = $this->preprocessToDom( $text );
4002                 $frame = $this->getPreprocessor()->newFrame();
4003                 $text = $frame->expand( $dom );
4004
4005                 if ( !$parsing ) {
4006                         $text = $this->mStripState->unstripBoth( $text );
4007                 }
4008
4009                 return $text;
4010         }
4011
4012         /**
4013          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4014          * @param string $text
4015          * @return string Signature text with /~{3,5}/ removed
4016          */
4017         function cleanSigInSig( $text ) {
4018                 $text = preg_replace( '/~{3,5}/', '', $text );
4019                 return $text;
4020         }
4021
4022         /**
4023          * Set up some variables which are usually set up in parse()
4024          * so that an external function can call some class members with confidence
4025          * @public
4026          */
4027         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4028                 $this->setTitle( $title );
4029                 $this->mOptions = $options;
4030                 $this->setOutputType( $outputType );
4031                 if ( $clearState ) {
4032                         $this->clearState();
4033                 }
4034         }
4035
4036         /**
4037          * Wrapper for preprocess()
4038          *
4039          * @param string $text the text to preprocess
4040          * @param ParserOptions $options  options
4041          * @return string
4042          * @public
4043          */
4044         function transformMsg( $text, $options ) {
4045                 global $wgTitle;
4046                 static $executing = false;
4047
4048                 # Guard against infinite recursion
4049                 if ( $executing ) {
4050                         return $text;
4051                 }
4052                 $executing = true;
4053
4054                 wfProfileIn(__METHOD__);
4055                 $text = $this->preprocess( $text, $wgTitle, $options );
4056
4057                 $executing = false;
4058                 wfProfileOut(__METHOD__);
4059                 return $text;
4060         }
4061
4062         /**
4063          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4064          * The callback should have the following form:
4065          *    function myParserHook( $text, $params, &$parser ) { ... }
4066          *
4067          * Transform and return $text. Use $parser for any required context, e.g. use
4068          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4069          *
4070          * @public
4071          *
4072          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4073          * @param mixed $callback The callback function (and object) to use for the tag
4074          *
4075          * @return The old value of the mTagHooks array associated with the hook
4076          */
4077         function setHook( $tag, $callback ) {
4078                 $tag = strtolower( $tag );
4079                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4080                 $this->mTagHooks[$tag] = $callback;
4081                 if( !in_array( $tag, $this->mStripList ) ) {
4082                         $this->mStripList[] = $tag;
4083                 }
4084
4085                 return $oldVal;
4086         }
4087
4088         function setTransparentTagHook( $tag, $callback ) {
4089                 $tag = strtolower( $tag );
4090                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4091                 $this->mTransparentTagHooks[$tag] = $callback;
4092
4093                 return $oldVal;
4094         }
4095
4096         /**
4097          * Remove all tag hooks
4098          */
4099         function clearTagHooks() {
4100                 $this->mTagHooks = array();
4101                 $this->mStripList = $this->mDefaultStripList;
4102         }
4103
4104         /**
4105          * Create a function, e.g. {{sum:1|2|3}}
4106          * The callback function should have the form:
4107          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4108          *
4109          * Or with SFH_OBJECT_ARGS:
4110          *    function myParserFunction( $parser, $frame, $args ) { ... }
4111          *
4112          * The callback may either return the text result of the function, or an array with the text
4113          * in element 0, and a number of flags in the other elements. The names of the flags are
4114          * specified in the keys. Valid flags are:
4115          *   found                     The text returned is valid, stop processing the template. This
4116          *                             is on by default.
4117          *   nowiki                    Wiki markup in the return value should be escaped
4118          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4119          *
4120          * @public
4121          *
4122          * @param string $id The magic word ID
4123          * @param mixed $callback The callback function (and object) to use
4124          * @param integer $flags a combination of the following flags:
4125          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4126          *
4127          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4128          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4129          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4130          *     the arguments, and to control the way they are expanded.
4131          *
4132          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4133          *     arguments, for instance:
4134          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4135          *
4136          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4137          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4138          *     working if/when this is changed.
4139          *
4140          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4141          *     expansion.
4142          *
4143          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4144          *     about the methods available in PPFrame and PPNode.
4145          *
4146          * @return The old callback function for this name, if any
4147          */
4148         function setFunctionHook( $id, $callback, $flags = 0 ) {
4149                 global $wgContLang;
4150
4151                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4152                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4153
4154                 # Add to function cache
4155                 $mw = MagicWord::get( $id );
4156                 if( !$mw )
4157                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4158
4159                 $synonyms = $mw->getSynonyms();
4160                 $sensitive = intval( $mw->isCaseSensitive() );
4161
4162                 foreach ( $synonyms as $syn ) {
4163                         # Case
4164                         if ( !$sensitive ) {
4165                                 $syn = $wgContLang->lc( $syn );
4166                         }
4167                         # Add leading hash
4168                         if ( !( $flags & SFH_NO_HASH ) ) {
4169                                 $syn = '#' . $syn;
4170                         }
4171                         # Remove trailing colon
4172                         if ( substr( $syn, -1, 1 ) === ':' ) {
4173                                 $syn = substr( $syn, 0, -1 );
4174                         }
4175                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4176                 }
4177                 return $oldVal;
4178         }
4179
4180         /**
4181          * Get all registered function hook identifiers
4182          *
4183          * @return array
4184          */
4185         function getFunctionHooks() {
4186                 return array_keys( $this->mFunctionHooks );
4187         }
4188
4189         /**
4190          * Create a tag function, e.g. <test>some stuff</test>.
4191          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4192          * Unlike parser functions, their content is not preprocessed.
4193          */
4194         function setFunctionTagHook( $tag, $callback, $flags ) {
4195                 $tag = strtolower( $tag );
4196                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4197                         $this->mFunctionTagHooks[$tag] : null;
4198                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4199
4200                 if( !in_array( $tag, $this->mStripList ) ) {
4201                         $this->mStripList[] = $tag;
4202                 }
4203
4204                 return $old;
4205         }
4206
4207         /**
4208          * FIXME: update documentation. makeLinkObj() is deprecated.
4209          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4210          * Placeholders created in Skin::makeLinkObj()
4211          * Returns an array of link CSS classes, indexed by PDBK.
4212          */
4213         function replaceLinkHolders( &$text, $options = 0 ) {
4214                 return $this->mLinkHolders->replace( $text );
4215         }
4216
4217         /**
4218          * Replace <!--LINK--> link placeholders with plain text of links
4219          * (not HTML-formatted).
4220          * @param string $text
4221          * @return string
4222          */
4223         function replaceLinkHoldersText( $text ) {
4224                 return $this->mLinkHolders->replaceText( $text );
4225         }
4226
4227         /**
4228          * Tag hook handler for 'pre'.
4229          */
4230         function renderPreTag( $text, $attribs ) {
4231                 // Backwards-compatibility hack
4232                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4233
4234                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4235                 return Xml::openElement( 'pre', $attribs ) .
4236                         Xml::escapeTagsOnly( $content ) .
4237                         '</pre>';
4238         }
4239
4240         /**
4241          * Renders an image gallery from a text with one line per image.
4242          * text labels may be given by using |-style alternative text. E.g.
4243          *   Image:one.jpg|The number "1"
4244          *   Image:tree.jpg|A tree
4245          * given as text will return the HTML of a gallery with two images,
4246          * labeled 'The number "1"' and
4247          * 'A tree'.
4248          */
4249         function renderImageGallery( $text, $params ) {
4250                 $ig = new ImageGallery();
4251                 $ig->setContextTitle( $this->mTitle );
4252                 $ig->setShowBytes( false );
4253                 $ig->setShowFilename( false );
4254                 $ig->setParser( $this );
4255                 $ig->setHideBadImages();
4256                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4257                 $ig->useSkin( $this->mOptions->getSkin() );
4258                 $ig->mRevisionId = $this->mRevisionId;
4259
4260                 if( isset( $params['caption'] ) ) {
4261                         $caption = $params['caption'];
4262                         $caption = htmlspecialchars( $caption );
4263                         $caption = $this->replaceInternalLinks( $caption );
4264                         $ig->setCaptionHtml( $caption );
4265                 }
4266                 if( isset( $params['perrow'] ) ) {
4267                         $ig->setPerRow( $params['perrow'] );
4268                 }
4269                 if( isset( $params['widths'] ) ) {
4270                         $ig->setWidths( $params['widths'] );
4271                 }
4272                 if( isset( $params['heights'] ) ) {
4273                         $ig->setHeights( $params['heights'] );
4274                 }
4275
4276                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4277
4278                 $lines = StringUtils::explode( "\n", $text );
4279                 foreach ( $lines as $line ) {
4280                         # match lines like these:
4281                         # Image:someimage.jpg|This is some image
4282                         $matches = array();
4283                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4284                         # Skip empty lines
4285                         if ( count( $matches ) == 0 ) {
4286                                 continue;
4287                         }
4288
4289                         if ( strpos( $matches[0], '%' ) !== false )
4290                                 $matches[1] = urldecode( $matches[1] );
4291                         $tp = Title::newFromText( $matches[1]/*, NS_FILE*/ );
4292                         $nt =& $tp;
4293                         if( is_null( $nt ) ) {
4294                                 # Bogus title. Ignore these so we don't bomb out later.
4295                                 continue;
4296                         }
4297                         if ( isset( $matches[3] ) ) {
4298                                 $label = $matches[3];
4299                         } else {
4300                                 $label = '';
4301                         }
4302
4303                         $html = $this->recursiveTagParse( trim( $label ) );
4304
4305                         $ig->add( $nt, $html );
4306
4307                         # Only add real images (bug #5586)
4308                         if ( $nt->getNamespace() == NS_FILE ) {
4309                                 $this->mOutput->addImage( $nt->getDBkey() );
4310                         }
4311                 }
4312                 return $ig->toHTML();
4313         }
4314
4315         function getImageParams( $handler ) {
4316                 if ( $handler ) {
4317                         $handlerClass = get_class( $handler );
4318                 } else {
4319                         $handlerClass = '';
4320                 }
4321                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4322                         // Initialise static lists
4323                         static $internalParamNames = array(
4324                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4325                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4326                                         'bottom', 'text-bottom' ),
4327                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4328                                         'upright', 'border', 'link', 'alt' ),
4329                         );
4330                         static $internalParamMap;
4331                         if ( !$internalParamMap ) {
4332                                 $internalParamMap = array();
4333                                 foreach ( $internalParamNames as $type => $names ) {
4334                                         foreach ( $names as $name ) {
4335                                                 $magicName = str_replace( '-', '_', "img_$name" );
4336                                                 $internalParamMap[$magicName] = array( $type, $name );
4337                                         }
4338                                 }
4339                         }
4340
4341                         // Add handler params
4342                         $paramMap = $internalParamMap;
4343                         if ( $handler ) {
4344                                 $handlerParamMap = $handler->getParamMap();
4345                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4346                                         $paramMap[$magic] = array( 'handler', $paramName );
4347                                 }
4348                         }
4349                         $this->mImageParams[$handlerClass] = $paramMap;
4350                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4351                 }
4352                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4353         }
4354
4355         /**
4356          * Parse image options text and use it to make an image
4357          * @param Title $title
4358          * @param string $options
4359          * @param LinkHolderArray $holders
4360          */
4361         function makeImage( $title, $options, $holders = false ) {
4362                 # Check if the options text is of the form "options|alt text"
4363                 # Options are:
4364                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4365                 #  * left       no resizing, just left align. label is used for alt= only
4366                 #  * right      same, but right aligned
4367                 #  * none       same, but not aligned
4368                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4369                 #  * center     center the image
4370                 #  * frame      Keep original image size, no magnify-button.
4371                 #  * framed     Same as "frame"
4372                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4373                 #  * upright    reduce width for upright images, rounded to full __0 px
4374                 #  * border     draw a 1px border around the image
4375                 #  * alt        Text for HTML alt attribute (defaults to empty)
4376                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4377                 # vertical-align values (no % or length right now):
4378                 #  * baseline
4379                 #  * sub
4380                 #  * super
4381                 #  * top
4382                 #  * text-top
4383                 #  * middle
4384                 #  * bottom
4385                 #  * text-bottom
4386
4387                 $parts = StringUtils::explode( "|", $options );
4388                 $sk = $this->mOptions->getSkin();
4389
4390                 # Give extensions a chance to select the file revision for us
4391                 $skip = $time = $descQuery = false;
4392                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4393
4394                 if ( $skip ) {
4395                         return $sk->link( $title );
4396                 }
4397
4398                 # Get the file
4399                 $imagename = $title->getDBkey();
4400                 $file = wfFindFile( $title, array( 'time' => $time ) );
4401                 # Get parameter map
4402                 $handler = $file ? $file->getHandler() : false;
4403
4404                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4405
4406                 # Process the input parameters
4407                 $caption = '';
4408                 $params = array( 'frame' => array(), 'handler' => array(),
4409                         'horizAlign' => array(), 'vertAlign' => array() );
4410                 foreach( $parts as $part ) {
4411                         $part = trim( $part );
4412                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4413                         $validated = false;
4414                         if( isset( $paramMap[$magicName] ) ) {
4415                                 list( $type, $paramName ) = $paramMap[$magicName];
4416
4417                                 // Special case; width and height come in one variable together
4418                                 if( $type === 'handler' && $paramName === 'width' ) {
4419                                         $m = array();
4420                                         # (bug 13500) In both cases (width/height and width only),
4421                                         # permit trailing "px" for backward compatibility.
4422                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4423                                                 $width = intval( $m[1] );
4424                                                 $height = intval( $m[2] );
4425                                                 if ( $handler->validateParam( 'width', $width ) ) {
4426                                                         $params[$type]['width'] = $width;
4427                                                         $validated = true;
4428                                                 }
4429                                                 if ( $handler->validateParam( 'height', $height ) ) {
4430                                                         $params[$type]['height'] = $height;
4431                                                         $validated = true;
4432                                                 }
4433                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4434                                                 $width = intval( $value );
4435                                                 if ( $handler->validateParam( 'width', $width ) ) {
4436                                                         $params[$type]['width'] = $width;
4437                                                         $validated = true;
4438                                                 }
4439                                         } // else no validation -- bug 13436
4440                                 } else {
4441                                         if ( $type === 'handler' ) {
4442                                                 # Validate handler parameter
4443                                                 $validated = $handler->validateParam( $paramName, $value );
4444                                         } else {
4445                                                 # Validate internal parameters
4446                                                 switch( $paramName ) {
4447                                                 case 'manualthumb':
4448                                                 case 'alt':
4449                                                         // @fixme - possibly check validity here for
4450                                                         // manualthumb? downstream behavior seems odd with
4451                                                         // missing manual thumbs.
4452                                                         $validated = true;
4453                                                         $value = $this->stripAltText( $value, $holders );
4454                                                         break;
4455                                                 case 'link':
4456                                                         $chars = self::EXT_LINK_URL_CLASS;
4457                                                         $prots = $this->mUrlProtocols;
4458                                                         if ( $value === '' ) {
4459                                                                 $paramName = 'no-link';
4460                                                                 $value = true;
4461                                                                 $validated = true;
4462                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4463                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4464                                                                         $paramName = 'link-url';
4465                                                                         $this->mOutput->addExternalLink( $value );
4466                                                                         $validated = true;
4467                                                                 }
4468                                                         } else {
4469                                                                 $linkTitle = Title::newFromText( $value );
4470                                                                 if ( $linkTitle ) {
4471                                                                         $paramName = 'link-title';
4472                                                                         $value = $linkTitle;
4473                                                                         $this->mOutput->addLink( $linkTitle );
4474                                                                         $validated = true;
4475                                                                 }
4476                                                         }
4477                                                         break;
4478                                                 default:
4479                                                         // Most other things appear to be empty or numeric...
4480                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4481                                                 }
4482                                         }
4483
4484                                         if ( $validated ) {
4485                                                 $params[$type][$paramName] = $value;
4486                                         }
4487                                 }
4488                         }
4489                         if ( !$validated ) {
4490                                 $caption = $part;
4491                         }
4492                 }
4493
4494                 # Process alignment parameters
4495                 if ( $params['horizAlign'] ) {
4496                         $params['frame']['align'] = key( $params['horizAlign'] );
4497                 }
4498                 if ( $params['vertAlign'] ) {
4499                         $params['frame']['valign'] = key( $params['vertAlign'] );
4500                 }
4501
4502                 $params['frame']['caption'] = $caption;
4503
4504                 # Will the image be presented in a frame, with the caption below?
4505                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4506                                  isset( $params['frame']['framed'] ) ||
4507                                  isset( $params['frame']['thumbnail'] ) ||
4508                                  isset( $params['frame']['manualthumb'] );
4509
4510                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4511                 # came to also set the caption, ordinary text after the image -- which
4512                 # makes no sense, because that just repeats the text multiple times in
4513                 # screen readers.  It *also* came to set the title attribute.
4514                 #
4515                 # Now that we have an alt attribute, we should not set the alt text to
4516                 # equal the caption: that's worse than useless, it just repeats the
4517                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4518                 # use the unnamed parameter for alt text as well, just for the time be-
4519                 # ing, if the unnamed param is set and the alt param is not.
4520                 #
4521                 # For the future, we need to figure out if we want to tweak this more,
4522                 # e.g., introducing a title= parameter for the title; ignoring the un-
4523                 # named parameter entirely for images without a caption; adding an ex-
4524                 # plicit caption= parameter and preserving the old magic unnamed para-
4525                 # meter for BC; ...
4526                 if ( $imageIsFramed ) { # Framed image
4527                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4528                                 # No caption or alt text, add the filename as the alt text so
4529                                 # that screen readers at least get some description of the image
4530                                 $params['frame']['alt'] = $title->getText();
4531                         }
4532                         # Do not set $params['frame']['title'] because tooltips don't make sense
4533                         # for framed images
4534                 } else { # Inline image
4535                         if ( !isset( $params['frame']['alt'] ) ) {
4536                                 # No alt text, use the "caption" for the alt text
4537                                 if ( $caption !== '') {
4538                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4539                                 } else {
4540                                         # No caption, fall back to using the filename for the
4541                                         # alt text
4542                                         $params['frame']['alt'] = $title->getText();
4543                                 }
4544                         }
4545                         # Use the "caption" for the tooltip text
4546                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4547                 }
4548
4549                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4550
4551                 # Linker does the rest
4552                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4553
4554                 # Give the handler a chance to modify the parser object
4555                 if ( $handler ) {
4556                         $handler->parserTransformHook( $this, $file );
4557                 }
4558
4559                 return $ret;
4560         }
4561
4562         protected function stripAltText( $caption, $holders ) {
4563                 # Strip bad stuff out of the title (tooltip).  We can't just use
4564                 # replaceLinkHoldersText() here, because if this function is called
4565                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4566                 if ( $holders ) {
4567                         $tooltip = $holders->replaceText( $caption );
4568                 } else {
4569                         $tooltip = $this->replaceLinkHoldersText( $caption );
4570                 }
4571
4572                 # make sure there are no placeholders in thumbnail attributes
4573                 # that are later expanded to html- so expand them now and
4574                 # remove the tags
4575                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4576                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4577
4578                 return $tooltip;
4579         }
4580
4581         /**
4582          * Set a flag in the output object indicating that the content is dynamic and
4583          * shouldn't be cached.
4584          */
4585         function disableCache() {
4586                 wfDebug( "Parser output marked as uncacheable.\n" );
4587                 $this->mOutput->mCacheTime = -1;
4588         }
4589
4590         /**#@+
4591          * Callback from the Sanitizer for expanding items found in HTML attribute
4592          * values, so they can be safely tested and escaped.
4593          * @param string $text
4594          * @param PPFrame $frame
4595          * @return string
4596          * @private
4597          */
4598         function attributeStripCallback( &$text, $frame = false ) {
4599                 $text = $this->replaceVariables( $text, $frame );
4600                 $text = $this->mStripState->unstripBoth( $text );
4601                 return $text;
4602         }
4603
4604         /**#@-*/
4605
4606         /**#@+
4607          * Accessor/mutator
4608          */
4609         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4610         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4611         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4612         /**#@-*/
4613
4614         /**#@+
4615          * Accessor
4616          */
4617         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4618         /**#@-*/
4619
4620
4621         /**
4622          * Break wikitext input into sections, and either pull or replace
4623          * some particular section's text.
4624          *
4625          * External callers should use the getSection and replaceSection methods.
4626          *
4627          * @param string $text Page wikitext
4628          * @param string $section A section identifier string of the form:
4629          *   <flag1> - <flag2> - ... - <section number>
4630          *
4631          * Currently the only recognised flag is "T", which means the target section number
4632          * was derived during a template inclusion parse, in other words this is a template
4633          * section edit link. If no flags are given, it was an ordinary section edit link.
4634          * This flag is required to avoid a section numbering mismatch when a section is
4635          * enclosed by <includeonly> (bug 6563).
4636          *
4637          * The section number 0 pulls the text before the first heading; other numbers will
4638          * pull the given section along with its lower-level subsections. If the section is
4639          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4640          *
4641          * @param string $mode One of "get" or "replace"
4642          * @param string $newText Replacement text for section data.
4643          * @return string for "get", the extracted section text.
4644          *                for "replace", the whole page with the section replaced.
4645          */
4646         private function extractSections( $text, $section, $mode, $newText='' ) {
4647                 global $wgTitle;
4648                 $this->clearState();
4649                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4650                 $this->mOptions = new ParserOptions;
4651                 $this->setOutputType( self::OT_WIKI );
4652                 $outText = '';
4653                 $frame = $this->getPreprocessor()->newFrame();
4654
4655                 // Process section extraction flags
4656                 $flags = 0;
4657                 $sectionParts = explode( '-', $section );
4658                 $sectionIndex = array_pop( $sectionParts );
4659                 foreach ( $sectionParts as $part ) {
4660                         if ( $part === 'T' ) {
4661                                 $flags |= self::PTD_FOR_INCLUSION;
4662                         }
4663                 }
4664                 // Preprocess the text
4665                 $root = $this->preprocessToDom( $text, $flags );
4666
4667                 // <h> nodes indicate section breaks
4668                 // They can only occur at the top level, so we can find them by iterating the root's children
4669                 $node = $root->getFirstChild();
4670
4671                 // Find the target section
4672                 if ( $sectionIndex == 0 ) {
4673                         // Section zero doesn't nest, level=big
4674                         $targetLevel = 1000;
4675                 } else {
4676             while ( $node ) {
4677                 if ( $node->getName() === 'h' ) {
4678                     $bits = $node->splitHeading();
4679                                         if ( $bits['i'] == $sectionIndex ) {
4680                                         $targetLevel = $bits['level'];
4681                                                 break;
4682                                         }
4683                                 }
4684                                 if ( $mode === 'replace' ) {
4685                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4686                                 }
4687                                 $node = $node->getNextSibling();
4688                         }
4689                 }
4690
4691                 if ( !$node ) {
4692                         // Not found
4693                         if ( $mode === 'get' ) {
4694                                 return $newText;
4695                         } else {
4696                                 return $text;
4697                         }
4698                 }
4699
4700                 // Find the end of the section, including nested sections
4701                 do {
4702                         if ( $node->getName() === 'h' ) {
4703                                 $bits = $node->splitHeading();
4704                                 $curLevel = $bits['level'];
4705                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4706                                         break;
4707                                 }
4708                         }
4709                         if ( $mode === 'get' ) {
4710                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4711                         }
4712                         $node = $node->getNextSibling();
4713                 } while ( $node );
4714
4715                 // Write out the remainder (in replace mode only)
4716                 if ( $mode === 'replace' ) {
4717                         // Output the replacement text
4718                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4719                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4720                         // Only add trailing whitespace if there is newText
4721                         if($newText != "") {
4722                                 $outText .= $newText . "\n\n";
4723                         }
4724
4725                         while ( $node ) {
4726                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4727                                 $node = $node->getNextSibling();
4728                         }
4729                 }
4730
4731                 if ( is_string( $outText ) ) {
4732                         // Re-insert stripped tags
4733                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4734                 }
4735
4736                 return $outText;
4737         }
4738
4739         /**
4740          * This function returns the text of a section, specified by a number ($section).
4741          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4742          * the first section before any such heading (section 0).
4743          *
4744          * If a section contains subsections, these are also returned.
4745          *
4746          * @param string $text text to look in
4747          * @param string $section section identifier
4748          * @param string $deftext default to return if section is not found
4749          * @return string text of the requested section
4750          */
4751         public function getSection( $text, $section, $deftext='' ) {
4752                 return $this->extractSections( $text, $section, "get", $deftext );
4753         }
4754
4755         public function replaceSection( $oldtext, $section, $text ) {
4756                 return $this->extractSections( $oldtext, $section, "replace", $text );
4757         }
4758
4759         /**
4760          * Get the timestamp associated with the current revision, adjusted for
4761          * the default server-local timestamp
4762          */
4763         function getRevisionTimestamp() {
4764                 if ( is_null( $this->mRevisionTimestamp ) ) {
4765                         wfProfileIn( __METHOD__ );
4766                         global $wgContLang;
4767                         $dbr = wfGetDB( DB_SLAVE );
4768                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4769                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4770
4771                         // Normalize timestamp to internal MW format for timezone processing.
4772                         // This has the added side-effect of replacing a null value with
4773                         // the current time, which gives us more sensible behavior for
4774                         // previews.
4775                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4776
4777                         // The cryptic '' timezone parameter tells to use the site-default
4778                         // timezone offset instead of the user settings.
4779                         //
4780                         // Since this value will be saved into the parser cache, served
4781                         // to other users, and potentially even used inside links and such,
4782                         // it needs to be consistent for all visitors.
4783                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4784
4785                         wfProfileOut( __METHOD__ );
4786                 }
4787                 return $this->mRevisionTimestamp;
4788         }
4789
4790         /**
4791          * Get the name of the user that edited the last revision
4792          */
4793         function getRevisionUser() {
4794                 // if this template is subst: the revision id will be blank,
4795                 // so just use the current user's name
4796                 if( $this->mRevisionId ) {
4797                         $revision = Revision::newFromId( $this->mRevisionId );
4798                         $revuser = $revision->getUserText();
4799                 } else {
4800                         global $wgUser;
4801                         $revuser = $wgUser->getName();
4802                 }
4803                 return $revuser;
4804         }
4805
4806         /**
4807          * Mutator for $mDefaultSort
4808          *
4809          * @param $sort New value
4810          */
4811         public function setDefaultSort( $sort ) {
4812                 $this->mDefaultSort = $sort;
4813         }
4814
4815         /**
4816          * Accessor for $mDefaultSort
4817          * Will use the title/prefixed title if none is set
4818          *
4819          * @return string
4820          */
4821         public function getDefaultSort() {
4822                 global $wgCategoryPrefixedDefaultSortkey;
4823                 if( $this->mDefaultSort !== false ) {
4824                         return $this->mDefaultSort;
4825                 } elseif ($this->mTitle->getNamespace() == NS_CATEGORY ||
4826                         !$wgCategoryPrefixedDefaultSortkey) {
4827                         return $this->mTitle->getText();
4828                 } else {
4829                         return $this->mTitle->getPrefixedText();
4830                 }
4831         }
4832
4833         /**
4834          * Accessor for $mDefaultSort
4835          * Unlike getDefaultSort(), will return false if none is set
4836          *
4837          * @return string or false
4838          */
4839         public function getCustomDefaultSort() {
4840                 return $this->mDefaultSort;
4841         }
4842
4843         /**
4844          * Try to guess the section anchor name based on a wikitext fragment
4845          * presumably extracted from a heading, for example "Header" from
4846          * "== Header ==".
4847          */
4848         public function guessSectionNameFromWikiText( $text ) {
4849                 # Strip out wikitext links(they break the anchor)
4850                 $text = $this->stripSectionName( $text );
4851                 $headline = Sanitizer::decodeCharReferences( $text );
4852                 # strip out HTML
4853                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
4854                 $headline = trim( $headline );
4855                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
4856                 $replacearray = array(
4857                         '%3A' => ':',
4858                         '%' => '.'
4859                 );
4860                 return str_replace(
4861                         array_keys( $replacearray ),
4862                         array_values( $replacearray ),
4863                         $sectionanchor );
4864         }
4865
4866         /**
4867          * Strips a text string of wikitext for use in a section anchor
4868          *
4869          * Accepts a text string and then removes all wikitext from the
4870          * string and leaves only the resultant text (i.e. the result of
4871          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
4872          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
4873          * to create valid section anchors by mimicing the output of the
4874          * parser when headings are parsed.
4875          *
4876          * @param $text string Text string to be stripped of wikitext
4877          * for use in a Section anchor
4878          * @return Filtered text string
4879          */
4880         public function stripSectionName( $text ) {
4881                 # Strip internal link markup
4882                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
4883                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
4884
4885                 # Strip external link markup (FIXME: Not Tolerant to blank link text
4886                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
4887                 # on how many empty links there are on the page - need to figure that out.
4888                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
4889
4890                 # Parse wikitext quotes (italics & bold)
4891                 $text = $this->doQuotes($text);
4892
4893                 # Strip HTML tags
4894                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
4895                 return $text;
4896         }
4897
4898         function srvus( $text ) {
4899                 return $this->testSrvus( $text, $this->mOutputType );
4900         }
4901
4902         /**
4903          * strip/replaceVariables/unstrip for preprocessor regression testing
4904          */
4905         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
4906                 $this->clearState();
4907                 if ( ! ( $title instanceof Title ) ) {
4908                         $title = Title::newFromText( $title );
4909                 }
4910                 $this->mTitle = $title;
4911                 $this->mOptions = $options;
4912                 $this->setOutputType( $outputType );
4913                 $text = $this->replaceVariables( $text );
4914                 $text = $this->mStripState->unstripBoth( $text );
4915                 $text = Sanitizer::removeHTMLtags( $text );
4916                 return $text;
4917         }
4918
4919         function testPst( $text, $title, $options ) {
4920                 global $wgUser;
4921                 if ( ! ( $title instanceof Title ) ) {
4922                         $title = Title::newFromText( $title );
4923                 }
4924                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
4925         }
4926
4927         function testPreprocess( $text, $title, $options ) {
4928                 if ( ! ( $title instanceof Title ) ) {
4929                         $title = Title::newFromText( $title );
4930                 }
4931                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
4932         }
4933
4934         function markerSkipCallback( $s, $callback ) {
4935                 $i = 0;
4936                 $out = '';
4937                 while ( $i < strlen( $s ) ) {
4938                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
4939                         if ( $markerStart === false ) {
4940                                 $out .= call_user_func( $callback, substr( $s, $i ) );
4941                                 break;
4942                         } else {
4943                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
4944                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
4945                                 if ( $markerEnd === false ) {
4946                                         $out .= substr( $s, $markerStart );
4947                                         break;
4948                                 } else {
4949                                         $markerEnd += strlen( self::MARKER_SUFFIX );
4950                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
4951                                         $i = $markerEnd;
4952                                 }
4953                         }
4954                 }
4955                 return $out;
4956         }
4957
4958         function serialiseHalfParsedText( $text ) {
4959                 $data = array();
4960                 $data['text'] = $text;
4961
4962                 // First, find all strip markers, and store their
4963                 //  data in an array.
4964                 $stripState = new StripState;
4965                 $pos = 0;
4966                 while( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) ) && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) ) {
4967                         $end_pos += strlen( self::MARKER_SUFFIX );
4968                         $marker = substr( $text, $start_pos, $end_pos-$start_pos );
4969
4970                         if ( !empty( $this->mStripState->general->data[$marker] ) ) {
4971                                 $replaceArray = $stripState->general;
4972                                 $stripText = $this->mStripState->general->data[$marker];
4973                         } elseif ( !empty( $this->mStripState->nowiki->data[$marker] ) ) {
4974                                 $replaceArray = $stripState->nowiki;
4975                                 $stripText = $this->mStripState->nowiki->data[$marker];
4976                         } else {
4977                                 throw new MWException( "Hanging strip marker: '$marker'." );
4978                         }
4979
4980                         $replaceArray->setPair( $marker, $stripText );
4981                         $pos = $end_pos;
4982                 }
4983                 $data['stripstate'] = $stripState;
4984
4985                 // Now, find all of our links, and store THEIR
4986                 //  data in an array! :)
4987                 $links = array( 'internal' => array(), 'interwiki' => array() );
4988                 $pos = 0;
4989
4990                 // Internal links
4991                 while( ( $start_pos = strpos( $text, '<!--LINK ', $pos ) ) ) {
4992                         list( $ns, $trail ) = explode( ':', substr( $text, $start_pos + strlen( '<!--LINK ' ) ), 2 );
4993
4994                         $ns = trim($ns);
4995                         if (empty( $links['internal'][$ns] )) {
4996                                 $links['internal'][$ns] = array();
4997                         }
4998
4999                         $key = trim( substr( $trail, 0, strpos( $trail, '-->' ) ) );
5000                         $links['internal'][$ns][] = $this->mLinkHolders->internals[$ns][$key];
5001                         $pos = $start_pos + strlen( "<!--LINK $ns:$key-->" );
5002                 }
5003
5004                 $pos = 0;
5005
5006                 // Interwiki links
5007                 while( ( $start_pos = strpos( $text, '<!--IWLINK ', $pos ) ) ) {
5008                         $data = substr( $text, $start_pos );
5009                         $key = trim( substr( $data, 0, strpos( $data, '-->' ) ) );
5010                         $links['interwiki'][] = $this->mLinkHolders->interwiki[$key];
5011                         $pos = $start_pos + strlen( "<!--IWLINK $key-->" );
5012                 }
5013
5014                 $data['linkholder'] = $links;
5015
5016                 return $data;
5017         }
5018
5019         function unserialiseHalfParsedText( $data, $intPrefix = null /* Unique identifying prefix */ ) {
5020                 if (!$intPrefix)
5021                         $intPrefix = $this->getRandomString();
5022
5023                 // First, extract the strip state.
5024                 $stripState = $data['stripstate'];
5025                 $this->mStripState->general->merge( $stripState->general );
5026                 $this->mStripState->nowiki->merge( $stripState->nowiki );
5027
5028                 // Now, extract the text, and renumber links
5029                 $text = $data['text'];
5030                 $links = $data['linkholder'];
5031
5032                 // Internal...
5033                 foreach( $links['internal'] as $ns => $nsLinks ) {
5034                         foreach( $nsLinks as $key => $entry ) {
5035                                 $newKey = $intPrefix . '-' . $key;
5036                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
5037
5038                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
5039                         }
5040                 }
5041
5042                 // Interwiki...
5043                 foreach( $links['interwiki'] as $key => $entry ) {
5044                         $newKey = "$intPrefix-$key";
5045                         $this->mLinkHolders->interwikis[$newKey] = $entry;
5046
5047                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
5048                 }
5049
5050                 // Should be good to go.
5051                 return $text;
5052         }
5053 }
5054
5055 /**
5056  * @todo document, briefly.
5057  * @ingroup Parser
5058  */
5059 class StripState {
5060         var $general, $nowiki;
5061
5062         function __construct() {
5063                 $this->general = new ReplacementArray;
5064                 $this->nowiki = new ReplacementArray;
5065         }
5066
5067         function unstripGeneral( $text ) {
5068                 wfProfileIn( __METHOD__ );
5069                 do {
5070                         $oldText = $text;
5071                         $text = $this->general->replace( $text );
5072                 } while ( $text !== $oldText );
5073                 wfProfileOut( __METHOD__ );
5074                 return $text;
5075         }
5076
5077         function unstripNoWiki( $text ) {
5078                 wfProfileIn( __METHOD__ );
5079                 do {
5080                         $oldText = $text;
5081                         $text = $this->nowiki->replace( $text );
5082                 } while ( $text !== $oldText );
5083                 wfProfileOut( __METHOD__ );
5084                 return $text;
5085         }
5086
5087         function unstripBoth( $text ) {
5088                 wfProfileIn( __METHOD__ );
5089                 do {
5090                         $oldText = $text;
5091                         $text = $this->general->replace( $text );
5092                         $text = $this->nowiki->replace( $text );
5093                 } while ( $text !== $oldText );
5094                 wfProfileOut( __METHOD__ );
5095                 return $text;
5096         }
5097 }
5098
5099 /**
5100  * @todo document, briefly.
5101  * @ingroup Parser
5102  */
5103 class OnlyIncludeReplacer {
5104         var $output = '';
5105
5106         function replace( $matches ) {
5107                 if ( substr( $matches[1], -1 ) === "\n" ) {
5108                         $this->output .= substr( $matches[1], 0, -1 );
5109                 } else {
5110                         $this->output .= $matches[1];
5111                 }
5112         }
5113 }