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                 $mSubsts, $mImageParams, $mImageParamsMagicArray, $mStripList, $mMarkerIndex,
  95                 $mPreprocessor, $mExtLinkBracketedRegex, $mUrlProtocols, $mDefaultStripList,
  96                 $mVarCache, $mConf, $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                 $this->mVarCache = array();
 205
 206                 /**
 207                  * Prefix for temporary replacement strings for the multipass parser.
 208                  * \x07 should never appear in input as it's disallowed in XML.
 209                  * Using it at the front also gives us a little extra robustness
 210                  * since it shouldn't match when butted up against identifier-like
 211                  * string constructs.
 212                  *
 213                  * Must not consist of all title characters, or else it will change
 214                  * the behaviour of <nowiki> in a link.
 215                  */
 216                 #$this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 217                 # Changed to \x7f to allow XML double-parsing -- TS
 218                 $this->mUniqPrefix = "\x7fUNIQ" . self::getRandomString();
 219
 220
 221                 # Clear these on every parse, bug 4549
 222                 $this->mTplExpandCache = $this->mTplRedirCache = $this->mTplDomCache = array();
 223
 224                 $this->mShowToc = true;
 225                 $this->mForceTocPosition = false;
 226                 $this->mIncludeSizes = array(
 227                         'post-expand' => 0,
 228                         'arg' => 0,
 229                 );
 230                 $this->mPPNodeCount = 0;
 231                 $this->mDefaultSort = false;
 232                 $this->mHeadings = array();
 233                 $this->mDoubleUnderscores = array();
 234                 $this->mExpensiveFunctionCount = 0;
 235
 236                 # Fix cloning
 237                 if ( isset( $this->mPreprocessor ) && $this->mPreprocessor->parser !== $this ) {
 238                         $this->mPreprocessor = null;
 239                 }
 240
 241                 wfRunHooks( 'ParserClearState', array( &$this ) );
 242                 wfProfileOut( __METHOD__ );
 243         }
 244
 245         function setOutputType( $ot ) {
 246                 $this->mOutputType = $ot;
 247                 // Shortcut alias
 248                 $this->ot = array(
 249                         'html' => $ot == self::OT_HTML,
 250                         'wiki' => $ot == self::OT_WIKI,
 251                         'pre' => $ot == self::OT_PREPROCESS,
 252                 );
 253         }
 254
 255         /**
 256          * Set the context title
 257          */
 258         function setTitle( $t ) {
 259                 if ( !$t || $t instanceof FakeTitle ) {
 260                         $t = Title::newFromText( 'NO TITLE' );
 261                 }
 262
 263                 if ( strval( $t->getFragment() ) !== '' ) {
 264                         # Strip the fragment to avoid various odd effects
 265                         $this->mTitle = clone $t;
 266                         $this->mTitle->setFragment( '' );
 267                 } else {
 268                         $this->mTitle = $t;
 269                 }
 270         }
 271
 272         /**
 273          * Accessor for mUniqPrefix.
 274          *
 275          * @public
 276          */
 277         function uniqPrefix() {
 278                 if( !isset( $this->mUniqPrefix ) ) {
 279                         // @todo Fixme: this is probably *horribly wrong*
 280                         // LanguageConverter seems to want $wgParser's uniqPrefix, however
 281                         // if this is called for a parser cache hit, the parser may not
 282                         // have ever been initialized in the first place.
 283                         // Not really sure what the heck is supposed to be going on here.
 284                         return '';
 285                         //throw new MWException( "Accessing uninitialized mUniqPrefix" );
 286                 }
 287                 return $this->mUniqPrefix;
 288         }
 289
 290         /**
 291          * Convert wikitext to HTML
 292          * Do not call this function recursively.
 293          *
 294          * @param $text String: text we want to parse
 295          * @param $title A title object
 296          * @param $options ParserOptions
 297          * @param $linestart boolean
 298          * @param $clearState boolean
 299          * @param $revid Int: number to pass in {{REVISIONID}}
 300          * @return ParserOutput a ParserOutput
 301          */
 302         public function parse( $text, Title $title, ParserOptions $options, $linestart = true, $clearState = true, $revid = null ) {
 303                 /**
 304                  * First pass--just handle <nowiki> sections, pass the rest off
 305                  * to internalParse() which does all the real work.
 306                  */
 307
 308                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang, $wgDisableLangConversion;
 309                 $fname = __METHOD__.'-' . wfGetCaller();
 310                 wfProfileIn( __METHOD__ );
 311                 wfProfileIn( $fname );
 312
 313                 if ( $clearState ) {
 314                         $this->clearState();
 315                 }
 316
 317                 $this->mOptions = $options;
 318                 $this->setTitle( $title ); // Page title has to be set for the pre-processor
 319
 320                 $oldRevisionId = $this->mRevisionId;
 321                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 322                 if( $revid !== null ) {
 323                         $this->mRevisionId = $revid;
 324                         $this->mRevisionTimestamp = null;
 325                 }
 326                 $this->setOutputType( self::OT_HTML );
 327                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 328                 # No more strip!
 329                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 330                 $text = $this->internalParse( $text );
 331
 332                 $text = $this->mStripState->unstripGeneral( $text );
 333
 334                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 335                 $fixtags = array(
 336                         # french spaces, last one Guillemet-left
 337                         # only if there is something before the space
 338                         '/(.) (?=\\?|:|;|!|%|\\302\\273)/' => '\\1&nbsp;\\2',
 339                         # french spaces, Guillemet-right
 340                         '/(\\302\\253) /' => '\\1&nbsp;',
 341                         '/&nbsp;(!\s*important)/' => ' \\1', #Beware of CSS magic word !important, bug #11874.
 342                 );
 343                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 344
 345                 $text = $this->doBlockLevels( $text, $linestart );
 346
 347                 $this->replaceLinkHolders( $text );
 348
 349                 // The position of the convert() call should not be changed. it
 350                 // assumes that the links are all replaced and the only thing left
 351                 // is the <nowiki> mark.
 352                 if ( !( $wgDisableLangConversion
 353                                 || isset( $this->mDoubleUnderscores['nocontentconvert'] )
 354                                 || $this->mTitle->isTalkPage()
 355                                 || $this->mTitle->isConversionTable() ) ) {
 356                         $text = $wgContLang->convert( $text );
 357                 }
 358
 359                 // A title may have been set in a conversion rule.
 360                 // Note that if a user tries to set a title in a conversion
 361                 // rule but content conversion was not done, then the parser
 362                 // won't pick it up.  This is probably expected behavior.
 363                 if ( $wgContLang->getConvRuleTitle() ) {
 364                         $this->mOutput->setTitleText( $wgContLang->getConvRuleTitle() );
 365                 } elseif ( !( $wgDisableLangConversion
 366                                           || isset( $this->mDoubleUnderscores['notitleconvert'] ) ) ) {
 367                         $this->mOutput->setTitleText( $wgContLang->convert( $title->getPrefixedText() ) );
 368                 }
 369
 370                 $text = $this->mStripState->unstripNoWiki( $text );
 371
 372                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 373
 374 //!JF Move to its own function
 375
 376                 $uniq_prefix = $this->mUniqPrefix;
 377                 $matches = array();
 378                 $elements = array_keys( $this->mTransparentTagHooks );
 379                 $text = self::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 380
 381                 foreach( $matches as $marker => $data ) {
 382                         list( $element, $content, $params, $tag ) = $data;
 383                         $tagName = strtolower( $element );
 384                         if( isset( $this->mTransparentTagHooks[$tagName] ) ) {
 385                                 $output = call_user_func_array( $this->mTransparentTagHooks[$tagName],
 386                                         array( $content, $params, $this ) );
 387                         } else {
 388                                 $output = $tag;
 389                         }
 390                         $this->mStripState->general->setPair( $marker, $output );
 391                 }
 392                 $text = $this->mStripState->unstripGeneral( $text );
 393
 394                 $text = Sanitizer::normalizeCharReferences( $text );
 395
 396                 if ( ( $wgUseTidy && $this->mOptions->mTidy ) || $wgAlwaysUseTidy ) {
 397                         $text = MWTidy::tidy( $text );
 398                 } else {
 399                         # attempt to sanitize at least some nesting problems
 400                         # (bug #2702 and quite a few others)
 401                         $tidyregs = array(
 402                                 # ''Something [http://www.cool.com cool''] -->
 403                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 404                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 405                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 406                                 # fix up an anchor inside another anchor, only
 407                                 # at least for a single single nested link (bug 3695)
 408                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 409                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 410                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 411                                 # contains a div, so fix it up here; replace
 412                                 # div with escaped text
 413                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 414                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 415                                 # remove empty italic or bold tag pairs, some
 416                                 # introduced by rules above
 417                                 '/<([bi])><\/\\1>/' => '',
 418                         );
 419
 420                         $text = preg_replace(
 421                                 array_keys( $tidyregs ),
 422                                 array_values( $tidyregs ),
 423                                 $text );
 424                 }
 425                 global $wgExpensiveParserFunctionLimit;
 426                 if ( $this->mExpensiveFunctionCount > $wgExpensiveParserFunctionLimit ) {
 427                         $this->limitationWarn( 'expensive-parserfunction', $this->mExpensiveFunctionCount, $wgExpensiveParserFunctionLimit );
 428                 }
 429
 430                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 431
 432                 # Information on include size limits, for the benefit of users who try to skirt them
 433                 if ( $this->mOptions->getEnableLimitReport() ) {
 434                         global $wgExpensiveParserFunctionLimit;
 435                         $max = $this->mOptions->getMaxIncludeSize();
 436                         $PFreport = "Expensive parser function count: {$this->mExpensiveFunctionCount}/$wgExpensiveParserFunctionLimit\n";
 437                         $limitReport =
 438                                 "NewPP limit report\n" .
 439                                 "Preprocessor node count: {$this->mPPNodeCount}/{$this->mOptions->mMaxPPNodeCount}\n" .
 440                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']}/$max bytes\n" .
 441                                 "Template argument size: {$this->mIncludeSizes['arg']}/$max bytes\n".
 442                                 $PFreport;
 443                         wfRunHooks( 'ParserLimitReport', array( $this, &$limitReport ) );
 444                         $text .= "\n<!-- \n$limitReport-->\n";
 445                 }
 446                 $this->mOutput->setText( $text );
 447
 448                 $this->mRevisionId = $oldRevisionId;
 449                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 450                 wfProfileOut( $fname );
 451                 wfProfileOut( __METHOD__ );
 452
 453                 return $this->mOutput;
 454         }
 455
 456         /**
 457          * Recursive parser entry point that can be called from an extension tag
 458          * hook.
 459          *
 460          * If $frame is not provided, then template variables (e.g., {{{1}}}) within $text are not expanded
 461          *
 462          * @param $text String: text extension wants to have parsed
 463          * @param PPFrame $frame: The frame to use for expanding any template variables
 464          */
 465         function recursiveTagParse( $text, $frame=false ) {
 466                 wfProfileIn( __METHOD__ );
 467                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 468                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 469                 $text = $this->internalParse( $text, false, $frame );
 470                 wfProfileOut( __METHOD__ );
 471                 return $text;
 472         }
 473
 474         /**
 475          * Expand templates and variables in the text, producing valid, static wikitext.
 476          * Also removes comments.
 477          */
 478         function preprocess( $text, $title, $options, $revid = null ) {
 479                 wfProfileIn( __METHOD__ );
 480                 $this->clearState();
 481                 $this->setOutputType( self::OT_PREPROCESS );
 482                 $this->mOptions = $options;
 483                 $this->setTitle( $title );
 484                 if( $revid !== null ) {
 485                         $this->mRevisionId = $revid;
 486                 }
 487                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 488                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 489                 $text = $this->replaceVariables( $text );
 490                 $text = $this->mStripState->unstripBoth( $text );
 491                 wfProfileOut( __METHOD__ );
 492                 return $text;
 493         }
 494
 495         /**
 496          * Get a random string
 497          *
 498          * @private
 499          * @static
 500          */
 501         function getRandomString() {
 502                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 503         }
 504
 505         function &getTitle() { return $this->mTitle; }
 506         function getOptions() { return $this->mOptions; }
 507         function getRevisionId() { return $this->mRevisionId; }
 508         function getOutput() { return $this->mOutput; }
 509         function nextLinkID() { return $this->mLinkID++; }
 510
 511         function getFunctionLang() {
 512                 global $wgLang, $wgContLang;
 513
 514                 $target = $this->mOptions->getTargetLanguage();
 515                 if ( $target !== null ) {
 516                         return $target;
 517                 } else {
 518                         return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 519                 }
 520         }
 521
 522         /**
 523          * Get a preprocessor object
 524          */
 525         function getPreprocessor() {
 526                 if ( !isset( $this->mPreprocessor ) ) {
 527                         $class = $this->mPreprocessorClass;
 528                         $this->mPreprocessor = new $class( $this );
 529                 }
 530                 return $this->mPreprocessor;
 531         }
 532
 533         /**
 534          * Replaces all occurrences of HTML-style comments and the given tags
 535          * in the text with a random marker and returns the next text. The output
 536          * parameter $matches will be an associative array filled with data in
 537          * the form:
 538          *   'UNIQ-xxxxx' => array(
 539          *     'element',
 540          *     'tag content',
 541          *     array( 'param' => 'x' ),
 542          *     '<element param="x">tag content</element>' ) )
 543          *
 544          * @param $elements list of element names. Comments are always extracted.
 545          * @param $text Source text string.
 546          * @param $uniq_prefix
 547          *
 548          * @public
 549          * @static
 550          */
 551         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 552                 static $n = 1;
 553                 $stripped = '';
 554                 $matches = array();
 555
 556                 $taglist = implode( '|', $elements );
 557                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?" . ">)|<(!--)/i";
 558
 559                 while ( $text != '' ) {
 560                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 561                         $stripped .= $p[0];
 562                         if( count( $p ) < 5 ) {
 563                                 break;
 564                         }
 565                         if( count( $p ) > 5 ) {
 566                                 // comment
 567                                 $element    = $p[4];
 568                                 $attributes = '';
 569                                 $close      = '';
 570                                 $inside     = $p[5];
 571                         } else {
 572                                 // tag
 573                                 $element    = $p[1];
 574                                 $attributes = $p[2];
 575                                 $close      = $p[3];
 576                                 $inside     = $p[4];
 577                         }
 578
 579                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . self::MARKER_SUFFIX;
 580                         $stripped .= $marker;
 581
 582                         if ( $close === '/>' ) {
 583                                 // Empty element tag, <tag />
 584                                 $content = null;
 585                                 $text = $inside;
 586                                 $tail = null;
 587                         } else {
 588                                 if( $element === '!--' ) {
 589                                         $end = '/(-->)/';
 590                                 } else {
 591                                         $end = "/(<\\/$element\\s*>)/i";
 592                                 }
 593                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 594                                 $content = $q[0];
 595                                 if( count( $q ) < 3 ) {
 596                                         # No end tag -- let it run out to the end of the text.
 597                                         $tail = '';
 598                                         $text = '';
 599                                 } else {
 600                                         $tail = $q[1];
 601                                         $text = $q[2];
 602                                 }
 603                         }
 604
 605                         $matches[$marker] = array( $element,
 606                                 $content,
 607                                 Sanitizer::decodeTagAttributes( $attributes ),
 608                                 "<$element$attributes$close$content$tail" );
 609                 }
 610                 return $stripped;
 611         }
 612
 613         /**
 614          * Get a list of strippable XML-like elements
 615          */
 616         function getStripList() {
 617                 global $wgRawHtml;
 618                 $elements = $this->mStripList;
 619                 if( $wgRawHtml ) {
 620                         $elements[] = 'html';
 621                 }
 622                 if( $this->mOptions->getUseTeX() ) {
 623                         $elements[] = 'math';
 624                 }
 625                 return $elements;
 626         }
 627
 628         /**
 629          * @deprecated use replaceVariables
 630          */
 631         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 632                 return $text;
 633         }
 634
 635         /**
 636          * Restores pre, math, and other extensions removed by strip()
 637          *
 638          * always call unstripNoWiki() after this one
 639          * @private
 640          * @deprecated use $this->mStripState->unstrip()
 641          */
 642         function unstrip( $text, $state ) {
 643                 return $state->unstripGeneral( $text );
 644         }
 645
 646         /**
 647          * Always call this after unstrip() to preserve the order
 648          *
 649          * @private
 650          * @deprecated use $this->mStripState->unstrip()
 651          */
 652         function unstripNoWiki( $text, $state ) {
 653                 return $state->unstripNoWiki( $text );
 654         }
 655
 656         /**
 657          * @deprecated use $this->mStripState->unstripBoth()
 658          */
 659         function unstripForHTML( $text ) {
 660                 return $this->mStripState->unstripBoth( $text );
 661         }
 662
 663         /**
 664          * Add an item to the strip state
 665          * Returns the unique tag which must be inserted into the stripped text
 666          * The tag will be replaced with the original text in unstrip()
 667          *
 668          * @private
 669          */
 670         function insertStripItem( $text ) {
 671                 $rnd = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}-" . self::MARKER_SUFFIX;
 672                 $this->mMarkerIndex++;
 673                 $this->mStripState->general->setPair( $rnd, $text );
 674                 return $rnd;
 675         }
 676
 677         /**
 678          * Interface with html tidy
 679          * @deprecated Use MWTidy::tidy()
 680          */
 681         public static function tidy( $text ) {
 682                 wfDeprecated( __METHOD__ );
 683                 return MWTidy::tidy( $text );
 684         }
 685
 686         /**
 687          * parse the wiki syntax used to render tables
 688          *
 689          * @private
 690          */
 691         function doTableStuff ( $text ) {
 692                 wfProfileIn( __METHOD__ );
 693
 694                 $lines = StringUtils::explode( "\n", $text );
 695                 $out = '';
 696                 $td_history = array (); // Is currently a td tag open?
 697                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 698                 $tr_history = array (); // Is currently a tr tag open?
 699                 $tr_attributes = array (); // history of tr attributes
 700                 $has_opened_tr = array(); // Did this table open a <tr> element?
 701                 $indent_level = 0; // indent level of the table
 702
 703                 foreach ( $lines as $outLine ) {
 704                         $line = trim( $outLine );
 705
 706                         if( $line == '' ) { // empty line, go to next line
 707                                 $out .= $outLine."\n";
 708                                 continue;
 709                         }
 710                         $first_character = $line[0];
 711                         $matches = array();
 712
 713                         if ( preg_match( '/^(:*)\{\|(.*)$/', $line , $matches ) ) {
 714                                 // First check if we are starting a new table
 715                                 $indent_level = strlen( $matches[1] );
 716
 717                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 718                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 719
 720                                 $outLine = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 721                                 array_push( $td_history , false );
 722                                 array_push( $last_tag_history , '' );
 723                                 array_push( $tr_history , false );
 724                                 array_push( $tr_attributes , '' );
 725                                 array_push( $has_opened_tr , false );
 726                         } else if ( count ( $td_history ) == 0 ) {
 727                                 // Don't do any of the following
 728                                 $out .= $outLine."\n";
 729                                 continue;
 730                         } else if ( substr ( $line , 0 , 2 ) === '|}' ) {
 731                                 // We are ending a table
 732                                 $line = '</table>' . substr ( $line , 2 );
 733                                 $last_tag = array_pop ( $last_tag_history );
 734
 735                                 if ( !array_pop ( $has_opened_tr ) ) {
 736                                         $line = "<tr><td></td></tr>{$line}";
 737                                 }
 738
 739                                 if ( array_pop ( $tr_history ) ) {
 740                                         $line = "</tr>{$line}";
 741                                 }
 742
 743                                 if ( array_pop ( $td_history ) ) {
 744                                         $line = "</{$last_tag}>{$line}";
 745                                 }
 746                                 array_pop ( $tr_attributes );
 747                                 $outLine = $line . str_repeat( '</dd></dl>' , $indent_level );
 748                         } else if ( substr ( $line , 0 , 2 ) === '|-' ) {
 749                                 // Now we have a table row
 750                                 $line = preg_replace( '#^\|-+#', '', $line );
 751
 752                                 // Whats after the tag is now only attributes
 753                                 $attributes = $this->mStripState->unstripBoth( $line );
 754                                 $attributes = Sanitizer::fixTagAttributes( $attributes, 'tr' );
 755                                 array_pop( $tr_attributes );
 756                                 array_push( $tr_attributes, $attributes );
 757
 758                                 $line = '';
 759                                 $last_tag = array_pop ( $last_tag_history );
 760                                 array_pop ( $has_opened_tr );
 761                                 array_push ( $has_opened_tr , true );
 762
 763                                 if ( array_pop ( $tr_history ) ) {
 764                                         $line = '</tr>';
 765                                 }
 766
 767                                 if ( array_pop ( $td_history ) ) {
 768                                         $line = "</{$last_tag}>{$line}";
 769                                 }
 770
 771                                 $outLine = $line;
 772                                 array_push ( $tr_history , false );
 773                                 array_push ( $td_history , false );
 774                                 array_push ( $last_tag_history , '' );
 775                         }
 776                         else if ( $first_character === '|' || $first_character === '!' || substr ( $line , 0 , 2 )  === '|+' ) {
 777                                 // This might be cell elements, td, th or captions
 778                                 if ( substr ( $line , 0 , 2 ) === '|+' ) {
 779                                         $first_character = '+';
 780                                         $line = substr ( $line , 1 );
 781                                 }
 782
 783                                 $line = substr ( $line , 1 );
 784
 785                                 if ( $first_character === '!' ) {
 786                                         $line = str_replace ( '!!' , '||' , $line );
 787                                 }
 788
 789                                 // Split up multiple cells on the same line.
 790                                 // FIXME : This can result in improper nesting of tags processed
 791                                 // by earlier parser steps, but should avoid splitting up eg
 792                                 // attribute values containing literal "||".
 793                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 794
 795                                 $outLine = '';
 796
 797                                 // Loop through each table cell
 798                                 foreach ( $cells as $cell )
 799                                 {
 800                                         $previous = '';
 801                                         if ( $first_character !== '+' )
 802                                         {
 803                                                 $tr_after = array_pop ( $tr_attributes );
 804                                                 if ( !array_pop ( $tr_history ) ) {
 805                                                         $previous = "<tr{$tr_after}>\n";
 806                                                 }
 807                                                 array_push ( $tr_history , true );
 808                                                 array_push ( $tr_attributes , '' );
 809                                                 array_pop ( $has_opened_tr );
 810                                                 array_push ( $has_opened_tr , true );
 811                                         }
 812
 813                                         $last_tag = array_pop ( $last_tag_history );
 814
 815                                         if ( array_pop ( $td_history ) ) {
 816                                                 $previous = "</{$last_tag}>{$previous}";
 817                                         }
 818
 819                                         if ( $first_character === '|' ) {
 820                                                 $last_tag = 'td';
 821                                         } else if ( $first_character === '!' ) {
 822                                                 $last_tag = 'th';
 823                                         } else if ( $first_character === '+' ) {
 824                                                 $last_tag = 'caption';
 825                                         } else {
 826                                                 $last_tag = '';
 827                                         }
 828
 829                                         array_push ( $last_tag_history , $last_tag );
 830
 831                                         // A cell could contain both parameters and data
 832                                         $cell_data = explode ( '|' , $cell , 2 );
 833
 834                                         // Bug 553: Note that a '|' inside an invalid link should not
 835                                         // be mistaken as delimiting cell parameters
 836                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 837                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 838                                         } else if ( count ( $cell_data ) == 1 )
 839                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 840                                         else {
 841                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 842                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 843                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 844                                         }
 845
 846                                         $outLine .= $cell;
 847                                         array_push ( $td_history , true );
 848                                 }
 849                         }
 850                         $out .= $outLine . "\n";
 851                 }
 852
 853                 // Closing open td, tr && table
 854                 while ( count ( $td_history ) > 0 )
 855                 {
 856                         if ( array_pop ( $td_history ) ) {
 857                                 $out .= "</td>\n";
 858                         }
 859                         if ( array_pop ( $tr_history ) ) {
 860                                 $out .= "</tr>\n";
 861                         }
 862                         if ( !array_pop ( $has_opened_tr ) ) {
 863                                 $out .= "<tr><td></td></tr>\n" ;
 864                         }
 865
 866                         $out .= "</table>\n";
 867                 }
 868
 869                 // Remove trailing line-ending (b/c)
 870                 if ( substr( $out, -1 ) === "\n" ) {
 871                         $out = substr( $out, 0, -1 );
 872                 }
 873
 874                 // special case: don't return empty table
 875                 if( $out === "<table>\n<tr><td></td></tr>\n</table>" ) {
 876                         $out = '';
 877                 }
 878
 879                 wfProfileOut( __METHOD__ );
 880
 881                 return $out;
 882         }
 883
 884         /**
 885          * Helper function for parse() that transforms wiki markup into
 886          * HTML. Only called for $mOutputType == self::OT_HTML.
 887          *
 888          * @private
 889          */
 890         function internalParse( $text, $isMain = true, $frame=false ) {
 891                 wfProfileIn( __METHOD__ );
 892
 893                 $origText = $text;
 894
 895                 # Hook to suspend the parser in this state
 896                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 897                         wfProfileOut( __METHOD__ );
 898                         return $text ;
 899                 }
 900
 901                 // if $frame is provided, then use $frame for replacing any variables
 902                 if ($frame) {
 903                         // use frame depth to infer how include/noinclude tags should be handled
 904                         // depth=0 means this is the top-level document; otherwise it's an included document
 905                         if( !$frame->depth )
 906                                 $flag = 0;
 907                         else
 908                                 $flag = Parser::PTD_FOR_INCLUSION;
 909                         $dom = $this->preprocessToDom( $text, $flag );
 910                         $text = $frame->expand( $dom );
 911                 }
 912                 // if $frame is not provided, then use old-style replaceVariables
 913                 else {
 914                         $text = $this->replaceVariables( $text );
 915                 }
 916
 917                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ), false, array_keys( $this->mTransparentTagHooks ) );
 918                 wfRunHooks( 'InternalParseBeforeLinks', array( &$this, &$text, &$this->mStripState ) );
 919
 920                 // Tables need to come after variable replacement for things to work
 921                 // properly; putting them before other transformations should keep
 922                 // exciting things like link expansions from showing up in surprising
 923                 // places.
 924                 $text = $this->doTableStuff( $text );
 925
 926                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 927
 928                 $text = $this->doDoubleUnderscore( $text );
 929
 930                 $text = $this->doHeadings( $text );
 931                 if( $this->mOptions->getUseDynamicDates() ) {
 932                         $df = DateFormatter::getInstance();
 933                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 934                 }
 935                 $text = $this->doAllQuotes( $text );
 936                 $text = $this->replaceInternalLinks( $text );
 937                 $text = $this->replaceExternalLinks( $text );
 938
 939                 # replaceInternalLinks may sometimes leave behind
 940                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 941                 $text = str_replace($this->mUniqPrefix.'NOPARSE', '', $text);
 942
 943                 $text = $this->doMagicLinks( $text );
 944                 $text = $this->formatHeadings( $text, $origText, $isMain );
 945
 946                 wfProfileOut( __METHOD__ );
 947                 return $text;
 948         }
 949
 950         /**
 951          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 952          * magic external links.
 953          *
 954          * DML
 955          * @private
 956          */
 957         function doMagicLinks( $text ) {
 958                 wfProfileIn( __METHOD__ );
 959                 $prots = $this->mUrlProtocols;
 960                 $urlChar = self::EXT_LINK_URL_CLASS;
 961                 $text = preg_replace_callback(
 962                         '!(?:                           # Start cases
 963                                 (<a.*?</a>) |               # m[1]: Skip link text
 964                                 (<.*?>) |                   # m[2]: Skip stuff inside HTML elements' . "
 965                                 (\\b(?:$prots)$urlChar+) |  # m[3]: Free external links" . '
 966                                 (?:RFC|PMID)\s+([0-9]+) |   # m[4]: RFC or PMID, capture number
 967                                 ISBN\s+(\b                  # m[5]: ISBN, capture number
 968                                     (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
 969                                     (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
 970                                     [0-9Xx]                 # check digit
 971                                     \b)
 972                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 973                 wfProfileOut( __METHOD__ );
 974                 return $text;
 975         }
 976
 977         function magicLinkCallback( $m ) {
 978                 if ( isset( $m[1] ) && $m[1] !== '' ) {
 979                         # Skip anchor
 980                         return $m[0];
 981                 } elseif ( isset( $m[2] ) && $m[2] !== '' ) {
 982                         # Skip HTML element
 983                         return $m[0];
 984                 } elseif ( isset( $m[3] ) && $m[3] !== '' ) {
 985                         # Free external link
 986                         return $this->makeFreeExternalLink( $m[0] );
 987                 } elseif ( isset( $m[4] ) && $m[4] !== '' ) {
 988                         # RFC or PMID
 989                         $CssClass = '';
 990                         if ( substr( $m[0], 0, 3 ) === 'RFC' ) {
 991                                 $keyword = 'RFC';
 992                                 $urlmsg = 'rfcurl';
 993                                 $CssClass = 'mw-magiclink-rfc';
 994                                 $id = $m[4];
 995                         } elseif ( substr( $m[0], 0, 4 ) === 'PMID' ) {
 996                                 $keyword = 'PMID';
 997                                 $urlmsg = 'pubmedurl';
 998                                 $CssClass = 'mw-magiclink-pmid';
 999                                 $id = $m[4];
1000                         } else {
1001                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1002                                         substr($m[0], 0, 20 ) . '"' );
1003                         }
1004                         $url = wfMsg( $urlmsg, $id);
1005                         $sk = $this->mOptions->getSkin();
1006                         $la = $sk->getExternalLinkAttributes( "external $CssClass" );
1007                         return "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1008                 } elseif ( isset( $m[5] ) && $m[5] !== '' ) {
1009                         # ISBN
1010                         $isbn = $m[5];
1011                         $num = strtr( $isbn, array(
1012                                 '-' => '',
1013                                 ' ' => '',
1014                                 'x' => 'X',
1015                         ));
1016                         $titleObj = SpecialPage::getTitleFor( 'Booksources', $num );
1017                         return'<a href="' .
1018                                 $titleObj->escapeLocalUrl() .
1019                                 "\" class=\"internal mw-magiclink-isbn\">ISBN $isbn</a>";
1020                 } else {
1021                         return $m[0];
1022                 }
1023         }
1024
1025         /**
1026          * Make a free external link, given a user-supplied URL
1027          * @return HTML
1028          * @private
1029          */
1030         function makeFreeExternalLink( $url ) {
1031                 global $wgContLang;
1032                 wfProfileIn( __METHOD__ );
1033
1034                 $sk = $this->mOptions->getSkin();
1035                 $trail = '';
1036
1037                 # The characters '<' and '>' (which were escaped by
1038                 # removeHTMLtags()) should not be included in
1039                 # URLs, per RFC 2396.
1040                 $m2 = array();
1041                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1042                         $trail = substr($url, $m2[0][1]) . $trail;
1043                         $url = substr($url, 0, $m2[0][1]);
1044                 }
1045
1046                 # Move trailing punctuation to $trail
1047                 $sep = ',;\.:!?';
1048                 # If there is no left bracket, then consider right brackets fair game too
1049                 if ( strpos( $url, '(' ) === false ) {
1050                         $sep .= ')';
1051                 }
1052
1053                 $numSepChars = strspn( strrev( $url ), $sep );
1054                 if ( $numSepChars ) {
1055                         $trail = substr( $url, -$numSepChars ) . $trail;
1056                         $url = substr( $url, 0, -$numSepChars );
1057                 }
1058
1059                 $url = Sanitizer::cleanUrl( $url );
1060
1061                 # Is this an external image?
1062                 $text = $this->maybeMakeExternalImage( $url );
1063                 if ( $text === false ) {
1064                         # Not an image, make a link
1065                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free',
1066                                 $this->getExternalLinkAttribs( $url ) );
1067                         # Register it in the output object...
1068                         # Replace unnecessary URL escape codes with their equivalent characters
1069                         $pasteurized = self::replaceUnusualEscapes( $url );
1070                         $this->mOutput->addExternalLink( $pasteurized );
1071                 }
1072                 wfProfileOut( __METHOD__ );
1073                 return $text . $trail;
1074         }
1075
1076
1077         /**
1078          * Parse headers and return html
1079          *
1080          * @private
1081          */
1082         function doHeadings( $text ) {
1083                 wfProfileIn( __METHOD__ );
1084                 for ( $i = 6; $i >= 1; --$i ) {
1085                         $h = str_repeat( '=', $i );
1086                         $text = preg_replace( "/^$h(.+)$h\\s*$/m",
1087                           "<h$i>\\1</h$i>", $text );
1088                 }
1089                 wfProfileOut( __METHOD__ );
1090                 return $text;
1091         }
1092
1093         /**
1094          * Replace single quotes with HTML markup
1095          * @private
1096          * @return string the altered text
1097          */
1098         function doAllQuotes( $text ) {
1099                 wfProfileIn( __METHOD__ );
1100                 $outtext = '';
1101                 $lines = StringUtils::explode( "\n", $text );
1102                 foreach ( $lines as $line ) {
1103                         $outtext .= $this->doQuotes( $line ) . "\n";
1104                 }
1105                 $outtext = substr($outtext, 0,-1);
1106                 wfProfileOut( __METHOD__ );
1107                 return $outtext;
1108         }
1109
1110         /**
1111          * Helper function for doAllQuotes()
1112          */
1113         public function doQuotes( $text ) {
1114                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1115                 if ( count( $arr ) == 1 )
1116                         return $text;
1117                 else
1118                 {
1119                         # First, do some preliminary work. This may shift some apostrophes from
1120                         # being mark-up to being text. It also counts the number of occurrences
1121                         # of bold and italics mark-ups.
1122                         $i = 0;
1123                         $numbold = 0;
1124                         $numitalics = 0;
1125                         foreach ( $arr as $r )
1126                         {
1127                                 if ( ( $i % 2 ) == 1 )
1128                                 {
1129                                         # If there are ever four apostrophes, assume the first is supposed to
1130                                         # be text, and the remaining three constitute mark-up for bold text.
1131                                         if ( strlen( $arr[$i] ) == 4 )
1132                                         {
1133                                                 $arr[$i-1] .= "'";
1134                                                 $arr[$i] = "'''";
1135                                         }
1136                                         # If there are more than 5 apostrophes in a row, assume they're all
1137                                         # text except for the last 5.
1138                                         else if ( strlen( $arr[$i] ) > 5 )
1139                                         {
1140                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1141                                                 $arr[$i] = "'''''";
1142                                         }
1143                                         # Count the number of occurrences of bold and italics mark-ups.
1144                                         # We are not counting sequences of five apostrophes.
1145                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1146                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1147                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1148                                 }
1149                                 $i++;
1150                         }
1151
1152                         # If there is an odd number of both bold and italics, it is likely
1153                         # that one of the bold ones was meant to be an apostrophe followed
1154                         # by italics. Which one we cannot know for certain, but it is more
1155                         # likely to be one that has a single-letter word before it.
1156                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1157                         {
1158                                 $i = 0;
1159                                 $firstsingleletterword = -1;
1160                                 $firstmultiletterword = -1;
1161                                 $firstspace = -1;
1162                                 foreach ( $arr as $r )
1163                                 {
1164                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1165                                         {
1166                                                 $x1 = substr ($arr[$i-1], -1);
1167                                                 $x2 = substr ($arr[$i-1], -2, 1);
1168                                                 if ($x1 === ' ') {
1169                                                         if ($firstspace == -1) $firstspace = $i;
1170                                                 } else if ($x2 === ' ') {
1171                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1172                                                 } else {
1173                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1174                                                 }
1175                                         }
1176                                         $i++;
1177                                 }
1178
1179                                 # If there is a single-letter word, use it!
1180                                 if ($firstsingleletterword > -1)
1181                                 {
1182                                         $arr [ $firstsingleletterword ] = "''";
1183                                         $arr [ $firstsingleletterword-1 ] .= "'";
1184                                 }
1185                                 # If not, but there's a multi-letter word, use that one.
1186                                 else if ($firstmultiletterword > -1)
1187                                 {
1188                                         $arr [ $firstmultiletterword ] = "''";
1189                                         $arr [ $firstmultiletterword-1 ] .= "'";
1190                                 }
1191                                 # ... otherwise use the first one that has neither.
1192                                 # (notice that it is possible for all three to be -1 if, for example,
1193                                 # there is only one pentuple-apostrophe in the line)
1194                                 else if ($firstspace > -1)
1195                                 {
1196                                         $arr [ $firstspace ] = "''";
1197                                         $arr [ $firstspace-1 ] .= "'";
1198                                 }
1199                         }
1200
1201                         # Now let's actually convert our apostrophic mush to HTML!
1202                         $output = '';
1203                         $buffer = '';
1204                         $state = '';
1205                         $i = 0;
1206                         foreach ($arr as $r)
1207                         {
1208                                 if (($i % 2) == 0)
1209                                 {
1210                                         if ($state === 'both')
1211                                                 $buffer .= $r;
1212                                         else
1213                                                 $output .= $r;
1214                                 }
1215                                 else
1216                                 {
1217                                         if (strlen ($r) == 2)
1218                                         {
1219                                                 if ($state === 'i')
1220                                                 { $output .= '</i>'; $state = ''; }
1221                                                 else if ($state === 'bi')
1222                                                 { $output .= '</i>'; $state = 'b'; }
1223                                                 else if ($state === 'ib')
1224                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1225                                                 else if ($state === 'both')
1226                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1227                                                 else # $state can be 'b' or ''
1228                                                 { $output .= '<i>'; $state .= 'i'; }
1229                                         }
1230                                         else if (strlen ($r) == 3)
1231                                         {
1232                                                 if ($state === 'b')
1233                                                 { $output .= '</b>'; $state = ''; }
1234                                                 else if ($state === 'bi')
1235                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1236                                                 else if ($state === 'ib')
1237                                                 { $output .= '</b>'; $state = 'i'; }
1238                                                 else if ($state === 'both')
1239                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1240                                                 else # $state can be 'i' or ''
1241                                                 { $output .= '<b>'; $state .= 'b'; }
1242                                         }
1243                                         else if (strlen ($r) == 5)
1244                                         {
1245                                                 if ($state === 'b')
1246                                                 { $output .= '</b><i>'; $state = 'i'; }
1247                                                 else if ($state === 'i')
1248                                                 { $output .= '</i><b>'; $state = 'b'; }
1249                                                 else if ($state === 'bi')
1250                                                 { $output .= '</i></b>'; $state = ''; }
1251                                                 else if ($state === 'ib')
1252                                                 { $output .= '</b></i>'; $state = ''; }
1253                                                 else if ($state === 'both')
1254                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1255                                                 else # ($state == '')
1256                                                 { $buffer = ''; $state = 'both'; }
1257                                         }
1258                                 }
1259                                 $i++;
1260                         }
1261                         # Now close all remaining tags.  Notice that the order is important.
1262                         if ($state === 'b' || $state === 'ib')
1263                                 $output .= '</b>';
1264                         if ($state === 'i' || $state === 'bi' || $state === 'ib')
1265                                 $output .= '</i>';
1266                         if ($state === 'bi')
1267                                 $output .= '</b>';
1268                         # There might be lonely ''''', so make sure we have a buffer
1269                         if ($state === 'both' && $buffer)
1270                                 $output .= '<b><i>'.$buffer.'</i></b>';
1271                         return $output;
1272                 }
1273         }
1274
1275         /**
1276          * Replace external links (REL)
1277          *
1278          * Note: this is all very hackish and the order of execution matters a lot.
1279          * Make sure to run maintenance/parserTests.php if you change this code.
1280          *
1281          * @private
1282          */
1283         function replaceExternalLinks( $text ) {
1284                 global $wgContLang;
1285                 wfProfileIn( __METHOD__ );
1286
1287                 $sk = $this->mOptions->getSkin();
1288
1289                 $bits = preg_split( $this->mExtLinkBracketedRegex, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1290                 $s = array_shift( $bits );
1291
1292                 $i = 0;
1293                 while ( $i<count( $bits ) ) {
1294                         $url = $bits[$i++];
1295                         $protocol = $bits[$i++];
1296                         $text = $bits[$i++];
1297                         $trail = $bits[$i++];
1298
1299                         # The characters '<' and '>' (which were escaped by
1300                         # removeHTMLtags()) should not be included in
1301                         # URLs, per RFC 2396.
1302                         $m2 = array();
1303                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1304                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1305                                 $url = substr($url, 0, $m2[0][1]);
1306                         }
1307
1308                         # If the link text is an image URL, replace it with an <img> tag
1309                         # This happened by accident in the original parser, but some people used it extensively
1310                         $img = $this->maybeMakeExternalImage( $text );
1311                         if ( $img !== false ) {
1312                                 $text = $img;
1313                         }
1314
1315                         $dtrail = '';
1316
1317                         # Set linktype for CSS - if URL==text, link is essentially free
1318                         $linktype = ($text === $url) ? 'free' : 'text';
1319
1320                         # No link text, e.g. [http://domain.tld/some.link]
1321                         if ( $text == '' ) {
1322                                 # Autonumber if allowed. See bug #5918
1323                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1324                                         $langObj = $this->getFunctionLang();
1325                                         $text = '[' . $langObj->formatNum( ++$this->mAutonumber ) . ']';
1326                                         $linktype = 'autonumber';
1327                                 } else {
1328                                         # Otherwise just use the URL
1329                                         $text = htmlspecialchars( $url );
1330                                         $linktype = 'free';
1331                                 }
1332                         } else {
1333                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1334                                 # Check for trail
1335                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1336                         }
1337
1338                         $text = $wgContLang->markNoConversion($text);
1339
1340                         $url = Sanitizer::cleanUrl( $url );
1341
1342                         # Use the encoded URL
1343                         # This means that users can paste URLs directly into the text
1344                         # Funny characters like &ouml; aren't valid in URLs anyway
1345                         # This was changed in August 2004
1346                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype,
1347                                 $this->getExternalLinkAttribs( $url ) ) . $dtrail . $trail;
1348
1349                         # Register link in the output object.
1350                         # Replace unnecessary URL escape codes with the referenced character
1351                         # This prevents spammers from hiding links from the filters
1352                         $pasteurized = self::replaceUnusualEscapes( $url );
1353                         $this->mOutput->addExternalLink( $pasteurized );
1354                 }
1355
1356                 wfProfileOut( __METHOD__ );
1357                 return $s;
1358         }
1359
1360         /**
1361          * Get an associative array of additional HTML attributes appropriate for a
1362          * particular external link.  This currently may include rel => nofollow
1363          * (depending on configuration, namespace, and the URL's domain) and/or a
1364          * target attribute (depending on configuration).
1365          *
1366          * @param string $url Optional URL, to extract the domain from for rel =>
1367          *   nofollow if appropriate
1368          * @return array Associative array of HTML attributes
1369          */
1370         function getExternalLinkAttribs( $url = false ) {
1371                 $attribs = array();
1372                 global $wgNoFollowLinks, $wgNoFollowNsExceptions;
1373                 $ns = $this->mTitle->getNamespace();
1374                 if( $wgNoFollowLinks && !in_array($ns, $wgNoFollowNsExceptions) ) {
1375                         $attribs['rel'] = 'nofollow';
1376
1377                         global $wgNoFollowDomainExceptions;
1378                         if ( $wgNoFollowDomainExceptions ) {
1379                                 $bits = wfParseUrl( $url );
1380                                 if ( is_array( $bits ) && isset( $bits['host'] ) ) {
1381                                         foreach ( $wgNoFollowDomainExceptions as $domain ) {
1382                                                 if( substr( $bits['host'], -strlen( $domain ) )
1383                                                 == $domain ) {
1384                                                         unset( $attribs['rel'] );
1385                                                         break;
1386                                                 }
1387                                         }
1388                                 }
1389                         }
1390                 }
1391                 if ( $this->mOptions->getExternalLinkTarget() ) {
1392                         $attribs['target'] = $this->mOptions->getExternalLinkTarget();
1393                 }
1394                 return $attribs;
1395         }
1396
1397
1398         /**
1399          * Replace unusual URL escape codes with their equivalent characters
1400          * @param string
1401          * @return string
1402          * @static
1403          * @todo  This can merge genuinely required bits in the path or query string,
1404          *        breaking legit URLs. A proper fix would treat the various parts of
1405          *        the URL differently; as a workaround, just use the output for
1406          *        statistical records, not for actual linking/output.
1407          */
1408         static function replaceUnusualEscapes( $url ) {
1409                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1410                         array( __CLASS__, 'replaceUnusualEscapesCallback' ), $url );
1411         }
1412
1413         /**
1414          * Callback function used in replaceUnusualEscapes().
1415          * Replaces unusual URL escape codes with their equivalent character
1416          * @static
1417          * @private
1418          */
1419         private static function replaceUnusualEscapesCallback( $matches ) {
1420                 $char = urldecode( $matches[0] );
1421                 $ord = ord( $char );
1422                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1423                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1424                         // No, shouldn't be escaped
1425                         return $char;
1426                 } else {
1427                         // Yes, leave it escaped
1428                         return $matches[0];
1429                 }
1430         }
1431
1432         /**
1433          * make an image if it's allowed, either through the global
1434          * option, through the exception, or through the on-wiki whitelist
1435          * @private
1436          */
1437         function maybeMakeExternalImage( $url ) {
1438                 $sk = $this->mOptions->getSkin();
1439                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1440                 $imagesexception = !empty($imagesfrom);
1441                 $text = false;
1442                 # $imagesfrom could be either a single string or an array of strings, parse out the latter
1443                 if( $imagesexception && is_array( $imagesfrom ) ) {
1444                         $imagematch = false;
1445                         foreach( $imagesfrom as $match ) {
1446                                 if( strpos( $url, $match ) === 0 ) {
1447                                         $imagematch = true;
1448                                         break;
1449                                 }
1450                         }
1451                 } elseif( $imagesexception ) {
1452                         $imagematch = (strpos( $url, $imagesfrom ) === 0);
1453                 } else {
1454                         $imagematch = false;
1455                 }
1456                 if ( $this->mOptions->getAllowExternalImages()
1457                      || ( $imagesexception && $imagematch ) ) {
1458                         if ( preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1459                                 # Image found
1460                                 $text = $sk->makeExternalImage( $url );
1461                         }
1462                 }
1463                 if( !$text && $this->mOptions->getEnableImageWhitelist()
1464                          && preg_match( self::EXT_IMAGE_REGEX, $url ) ) {
1465                         $whitelist = explode( "\n", wfMsgForContent( 'external_image_whitelist' ) );
1466                         foreach( $whitelist as $entry ) {
1467                                 # Sanitize the regex fragment, make it case-insensitive, ignore blank entries/comments
1468                                 if( strpos( $entry, '#' ) === 0 || $entry === '' )
1469                                         continue;
1470                                 if( preg_match( '/' . str_replace( '/', '\\/', $entry ) . '/i', $url ) ) {
1471                                         # Image matches a whitelist entry
1472                                         $text = $sk->makeExternalImage( $url );
1473                                         break;
1474                                 }
1475                         }
1476                 }
1477                 return $text;
1478         }
1479
1480         /**
1481          * Process [[ ]] wikilinks
1482          * @return processed text
1483          *
1484          * @private
1485          */
1486         function replaceInternalLinks( $s ) {
1487                 $this->mLinkHolders->merge( $this->replaceInternalLinks2( $s ) );
1488                 return $s;
1489         }
1490
1491         /**
1492          * Process [[ ]] wikilinks (RIL)
1493          * @return LinkHolderArray
1494          *
1495          * @private
1496          */
1497         function replaceInternalLinks2( &$s ) {
1498                 global $wgContLang;
1499
1500                 wfProfileIn( __METHOD__ );
1501
1502                 wfProfileIn( __METHOD__.'-setup' );
1503                 static $tc = FALSE, $e1, $e1_img;
1504                 # the % is needed to support urlencoded titles as well
1505                 if ( !$tc ) {
1506                         $tc = Title::legalChars() . '#%';
1507                         # Match a link having the form [[namespace:link|alternate]]trail
1508                         $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD";
1509                         # Match cases where there is no "]]", which might still be images
1510                         $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD";
1511                 }
1512
1513                 $sk = $this->mOptions->getSkin();
1514                 $holders = new LinkHolderArray( $this );
1515
1516                 #split the entire text string on occurences of [[
1517                 $a = StringUtils::explode( '[[', ' ' . $s );
1518                 #get the first element (all text up to first [[), and remove the space we added
1519                 $s = $a->current();
1520                 $a->next();
1521                 $line = $a->current(); # Workaround for broken ArrayIterator::next() that returns "void"
1522                 $s = substr( $s, 1 );
1523
1524                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1525                 $e2 = null;
1526                 if ( $useLinkPrefixExtension ) {
1527                         # Match the end of a line for a word that's not followed by whitespace,
1528                         # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1529                         $e2 = wfMsgForContent( 'linkprefix' );
1530                 }
1531
1532                 if( is_null( $this->mTitle ) ) {
1533                         wfProfileOut( __METHOD__.'-setup' );
1534                         wfProfileOut( __METHOD__ );
1535                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1536                 }
1537                 $nottalk = !$this->mTitle->isTalkPage();
1538
1539                 if ( $useLinkPrefixExtension ) {
1540                         $m = array();
1541                         if ( preg_match( $e2, $s, $m ) ) {
1542                                 $first_prefix = $m[2];
1543                         } else {
1544                                 $first_prefix = false;
1545                         }
1546                 } else {
1547                         $prefix = '';
1548                 }
1549
1550                 if($wgContLang->hasVariants()) {
1551                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1552                 } else {
1553                         $selflink = array($this->mTitle->getPrefixedText());
1554                 }
1555                 $useSubpages = $this->areSubpagesAllowed();
1556                 wfProfileOut( __METHOD__.'-setup' );
1557
1558                 # Loop for each link
1559                 for ( ; $line !== false && $line !== null ; $a->next(), $line = $a->current() ) {
1560                         # Check for excessive memory usage
1561                         if ( $holders->isBig() ) {
1562                                 # Too big
1563                                 # Do the existence check, replace the link holders and clear the array
1564                                 $holders->replace( $s );
1565                                 $holders->clear();
1566                         }
1567
1568                         if ( $useLinkPrefixExtension ) {
1569                                 wfProfileIn( __METHOD__.'-prefixhandling' );
1570                                 if ( preg_match( $e2, $s, $m ) ) {
1571                                         $prefix = $m[2];
1572                                         $s = $m[1];
1573                                 } else {
1574                                         $prefix='';
1575                                 }
1576                                 # first link
1577                                 if($first_prefix) {
1578                                         $prefix = $first_prefix;
1579                                         $first_prefix = false;
1580                                 }
1581                                 wfProfileOut( __METHOD__.'-prefixhandling' );
1582                         }
1583
1584                         $might_be_img = false;
1585
1586                         wfProfileIn( __METHOD__."-e1" );
1587                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1588                                 $text = $m[2];
1589                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1590                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1591                                 # the real problem is with the $e1 regex
1592                                 # See bug 1300.
1593                                 #
1594                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1595                                 # and no image is in sight. See bug 2095.
1596                                 #
1597                                 if( $text !== '' &&
1598                                         substr( $m[3], 0, 1 ) === ']' &&
1599                                         strpos($text, '[') !== false
1600                                 )
1601                                 {
1602                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1603                                         $m[3] = substr( $m[3], 1 );
1604                                 }
1605                                 # fix up urlencoded title texts
1606                                 if( strpos( $m[1], '%' ) !== false ) {
1607                                         # Should anchors '#' also be rejected?
1608                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1609                                 }
1610                                 $trail = $m[3];
1611                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1612                                 $might_be_img = true;
1613                                 $text = $m[2];
1614                                 if ( strpos( $m[1], '%' ) !== false ) {
1615                                         $m[1] = urldecode($m[1]);
1616                                 }
1617                                 $trail = "";
1618                         } else { # Invalid form; output directly
1619                                 $s .= $prefix . '[[' . $line ;
1620                                 wfProfileOut( __METHOD__."-e1" );
1621                                 continue;
1622                         }
1623                         wfProfileOut( __METHOD__."-e1" );
1624                         wfProfileIn( __METHOD__."-misc" );
1625
1626                         # Don't allow internal links to pages containing
1627                         # PROTO: where PROTO is a valid URL protocol; these
1628                         # should be external links.
1629                         if ( preg_match( '/^\b(?:' . wfUrlProtocols() . ')/', $m[1] ) ) {
1630                                 $s .= $prefix . '[[' . $line ;
1631                                 wfProfileOut( __METHOD__."-misc" );
1632                                 continue;
1633                         }
1634
1635                         # Make subpage if necessary
1636                         if ( $useSubpages ) {
1637                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1638                         } else {
1639                                 $link = $m[1];
1640                         }
1641
1642                         $noforce = (substr( $m[1], 0, 1 ) !== ':');
1643                         if (!$noforce) {
1644                                 # Strip off leading ':'
1645                                 $link = substr( $link, 1 );
1646                         }
1647
1648                         wfProfileOut( __METHOD__."-misc" );
1649                         wfProfileIn( __METHOD__."-title" );
1650                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki( $link ) );
1651                         if ( $nt === null ) {
1652                                 $s .= $prefix . '[[' . $line;
1653                                 wfProfileOut( __METHOD__."-title" );
1654                                 continue;
1655                         }
1656
1657                         $ns = $nt->getNamespace();
1658                         $iw = $nt->getInterWiki();
1659                         wfProfileOut( __METHOD__."-title" );
1660
1661                         if ( $might_be_img ) { # if this is actually an invalid link
1662                                 wfProfileIn( __METHOD__."-might_be_img" );
1663                                 if ( $ns == NS_FILE && $noforce ) { #but might be an image
1664                                         $found = false;
1665                                         while ( true ) {
1666                                                 #look at the next 'line' to see if we can close it there
1667                                                 $a->next();
1668                                                 $next_line = $a->current();
1669                                                 if ( $next_line === false || $next_line === null ) {
1670                                                         break;
1671                                                 }
1672                                                 $m = explode( ']]', $next_line, 3 );
1673                                                 if ( count( $m ) == 3 ) {
1674                                                         # the first ]] closes the inner link, the second the image
1675                                                         $found = true;
1676                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1677                                                         $trail = $m[2];
1678                                                         break;
1679                                                 } elseif ( count( $m ) == 2 ) {
1680                                                         #if there's exactly one ]] that's fine, we'll keep looking
1681                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1682                                                 } else {
1683                                                         #if $next_line is invalid too, we need look no further
1684                                                         $text .= '[[' . $next_line;
1685                                                         break;
1686                                                 }
1687                                         }
1688                                         if ( !$found ) {
1689                                                 # we couldn't find the end of this imageLink, so output it raw
1690                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1691                                                 $holders->merge( $this->replaceInternalLinks2( $text ) );
1692                                                 $s .= "{$prefix}[[$link|$text";
1693                                                 # note: no $trail, because without an end, there *is* no trail
1694                                                 wfProfileOut( __METHOD__."-might_be_img" );
1695                                                 continue;
1696                                         }
1697                                 } else { #it's not an image, so output it raw
1698                                         $s .= "{$prefix}[[$link|$text";
1699                                         # note: no $trail, because without an end, there *is* no trail
1700                                         wfProfileOut( __METHOD__."-might_be_img" );
1701                                         continue;
1702                                 }
1703                                 wfProfileOut( __METHOD__."-might_be_img" );
1704                         }
1705
1706                         $wasblank = ( $text  == '' );
1707                         if ( $wasblank ) $text = $link;
1708
1709                         # Link not escaped by : , create the various objects
1710                         if ( $noforce ) {
1711
1712                                 # Interwikis
1713                                 wfProfileIn( __METHOD__."-interwiki" );
1714                                 if ( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1715                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1716                                         $s = rtrim($s . $prefix);
1717                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1718                                         wfProfileOut( __METHOD__."-interwiki" );
1719                                         continue;
1720                                 }
1721                                 wfProfileOut( __METHOD__."-interwiki" );
1722
1723                                 if ( $ns == NS_FILE ) {
1724                                         wfProfileIn( __METHOD__."-image" );
1725                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1726                                                 if ( $wasblank ) {
1727                                                         # if no parameters were passed, $text
1728                                                         # becomes something like "File:Foo.png",
1729                                                         # which we don't want to pass on to the
1730                                                         # image generator
1731                                                         $text = '';
1732                                                 } else {
1733                                                         # recursively parse links inside the image caption
1734                                                         # actually, this will parse them in any other parameters, too,
1735                                                         # but it might be hard to fix that, and it doesn't matter ATM
1736                                                         $text = $this->replaceExternalLinks($text);
1737                                                         $holders->merge( $this->replaceInternalLinks2( $text ) );
1738                                                 }
1739                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1740                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text, $holders ) ) . $trail;
1741                                         } else {
1742                                                 $s .= $prefix . $trail;
1743                                         }
1744                                         $this->mOutput->addImage( $nt->getDBkey() );
1745                                         wfProfileOut( __METHOD__."-image" );
1746                                         continue;
1747
1748                                 }
1749
1750                                 if ( $ns == NS_CATEGORY ) {
1751                                         wfProfileIn( __METHOD__."-category" );
1752                                         $s = rtrim($s . "\n"); # bug 87
1753
1754                                         if ( $wasblank ) {
1755                                                 $sortkey = $this->getDefaultSort();
1756                                         } else {
1757                                                 $sortkey = $text;
1758                                         }
1759                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1760                                         $sortkey = str_replace( "\n", '', $sortkey );
1761                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1762                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1763
1764                                         /**
1765                                          * Strip the whitespace Category links produce, see bug 87
1766                                          * @todo We might want to use trim($tmp, "\n") here.
1767                                          */
1768                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1769
1770                                         wfProfileOut( __METHOD__."-category" );
1771                                         continue;
1772                                 }
1773                         }
1774
1775                         # Self-link checking
1776                         if( $nt->getFragment() === '' && $ns != NS_SPECIAL ) {
1777                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1778                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1779                                         continue;
1780                                 }
1781                         }
1782
1783                         # NS_MEDIA is a pseudo-namespace for linking directly to a file
1784                         # FIXME: Should do batch file existence checks, see comment below
1785                         if( $ns == NS_MEDIA ) {
1786                                 wfProfileIn( __METHOD__."-media" );
1787                                 # Give extensions a chance to select the file revision for us
1788                                 $skip = $time = false;
1789                                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$nt, &$skip, &$time ) );
1790                                 if ( $skip ) {
1791                                         $link = $sk->link( $nt );
1792                                 } else {
1793                                         $link = $sk->makeMediaLinkObj( $nt, $text, $time );
1794                                 }
1795                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1796                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1797                                 $this->mOutput->addImage( $nt->getDBkey() );
1798                                 wfProfileOut( __METHOD__."-media" );
1799                                 continue;
1800                         }
1801
1802                         wfProfileIn( __METHOD__."-always_known" );
1803                         # Some titles, such as valid special pages or files in foreign repos, should
1804                         # be shown as bluelinks even though they're not included in the page table
1805                         #
1806                         # FIXME: isAlwaysKnown() can be expensive for file links; we should really do
1807                         # batch file existence checks for NS_FILE and NS_MEDIA
1808                         if( $iw == '' && $nt->isAlwaysKnown() ) {
1809                                 $this->mOutput->addLink( $nt );
1810                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1811                         } else {
1812                                 # Links will be added to the output link list after checking
1813                                 $s .= $holders->makeHolder( $nt, $text, '', $trail, $prefix );
1814                         }
1815                         wfProfileOut( __METHOD__."-always_known" );
1816                 }
1817                 wfProfileOut( __METHOD__ );
1818                 return $holders;
1819         }
1820
1821         /**
1822          * Make a link placeholder. The text returned can be later resolved to a real link with
1823          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1824          * parsing of interwiki links, and secondly to allow all existence checks and
1825          * article length checks (for stub links) to be bundled into a single query.
1826          *
1827          * @deprecated
1828          */
1829         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1830                 return $this->mLinkHolders->makeHolder( $nt, $text, $query, $trail, $prefix );
1831         }
1832
1833         /**
1834          * Render a forced-blue link inline; protect against double expansion of
1835          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1836          * Since this little disaster has to split off the trail text to avoid
1837          * breaking URLs in the following text without breaking trails on the
1838          * wiki links, it's been made into a horrible function.
1839          *
1840          * @param Title $nt
1841          * @param string $text
1842          * @param string $query
1843          * @param string $trail
1844          * @param string $prefix
1845          * @return string HTML-wikitext mix oh yuck
1846          */
1847         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1848                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1849                 $sk = $this->mOptions->getSkin();
1850                 // FIXME: use link() instead of deprecated makeKnownLinkObj()
1851                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1852                 return $this->armorLinks( $link ) . $trail;
1853         }
1854
1855         /**
1856          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1857          * going to go through further parsing steps before inline URL expansion.
1858          *
1859          * Not needed quite as much as it used to be since free links are a bit
1860          * more sensible these days. But bracketed links are still an issue.
1861          *
1862          * @param string more-or-less HTML
1863          * @return string less-or-more HTML with NOPARSE bits
1864          */
1865         function armorLinks( $text ) {
1866                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1867                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1868         }
1869
1870         /**
1871          * Return true if subpage links should be expanded on this page.
1872          * @return bool
1873          */
1874         function areSubpagesAllowed() {
1875                 # Some namespaces don't allow subpages
1876                 return MWNamespace::hasSubpages( $this->mTitle->getNamespace() );
1877         }
1878
1879         /**
1880          * Handle link to subpage if necessary
1881          * @param string $target the source of the link
1882          * @param string &$text the link text, modified as necessary
1883          * @return string the full name of the link
1884          * @private
1885          */
1886         function maybeDoSubpageLink($target, &$text) {
1887                 return Linker::normalizeSubpageLink( $this->mTitle, $target, $text );
1888         }
1889
1890         /**#@+
1891          * Used by doBlockLevels()
1892          * @private
1893          */
1894         /* private */ function closeParagraph() {
1895                 $result = '';
1896                 if ( $this->mLastSection != '' ) {
1897                         $result = '</' . $this->mLastSection  . ">\n";
1898                 }
1899                 $this->mInPre = false;
1900                 $this->mLastSection = '';
1901                 return $result;
1902         }
1903         # getCommon() returns the length of the longest common substring
1904         # of both arguments, starting at the beginning of both.
1905         #
1906         /* private */ function getCommon( $st1, $st2 ) {
1907                 $fl = strlen( $st1 );
1908                 $shorter = strlen( $st2 );
1909                 if ( $fl < $shorter ) { $shorter = $fl; }
1910
1911                 for ( $i = 0; $i < $shorter; ++$i ) {
1912                         if ( $st1{$i} != $st2{$i} ) { break; }
1913                 }
1914                 return $i;
1915         }
1916         # These next three functions open, continue, and close the list
1917         # element appropriate to the prefix character passed into them.
1918         #
1919         /* private */ function openList( $char ) {
1920                 $result = $this->closeParagraph();
1921
1922                 if ( '*' === $char ) { $result .= '<ul><li>'; }
1923                 elseif ( '#' === $char ) { $result .= '<ol><li>'; }
1924                 elseif ( ':' === $char ) { $result .= '<dl><dd>'; }
1925                 elseif ( ';' === $char ) {
1926                         $result .= '<dl><dt>';
1927                         $this->mDTopen = true;
1928                 }
1929                 else { $result = '<!-- ERR 1 -->'; }
1930
1931                 return $result;
1932         }
1933
1934         /* private */ function nextItem( $char ) {
1935                 if ( '*' === $char || '#' === $char ) { return '</li><li>'; }
1936                 elseif ( ':' === $char || ';' === $char ) {
1937                         $close = '</dd>';
1938                         if ( $this->mDTopen ) { $close = '</dt>'; }
1939                         if ( ';' === $char ) {
1940                                 $this->mDTopen = true;
1941                                 return $close . '<dt>';
1942                         } else {
1943                                 $this->mDTopen = false;
1944                                 return $close . '<dd>';
1945                         }
1946                 }
1947                 return '<!-- ERR 2 -->';
1948         }
1949
1950         /* private */ function closeList( $char ) {
1951                 if ( '*' === $char ) { $text = '</li></ul>'; }
1952                 elseif ( '#' === $char ) { $text = '</li></ol>'; }
1953                 elseif ( ':' === $char ) {
1954                         if ( $this->mDTopen ) {
1955                                 $this->mDTopen = false;
1956                                 $text = '</dt></dl>';
1957                         } else {
1958                                 $text = '</dd></dl>';
1959                         }
1960                 }
1961                 else {  return '<!-- ERR 3 -->'; }
1962                 return $text."\n";
1963         }
1964         /**#@-*/
1965
1966         /**
1967          * Make lists from lines starting with ':', '*', '#', etc. (DBL)
1968          *
1969          * @param $linestart bool whether or not this is at the start of a line.
1970          * @private
1971          * @return string the lists rendered as HTML
1972          */
1973         function doBlockLevels( $text, $linestart ) {
1974                 wfProfileIn( __METHOD__ );
1975
1976                 # Parsing through the text line by line.  The main thing
1977                 # happening here is handling of block-level elements p, pre,
1978                 # and making lists from lines starting with * # : etc.
1979                 #
1980                 $textLines = StringUtils::explode( "\n", $text );
1981
1982                 $lastPrefix = $output = '';
1983                 $this->mDTopen = $inBlockElem = false;
1984                 $prefixLength = 0;
1985                 $paragraphStack = false;
1986
1987                 foreach ( $textLines as $oLine ) {
1988                         # Fix up $linestart
1989                         if ( !$linestart ) {
1990                                 $output .= $oLine;
1991                                 $linestart = true;
1992                                 continue;
1993                         }
1994                         // * = ul
1995                         // # = ol
1996                         // ; = dt
1997                         // : = dd
1998
1999                         $lastPrefixLength = strlen( $lastPrefix );
2000                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2001                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2002                         // If not in a <pre> element, scan for and figure out what prefixes are there.
2003                         if ( !$this->mInPre ) {
2004                                 # Multiple prefixes may abut each other for nested lists.
2005                                 $prefixLength = strspn( $oLine, '*#:;' );
2006                                 $prefix = substr( $oLine, 0, $prefixLength );
2007
2008                                 # eh?
2009                                 // ; and : are both from definition-lists, so they're equivalent
2010                                 //  for the purposes of determining whether or not we need to open/close
2011                                 //  elements.
2012                                 $prefix2 = str_replace( ';', ':', $prefix );
2013                                 $t = substr( $oLine, $prefixLength );
2014                                 $this->mInPre = (bool)$preOpenMatch;
2015                         } else {
2016                                 # Don't interpret any other prefixes in preformatted text
2017                                 $prefixLength = 0;
2018                                 $prefix = $prefix2 = '';
2019                                 $t = $oLine;
2020                         }
2021
2022                         # List generation
2023                         if( $prefixLength && $lastPrefix === $prefix2 ) {
2024                                 # Same as the last item, so no need to deal with nesting or opening stuff
2025                                 $output .= $this->nextItem( substr( $prefix, -1 ) );
2026                                 $paragraphStack = false;
2027
2028                                 if ( substr( $prefix, -1 ) === ';') {
2029                                         # The one nasty exception: definition lists work like this:
2030                                         # ; title : definition text
2031                                         # So we check for : in the remainder text to split up the
2032                                         # title and definition, without b0rking links.
2033                                         $term = $t2 = '';
2034                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2035                                                 $t = $t2;
2036                                                 $output .= $term . $this->nextItem( ':' );
2037                                         }
2038                                 }
2039                         } elseif( $prefixLength || $lastPrefixLength ) {
2040                                 // We need to open or close prefixes, or both.
2041
2042                                 # Either open or close a level...
2043                                 $commonPrefixLength = $this->getCommon( $prefix, $lastPrefix );
2044                                 $paragraphStack = false;
2045
2046                                 // Close all the prefixes which aren't shared.
2047                                 while( $commonPrefixLength < $lastPrefixLength ) {
2048                                         $output .= $this->closeList( $lastPrefix[$lastPrefixLength-1] );
2049                                         --$lastPrefixLength;
2050                                 }
2051
2052                                 // Continue the current prefix if appropriate.
2053                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2054                                         $output .= $this->nextItem( $prefix[$commonPrefixLength-1] );
2055                                 }
2056
2057                                 // Open prefixes where appropriate.
2058                                 while ( $prefixLength > $commonPrefixLength ) {
2059                                         $char = substr( $prefix, $commonPrefixLength, 1 );
2060                                         $output .= $this->openList( $char );
2061
2062                                         if ( ';' === $char ) {
2063                                                 # FIXME: This is dupe of code above
2064                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2065                                                         $t = $t2;
2066                                                         $output .= $term . $this->nextItem( ':' );
2067                                                 }
2068                                         }
2069                                         ++$commonPrefixLength;
2070                                 }
2071                                 $lastPrefix = $prefix2;
2072                         }
2073
2074                         // If we have no prefixes, go to paragraph mode.
2075                         if( 0 == $prefixLength ) {
2076                                 wfProfileIn( __METHOD__."-paragraph" );
2077                                 # No prefix (not in list)--go to paragraph mode
2078                                 // XXX: use a stack for nestable elements like span, table and div
2079                                 $openmatch = preg_match('/(?:<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2080                                 $closematch = preg_match(
2081                                         '/(?:<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2082                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2083                                 if ( $openmatch or $closematch ) {
2084                                         $paragraphStack = false;
2085                                         # TODO bug 5718: paragraph closed
2086                                         $output .= $this->closeParagraph();
2087                                         if ( $preOpenMatch and !$preCloseMatch ) {
2088                                                 $this->mInPre = true;
2089                                         }
2090                                         if ( $closematch ) {
2091                                                 $inBlockElem = false;
2092                                         } else {
2093                                                 $inBlockElem = true;
2094                                         }
2095                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2096                                         if ( ' ' == substr( $t, 0, 1 ) and ( $this->mLastSection === 'pre' or trim($t) != '' ) ) {
2097                                                 // pre
2098                                                 if ($this->mLastSection !== 'pre') {
2099                                                         $paragraphStack = false;
2100                                                         $output .= $this->closeParagraph().'<pre>';
2101                                                         $this->mLastSection = 'pre';
2102                                                 }
2103                                                 $t = substr( $t, 1 );
2104                                         } else {
2105                                                 // paragraph
2106                                                 if ( trim($t) == '' ) {
2107                                                         if ( $paragraphStack ) {
2108                                                                 $output .= $paragraphStack.'<br />';
2109                                                                 $paragraphStack = false;
2110                                                                 $this->mLastSection = 'p';
2111                                                         } else {
2112                                                                 if ($this->mLastSection !== 'p' ) {
2113                                                                         $output .= $this->closeParagraph();
2114                                                                         $this->mLastSection = '';
2115                                                                         $paragraphStack = '<p>';
2116                                                                 } else {
2117                                                                         $paragraphStack = '</p><p>';
2118                                                                 }
2119                                                         }
2120                                                 } else {
2121                                                         if ( $paragraphStack ) {
2122                                                                 $output .= $paragraphStack;
2123                                                                 $paragraphStack = false;
2124                                                                 $this->mLastSection = 'p';
2125                                                         } else if ($this->mLastSection !== 'p') {
2126                                                                 $output .= $this->closeParagraph().'<p>';
2127                                                                 $this->mLastSection = 'p';
2128                                                         }
2129                                                 }
2130                                         }
2131                                 }
2132                                 wfProfileOut( __METHOD__."-paragraph" );
2133                         }
2134                         // somewhere above we forget to get out of pre block (bug 785)
2135                         if($preCloseMatch && $this->mInPre) {
2136                                 $this->mInPre = false;
2137                         }
2138                         if ($paragraphStack === false) {
2139                                 $output .= $t."\n";
2140                         }
2141                 }
2142                 while ( $prefixLength ) {
2143                         $output .= $this->closeList( $prefix2[$prefixLength-1] );
2144                         --$prefixLength;
2145                 }
2146                 if ( $this->mLastSection != '' ) {
2147                         $output .= '</' . $this->mLastSection . '>';
2148                         $this->mLastSection = '';
2149                 }
2150
2151                 wfProfileOut( __METHOD__ );
2152                 return $output;
2153         }
2154
2155         /**
2156          * Split up a string on ':', ignoring any occurences inside tags
2157          * to prevent illegal overlapping.
2158          * @param string $str the string to split
2159          * @param string &$before set to everything before the ':'
2160          * @param string &$after set to everything after the ':'
2161          * return string the position of the ':', or false if none found
2162          */
2163         function findColonNoLinks($str, &$before, &$after) {
2164                 wfProfileIn( __METHOD__ );
2165
2166                 $pos = strpos( $str, ':' );
2167                 if( $pos === false ) {
2168                         // Nothing to find!
2169                         wfProfileOut( __METHOD__ );
2170                         return false;
2171                 }
2172
2173                 $lt = strpos( $str, '<' );
2174                 if( $lt === false || $lt > $pos ) {
2175                         // Easy; no tag nesting to worry about
2176                         $before = substr( $str, 0, $pos );
2177                         $after = substr( $str, $pos+1 );
2178                         wfProfileOut( __METHOD__ );
2179                         return $pos;
2180                 }
2181
2182                 // Ugly state machine to walk through avoiding tags.
2183                 $state = self::COLON_STATE_TEXT;
2184                 $stack = 0;
2185                 $len = strlen( $str );
2186                 for( $i = 0; $i < $len; $i++ ) {
2187                         $c = $str{$i};
2188
2189                         switch( $state ) {
2190                         // (Using the number is a performance hack for common cases)
2191                         case 0: // self::COLON_STATE_TEXT:
2192                                 switch( $c ) {
2193                                 case "<":
2194                                         // Could be either a <start> tag or an </end> tag
2195                                         $state = self::COLON_STATE_TAGSTART;
2196                                         break;
2197                                 case ":":
2198                                         if( $stack == 0 ) {
2199                                                 // We found it!
2200                                                 $before = substr( $str, 0, $i );
2201                                                 $after = substr( $str, $i + 1 );
2202                                                 wfProfileOut( __METHOD__ );
2203                                                 return $i;
2204                                         }
2205                                         // Embedded in a tag; don't break it.
2206                                         break;
2207                                 default:
2208                                         // Skip ahead looking for something interesting
2209                                         $colon = strpos( $str, ':', $i );
2210                                         if( $colon === false ) {
2211                                                 // Nothing else interesting
2212                                                 wfProfileOut( __METHOD__ );
2213                                                 return false;
2214                                         }
2215                                         $lt = strpos( $str, '<', $i );
2216                                         if( $stack === 0 ) {
2217                                                 if( $lt === false || $colon < $lt ) {
2218                                                         // We found it!
2219                                                         $before = substr( $str, 0, $colon );
2220                                                         $after = substr( $str, $colon + 1 );
2221                                                         wfProfileOut( __METHOD__ );
2222                                                         return $i;
2223                                                 }
2224                                         }
2225                                         if( $lt === false ) {
2226                                                 // Nothing else interesting to find; abort!
2227                                                 // We're nested, but there's no close tags left. Abort!
2228                                                 break 2;
2229                                         }
2230                                         // Skip ahead to next tag start
2231                                         $i = $lt;
2232                                         $state = self::COLON_STATE_TAGSTART;
2233                                 }
2234                                 break;
2235                         case 1: // self::COLON_STATE_TAG:
2236                                 // In a <tag>
2237                                 switch( $c ) {
2238                                 case ">":
2239                                         $stack++;
2240                                         $state = self::COLON_STATE_TEXT;
2241                                         break;
2242                                 case "/":
2243                                         // Slash may be followed by >?
2244                                         $state = self::COLON_STATE_TAGSLASH;
2245                                         break;
2246                                 default:
2247                                         // ignore
2248                                 }
2249                                 break;
2250                         case 2: // self::COLON_STATE_TAGSTART:
2251                                 switch( $c ) {
2252                                 case "/":
2253                                         $state = self::COLON_STATE_CLOSETAG;
2254                                         break;
2255                                 case "!":
2256                                         $state = self::COLON_STATE_COMMENT;
2257                                         break;
2258                                 case ">":
2259                                         // Illegal early close? This shouldn't happen D:
2260                                         $state = self::COLON_STATE_TEXT;
2261                                         break;
2262                                 default:
2263                                         $state = self::COLON_STATE_TAG;
2264                                 }
2265                                 break;
2266                         case 3: // self::COLON_STATE_CLOSETAG:
2267                                 // In a </tag>
2268                                 if( $c === ">" ) {
2269                                         $stack--;
2270                                         if( $stack < 0 ) {
2271                                                 wfDebug( __METHOD__.": Invalid input; too many close tags\n" );
2272                                                 wfProfileOut( __METHOD__ );
2273                                                 return false;
2274                                         }
2275                                         $state = self::COLON_STATE_TEXT;
2276                                 }
2277                                 break;
2278                         case self::COLON_STATE_TAGSLASH:
2279                                 if( $c === ">" ) {
2280                                         // Yes, a self-closed tag <blah/>
2281                                         $state = self::COLON_STATE_TEXT;
2282                                 } else {
2283                                         // Probably we're jumping the gun, and this is an attribute
2284                                         $state = self::COLON_STATE_TAG;
2285                                 }
2286                                 break;
2287                         case 5: // self::COLON_STATE_COMMENT:
2288                                 if( $c === "-" ) {
2289                                         $state = self::COLON_STATE_COMMENTDASH;
2290                                 }
2291                                 break;
2292                         case self::COLON_STATE_COMMENTDASH:
2293                                 if( $c === "-" ) {
2294                                         $state = self::COLON_STATE_COMMENTDASHDASH;
2295                                 } else {
2296                                         $state = self::COLON_STATE_COMMENT;
2297                                 }
2298                                 break;
2299                         case self::COLON_STATE_COMMENTDASHDASH:
2300                                 if( $c === ">" ) {
2301                                         $state = self::COLON_STATE_TEXT;
2302                                 } else {
2303                                         $state = self::COLON_STATE_COMMENT;
2304                                 }
2305                                 break;
2306                         default:
2307                                 throw new MWException( "State machine error in " . __METHOD__ );
2308                         }
2309                 }
2310                 if( $stack > 0 ) {
2311                         wfDebug( __METHOD__.": Invalid input; not enough close tags (stack $stack, state $state)\n" );
2312                         return false;
2313                 }
2314                 wfProfileOut( __METHOD__ );
2315                 return false;
2316         }
2317
2318         /**
2319          * Return value of a magic variable (like PAGENAME)
2320          *
2321          * @private
2322          */
2323         function getVariableValue( $index, $frame=false ) {
2324                 global $wgContLang, $wgSitename, $wgServer, $wgServerName;
2325                 global $wgScriptPath, $wgStylePath;
2326
2327                 /**
2328                  * Some of these require message or data lookups and can be
2329                  * expensive to check many times.
2330                  */
2331                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$this->mVarCache ) ) ) {
2332                         if ( isset( $this->mVarCache[$index] ) ) {
2333                                 return $this->mVarCache[$index];
2334                         }
2335                 }
2336
2337                 $ts = wfTimestamp( TS_UNIX, $this->mOptions->getTimestamp() );
2338                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2339
2340                 # Use the time zone
2341                 global $wgLocaltimezone;
2342                 if ( isset( $wgLocaltimezone ) ) {
2343                         $oldtz = date_default_timezone_get();
2344                         date_default_timezone_set( $wgLocaltimezone );
2345                 }
2346
2347                 $localTimestamp = date( 'YmdHis', $ts );
2348                 $localMonth = date( 'm', $ts );
2349                 $localMonth1 = date( 'n', $ts );
2350                 $localMonthName = date( 'n', $ts );
2351                 $localDay = date( 'j', $ts );
2352                 $localDay2 = date( 'd', $ts );
2353                 $localDayOfWeek = date( 'w', $ts );
2354                 $localWeek = date( 'W', $ts );
2355                 $localYear = date( 'Y', $ts );
2356                 $localHour = date( 'H', $ts );
2357                 if ( isset( $wgLocaltimezone ) ) {
2358                         date_default_timezone_set( $oldtz );
2359                 }
2360
2361                 switch ( $index ) {
2362                         case 'currentmonth':
2363                                 $value = $wgContLang->formatNum( gmdate( 'm', $ts ) );
2364                                 break;
2365                         case 'currentmonth1':
2366                                 $value = $wgContLang->formatNum( gmdate( 'n', $ts ) );
2367                                 break;
2368                         case 'currentmonthname':
2369                                 $value = $wgContLang->getMonthName( gmdate( 'n', $ts ) );
2370                                 break;
2371                         case 'currentmonthnamegen':
2372                                 $value = $wgContLang->getMonthNameGen( gmdate( 'n', $ts ) );
2373                                 break;
2374                         case 'currentmonthabbrev':
2375                                 $value = $wgContLang->getMonthAbbreviation( gmdate( 'n', $ts ) );
2376                                 break;
2377                         case 'currentday':
2378                                 $value = $wgContLang->formatNum( gmdate( 'j', $ts ) );
2379                                 break;
2380                         case 'currentday2':
2381                                 $value = $wgContLang->formatNum( gmdate( 'd', $ts ) );
2382                                 break;
2383                         case 'localmonth':
2384                                 $value = $wgContLang->formatNum( $localMonth );
2385                                 break;
2386                         case 'localmonth1':
2387                                 $value = $wgContLang->formatNum( $localMonth1 );
2388                                 break;
2389                         case 'localmonthname':
2390                                 $value = $wgContLang->getMonthName( $localMonthName );
2391                                 break;
2392                         case 'localmonthnamegen':
2393                                 $value = $wgContLang->getMonthNameGen( $localMonthName );
2394                                 break;
2395                         case 'localmonthabbrev':
2396                                 $value = $wgContLang->getMonthAbbreviation( $localMonthName );
2397                                 break;
2398                         case 'localday':
2399                                 $value = $wgContLang->formatNum( $localDay );
2400                                 break;
2401                         case 'localday2':
2402                                 $value = $wgContLang->formatNum( $localDay2 );
2403                                 break;
2404                         case 'pagename':
2405                                 $value = wfEscapeWikiText( $this->mTitle->getText() );
2406                                 break;
2407                         case 'pagenamee':
2408                                 $value = $this->mTitle->getPartialURL();
2409                                 break;
2410                         case 'fullpagename':
2411                                 $value = wfEscapeWikiText( $this->mTitle->getPrefixedText() );
2412                                 break;
2413                         case 'fullpagenamee':
2414                                 $value = $this->mTitle->getPrefixedURL();
2415                                 break;
2416                         case 'subpagename':
2417                                 $value = wfEscapeWikiText( $this->mTitle->getSubpageText() );
2418                                 break;
2419                         case 'subpagenamee':
2420                                 $value = $this->mTitle->getSubpageUrlForm();
2421                                 break;
2422                         case 'basepagename':
2423                                 $value = wfEscapeWikiText( $this->mTitle->getBaseText() );
2424                                 break;
2425                         case 'basepagenamee':
2426                                 $value = wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2427                                 break;
2428                         case 'talkpagename':
2429                                 if( $this->mTitle->canTalk() ) {
2430                                         $talkPage = $this->mTitle->getTalkPage();
2431                                         $value = wfEscapeWikiText( $talkPage->getPrefixedText() );
2432                                 } else {
2433                                         $value = '';
2434                                 }
2435                                 break;
2436                         case 'talkpagenamee':
2437                                 if( $this->mTitle->canTalk() ) {
2438                                         $talkPage = $this->mTitle->getTalkPage();
2439                                         $value = $talkPage->getPrefixedUrl();
2440                                 } else {
2441                                         $value = '';
2442                                 }
2443                                 break;
2444                         case 'subjectpagename':
2445                                 $subjPage = $this->mTitle->getSubjectPage();
2446                                 $value = wfEscapeWikiText( $subjPage->getPrefixedText() );
2447                                 break;
2448                         case 'subjectpagenamee':
2449                                 $subjPage = $this->mTitle->getSubjectPage();
2450                                 $value = $subjPage->getPrefixedUrl();
2451                                 break;
2452                         case 'revisionid':
2453                                 // Let the edit saving system know we should parse the page
2454                                 // *after* a revision ID has been assigned.
2455                                 $this->mOutput->setFlag( 'vary-revision' );
2456                                 wfDebug( __METHOD__ . ": {{REVISIONID}} used, setting vary-revision...\n" );
2457                                 $value = $this->mRevisionId;
2458                                 break;
2459                         case 'revisionday':
2460                                 // Let the edit saving system know we should parse the page
2461                                 // *after* a revision ID has been assigned. This is for null edits.
2462                                 $this->mOutput->setFlag( 'vary-revision' );
2463                                 wfDebug( __METHOD__ . ": {{REVISIONDAY}} used, setting vary-revision...\n" );
2464                                 $value = intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2465                                 break;
2466                         case 'revisionday2':
2467                                 // Let the edit saving system know we should parse the page
2468                                 // *after* a revision ID has been assigned. This is for null edits.
2469                                 $this->mOutput->setFlag( 'vary-revision' );
2470                                 wfDebug( __METHOD__ . ": {{REVISIONDAY2}} used, setting vary-revision...\n" );
2471                                 $value = substr( $this->getRevisionTimestamp(), 6, 2 );
2472                                 break;
2473                         case 'revisionmonth':
2474                                 // Let the edit saving system know we should parse the page
2475                                 // *after* a revision ID has been assigned. This is for null edits.
2476                                 $this->mOutput->setFlag( 'vary-revision' );
2477                                 wfDebug( __METHOD__ . ": {{REVISIONMONTH}} used, setting vary-revision...\n" );
2478                                 $value = intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2479                                 break;
2480                         case 'revisionyear':
2481                                 // Let the edit saving system know we should parse the page
2482                                 // *after* a revision ID has been assigned. This is for null edits.
2483                                 $this->mOutput->setFlag( 'vary-revision' );
2484                                 wfDebug( __METHOD__ . ": {{REVISIONYEAR}} used, setting vary-revision...\n" );
2485                                 $value = substr( $this->getRevisionTimestamp(), 0, 4 );
2486                                 break;
2487                         case 'revisiontimestamp':
2488                                 // Let the edit saving system know we should parse the page
2489                                 // *after* a revision ID has been assigned. This is for null edits.
2490                                 $this->mOutput->setFlag( 'vary-revision' );
2491                                 wfDebug( __METHOD__ . ": {{REVISIONTIMESTAMP}} used, setting vary-revision...\n" );
2492                                 $value = $this->getRevisionTimestamp();
2493                                 break;
2494                         case 'revisionuser':
2495                                 // Let the edit saving system know we should parse the page
2496                                 // *after* a revision ID has been assigned. This is for null edits.
2497                                 $this->mOutput->setFlag( 'vary-revision' );
2498                                 wfDebug( __METHOD__ . ": {{REVISIONUSER}} used, setting vary-revision...\n" );
2499                                 $value = $this->getRevisionUser();
2500                                 break;
2501                         case 'namespace':
2502                                 $value = str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2503                                 break;
2504                         case 'namespacee':
2505                                 $value = wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2506                                 break;
2507                         case 'talkspace':
2508                                 $value = $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2509                                 break;
2510                         case 'talkspacee':
2511                                 $value = $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2512                                 break;
2513                         case 'subjectspace':
2514                                 $value = $this->mTitle->getSubjectNsText();
2515                                 break;
2516                         case 'subjectspacee':
2517                                 $value = ( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2518                                 break;
2519                         case 'currentdayname':
2520                                 $value = $wgContLang->getWeekdayName( gmdate( 'w', $ts ) + 1 );
2521                                 break;
2522                         case 'currentyear':
2523                                 $value = $wgContLang->formatNum( gmdate( 'Y', $ts ), true );
2524                                 break;
2525                         case 'currenttime':
2526                                 $value = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2527                                 break;
2528                         case 'currenthour':
2529                                 $value = $wgContLang->formatNum( gmdate( 'H', $ts ), true );
2530                                 break;
2531                         case 'currentweek':
2532                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2533                                 // int to remove the padding
2534                                 $value = $wgContLang->formatNum( (int)gmdate( 'W', $ts ) );
2535                                 break;
2536                         case 'currentdow':
2537                                 $value = $wgContLang->formatNum( gmdate( 'w', $ts ) );
2538                                 break;
2539                         case 'localdayname':
2540                                 $value = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2541                                 break;
2542                         case 'localyear':
2543                                 $value = $wgContLang->formatNum( $localYear, true );
2544                                 break;
2545                         case 'localtime':
2546                                 $value = $wgContLang->time( $localTimestamp, false, false );
2547                                 break;
2548                         case 'localhour':
2549                                 $value = $wgContLang->formatNum( $localHour, true );
2550                                 break;
2551                         case 'localweek':
2552                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2553                                 // int to remove the padding
2554                                 $value = $wgContLang->formatNum( (int)$localWeek );
2555                                 break;
2556                         case 'localdow':
2557                                 $value = $wgContLang->formatNum( $localDayOfWeek );
2558                                 break;
2559                         case 'numberofarticles':
2560                                 $value = $wgContLang->formatNum( SiteStats::articles() );
2561                                 break;
2562                         case 'numberoffiles':
2563                                 $value = $wgContLang->formatNum( SiteStats::images() );
2564                                 break;
2565                         case 'numberofusers':
2566                                 $value = $wgContLang->formatNum( SiteStats::users() );
2567                                 break;
2568                         case 'numberofactiveusers':
2569                                 $value = $wgContLang->formatNum( SiteStats::activeUsers() );
2570                                 break;
2571                         case 'numberofpages':
2572                                 $value = $wgContLang->formatNum( SiteStats::pages() );
2573                                 break;
2574                         case 'numberofadmins':
2575                                 $value = $wgContLang->formatNum( SiteStats::numberingroup('sysop') );
2576                                 break;
2577                         case 'numberofedits':
2578                                 $value = $wgContLang->formatNum( SiteStats::edits() );
2579                                 break;
2580                         case 'numberofviews':
2581                                 $value = $wgContLang->formatNum( SiteStats::views() );
2582                                 break;
2583                         case 'currenttimestamp':
2584                                 $value = wfTimestamp( TS_MW, $ts );
2585                                 break;
2586                         case 'localtimestamp':
2587                                 $value = $localTimestamp;
2588                                 break;
2589                         case 'currentversion':
2590                                 $value = SpecialVersion::getVersion();
2591                                 break;
2592                         case 'sitename':
2593                                 return $wgSitename;
2594                         case 'server':
2595                                 return $wgServer;
2596                         case 'servername':
2597                                 return $wgServerName;
2598                         case 'scriptpath':
2599                                 return $wgScriptPath;
2600                         case 'stylepath':
2601                                 return $wgStylePath;
2602                         case 'directionmark':
2603                                 return $wgContLang->getDirMark();
2604                         case 'contentlanguage':
2605                                 global $wgContLanguageCode;
2606                                 return $wgContLanguageCode;
2607                         default:
2608                                 $ret = null;
2609                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$this->mVarCache, &$index, &$ret, &$frame ) ) )
2610                                         return $ret;
2611                                 else
2612                                         return null;
2613                 }
2614
2615                 if ( $index )
2616                         $this->mVarCache[$index] = $value;
2617
2618                 return $value;
2619         }
2620
2621         /**
2622          * initialise the magic variables (like CURRENTMONTHNAME) and substitution modifiers
2623          *
2624          * @private
2625          */
2626         function initialiseVariables() {
2627                 wfProfileIn( __METHOD__ );
2628                 $variableIDs = MagicWord::getVariableIDs();
2629                 $substIDs = MagicWord::getSubstIDs();
2630
2631                 $this->mVariables = new MagicWordArray( $variableIDs );
2632                 $this->mSubsts = new MagicWordArray( $substIDs );
2633                 wfProfileOut( __METHOD__ );
2634         }
2635
2636         /**
2637          * Preprocess some wikitext and return the document tree.
2638          * This is the ghost of replace_variables().
2639          *
2640          * @param string $text The text to parse
2641          * @param integer flags Bitwise combination of:
2642          *          self::PTD_FOR_INCLUSION    Handle <noinclude>/<includeonly> as if the text is being
2643          *                                     included. Default is to assume a direct page view.
2644          *
2645          * The generated DOM tree must depend only on the input text and the flags.
2646          * The DOM tree must be the same in OT_HTML and OT_WIKI mode, to avoid a regression of bug 4899.
2647          *
2648          * Any flag added to the $flags parameter here, or any other parameter liable to cause a
2649          * change in the DOM tree for a given text, must be passed through the section identifier
2650          * in the section edit link and thus back to extractSections().
2651          *
2652          * The output of this function is currently only cached in process memory, but a persistent
2653          * cache may be implemented at a later date which takes further advantage of these strict
2654          * dependency requirements.
2655          *
2656          * @private
2657          */
2658         function preprocessToDom ( $text, $flags = 0 ) {
2659                 $dom = $this->getPreprocessor()->preprocessToObj( $text, $flags );
2660                 return $dom;
2661         }
2662
2663         /*
2664          * Return a three-element array: leading whitespace, string contents, trailing whitespace
2665          */
2666         public static function splitWhitespace( $s ) {
2667                 $ltrimmed = ltrim( $s );
2668                 $w1 = substr( $s, 0, strlen( $s ) - strlen( $ltrimmed ) );
2669                 $trimmed = rtrim( $ltrimmed );
2670                 $diff = strlen( $ltrimmed ) - strlen( $trimmed );
2671                 if ( $diff > 0 ) {
2672                         $w2 = substr( $ltrimmed, -$diff );
2673                 } else {
2674                         $w2 = '';
2675                 }
2676                 return array( $w1, $trimmed, $w2 );
2677         }
2678
2679         /**
2680          * Replace magic variables, templates, and template arguments
2681          * with the appropriate text. Templates are substituted recursively,
2682          * taking care to avoid infinite loops.
2683          *
2684          * Note that the substitution depends on value of $mOutputType:
2685          *  self::OT_WIKI: only {{subst:}} templates
2686          *  self::OT_PREPROCESS: templates but not extension tags
2687          *  self::OT_HTML: all templates and extension tags
2688          *
2689          * @param string $tex The text to transform
2690          * @param PPFrame $frame Object describing the arguments passed to the template.
2691          *        Arguments may also be provided as an associative array, as was the usual case before MW1.12.
2692          *        Providing arguments this way may be useful for extensions wishing to perform variable replacement explicitly.
2693          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2694          * @private
2695          */
2696         function replaceVariables( $text, $frame = false, $argsOnly = false ) {
2697                 # Is there any text? Also, Prevent too big inclusions!
2698                 if ( strlen( $text ) < 1 || strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2699                         return $text;
2700                 }
2701                 wfProfileIn( __METHOD__ );
2702
2703                 if ( $frame === false ) {
2704                         $frame = $this->getPreprocessor()->newFrame();
2705                 } elseif ( !( $frame instanceof PPFrame ) ) {
2706                         wfDebug( __METHOD__." called using plain parameters instead of a PPFrame instance. Creating custom frame.\n" );
2707                         $frame = $this->getPreprocessor()->newCustomFrame($frame);
2708                 }
2709
2710                 $dom = $this->preprocessToDom( $text );
2711                 $flags = $argsOnly ? PPFrame::NO_TEMPLATES : 0;
2712                 $text = $frame->expand( $dom, $flags );
2713
2714                 wfProfileOut( __METHOD__ );
2715                 return $text;
2716         }
2717
2718         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2719         static function createAssocArgs( $args ) {
2720                 $assocArgs = array();
2721                 $index = 1;
2722                 foreach( $args as $arg ) {
2723                         $eqpos = strpos( $arg, '=' );
2724                         if ( $eqpos === false ) {
2725                                 $assocArgs[$index++] = $arg;
2726                         } else {
2727                                 $name = trim( substr( $arg, 0, $eqpos ) );
2728                                 $value = trim( substr( $arg, $eqpos+1 ) );
2729                                 if ( $value === false ) {
2730                                         $value = '';
2731                                 }
2732                                 if ( $name !== false ) {
2733                                         $assocArgs[$name] = $value;
2734                                 }
2735                         }
2736                 }
2737
2738                 return $assocArgs;
2739         }
2740
2741         /**
2742          * Warn the user when a parser limitation is reached
2743          * Will warn at most once the user per limitation type
2744          *
2745          * @param string $limitationType, should be one of:
2746          *   'expensive-parserfunction' (corresponding messages: 'expensive-parserfunction-warning', 'expensive-parserfunction-category')
2747          *   'post-expand-template-argument' (corresponding messages: 'post-expand-template-argument-warning', 'post-expand-template-argument-category')
2748          *   'post-expand-template-inclusion' (corresponding messages: 'post-expand-template-inclusion-warning', 'post-expand-template-inclusion-category')
2749          * @params int $current, $max When an explicit limit has been
2750          *       exceeded, provide the values (optional)
2751          */
2752         function limitationWarn( $limitationType, $current=null, $max=null) {
2753                 //does no harm if $current and $max are present but are unnecessary for the message
2754                 $warning = wfMsgExt( "$limitationType-warning", array( 'parsemag', 'escape' ), $current, $max );
2755                 $this->mOutput->addWarning( $warning );
2756                 $this->addTrackingCategory( "$limitationType-category" );
2757         }
2758
2759         /**
2760          * Return the text of a template, after recursively
2761          * replacing any variables or templates within the template.
2762          *
2763          * @param array $piece The parts of the template
2764          *  $piece['title']: the title, i.e. the part before the |
2765          *  $piece['parts']: the parameter array
2766          *  $piece['lineStart']: whether the brace was at the start of a line
2767          * @param PPFrame The current frame, contains template arguments
2768          * @return string the text of the template
2769          * @private
2770          */
2771         function braceSubstitution( $piece, $frame ) {
2772                 global $wgContLang, $wgNonincludableNamespaces;
2773                 wfProfileIn( __METHOD__ );
2774                 wfProfileIn( __METHOD__.'-setup' );
2775
2776                 # Flags
2777                 $found = false;             # $text has been filled
2778                 $nowiki = false;            # wiki markup in $text should be escaped
2779                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2780                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2781                 $isChildObj = false;        # $text is a DOM node needing expansion in a child frame
2782                 $isLocalObj = false;        # $text is a DOM node needing expansion in the current frame
2783
2784                 # Title object, where $text came from
2785                 $title = null;
2786
2787                 # $part1 is the bit before the first |, and must contain only title characters.
2788                 # Various prefixes will be stripped from it later.
2789                 $titleWithSpaces = $frame->expand( $piece['title'] );
2790                 $part1 = trim( $titleWithSpaces );
2791                 $titleText = false;
2792
2793                 # Original title text preserved for various purposes
2794                 $originalTitle = $part1;
2795
2796                 # $args is a list of argument nodes, starting from index 0, not including $part1
2797                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2798                 wfProfileOut( __METHOD__.'-setup' );
2799
2800                 # SUBST
2801                 wfProfileIn( __METHOD__.'-modifiers' );
2802                 if ( !$found ) {
2803
2804                         $substMatch = $this->mSubsts->matchStartAndRemove( $part1 );
2805
2806                         # Possibilities for substMatch: "subst", "safesubst" or FALSE
2807                         # Whether to include depends also on whether we are in the pre-save-transform
2808                         #
2809                         # safesubst || (subst && PST) || (false && !PST) => transclude (skip the if)
2810                         # (false && PST) || (subst && !PST)  => return input (handled by if)
2811                         if ( $substMatch != 'safesubst' && ($substMatch == 'subst' xor $this->ot['wiki']) ) {
2812                                 $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
2813                                 $isLocalObj = true;
2814                                 $found = true;
2815                         }
2816                 }
2817
2818                 # Variables
2819                 if ( !$found && $args->getLength() == 0 ) {
2820                         $id = $this->mVariables->matchStartToEnd( $part1 );
2821                         if ( $id !== false ) {
2822                                 $text = $this->getVariableValue( $id, $frame );
2823                                 if (MagicWord::getCacheTTL($id)>-1)
2824                                         $this->mOutput->mContainsOldMagic = true;
2825                                 $found = true;
2826                         }
2827                 }
2828
2829                 # MSG, MSGNW and RAW
2830                 if ( !$found ) {
2831                         # Check for MSGNW:
2832                         $mwMsgnw = MagicWord::get( 'msgnw' );
2833                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2834                                 $nowiki = true;
2835                         } else {
2836                                 # Remove obsolete MSG:
2837                                 $mwMsg = MagicWord::get( 'msg' );
2838                                 $mwMsg->matchStartAndRemove( $part1 );
2839                         }
2840
2841                         # Check for RAW:
2842                         $mwRaw = MagicWord::get( 'raw' );
2843                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2844                                 $forceRawInterwiki = true;
2845                         }
2846                 }
2847                 wfProfileOut( __METHOD__.'-modifiers' );
2848
2849                 # Parser functions
2850                 if ( !$found ) {
2851                         wfProfileIn( __METHOD__ . '-pfunc' );
2852
2853                         $colonPos = strpos( $part1, ':' );
2854                         if ( $colonPos !== false ) {
2855                                 # Case sensitive functions
2856                                 $function = substr( $part1, 0, $colonPos );
2857                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2858                                         $function = $this->mFunctionSynonyms[1][$function];
2859                                 } else {
2860                                         # Case insensitive functions
2861                                         $function = $wgContLang->lc( $function );
2862                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2863                                                 $function = $this->mFunctionSynonyms[0][$function];
2864                                         } else {
2865                                                 $function = false;
2866                                         }
2867                                 }
2868                                 if ( $function ) {
2869                                         list( $callback, $flags ) = $this->mFunctionHooks[$function];
2870                                         $initialArgs = array( &$this );
2871                                         $funcArgs = array( trim( substr( $part1, $colonPos + 1 ) ) );
2872                                         if ( $flags & SFH_OBJECT_ARGS ) {
2873                                                 # Add a frame parameter, and pass the arguments as an array
2874                                                 $allArgs = $initialArgs;
2875                                                 $allArgs[] = $frame;
2876                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2877                                                         $funcArgs[] = $args->item( $i );
2878                                                 }
2879                                                 $allArgs[] = $funcArgs;
2880                                         } else {
2881                                                 # Convert arguments to plain text
2882                                                 for ( $i = 0; $i < $args->getLength(); $i++ ) {
2883                                                         $funcArgs[] = trim( $frame->expand( $args->item( $i ) ) );
2884                                                 }
2885                                                 $allArgs = array_merge( $initialArgs, $funcArgs );
2886                                         }
2887
2888                                         # Workaround for PHP bug 35229 and similar
2889                                         if ( !is_callable( $callback ) ) {
2890                                                 wfProfileOut( __METHOD__ . '-pfunc' );
2891                                                 wfProfileOut( __METHOD__ );
2892                                                 throw new MWException( "Tag hook for $function is not callable\n" );
2893                                         }
2894                                         $result = call_user_func_array( $callback, $allArgs );
2895                                         $found = true;
2896                                         $noparse = true;
2897                                         $preprocessFlags = 0;
2898
2899                                         if ( is_array( $result ) ) {
2900                                                 if ( isset( $result[0] ) ) {
2901                                                         $text = $result[0];
2902                                                         unset( $result[0] );
2903                                                 }
2904
2905                                                 // Extract flags into the local scope
2906                                                 // This allows callers to set flags such as nowiki, found, etc.
2907                                                 extract( $result );
2908                                         } else {
2909                                                 $text = $result;
2910                                         }
2911                                         if ( !$noparse ) {
2912                                                 $text = $this->preprocessToDom( $text, $preprocessFlags );
2913                                                 $isChildObj = true;
2914                                         }
2915                                 }
2916                         }
2917                         wfProfileOut( __METHOD__ . '-pfunc' );
2918                 }
2919
2920                 # Finish mangling title and then check for loops.
2921                 # Set $title to a Title object and $titleText to the PDBK
2922                 if ( !$found ) {
2923                         $ns = NS_TEMPLATE;
2924                         # Split the title into page and subpage
2925                         $subpage = '';
2926                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2927                         if ($subpage !== '') {
2928                                 $ns = $this->mTitle->getNamespace();
2929                         }
2930                         $title = Title::newFromText( $part1, $ns );
2931                         if ( $title ) {
2932                                 $titleText = $title->getPrefixedText();
2933                                 # Check for language variants if the template is not found
2934                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
2935                                         $wgContLang->findVariantLink( $part1, $title, true );
2936                                 }
2937                                 # Do recursion depth check
2938                                 $limit = $this->mOptions->getMaxTemplateDepth();
2939                                 if ( $frame->depth >= $limit ) {
2940                                         $found = true;
2941                                         $text = '<span class="error">' . wfMsgForContent( 'parser-template-recursion-depth-warning', $limit ) . '</span>';
2942                                 }
2943                         }
2944                 }
2945
2946                 # Load from database
2947                 if ( !$found && $title ) {
2948                         wfProfileIn( __METHOD__ . '-loadtpl' );
2949                         if ( !$title->isExternal() ) {
2950                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
2951                                         $text = SpecialPage::capturePath( $title );
2952                                         if ( is_string( $text ) ) {
2953                                                 $found = true;
2954                                                 $isHTML = true;
2955                                                 $this->disableCache();
2956                                         }
2957                                 } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
2958                                         $found = false; //access denied
2959                                         wfDebug( __METHOD__.": template inclusion denied for " . $title->getPrefixedDBkey() );
2960                                 } else {
2961                                         list( $text, $title ) = $this->getTemplateDom( $title );
2962                                         if ( $text !== false ) {
2963                                                 $found = true;
2964                                                 $isChildObj = true;
2965                                         }
2966                                 }
2967
2968                                 # If the title is valid but undisplayable, make a link to it
2969                                 if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
2970                                         $text = "[[:$titleText]]";
2971                                         $found = true;
2972                                 }
2973                         } elseif ( $title->isTrans() ) {
2974                                 // Interwiki transclusion
2975                                 if ( $this->ot['html'] && !$forceRawInterwiki ) {
2976                                         $text = $this->interwikiTransclude( $title, 'render' );
2977                                         $isHTML = true;
2978                                 } else {
2979                                         $text = $this->interwikiTransclude( $title, 'raw' );
2980                                         // Preprocess it like a template
2981                                         $text = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
2982                                         $isChildObj = true;
2983                                 }
2984                                 $found = true;
2985                         }
2986
2987                         # Do infinite loop check
2988                         # This has to be done after redirect resolution to avoid infinite loops via redirects
2989                         if ( !$frame->loopCheck( $title ) ) {
2990                                 $found = true;
2991                                 $text = '<span class="error">' . wfMsgForContent( 'parser-template-loop-warning', $titleText ) . '</span>';
2992                                 wfDebug( __METHOD__.": template loop broken at '$titleText'\n" );
2993                         }
2994                         wfProfileOut( __METHOD__ . '-loadtpl' );
2995                 }
2996
2997                 # If we haven't found text to substitute by now, we're done
2998                 # Recover the source wikitext and return it
2999                 if ( !$found ) {
3000                         $text = $frame->virtualBracketedImplode( '{{', '|', '}}', $titleWithSpaces, $args );
3001                         wfProfileOut( __METHOD__ );
3002                         return array( 'object' => $text );
3003                 }
3004
3005                 # Expand DOM-style return values in a child frame
3006                 if ( $isChildObj ) {
3007                         # Clean up argument array
3008                         $newFrame = $frame->newChild( $args, $title );
3009
3010                         if ( $nowiki ) {
3011                                 $text = $newFrame->expand( $text, PPFrame::RECOVER_ORIG );
3012                         } elseif ( $titleText !== false && $newFrame->isEmpty() ) {
3013                                 # Expansion is eligible for the empty-frame cache
3014                                 if ( isset( $this->mTplExpandCache[$titleText] ) ) {
3015                                         $text = $this->mTplExpandCache[$titleText];
3016                                 } else {
3017                                         $text = $newFrame->expand( $text );
3018                                         $this->mTplExpandCache[$titleText] = $text;
3019                                 }
3020                         } else {
3021                                 # Uncached expansion
3022                                 $text = $newFrame->expand( $text );
3023                         }
3024                 }
3025                 if ( $isLocalObj && $nowiki ) {
3026                         $text = $frame->expand( $text, PPFrame::RECOVER_ORIG );
3027                         $isLocalObj = false;
3028                 }
3029
3030                 # Replace raw HTML by a placeholder
3031                 # Add a blank line preceding, to prevent it from mucking up
3032                 # immediately preceding headings
3033                 if ( $isHTML ) {
3034                         $text = "\n\n" . $this->insertStripItem( $text );
3035                 }
3036                 # Escape nowiki-style return values
3037                 elseif ( $nowiki && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3038                         $text = wfEscapeWikiText( $text );
3039                 }
3040                 # Bug 529: if the template begins with a table or block-level
3041                 # element, it should be treated as beginning a new line.
3042                 # This behaviour is somewhat controversial.
3043                 elseif ( is_string( $text ) && !$piece['lineStart'] && preg_match('/^(?:{\\||:|;|#|\*)/', $text)) /*}*/{
3044                         $text = "\n" . $text;
3045                 }
3046
3047                 if ( is_string( $text ) && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3048                         # Error, oversize inclusion
3049                         $text = "[[$originalTitle]]" .
3050                                 $this->insertStripItem( '<!-- WARNING: template omitted, post-expand include size too large -->' );
3051                         $this->limitationWarn( 'post-expand-template-inclusion' );
3052                 }
3053
3054                 if ( $isLocalObj ) {
3055                         $ret = array( 'object' => $text );
3056                 } else {
3057                         $ret = array( 'text' => $text );
3058                 }
3059
3060                 wfProfileOut( __METHOD__ );
3061                 return $ret;
3062         }
3063
3064         /**
3065          * Get the semi-parsed DOM representation of a template with a given title,
3066          * and its redirect destination title. Cached.
3067          */
3068         function getTemplateDom( $title ) {
3069                 $cacheTitle = $title;
3070                 $titleText = $title->getPrefixedDBkey();
3071
3072                 if ( isset( $this->mTplRedirCache[$titleText] ) ) {
3073                         list( $ns, $dbk ) = $this->mTplRedirCache[$titleText];
3074                         $title = Title::makeTitle( $ns, $dbk );
3075                         $titleText = $title->getPrefixedDBkey();
3076                 }
3077                 if ( isset( $this->mTplDomCache[$titleText] ) ) {
3078                         return array( $this->mTplDomCache[$titleText], $title );
3079                 }
3080
3081                 // Cache miss, go to the database
3082                 list( $text, $title ) = $this->fetchTemplateAndTitle( $title );
3083
3084                 if ( $text === false ) {
3085                         $this->mTplDomCache[$titleText] = false;
3086                         return array( false, $title );
3087                 }
3088
3089                 $dom = $this->preprocessToDom( $text, self::PTD_FOR_INCLUSION );
3090                 $this->mTplDomCache[ $titleText ] = $dom;
3091
3092                 if (! $title->equals($cacheTitle)) {
3093                         $this->mTplRedirCache[$cacheTitle->getPrefixedDBkey()] =
3094                                 array( $title->getNamespace(),$cdb = $title->getDBkey() );
3095                 }
3096
3097                 return array( $dom, $title );
3098         }
3099
3100         /**
3101          * Fetch the unparsed text of a template and register a reference to it.
3102          */
3103         function fetchTemplateAndTitle( $title ) {
3104                 $templateCb = $this->mOptions->getTemplateCallback();
3105                 $stuff = call_user_func( $templateCb, $title, $this );
3106                 $text = $stuff['text'];
3107                 $finalTitle = isset( $stuff['finalTitle'] ) ? $stuff['finalTitle'] : $title;
3108                 if ( isset( $stuff['deps'] ) ) {
3109                         foreach ( $stuff['deps'] as $dep ) {
3110                                 $this->mOutput->addTemplate( $dep['title'], $dep['page_id'], $dep['rev_id'] );
3111                         }
3112                 }
3113                 return array($text,$finalTitle);
3114         }
3115
3116         function fetchTemplate( $title ) {
3117                 $rv = $this->fetchTemplateAndTitle($title);
3118                 return $rv[0];
3119         }
3120
3121         /**
3122          * Static function to get a template
3123          * Can be overridden via ParserOptions::setTemplateCallback().
3124          */
3125         static function statelessFetchTemplate( $title, $parser=false ) {
3126                 $text = $skip = false;
3127                 $finalTitle = $title;
3128                 $deps = array();
3129
3130                 // Loop to fetch the article, with up to 1 redirect
3131                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3132                         # Give extensions a chance to select the revision instead
3133                         $id = false; // Assume current
3134                         wfRunHooks( 'BeforeParserFetchTemplateAndtitle', array( $parser, &$title, &$skip, &$id ) );
3135
3136                         if( $skip ) {
3137                                 $text = false;
3138                                 $deps[] = array(
3139                                         'title' => $title,
3140                                         'page_id' => $title->getArticleID(),
3141                                         'rev_id' => null );
3142                                 break;
3143                         }
3144                         $rev = $id ? Revision::newFromId( $id ) : Revision::newFromTitle( $title );
3145                         $rev_id = $rev ? $rev->getId() : 0;
3146                         // If there is no current revision, there is no page
3147                         if( $id === false && !$rev ) {
3148                                 $linkCache = LinkCache::singleton();
3149                                 $linkCache->addBadLinkObj( $title );
3150                         }
3151
3152                         $deps[] = array(
3153                                 'title' => $title,
3154                                 'page_id' => $title->getArticleID(),
3155                                 'rev_id' => $rev_id );
3156
3157                         if( $rev ) {
3158                                 $text = $rev->getText();
3159                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3160                                 global $wgContLang;
3161                                 $message = $wgContLang->lcfirst( $title->getText() );
3162                                 $text = wfMsgForContentNoTrans( $message );
3163                                 if( wfEmptyMsg( $message, $text ) ) {
3164                                         $text = false;
3165                                         break;
3166                                 }
3167                         } else {
3168                                 break;
3169                         }
3170                         if ( $text === false ) {
3171                                 break;
3172                         }
3173                         // Redirect?
3174                         $finalTitle = $title;
3175                         $title = Title::newFromRedirect( $text );
3176                 }
3177                 return array(
3178                         'text' => $text,
3179                         'finalTitle' => $finalTitle,
3180                         'deps' => $deps );
3181         }
3182
3183         /**
3184          * Transclude an interwiki link.
3185          */
3186         function interwikiTransclude( $title, $action ) {
3187                 global $wgEnableScaryTranscluding;
3188
3189                 if (!$wgEnableScaryTranscluding)
3190                         return wfMsg('scarytranscludedisabled');
3191
3192                 $url = $title->getFullUrl( "action=$action" );
3193
3194                 if (strlen($url) > 255)
3195                         return wfMsg('scarytranscludetoolong');
3196                 return $this->fetchScaryTemplateMaybeFromCache($url);
3197         }
3198
3199         function fetchScaryTemplateMaybeFromCache($url) {
3200                 global $wgTranscludeCacheExpiry;
3201                 $dbr = wfGetDB(DB_SLAVE);
3202                 $tsCond = $dbr->timestamp( time() - $wgTranscludeCacheExpiry );
3203                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3204                                 array('tc_url' => $url, "tc_time >= " . $dbr->addQuotes( $tsCond ) ) );
3205                 if ($obj) {
3206                         return $obj->tc_contents;
3207                 }
3208
3209                 $text = Http::get($url);
3210                 if (!$text)
3211                         return wfMsg('scarytranscludefailed', $url);
3212
3213                 $dbw = wfGetDB(DB_MASTER);
3214                 $dbw->replace('transcache', array('tc_url'), array(
3215                         'tc_url' => $url,
3216                         'tc_time' => $dbw->timestamp( time() ),
3217                         'tc_contents' => $text));
3218                 return $text;
3219         }
3220
3221
3222         /**
3223          * Triple brace replacement -- used for template arguments
3224          * @private
3225          */
3226         function argSubstitution( $piece, $frame ) {
3227                 wfProfileIn( __METHOD__ );
3228
3229                 $error = false;
3230                 $parts = $piece['parts'];
3231                 $nameWithSpaces = $frame->expand( $piece['title'] );
3232                 $argName = trim( $nameWithSpaces );
3233                 $object = false;
3234                 $text = $frame->getArgument( $argName );
3235                 if (  $text === false && $parts->getLength() > 0
3236                   && (
3237                     $this->ot['html']
3238                     || $this->ot['pre']
3239                     || ( $this->ot['wiki'] && $frame->isTemplate() )
3240                   )
3241                 ) {
3242                         # No match in frame, use the supplied default
3243                         $object = $parts->item( 0 )->getChildren();
3244                 }
3245                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3246                         $error = '<!-- WARNING: argument omitted, expansion size too large -->';
3247                         $this->limitationWarn( 'post-expand-template-argument' );
3248                 }
3249
3250                 if ( $text === false && $object === false ) {
3251                         # No match anywhere
3252                         $object = $frame->virtualBracketedImplode( '{{{', '|', '}}}', $nameWithSpaces, $parts );
3253                 }
3254                 if ( $error !== false ) {
3255                         $text .= $error;
3256                 }
3257                 if ( $object !== false ) {
3258                         $ret = array( 'object' => $object );
3259                 } else {
3260                         $ret = array( 'text' => $text );
3261                 }
3262
3263                 wfProfileOut( __METHOD__ );
3264                 return $ret;
3265         }
3266
3267         /**
3268          * Return the text to be used for a given extension tag.
3269          * This is the ghost of strip().
3270          *
3271          * @param array $params Associative array of parameters:
3272          *     name       PPNode for the tag name
3273          *     attr       PPNode for unparsed text where tag attributes are thought to be
3274          *     attributes Optional associative array of parsed attributes
3275          *     inner      Contents of extension element
3276          *     noClose    Original text did not have a close tag
3277          * @param PPFrame $frame
3278          */
3279         function extensionSubstitution( $params, $frame ) {
3280                 global $wgRawHtml, $wgContLang;
3281
3282                 $name = $frame->expand( $params['name'] );
3283                 $attrText = !isset( $params['attr'] ) ? null : $frame->expand( $params['attr'] );
3284                 $content = !isset( $params['inner'] ) ? null : $frame->expand( $params['inner'] );
3285
3286                 $marker = "{$this->mUniqPrefix}-$name-" . sprintf('%08X', $this->mMarkerIndex++) . self::MARKER_SUFFIX;
3287
3288                 $isFunctionTag = isset( $this->mFunctionTagHooks[strtolower($name)] ) &&
3289                         ( $this->ot['html'] || $this->ot['pre'] );
3290                 if ( $this->ot['html'] || $isFunctionTag ) {
3291                         $name = strtolower( $name );
3292                         $attributes = Sanitizer::decodeTagAttributes( $attrText );
3293                         if ( isset( $params['attributes'] ) ) {
3294                                 $attributes = $attributes + $params['attributes'];
3295                         }
3296                         switch ( $name ) {
3297                                 case 'html':
3298                                         if( $wgRawHtml ) {
3299                                                 $output = $content;
3300                                                 break;
3301                                         } else {
3302                                                 throw new MWException( '<html> extension tag encountered unexpectedly' );
3303                                         }
3304                                 case 'nowiki':
3305                                         $content = strtr($content, array('-{' => '-&#123;', '}-' => '&#125;-'));
3306                                         $output = Xml::escapeTagsOnly( $content );
3307                                         break;
3308                                 case 'gallery':
3309                                         $output = $this->renderImageGallery( $content, $attributes );
3310                                         break;
3311                                 case 'math':
3312                                         if ( $this->mOptions->getUseTeX() ) {
3313                                                 $output = $wgContLang->armourMath(
3314                                                         MathRenderer::renderMath( $content, $attributes ) );
3315                                                 break;
3316                                         }
3317                                         /* else let a tag hook handle it (bug 21222) */
3318                                 default:
3319                                         if( isset( $this->mTagHooks[$name] ) ) {
3320                                                 # Workaround for PHP bug 35229 and similar
3321                                                 if ( !is_callable( $this->mTagHooks[$name] ) ) {
3322                                                         throw new MWException( "Tag hook for $name is not callable\n" );
3323                                                 }
3324                                                 $output = call_user_func_array( $this->mTagHooks[$name],
3325                                                         array( $content, $attributes, $this, $frame ) );
3326                                         } elseif( isset( $this->mFunctionTagHooks[$name] ) ) {
3327                                                 list( $callback, $flags ) = $this->mFunctionTagHooks[$name];
3328                                                 if( !is_callable( $callback ) )
3329                                                         throw new MWException( "Tag hook for $name is not callable\n" );
3330
3331                                                 $output = call_user_func_array( $callback,
3332                                                         array( &$this, $frame, $content, $attributes ) );
3333                                         } else {
3334                                                 $output = '<span class="error">Invalid tag extension name: ' .
3335                                                         htmlspecialchars( $name ) . '</span>';
3336                                         }
3337                         }
3338                 } else {
3339                         if ( is_null( $attrText ) ) {
3340                                 $attrText = '';
3341                         }
3342                         if ( isset( $params['attributes'] ) ) {
3343                                 foreach ( $params['attributes'] as $attrName => $attrValue ) {
3344                                         $attrText .= ' ' . htmlspecialchars( $attrName ) . '="' .
3345                                                 htmlspecialchars( $attrValue ) . '"';
3346                                 }
3347                         }
3348                         if ( $content === null ) {
3349                                 $output = "<$name$attrText/>";
3350                         } else {
3351                                 $close = is_null( $params['close'] ) ? '' : $frame->expand( $params['close'] );
3352                                 $output = "<$name$attrText>$content$close";
3353                         }
3354                 }
3355
3356                 if( $isFunctionTag ) {
3357                         return $output;
3358                 } elseif ( $name === 'html' || $name === 'nowiki' ) {
3359                         $this->mStripState->nowiki->setPair( $marker, $output );
3360                 } else {
3361                         $this->mStripState->general->setPair( $marker, $output );
3362                 }
3363                 return $marker;
3364         }
3365
3366         /**
3367          * Increment an include size counter
3368          *
3369          * @param string $type The type of expansion
3370          * @param integer $size The size of the text
3371          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3372          */
3373         function incrementIncludeSize( $type, $size ) {
3374                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize( $type ) ) {
3375                         return false;
3376                 } else {
3377                         $this->mIncludeSizes[$type] += $size;
3378                         return true;
3379                 }
3380         }
3381
3382         /**
3383          * Increment the expensive function count
3384          *
3385          * @return boolean False if the limit has been exceeded
3386          */
3387         function incrementExpensiveFunctionCount() {
3388                 global $wgExpensiveParserFunctionLimit;
3389                 $this->mExpensiveFunctionCount++;
3390                 if($this->mExpensiveFunctionCount <= $wgExpensiveParserFunctionLimit) {
3391                         return true;
3392                 }
3393                 return false;
3394         }
3395
3396         /**
3397          * Strip double-underscore items like __NOGALLERY__ and __NOTOC__
3398          * Fills $this->mDoubleUnderscores, returns the modified text
3399          */
3400         function doDoubleUnderscore( $text ) {
3401                 wfProfileIn( __METHOD__ );
3402
3403                 // The position of __TOC__ needs to be recorded
3404                 $mw = MagicWord::get( 'toc' );
3405                 if( $mw->match( $text ) ) {
3406                         $this->mShowToc = true;
3407                         $this->mForceTocPosition = true;
3408
3409                         // Set a placeholder. At the end we'll fill it in with the TOC.
3410                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3411
3412                         // Only keep the first one.
3413                         $text = $mw->replace( '', $text );
3414                 }
3415
3416                 // Now match and remove the rest of them
3417                 $mwa = MagicWord::getDoubleUnderscoreArray();
3418                 $this->mDoubleUnderscores = $mwa->matchAndRemove( $text );
3419
3420                 if ( isset( $this->mDoubleUnderscores['nogallery'] ) ) {
3421                         $this->mOutput->mNoGallery = true;
3422                 }
3423                 if ( isset( $this->mDoubleUnderscores['notoc'] ) && !$this->mForceTocPosition ) {
3424                         $this->mShowToc = false;
3425                 }
3426                 if ( isset( $this->mDoubleUnderscores['hiddencat'] ) && $this->mTitle->getNamespace() == NS_CATEGORY ) {
3427                         $this->mOutput->setProperty( 'hiddencat', 'y' );
3428                         $this->addTrackingCategory( 'hidden-category-category' );
3429                 }
3430                 # (bug 8068) Allow control over whether robots index a page.
3431                 #
3432                 # FIXME (bug 14899): __INDEX__ always overrides __NOINDEX__ here!  This
3433                 # is not desirable, the last one on the page should win.
3434                 if( isset( $this->mDoubleUnderscores['noindex'] ) && $this->mTitle->canUseNoindex() ) {
3435                         $this->mOutput->setIndexPolicy( 'noindex' );
3436                         $this->addTrackingCategory( 'noindex-category' );
3437                 }
3438                 if( isset( $this->mDoubleUnderscores['index'] ) && $this->mTitle->canUseNoindex() ){
3439                         $this->mOutput->setIndexPolicy( 'index' );
3440                         $this->addTrackingCategory( 'index-category' );
3441                 }
3442
3443                 wfProfileOut( __METHOD__ );
3444                 return $text;
3445         }
3446
3447         /**
3448          * Add a tracking category, getting the title from a system message,
3449          * or print a debug message if the title is invalid.
3450          * @param $msg String message key
3451          * @return Bool whether the addition was successful
3452          */
3453         protected function addTrackingCategory( $msg ){
3454                 $cat = wfMsgForContent( $msg );
3455
3456                 # Allow tracking categories to be disabled by setting them to "-"
3457                 if( $cat === '-' ) return false;
3458
3459                 $containerCategory = Title::makeTitleSafe( NS_CATEGORY, $cat );
3460                 if ( $containerCategory ) {
3461                         $this->mOutput->addCategory( $containerCategory->getDBkey(), $this->getDefaultSort() );
3462                         return true;
3463                 } else {
3464                         wfDebug( __METHOD__.": [[MediaWiki:$msg]] is not a valid title!\n" );
3465                         return false;
3466                 }
3467         }
3468
3469         /**
3470          * This function accomplishes several tasks:
3471          * 1) Auto-number headings if that option is enabled
3472          * 2) Add an [edit] link to sections for users who have enabled the option and can edit the page
3473          * 3) Add a Table of contents on the top for users who have enabled the option
3474          * 4) Auto-anchor headings
3475          *
3476          * It loops through all headlines, collects the necessary data, then splits up the
3477          * string and re-inserts the newly formatted headlines.
3478          *
3479          * @param string $text
3480          * @param string $origText Original, untouched wikitext
3481          * @param boolean $isMain
3482          * @private
3483          */
3484         function formatHeadings( $text, $origText, $isMain=true ) {
3485                 global $wgMaxTocLevel, $wgContLang, $wgHtml5, $wgExperimentalHtmlIds;
3486
3487                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3488                 $showEditLink = $this->mOptions->getEditSection();
3489
3490                 // Do not call quickUserCan unless necessary
3491                 if( $showEditLink && !$this->mTitle->quickUserCan( 'edit' ) ) {
3492                         $showEditLink = 0;
3493                 }
3494
3495                 # Inhibit editsection links if requested in the page
3496                 if ( isset( $this->mDoubleUnderscores['noeditsection'] )  || $this->mOptions->getIsPrintable() ) {
3497                         $showEditLink = 0;
3498                 }
3499
3500                 # Get all headlines for numbering them and adding funky stuff like [edit]
3501                 # links - this is for later, but we need the number of headlines right now
3502                 $matches = array();
3503                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3504
3505                 # if there are fewer than 4 headlines in the article, do not show TOC
3506                 # unless it's been explicitly enabled.
3507                 $enoughToc = $this->mShowToc &&
3508                         (($numMatches >= 4) || $this->mForceTocPosition);
3509
3510                 # Allow user to stipulate that a page should have a "new section"
3511                 # link added via __NEWSECTIONLINK__
3512                 if ( isset( $this->mDoubleUnderscores['newsectionlink'] ) ) {
3513                         $this->mOutput->setNewSection( true );
3514                 }
3515
3516                 # Allow user to remove the "new section"
3517                 # link via __NONEWSECTIONLINK__
3518                 if ( isset( $this->mDoubleUnderscores['nonewsectionlink'] ) ) {
3519                         $this->mOutput->hideNewSection( true );
3520                 }
3521
3522                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3523                 # override above conditions and always show TOC above first header
3524                 if ( isset( $this->mDoubleUnderscores['forcetoc'] ) ) {
3525                         $this->mShowToc = true;
3526                         $enoughToc = true;
3527                 }
3528
3529                 # We need this to perform operations on the HTML
3530                 $sk = $this->mOptions->getSkin();
3531
3532                 # headline counter
3533                 $headlineCount = 0;
3534                 $numVisible = 0;
3535
3536                 # Ugh .. the TOC should have neat indentation levels which can be
3537                 # passed to the skin functions. These are determined here
3538                 $toc = '';
3539                 $full = '';
3540                 $head = array();
3541                 $sublevelCount = array();
3542                 $levelCount = array();
3543                 $toclevel = 0;
3544                 $level = 0;
3545                 $prevlevel = 0;
3546                 $toclevel = 0;
3547                 $prevtoclevel = 0;
3548                 $markerRegex = "{$this->mUniqPrefix}-h-(\d+)-" . self::MARKER_SUFFIX;
3549                 $baseTitleText = $this->mTitle->getPrefixedDBkey();
3550                 $oldType = $this->mOutputType;
3551                 $this->setOutputType( self::OT_WIKI );
3552                 $frame = $this->getPreprocessor()->newFrame();
3553                 $root = $this->preprocessToDom( $origText );
3554                 $node = $root->getFirstChild();
3555                 $byteOffset = 0;
3556                 $tocraw = array();
3557
3558                 foreach( $matches[3] as $headline ) {
3559                         $isTemplate = false;
3560                         $titleText = false;
3561                         $sectionIndex = false;
3562                         $numbering = '';
3563                         $markerMatches = array();
3564                         if (preg_match("/^$markerRegex/", $headline, $markerMatches)) {
3565                                 $serial = $markerMatches[1];
3566                                 list( $titleText, $sectionIndex ) = $this->mHeadings[$serial];
3567                                 $isTemplate = ($titleText != $baseTitleText);
3568                                 $headline = preg_replace("/^$markerRegex/", "", $headline);
3569                         }
3570
3571                         if( $toclevel ) {
3572                                 $prevlevel = $level;
3573                                 $prevtoclevel = $toclevel;
3574                         }
3575                         $level = $matches[1][$headlineCount];
3576
3577                         if ( $level > $prevlevel ) {
3578                                 # Increase TOC level
3579                                 $toclevel++;
3580                                 $sublevelCount[$toclevel] = 0;
3581                                 if( $toclevel<$wgMaxTocLevel ) {
3582                                         $prevtoclevel = $toclevel;
3583                                         $toc .= $sk->tocIndent();
3584                                         $numVisible++;
3585                                 }
3586                         }
3587                         elseif ( $level < $prevlevel && $toclevel > 1 ) {
3588                                 # Decrease TOC level, find level to jump to
3589
3590                                 for ($i = $toclevel; $i > 0; $i--) {
3591                                         if ( $levelCount[$i] == $level ) {
3592                                                 # Found last matching level
3593                                                 $toclevel = $i;
3594                                                 break;
3595                                         }
3596                                         elseif ( $levelCount[$i] < $level ) {
3597                                                 # Found first matching level below current level
3598                                                 $toclevel = $i + 1;
3599                                                 break;
3600                                         }
3601                                 }
3602                                 if( $i == 0 ) $toclevel = 1;
3603                                 if( $toclevel<$wgMaxTocLevel ) {
3604                                         if($prevtoclevel < $wgMaxTocLevel) {
3605                                                 # Unindent only if the previous toc level was shown :p
3606                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3607                                                 $prevtoclevel = $toclevel;
3608                                         } else {
3609                                                 $toc .= $sk->tocLineEnd();
3610                                         }
3611                                 }
3612                         }
3613                         else {
3614                                 # No change in level, end TOC line
3615                                 if( $toclevel<$wgMaxTocLevel ) {
3616                                         $toc .= $sk->tocLineEnd();
3617                                 }
3618                         }
3619
3620                         $levelCount[$toclevel] = $level;
3621
3622                         # count number of headlines for each level
3623                         @$sublevelCount[$toclevel]++;
3624                         $dot = 0;
3625                         for( $i = 1; $i <= $toclevel; $i++ ) {
3626                                 if( !empty( $sublevelCount[$i] ) ) {
3627                                         if( $dot ) {
3628                                                 $numbering .= '.';
3629                                         }
3630                                         $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3631                                         $dot = 1;
3632                                 }
3633                         }
3634
3635                         # The safe header is a version of the header text safe to use for links
3636                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3637                         $safeHeadline = $this->mStripState->unstripBoth( $headline );
3638
3639                         # Remove link placeholders by the link text.
3640                         #     <!--LINK number-->
3641                         # turns into
3642                         #     link text with suffix
3643                         $safeHeadline = $this->replaceLinkHoldersText( $safeHeadline );
3644
3645                         # Strip out HTML (other than plain <sup> and <sub>: bug 8393)
3646                         $tocline = preg_replace(
3647                                 array( '#<(?!/?(sup|sub)).*?'.'>#', '#<(/?(sup|sub)).*?'.'>#' ),
3648                                 array( '',                          '<$1>'),
3649                                 $safeHeadline
3650                         );
3651                         $tocline = trim( $tocline );
3652
3653                         # For the anchor, strip out HTML-y stuff period
3654                         $safeHeadline = preg_replace( '/<.*?'.'>/', '', $safeHeadline );
3655                         $safeHeadline = preg_replace( '/[ _]+/', ' ', $safeHeadline );
3656                         $safeHeadline = trim( $safeHeadline );
3657
3658                         # Save headline for section edit hint before it's escaped
3659                         $headlineHint = $safeHeadline;
3660
3661                         if ( $wgHtml5 && $wgExperimentalHtmlIds ) {
3662                                 # For reverse compatibility, provide an id that's
3663                                 # HTML4-compatible, like we used to.
3664                                 #
3665                                 # It may be worth noting, academically, that it's possible for
3666                                 # the legacy anchor to conflict with a non-legacy headline
3667                                 # anchor on the page.  In this case likely the "correct" thing
3668                                 # would be to either drop the legacy anchors or make sure
3669                                 # they're numbered first.  However, this would require people
3670                                 # to type in section names like "abc_.D7.93.D7.90.D7.A4"
3671                                 # manually, so let's not bother worrying about it.
3672                                 $legacyHeadline = Sanitizer::escapeId( $safeHeadline,
3673                                         array( 'noninitial', 'legacy' ) );
3674                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline );
3675
3676                                 if ( $legacyHeadline == $safeHeadline ) {
3677                                         # No reason to have both (in fact, we can't)
3678                                         $legacyHeadline = false;
3679                                 }
3680                         } else {
3681                                 $legacyHeadline = false;
3682                                 $safeHeadline = Sanitizer::escapeId( $safeHeadline,
3683                                         'noninitial' );
3684                         }
3685
3686                         # HTML names must be case-insensitively unique (bug 10721).  FIXME:
3687                         # Does this apply to Unicode characters?  Because we aren't
3688                         # handling those here.
3689                         $arrayKey = strtolower( $safeHeadline );
3690                         if ( $legacyHeadline === false ) {
3691                                 $legacyArrayKey = false;
3692                         } else {
3693                                 $legacyArrayKey = strtolower( $legacyHeadline );
3694                         }
3695
3696                         # count how many in assoc. array so we can track dupes in anchors
3697                         if ( isset( $refers[$arrayKey] ) ) {
3698                                 $refers[$arrayKey]++;
3699                         } else {
3700                                 $refers[$arrayKey] = 1;
3701                         }
3702                         if ( isset( $refers[$legacyArrayKey] ) ) {
3703                                 $refers[$legacyArrayKey]++;
3704                         } else {
3705                                 $refers[$legacyArrayKey] = 1;
3706                         }
3707
3708                         # Don't number the heading if it is the only one (looks silly)
3709                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3710                                 # the two are different if the line contains a link
3711                                 $headline = $numbering . ' ' . $headline;
3712                         }
3713
3714                         # Create the anchor for linking from the TOC to the section
3715                         $anchor = $safeHeadline;
3716                         $legacyAnchor = $legacyHeadline;
3717                         if ( $refers[$arrayKey] > 1 ) {
3718                                 $anchor .= '_' . $refers[$arrayKey];
3719                         }
3720                         if ( $legacyHeadline !== false && $refers[$legacyArrayKey] > 1 ) {
3721                                 $legacyAnchor .= '_' . $refers[$legacyArrayKey];
3722                         }
3723                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3724                                 $toc .= $sk->tocLine($anchor, $tocline,
3725                                         $numbering, $toclevel, ($isTemplate ? false : $sectionIndex));
3726                         }
3727
3728                         # Add the section to the section tree
3729                         # Find the DOM node for this header
3730                         while ( $node && !$isTemplate ) {
3731                                 if ( $node->getName() === 'h' ) {
3732                                         $bits = $node->splitHeading();
3733                                         if ( $bits['i'] == $sectionIndex )
3734                                                 break;
3735                                 }
3736                                 $byteOffset += mb_strlen( $this->mStripState->unstripBoth(
3737                                         $frame->expand( $node, PPFrame::RECOVER_ORIG ) ) );
3738                                 $node = $node->getNextSibling();
3739                         }
3740                         $tocraw[] = array(
3741                                 'toclevel' => $toclevel,
3742                                 'level' => $level,
3743                                 'line' => $tocline,
3744                                 'number' => $numbering,
3745                                 'index' => ($isTemplate ? 'T-' : '' ) . $sectionIndex,
3746                                 'fromtitle' => $titleText,
3747                                 'byteoffset' => ( $isTemplate ? null : $byteOffset ),
3748                                 'anchor' => $anchor,
3749                         );
3750
3751                         # give headline the correct <h#> tag
3752                         if( $showEditLink && $sectionIndex !== false ) {
3753                                 if( $isTemplate ) {
3754                                         # Put a T flag in the section identifier, to indicate to extractSections()
3755                                         # that sections inside <includeonly> should be counted.
3756                                         $editlink = $sk->doEditSectionLink(Title::newFromText( $titleText ), "T-$sectionIndex");
3757                                 } else {
3758                                         $editlink = $sk->doEditSectionLink($this->mTitle, $sectionIndex, $headlineHint);
3759                                 }
3760                         } else {
3761                                 $editlink = '';
3762                         }
3763                         $head[$headlineCount] = $sk->makeHeadline( $level,
3764                                 $matches['attrib'][$headlineCount], $anchor, $headline,
3765                                 $editlink, $legacyAnchor );
3766
3767                         $headlineCount++;
3768                 }
3769
3770                 $this->setOutputType( $oldType );
3771
3772                 # Never ever show TOC if no headers
3773                 if( $numVisible < 1 ) {
3774                         $enoughToc = false;
3775                 }
3776
3777                 if( $enoughToc ) {
3778                         if( $prevtoclevel > 0 && $prevtoclevel < $wgMaxTocLevel ) {
3779                                 $toc .= $sk->tocUnindent( $prevtoclevel - 1 );
3780                         }
3781                         $toc = $sk->tocList( $toc );
3782                         $this->mOutput->setTOCHTML( $toc );
3783                 }
3784
3785                 if ( $isMain ) {
3786                         $this->mOutput->setSections( $tocraw );
3787                 }
3788
3789                 # split up and insert constructed headlines
3790
3791                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3792                 $i = 0;
3793
3794                 foreach( $blocks as $block ) {
3795                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block !== "\n" ) {
3796                                 # This is the [edit] link that appears for the top block of text when
3797                                 # section editing is enabled
3798
3799                                 # Disabled because it broke block formatting
3800                                 # For example, a bullet point in the top line
3801                                 # $full .= $sk->editSectionLink(0);
3802                         }
3803                         $full .= $block;
3804                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3805                                 # Top anchor now in skin
3806                                 $full = $full.$toc;
3807                         }
3808
3809                         if( !empty( $head[$i] ) ) {
3810                                 $full .= $head[$i];
3811                         }
3812                         $i++;
3813                 }
3814                 if( $this->mForceTocPosition ) {
3815                         return str_replace( '<!--MWTOC-->', $toc, $full );
3816                 } else {
3817                         return $full;
3818                 }
3819         }
3820
3821         /**
3822          * Merge $tree2 into $tree1 by replacing the section with index
3823          * $section in $tree1 and its descendants with the sections in $tree2.
3824          * Note that in the returned section tree, only the 'index' and
3825          * 'byteoffset' fields are guaranteed to be correct.
3826          * @param $tree1 array Section tree from ParserOutput::getSectons()
3827         *  @param $tree2 array Section tree
3828          * @param $section int Section index
3829          * @param $title Title Title both section trees come from
3830          * @param $len2 int Length of the original wikitext for $tree2
3831          * @return array Merged section tree
3832          */
3833         public static function mergeSectionTrees( $tree1, $tree2, $section, $title, $len2 ) {
3834                 global $wgContLang;
3835                 $newTree = array();
3836                 $targetLevel = false;
3837                 $merged = false;
3838                 $lastLevel = 1;
3839                 $nextIndex = 1;
3840                 $numbering = array( 0 );
3841                 $titletext = $title->getPrefixedDBkey();
3842                 foreach ( $tree1 as $s ) {
3843                         if ( $targetLevel !== false ) {
3844                                 if ( $s['level'] <= $targetLevel )
3845                                         // We've skipped enough
3846                                         $targetLevel = false;
3847                                 else
3848                                         continue;
3849                         }
3850                         if ( $s['index'] != $section ||
3851                                         $s['fromtitle'] != $titletext ) {
3852                                 self::incrementNumbering( $numbering,
3853                                         $s['toclevel'], $lastLevel );
3854
3855                                 // Rewrite index, byteoffset and number
3856                                 if ( $s['fromtitle'] == $titletext ) {
3857                                         $s['index'] = $nextIndex++;
3858                                         if ( $merged )
3859                                                 $s['byteoffset'] += $len2;
3860                                 }
3861                                 $s['number']  = implode( '.', array_map(
3862                                         array( $wgContLang, 'formatnum' ),
3863                                         $numbering ) );
3864                                 $lastLevel = $s['toclevel'];
3865                                 $newTree[] = $s;
3866                         } else {
3867                                 // We're at $section
3868                                 // Insert sections from $tree2 here
3869                                 foreach ( $tree2 as $s2 ) {
3870                                         // Rewrite the fields in $s2
3871                                         // before inserting it
3872                                         $s2['toclevel'] += $s['toclevel'] - 1;
3873                                         $s2['level'] += $s['level'] - 1;
3874                                         $s2['index'] = $nextIndex++;
3875                                         $s2['byteoffset'] += $s['byteoffset'];
3876
3877                                         self::incrementNumbering( $numbering,
3878                                                 $s2['toclevel'], $lastLevel );
3879                                         $s2['number']  = implode( '.', array_map(
3880                                                 array( $wgContLang, 'formatnum' ),
3881                                                 $numbering ) );
3882                                         $lastLevel = $s2['toclevel'];
3883                                         $newTree[] = $s2;
3884                                 }
3885                                 // Skip all descendants of $section in $tree1
3886                                 $targetLevel = $s['level'];
3887                                 $merged = true;
3888                         }
3889                 }
3890                 return $newTree;
3891         }
3892
3893         /**
3894          * Increment a section number. Helper function for mergeSectionTrees()
3895          * @param $number array Array representing a section number
3896          * @param $level int Current TOC level (depth)
3897          * @param $lastLevel int Level of previous TOC entry
3898          */
3899         private static function incrementNumbering( &$number, $level, $lastLevel ) {
3900                 if ( $level > $lastLevel )
3901                         $number[$level - 1] = 1;
3902                 else if ( $level < $lastLevel ) {
3903                         foreach ( $number as $key => $unused )
3904                                 if ( $key >= $level )
3905                                         unset( $number[$key] );
3906                         $number[$level - 1]++;
3907                 } else
3908                         $number[$level - 1]++;
3909         }
3910
3911         /**
3912          * Transform wiki markup when saving a page by doing \r\n -> \n
3913          * conversion, substitting signatures, {{subst:}} templates, etc.
3914          *
3915          * @param string $text the text to transform
3916          * @param Title &$title the Title object for the current article
3917          * @param User $user the User object describing the current user
3918          * @param ParserOptions $options parsing options
3919          * @param bool $clearState whether to clear the parser state first
3920          * @return string the altered wiki markup
3921          * @public
3922          */
3923         function preSaveTransform( $text, Title $title, $user, $options, $clearState = true ) {
3924                 $this->mOptions = $options;
3925                 $this->setTitle( $title );
3926                 $this->setOutputType( self::OT_WIKI );
3927
3928                 if ( $clearState ) {
3929                         $this->clearState();
3930                 }
3931
3932                 $pairs = array(
3933                         "\r\n" => "\n",
3934                 );
3935                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3936                 $text = $this->pstPass2( $text, $user );
3937                 $text = $this->mStripState->unstripBoth( $text );
3938                 return $text;
3939         }
3940
3941         /**
3942          * Pre-save transform helper function
3943          * @private
3944          */
3945         function pstPass2( $text, $user ) {
3946                 global $wgContLang, $wgLocaltimezone;
3947
3948                 /* Note: This is the timestamp saved as hardcoded wikitext to
3949                  * the database, we use $wgContLang here in order to give
3950                  * everyone the same signature and use the default one rather
3951                  * than the one selected in each user's preferences.
3952                  *
3953                  * (see also bug 12815)
3954                  */
3955                 $ts = $this->mOptions->getTimestamp();
3956                 if ( isset( $wgLocaltimezone ) ) {
3957                         $tz = $wgLocaltimezone;
3958                 } else {
3959                         $tz = date_default_timezone_get();
3960                 }
3961
3962                 $unixts = wfTimestamp( TS_UNIX, $ts );
3963                 $oldtz = date_default_timezone_get();
3964                 date_default_timezone_set( $tz );
3965                 $ts = date( 'YmdHis', $unixts );
3966                 $tzMsg = date( 'T', $unixts );  # might vary on DST changeover!
3967
3968                 /* Allow translation of timezones trough wiki. date() can return
3969                  * whatever crap the system uses, localised or not, so we cannot
3970                  * ship premade translations.
3971                  */
3972                 $key = 'timezone-' . strtolower( trim( $tzMsg ) );
3973                 $value = wfMsgForContent( $key );
3974                 if ( !wfEmptyMsg( $key, $value ) ) $tzMsg = $value;
3975
3976                 date_default_timezone_set( $oldtz );
3977
3978                 $d = $wgContLang->timeanddate( $ts, false, false ) . " ($tzMsg)";
3979
3980                 # Variable replacement
3981                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3982                 $text = $this->replaceVariables( $text );
3983
3984                 # Signatures
3985                 $sigText = $this->getUserSig( $user );
3986                 $text = strtr( $text, array(
3987                         '~~~~~' => $d,
3988                         '~~~~' => "$sigText $d",
3989                         '~~~' => $sigText
3990                 ) );
3991
3992                 # Context links: [[|name]] and [[name (context)|]]
3993                 #
3994                 global $wgLegalTitleChars;
3995                 $tc = "[$wgLegalTitleChars]";
3996                 $nc = '[ _0-9A-Za-z\x80-\xff-]'; # Namespaces can use non-ascii!
3997
3998                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3999                 $p4 = "/\[\[(:?$nc+:|:|)($tc+?)(（$tc+）)\\|]]/";             # [[ns:page（context）|]]
4000                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
4001                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
4002
4003                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
4004                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
4005                 $text = preg_replace( $p4, '[[\\1\\2\\3|\\2]]', $text );
4006                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
4007
4008                 $t = $this->mTitle->getText();
4009                 $m = array();
4010                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
4011                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4012                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && "$m[1]$m[2]" != '' ) {
4013                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
4014                 } else {
4015                         # if there's no context, don't bother duplicating the title
4016                         $text = preg_replace( $p2, '[[\\1]]', $text );
4017                 }
4018
4019                 # Trim trailing whitespace
4020                 $text = rtrim( $text );
4021
4022                 return $text;
4023         }
4024
4025         /**
4026          * Fetch the user's signature text, if any, and normalize to
4027          * validated, ready-to-insert wikitext.
4028          * If you have pre-fetched the nickname or the fancySig option, you can
4029          * specify them here to save a database query.
4030          *
4031          * @param User $user
4032          * @return string
4033          */
4034         function getUserSig( &$user, $nickname = false, $fancySig = null ) {
4035                 global $wgMaxSigChars;
4036
4037                 $username = $user->getName();
4038
4039                 // If not given, retrieve from the user object.
4040                 if ( $nickname === false )
4041                         $nickname = $user->getOption( 'nickname' );
4042
4043                 if ( is_null( $fancySig ) )
4044                         $fancySig = $user->getBoolOption( 'fancysig' );
4045
4046                 $nickname = $nickname == null ? $username : $nickname;
4047
4048                 if( mb_strlen( $nickname ) > $wgMaxSigChars ) {
4049                         $nickname = $username;
4050                         wfDebug( __METHOD__ . ": $username has overlong signature.\n" );
4051                 } elseif( $fancySig !== false ) {
4052                         # Sig. might contain markup; validate this
4053                         if( $this->validateSig( $nickname ) !== false ) {
4054                                 # Validated; clean up (if needed) and return it
4055                                 return $this->cleanSig( $nickname, true );
4056                         } else {
4057                                 # Failed to validate; fall back to the default
4058                                 $nickname = $username;
4059                                 wfDebug( __METHOD__.": $username has bad XML tags in signature.\n" );
4060                         }
4061                 }
4062
4063                 // Make sure nickname doesnt get a sig in a sig
4064                 $nickname = $this->cleanSigInSig( $nickname );
4065
4066                 # If we're still here, make it a link to the user page
4067                 $userText = wfEscapeWikiText( $username );
4068                 $nickText = wfEscapeWikiText( $nickname );
4069                 if ( $user->isAnon() )  {
4070                         return wfMsgExt( 'signature-anon', array( 'content', 'parsemag' ), $userText, $nickText );
4071                 } else {
4072                         return wfMsgExt( 'signature', array( 'content', 'parsemag' ), $userText, $nickText );
4073                 }
4074         }
4075
4076         /**
4077          * Check that the user's signature contains no bad XML
4078          *
4079          * @param string $text
4080          * @return mixed An expanded string, or false if invalid.
4081          */
4082         function validateSig( $text ) {
4083                 return( Xml::isWellFormedXmlFragment( $text ) ? $text : false );
4084         }
4085
4086         /**
4087          * Clean up signature text
4088          *
4089          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
4090          * 2) Substitute all transclusions
4091          *
4092          * @param string $text
4093          * @param $parsing Whether we're cleaning (preferences save) or parsing
4094          * @return string Signature text
4095          */
4096         function cleanSig( $text, $parsing = false ) {
4097                 if ( !$parsing ) {
4098                         global $wgTitle;
4099                         $this->clearState();
4100                         $this->setTitle( $wgTitle );
4101                         $this->mOptions = new ParserOptions;
4102                         $this->setOutputType = self::OT_PREPROCESS;
4103                 }
4104
4105                 # Option to disable this feature
4106                 if ( !$this->mOptions->getCleanSignatures() ) {
4107                         return $text;
4108                 }
4109
4110                 # FIXME: regex doesn't respect extension tags or nowiki
4111                 #  => Move this logic to braceSubstitution()
4112                 $substWord = MagicWord::get( 'subst' );
4113                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
4114                 $substText = '{{' . $substWord->getSynonym( 0 );
4115
4116                 $text = preg_replace( $substRegex, $substText, $text );
4117                 $text = $this->cleanSigInSig( $text );
4118                 $dom = $this->preprocessToDom( $text );
4119                 $frame = $this->getPreprocessor()->newFrame();
4120                 $text = $frame->expand( $dom );
4121
4122                 if ( !$parsing ) {
4123                         $text = $this->mStripState->unstripBoth( $text );
4124                 }
4125
4126                 return $text;
4127         }
4128
4129         /**
4130          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
4131          * @param string $text
4132          * @return string Signature text with /~{3,5}/ removed
4133          */
4134         function cleanSigInSig( $text ) {
4135                 $text = preg_replace( '/~{3,5}/', '', $text );
4136                 return $text;
4137         }
4138
4139         /**
4140          * Set up some variables which are usually set up in parse()
4141          * so that an external function can call some class members with confidence
4142          * @public
4143          */
4144         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
4145                 $this->setTitle( $title );
4146                 $this->mOptions = $options;
4147                 $this->setOutputType( $outputType );
4148                 if ( $clearState ) {
4149                         $this->clearState();
4150                 }
4151         }
4152
4153         /**
4154          * Wrapper for preprocess()
4155          *
4156          * @param string $text the text to preprocess
4157          * @param ParserOptions $options  options
4158          * @return string
4159          * @public
4160          */
4161         function transformMsg( $text, $options ) {
4162                 global $wgTitle;
4163                 static $executing = false;
4164
4165                 # Guard against infinite recursion
4166                 if ( $executing ) {
4167                         return $text;
4168                 }
4169                 $executing = true;
4170
4171                 wfProfileIn(__METHOD__);
4172                 $text = $this->preprocess( $text, $wgTitle, $options );
4173
4174                 $executing = false;
4175                 wfProfileOut(__METHOD__);
4176                 return $text;
4177         }
4178
4179         /**
4180          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
4181          * The callback should have the following form:
4182          *    function myParserHook( $text, $params, &$parser ) { ... }
4183          *
4184          * Transform and return $text. Use $parser for any required context, e.g. use
4185          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
4186          *
4187          * @public
4188          *
4189          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
4190          * @param mixed $callback The callback function (and object) to use for the tag
4191          *
4192          * @return The old value of the mTagHooks array associated with the hook
4193          */
4194         function setHook( $tag, $callback ) {
4195                 $tag = strtolower( $tag );
4196                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
4197                 $this->mTagHooks[$tag] = $callback;
4198                 if( !in_array( $tag, $this->mStripList ) ) {
4199                         $this->mStripList[] = $tag;
4200                 }
4201
4202                 return $oldVal;
4203         }
4204
4205         function setTransparentTagHook( $tag, $callback ) {
4206                 $tag = strtolower( $tag );
4207                 $oldVal = isset( $this->mTransparentTagHooks[$tag] ) ? $this->mTransparentTagHooks[$tag] : null;
4208                 $this->mTransparentTagHooks[$tag] = $callback;
4209
4210                 return $oldVal;
4211         }
4212
4213         /**
4214          * Remove all tag hooks
4215          */
4216         function clearTagHooks() {
4217                 $this->mTagHooks = array();
4218                 $this->mStripList = $this->mDefaultStripList;
4219         }
4220
4221         /**
4222          * Create a function, e.g. {{sum:1|2|3}}
4223          * The callback function should have the form:
4224          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
4225          *
4226          * Or with SFH_OBJECT_ARGS:
4227          *    function myParserFunction( $parser, $frame, $args ) { ... }
4228          *
4229          * The callback may either return the text result of the function, or an array with the text
4230          * in element 0, and a number of flags in the other elements. The names of the flags are
4231          * specified in the keys. Valid flags are:
4232          *   found                     The text returned is valid, stop processing the template. This
4233          *                             is on by default.
4234          *   nowiki                    Wiki markup in the return value should be escaped
4235          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
4236          *
4237          * @public
4238          *
4239          * @param string $id The magic word ID
4240          * @param mixed $callback The callback function (and object) to use
4241          * @param integer $flags a combination of the following flags:
4242          *     SFH_NO_HASH   No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
4243          *
4244          *     SFH_OBJECT_ARGS   Pass the template arguments as PPNode objects instead of text. This
4245          *     allows for conditional expansion of the parse tree, allowing you to eliminate dead
4246          *     branches and thus speed up parsing. It is also possible to analyse the parse tree of
4247          *     the arguments, and to control the way they are expanded.
4248          *
4249          *     The $frame parameter is a PPFrame. This can be used to produce expanded text from the
4250          *     arguments, for instance:
4251          *         $text = isset( $args[0] ) ? $frame->expand( $args[0] ) : '';
4252          *
4253          *     For technical reasons, $args[0] is pre-expanded and will be a string. This may change in
4254          *     future versions. Please call $frame->expand() on it anyway so that your code keeps
4255          *     working if/when this is changed.
4256          *
4257          *     If you want whitespace to be trimmed from $args, you need to do it yourself, post-
4258          *     expansion.
4259          *
4260          *     Please read the documentation in includes/parser/Preprocessor.php for more information
4261          *     about the methods available in PPFrame and PPNode.
4262          *
4263          * @return The old callback function for this name, if any
4264          */
4265         function setFunctionHook( $id, $callback, $flags = 0 ) {
4266                 global $wgContLang;
4267
4268                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id][0] : null;
4269                 $this->mFunctionHooks[$id] = array( $callback, $flags );
4270
4271                 # Add to function cache
4272                 $mw = MagicWord::get( $id );
4273                 if( !$mw )
4274                         throw new MWException( __METHOD__.'() expecting a magic word identifier.' );
4275
4276                 $synonyms = $mw->getSynonyms();
4277                 $sensitive = intval( $mw->isCaseSensitive() );
4278
4279                 foreach ( $synonyms as $syn ) {
4280                         # Case
4281                         if ( !$sensitive ) {
4282                                 $syn = $wgContLang->lc( $syn );
4283                         }
4284                         # Add leading hash
4285                         if ( !( $flags & SFH_NO_HASH ) ) {
4286                                 $syn = '#' . $syn;
4287                         }
4288                         # Remove trailing colon
4289                         if ( substr( $syn, -1, 1 ) === ':' ) {
4290                                 $syn = substr( $syn, 0, -1 );
4291                         }
4292                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
4293                 }
4294                 return $oldVal;
4295         }
4296
4297         /**
4298          * Get all registered function hook identifiers
4299          *
4300          * @return array
4301          */
4302         function getFunctionHooks() {
4303                 return array_keys( $this->mFunctionHooks );
4304         }
4305
4306         /**
4307          * Create a tag function, e.g. <test>some stuff</test>.
4308          * Unlike tag hooks, tag functions are parsed at preprocessor level.
4309          * Unlike parser functions, their content is not preprocessed.
4310          */
4311         function setFunctionTagHook( $tag, $callback, $flags ) {
4312                 $tag = strtolower( $tag );
4313                 $old = isset( $this->mFunctionTagHooks[$tag] ) ?
4314                         $this->mFunctionTagHooks[$tag] : null;
4315                 $this->mFunctionTagHooks[$tag] = array( $callback, $flags );
4316
4317                 if( !in_array( $tag, $this->mStripList ) ) {
4318                         $this->mStripList[] = $tag;
4319                 }
4320
4321                 return $old;
4322         }
4323
4324         /**
4325          * FIXME: update documentation. makeLinkObj() is deprecated.
4326          * Replace <!--LINK--> link placeholders with actual links, in the buffer
4327          * Placeholders created in Skin::makeLinkObj()
4328          * Returns an array of link CSS classes, indexed by PDBK.
4329          */
4330         function replaceLinkHolders( &$text, $options = 0 ) {
4331                 return $this->mLinkHolders->replace( $text );
4332         }
4333
4334         /**
4335          * Replace <!--LINK--> link placeholders with plain text of links
4336          * (not HTML-formatted).
4337          * @param string $text
4338          * @return string
4339          */
4340         function replaceLinkHoldersText( $text ) {
4341                 return $this->mLinkHolders->replaceText( $text );
4342         }
4343
4344         /**
4345          * Tag hook handler for 'pre'.
4346          */
4347         function renderPreTag( $text, $attribs ) {
4348                 // Backwards-compatibility hack
4349                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4350
4351                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4352                 return Xml::openElement( 'pre', $attribs ) .
4353                         Xml::escapeTagsOnly( $content ) .
4354                         '</pre>';
4355         }
4356
4357         /**
4358          * Renders an image gallery from a text with one line per image.
4359          * text labels may be given by using |-style alternative text. E.g.
4360          *   Image:one.jpg|The number "1"
4361          *   Image:tree.jpg|A tree
4362          * given as text will return the HTML of a gallery with two images,
4363          * labeled 'The number "1"' and
4364          * 'A tree'.
4365          */
4366         function renderImageGallery( $text, $params ) {
4367                 $ig = new ImageGallery();
4368                 $ig->setContextTitle( $this->mTitle );
4369                 $ig->setShowBytes( false );
4370                 $ig->setShowFilename( false );
4371                 $ig->setParser( $this );
4372                 $ig->setHideBadImages();
4373                 $ig->setAttributes( Sanitizer::validateTagAttributes( $params, 'table' ) );
4374                 $ig->useSkin( $this->mOptions->getSkin() );
4375                 $ig->mRevisionId = $this->mRevisionId;
4376
4377                 if( isset( $params['caption'] ) ) {
4378                         $caption = $params['caption'];
4379                         $caption = htmlspecialchars( $caption );
4380                         $caption = $this->replaceInternalLinks( $caption );
4381                         $ig->setCaptionHtml( $caption );
4382                 }
4383                 if( isset( $params['perrow'] ) ) {
4384                         $ig->setPerRow( $params['perrow'] );
4385                 }
4386                 if( isset( $params['widths'] ) ) {
4387                         $ig->setWidths( $params['widths'] );
4388                 }
4389                 if( isset( $params['heights'] ) ) {
4390                         $ig->setHeights( $params['heights'] );
4391                 }
4392
4393                 wfRunHooks( 'BeforeParserrenderImageGallery', array( &$this, &$ig ) );
4394
4395                 $lines = StringUtils::explode( "\n", $text );
4396                 foreach ( $lines as $line ) {
4397                         # match lines like these:
4398                         # Image:someimage.jpg|This is some image
4399                         $matches = array();
4400                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4401                         # Skip empty lines
4402                         if ( count( $matches ) == 0 ) {
4403                                 continue;
4404                         }
4405
4406                         if ( strpos( $matches[0], '%' ) !== false )
4407                                 $matches[1] = urldecode( $matches[1] );
4408                         $tp = Title::newFromText( $matches[1]/*, NS_FILE*/ );
4409                         $nt =& $tp;
4410                         if( is_null( $nt ) ) {
4411                                 # Bogus title. Ignore these so we don't bomb out later.
4412                                 continue;
4413                         }
4414                         if ( isset( $matches[3] ) ) {
4415                                 $label = $matches[3];
4416                         } else {
4417                                 $label = '';
4418                         }
4419
4420                         $html = $this->recursiveTagParse( trim( $label ) );
4421
4422                         $ig->add( $nt, $html );
4423
4424                         # Only add real images (bug #5586)
4425                         if ( $nt->getNamespace() == NS_FILE ) {
4426                                 $this->mOutput->addImage( $nt->getDBkey() );
4427                         }
4428                 }
4429                 return $ig->toHTML();
4430         }
4431
4432         function getImageParams( $handler ) {
4433                 if ( $handler ) {
4434                         $handlerClass = get_class( $handler );
4435                 } else {
4436                         $handlerClass = '';
4437                 }
4438                 if ( !isset( $this->mImageParams[$handlerClass]  ) ) {
4439                         // Initialise static lists
4440                         static $internalParamNames = array(
4441                                 'horizAlign' => array( 'left', 'right', 'center', 'none' ),
4442                                 'vertAlign' => array( 'baseline', 'sub', 'super', 'top', 'text-top', 'middle',
4443                                         'bottom', 'text-bottom' ),
4444                                 'frame' => array( 'thumbnail', 'manualthumb', 'framed', 'frameless',
4445                                         'upright', 'border', 'link', 'alt' ),
4446                         );
4447                         static $internalParamMap;
4448                         if ( !$internalParamMap ) {
4449                                 $internalParamMap = array();
4450                                 foreach ( $internalParamNames as $type => $names ) {
4451                                         foreach ( $names as $name ) {
4452                                                 $magicName = str_replace( '-', '_', "img_$name" );
4453                                                 $internalParamMap[$magicName] = array( $type, $name );
4454                                         }
4455                                 }
4456                         }
4457
4458                         // Add handler params
4459                         $paramMap = $internalParamMap;
4460                         if ( $handler ) {
4461                                 $handlerParamMap = $handler->getParamMap();
4462                                 foreach ( $handlerParamMap as $magic => $paramName ) {
4463                                         $paramMap[$magic] = array( 'handler', $paramName );
4464                                 }
4465                         }
4466                         $this->mImageParams[$handlerClass] = $paramMap;
4467                         $this->mImageParamsMagicArray[$handlerClass] = new MagicWordArray( array_keys( $paramMap ) );
4468                 }
4469                 return array( $this->mImageParams[$handlerClass], $this->mImageParamsMagicArray[$handlerClass] );
4470         }
4471
4472         /**
4473          * Parse image options text and use it to make an image
4474          * @param Title $title
4475          * @param string $options
4476          * @param LinkHolderArray $holders
4477          */
4478         function makeImage( $title, $options, $holders = false ) {
4479                 # Check if the options text is of the form "options|alt text"
4480                 # Options are:
4481                 #  * thumbnail  make a thumbnail with enlarge-icon and caption, alignment depends on lang
4482                 #  * left       no resizing, just left align. label is used for alt= only
4483                 #  * right      same, but right aligned
4484                 #  * none       same, but not aligned
4485                 #  * ___px      scale to ___ pixels width, no aligning. e.g. use in taxobox
4486                 #  * center     center the image
4487                 #  * frame      Keep original image size, no magnify-button.
4488                 #  * framed     Same as "frame"
4489                 #  * frameless  like 'thumb' but without a frame. Keeps user preferences for width
4490                 #  * upright    reduce width for upright images, rounded to full __0 px
4491                 #  * border     draw a 1px border around the image
4492                 #  * alt        Text for HTML alt attribute (defaults to empty)
4493                 #  * link       Set the target of the image link. Can be external, interwiki, or local
4494                 # vertical-align values (no % or length right now):
4495                 #  * baseline
4496                 #  * sub
4497                 #  * super
4498                 #  * top
4499                 #  * text-top
4500                 #  * middle
4501                 #  * bottom
4502                 #  * text-bottom
4503
4504                 $parts = StringUtils::explode( "|", $options );
4505                 $sk = $this->mOptions->getSkin();
4506
4507                 # Give extensions a chance to select the file revision for us
4508                 $skip = $time = $descQuery = false;
4509                 wfRunHooks( 'BeforeParserMakeImageLinkObj', array( &$this, &$title, &$skip, &$time, &$descQuery ) );
4510
4511                 if ( $skip ) {
4512                         return $sk->link( $title );
4513                 }
4514
4515                 # Get the file
4516                 $imagename = $title->getDBkey();
4517                 $file = wfFindFile( $title, array( 'time' => $time ) );
4518                 # Get parameter map
4519                 $handler = $file ? $file->getHandler() : false;
4520
4521                 list( $paramMap, $mwArray ) = $this->getImageParams( $handler );
4522
4523                 # Process the input parameters
4524                 $caption = '';
4525                 $params = array( 'frame' => array(), 'handler' => array(),
4526                         'horizAlign' => array(), 'vertAlign' => array() );
4527                 foreach( $parts as $part ) {
4528                         $part = trim( $part );
4529                         list( $magicName, $value ) = $mwArray->matchVariableStartToEnd( $part );
4530                         $validated = false;
4531                         if( isset( $paramMap[$magicName] ) ) {
4532                                 list( $type, $paramName ) = $paramMap[$magicName];
4533
4534                                 // Special case; width and height come in one variable together
4535                                 if( $type === 'handler' && $paramName === 'width' ) {
4536                                         $m = array();
4537                                         # (bug 13500) In both cases (width/height and width only),
4538                                         # permit trailing "px" for backward compatibility.
4539                                         if ( preg_match( '/^([0-9]*)x([0-9]*)\s*(?:px)?\s*$/', $value, $m ) ) {
4540                                                 $width = intval( $m[1] );
4541                                                 $height = intval( $m[2] );
4542                                                 if ( $handler->validateParam( 'width', $width ) ) {
4543                                                         $params[$type]['width'] = $width;
4544                                                         $validated = true;
4545                                                 }
4546                                                 if ( $handler->validateParam( 'height', $height ) ) {
4547                                                         $params[$type]['height'] = $height;
4548                                                         $validated = true;
4549                                                 }
4550                                         } elseif ( preg_match( '/^[0-9]*\s*(?:px)?\s*$/', $value ) ) {
4551                                                 $width = intval( $value );
4552                                                 if ( $handler->validateParam( 'width', $width ) ) {
4553                                                         $params[$type]['width'] = $width;
4554                                                         $validated = true;
4555                                                 }
4556                                         } // else no validation -- bug 13436
4557                                 } else {
4558                                         if ( $type === 'handler' ) {
4559                                                 # Validate handler parameter
4560                                                 $validated = $handler->validateParam( $paramName, $value );
4561                                         } else {
4562                                                 # Validate internal parameters
4563                                                 switch( $paramName ) {
4564                                                 case 'manualthumb':
4565                                                 case 'alt':
4566                                                         // @todo Fixme: possibly check validity here for
4567                                                         // manualthumb? downstream behavior seems odd with
4568                                                         // missing manual thumbs.
4569                                                         $validated = true;
4570                                                         $value = $this->stripAltText( $value, $holders );
4571                                                         break;
4572                                                 case 'link':
4573                                                         $chars = self::EXT_LINK_URL_CLASS;
4574                                                         $prots = $this->mUrlProtocols;
4575                                                         if ( $value === '' ) {
4576                                                                 $paramName = 'no-link';
4577                                                                 $value = true;
4578                                                                 $validated = true;
4579                                                         } elseif ( preg_match( "/^$prots/", $value ) ) {
4580                                                                 if ( preg_match( "/^($prots)$chars+$/", $value, $m ) ) {
4581                                                                         $paramName = 'link-url';
4582                                                                         $this->mOutput->addExternalLink( $value );
4583                                                                         $validated = true;
4584                                                                 }
4585                                                         } else {
4586                                                                 $linkTitle = Title::newFromText( $value );
4587                                                                 if ( $linkTitle ) {
4588                                                                         $paramName = 'link-title';
4589                                                                         $value = $linkTitle;
4590                                                                         $this->mOutput->addLink( $linkTitle );
4591                                                                         $validated = true;
4592                                                                 }
4593                                                         }
4594                                                         break;
4595                                                 default:
4596                                                         // Most other things appear to be empty or numeric...
4597                                                         $validated = ( $value === false || is_numeric( trim( $value ) ) );
4598                                                 }
4599                                         }
4600
4601                                         if ( $validated ) {
4602                                                 $params[$type][$paramName] = $value;
4603                                         }
4604                                 }
4605                         }
4606                         if ( !$validated ) {
4607                                 $caption = $part;
4608                         }
4609                 }
4610
4611                 # Process alignment parameters
4612                 if ( $params['horizAlign'] ) {
4613                         $params['frame']['align'] = key( $params['horizAlign'] );
4614                 }
4615                 if ( $params['vertAlign'] ) {
4616                         $params['frame']['valign'] = key( $params['vertAlign'] );
4617                 }
4618
4619                 $params['frame']['caption'] = $caption;
4620
4621                 # Will the image be presented in a frame, with the caption below?
4622                 $imageIsFramed = isset( $params['frame']['frame'] ) ||
4623                                  isset( $params['frame']['framed'] ) ||
4624                                  isset( $params['frame']['thumbnail'] ) ||
4625                                  isset( $params['frame']['manualthumb'] );
4626
4627                 # In the old days, [[Image:Foo|text...]] would set alt text.  Later it
4628                 # came to also set the caption, ordinary text after the image -- which
4629                 # makes no sense, because that just repeats the text multiple times in
4630                 # screen readers.  It *also* came to set the title attribute.
4631                 #
4632                 # Now that we have an alt attribute, we should not set the alt text to
4633                 # equal the caption: that's worse than useless, it just repeats the
4634                 # text.  This is the framed/thumbnail case.  If there's no caption, we
4635                 # use the unnamed parameter for alt text as well, just for the time be-
4636                 # ing, if the unnamed param is set and the alt param is not.
4637                 #
4638                 # For the future, we need to figure out if we want to tweak this more,
4639                 # e.g., introducing a title= parameter for the title; ignoring the un-
4640                 # named parameter entirely for images without a caption; adding an ex-
4641                 # plicit caption= parameter and preserving the old magic unnamed para-
4642                 # meter for BC; ...
4643                 if ( $imageIsFramed ) { # Framed image
4644                         if ( $caption === '' && !isset( $params['frame']['alt'] ) ) {
4645                                 # No caption or alt text, add the filename as the alt text so
4646                                 # that screen readers at least get some description of the image
4647                                 $params['frame']['alt'] = $title->getText();
4648                         }
4649                         # Do not set $params['frame']['title'] because tooltips don't make sense
4650                         # for framed images
4651                 } else { # Inline image
4652                         if ( !isset( $params['frame']['alt'] ) ) {
4653                                 # No alt text, use the "caption" for the alt text
4654                                 if ( $caption !== '') {
4655                                         $params['frame']['alt'] = $this->stripAltText( $caption, $holders );
4656                                 } else {
4657                                         # No caption, fall back to using the filename for the
4658                                         # alt text
4659                                         $params['frame']['alt'] = $title->getText();
4660                                 }
4661                         }
4662                         # Use the "caption" for the tooltip text
4663                         $params['frame']['title'] = $this->stripAltText( $caption, $holders );
4664                 }
4665
4666                 wfRunHooks( 'ParserMakeImageParams', array( $title, $file, &$params ) );
4667
4668                 # Linker does the rest
4669                 $ret = $sk->makeImageLink2( $title, $file, $params['frame'], $params['handler'], $time, $descQuery );
4670
4671                 # Give the handler a chance to modify the parser object
4672                 if ( $handler ) {
4673                         $handler->parserTransformHook( $this, $file );
4674                 }
4675
4676                 return $ret;
4677         }
4678
4679         protected function stripAltText( $caption, $holders ) {
4680                 # Strip bad stuff out of the title (tooltip).  We can't just use
4681                 # replaceLinkHoldersText() here, because if this function is called
4682                 # from replaceInternalLinks2(), mLinkHolders won't be up-to-date.
4683                 if ( $holders ) {
4684                         $tooltip = $holders->replaceText( $caption );
4685                 } else {
4686                         $tooltip = $this->replaceLinkHoldersText( $caption );
4687                 }
4688
4689                 # make sure there are no placeholders in thumbnail attributes
4690                 # that are later expanded to html- so expand them now and
4691                 # remove the tags
4692                 $tooltip = $this->mStripState->unstripBoth( $tooltip );
4693                 $tooltip = Sanitizer::stripAllTags( $tooltip );
4694
4695                 return $tooltip;
4696         }
4697
4698         /**
4699          * Set a flag in the output object indicating that the content is dynamic and
4700          * shouldn't be cached.
4701          */
4702         function disableCache() {
4703                 wfDebug( "Parser output marked as uncacheable.\n" );
4704                 $this->mOutput->mCacheTime = -1;
4705         }
4706
4707         /**#@+
4708          * Callback from the Sanitizer for expanding items found in HTML attribute
4709          * values, so they can be safely tested and escaped.
4710          * @param string $text
4711          * @param PPFrame $frame
4712          * @return string
4713          * @private
4714          */
4715         function attributeStripCallback( &$text, $frame = false ) {
4716                 $text = $this->replaceVariables( $text, $frame );
4717                 $text = $this->mStripState->unstripBoth( $text );
4718                 return $text;
4719         }
4720
4721         /**#@-*/
4722
4723         /**#@+
4724          * Accessor/mutator
4725          */
4726         function Title( $x = null ) { return wfSetVar( $this->mTitle, $x ); }
4727         function Options( $x = null ) { return wfSetVar( $this->mOptions, $x ); }
4728         function OutputType( $x = null ) { return wfSetVar( $this->mOutputType, $x ); }
4729         /**#@-*/
4730
4731         /**#@+
4732          * Accessor
4733          */
4734         function getTags() { return array_merge( array_keys($this->mTransparentTagHooks), array_keys( $this->mTagHooks ) ); }
4735         /**#@-*/
4736
4737
4738         /**
4739          * Break wikitext input into sections, and either pull or replace
4740          * some particular section's text.
4741          *
4742          * External callers should use the getSection and replaceSection methods.
4743          *
4744          * @param string $text Page wikitext
4745          * @param string $section A section identifier string of the form:
4746          *   <flag1> - <flag2> - ... - <section number>
4747          *
4748          * Currently the only recognised flag is "T", which means the target section number
4749          * was derived during a template inclusion parse, in other words this is a template
4750          * section edit link. If no flags are given, it was an ordinary section edit link.
4751          * This flag is required to avoid a section numbering mismatch when a section is
4752          * enclosed by <includeonly> (bug 6563).
4753          *
4754          * The section number 0 pulls the text before the first heading; other numbers will
4755          * pull the given section along with its lower-level subsections. If the section is
4756          * not found, $mode=get will return $newtext, and $mode=replace will return $text.
4757          *
4758          * @param string $mode One of "get" or "replace"
4759          * @param string $newText Replacement text for section data.
4760          * @return string for "get", the extracted section text.
4761          *                for "replace", the whole page with the section replaced.
4762          */
4763         private function extractSections( $text, $section, $mode, $newText='' ) {
4764                 global $wgTitle;
4765                 $this->clearState();
4766                 $this->setTitle( $wgTitle ); // not generally used but removes an ugly failure mode
4767                 $this->mOptions = new ParserOptions;
4768                 $this->setOutputType( self::OT_WIKI );
4769                 $outText = '';
4770                 $frame = $this->getPreprocessor()->newFrame();
4771
4772                 // Process section extraction flags
4773                 $flags = 0;
4774                 $sectionParts = explode( '-', $section );
4775                 $sectionIndex = array_pop( $sectionParts );
4776                 foreach ( $sectionParts as $part ) {
4777                         if ( $part === 'T' ) {
4778                                 $flags |= self::PTD_FOR_INCLUSION;
4779                         }
4780                 }
4781                 // Preprocess the text
4782                 $root = $this->preprocessToDom( $text, $flags );
4783
4784                 // <h> nodes indicate section breaks
4785                 // They can only occur at the top level, so we can find them by iterating the root's children
4786                 $node = $root->getFirstChild();
4787
4788                 // Find the target section
4789                 if ( $sectionIndex == 0 ) {
4790                         // Section zero doesn't nest, level=big
4791                         $targetLevel = 1000;
4792                 } else {
4793             while ( $node ) {
4794                 if ( $node->getName() === 'h' ) {
4795                     $bits = $node->splitHeading();
4796                                         if ( $bits['i'] == $sectionIndex ) {
4797                                         $targetLevel = $bits['level'];
4798                                                 break;
4799                                         }
4800                                 }
4801                                 if ( $mode === 'replace' ) {
4802                                         $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4803                                 }
4804                                 $node = $node->getNextSibling();
4805                         }
4806                 }
4807
4808                 if ( !$node ) {
4809                         // Not found
4810                         if ( $mode === 'get' ) {
4811                                 return $newText;
4812                         } else {
4813                                 return $text;
4814                         }
4815                 }
4816
4817                 // Find the end of the section, including nested sections
4818                 do {
4819                         if ( $node->getName() === 'h' ) {
4820                                 $bits = $node->splitHeading();
4821                                 $curLevel = $bits['level'];
4822                                 if ( $bits['i'] != $sectionIndex && $curLevel <= $targetLevel ) {
4823                                         break;
4824                                 }
4825                         }
4826                         if ( $mode === 'get' ) {
4827                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4828                         }
4829                         $node = $node->getNextSibling();
4830                 } while ( $node );
4831
4832                 // Write out the remainder (in replace mode only)
4833                 if ( $mode === 'replace' ) {
4834                         // Output the replacement text
4835                         // Add two newlines on -- trailing whitespace in $newText is conventionally
4836                         // stripped by the editor, so we need both newlines to restore the paragraph gap
4837                         // Only add trailing whitespace if there is newText
4838                         if($newText != "") {
4839                                 $outText .= $newText . "\n\n";
4840                         }
4841
4842                         while ( $node ) {
4843                                 $outText .= $frame->expand( $node, PPFrame::RECOVER_ORIG );
4844                                 $node = $node->getNextSibling();
4845                         }
4846                 }
4847
4848                 if ( is_string( $outText ) ) {
4849                         // Re-insert stripped tags
4850                         $outText = rtrim( $this->mStripState->unstripBoth( $outText ) );
4851                 }
4852
4853                 return $outText;
4854         }
4855
4856         /**
4857          * This function returns the text of a section, specified by a number ($section).
4858          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4859          * the first section before any such heading (section 0).
4860          *
4861          * If a section contains subsections, these are also returned.
4862          *
4863          * @param string $text text to look in
4864          * @param string $section section identifier
4865          * @param string $deftext default to return if section is not found
4866          * @return string text of the requested section
4867          */
4868         public function getSection( $text, $section, $deftext='' ) {
4869                 return $this->extractSections( $text, $section, "get", $deftext );
4870         }
4871
4872         public function replaceSection( $oldtext, $section, $text ) {
4873                 return $this->extractSections( $oldtext, $section, "replace", $text );
4874         }
4875
4876         /**
4877          * Get the timestamp associated with the current revision, adjusted for
4878          * the default server-local timestamp
4879          */
4880         function getRevisionTimestamp() {
4881                 if ( is_null( $this->mRevisionTimestamp ) ) {
4882                         wfProfileIn( __METHOD__ );
4883                         global $wgContLang;
4884                         $dbr = wfGetDB( DB_SLAVE );
4885                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4886                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4887
4888                         // Normalize timestamp to internal MW format for timezone processing.
4889                         // This has the added side-effect of replacing a null value with
4890                         // the current time, which gives us more sensible behavior for
4891                         // previews.
4892                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4893
4894                         // The cryptic '' timezone parameter tells to use the site-default
4895                         // timezone offset instead of the user settings.
4896                         //
4897                         // Since this value will be saved into the parser cache, served
4898                         // to other users, and potentially even used inside links and such,
4899                         // it needs to be consistent for all visitors.
4900                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4901
4902                         wfProfileOut( __METHOD__ );
4903                 }
4904                 return $this->mRevisionTimestamp;
4905         }
4906
4907         /**
4908          * Get the name of the user that edited the last revision
4909          */
4910         function getRevisionUser() {
4911                 // if this template is subst: the revision id will be blank,
4912                 // so just use the current user's name
4913                 if( $this->mRevisionId ) {
4914                         $revision = Revision::newFromId( $this->mRevisionId );
4915                         $revuser = $revision->getUserText();
4916                 } else {
4917                         global $wgUser;
4918                         $revuser = $wgUser->getName();
4919                 }
4920                 return $revuser;
4921         }
4922
4923         /**
4924          * Mutator for $mDefaultSort
4925          *
4926          * @param $sort New value
4927          */
4928         public function setDefaultSort( $sort ) {
4929                 $this->mDefaultSort = $sort;
4930         }
4931
4932         /**
4933          * Accessor for $mDefaultSort
4934          * Will use the title/prefixed title if none is set
4935          *
4936          * @return string
4937          */
4938         public function getDefaultSort() {
4939                 global $wgCategoryPrefixedDefaultSortkey;
4940                 if( $this->mDefaultSort !== false ) {
4941                         return $this->mDefaultSort;
4942                 } elseif ($this->mTitle->getNamespace() == NS_CATEGORY ||
4943                         !$wgCategoryPrefixedDefaultSortkey) {
4944                         return $this->mTitle->getText();
4945                 } else {
4946                         return $this->mTitle->getPrefixedText();
4947                 }
4948         }
4949
4950         /**
4951          * Accessor for $mDefaultSort
4952          * Unlike getDefaultSort(), will return false if none is set
4953          *
4954          * @return string or false
4955          */
4956         public function getCustomDefaultSort() {
4957                 return $this->mDefaultSort;
4958         }
4959
4960         /**
4961          * Try to guess the section anchor name based on a wikitext fragment
4962          * presumably extracted from a heading, for example "Header" from
4963          * "== Header ==".
4964          */
4965         public function guessSectionNameFromWikiText( $text ) {
4966                 # Strip out wikitext links(they break the anchor)
4967                 $text = $this->stripSectionName( $text );
4968                 $headline = Sanitizer::decodeCharReferences( $text );
4969                 # strip out HTML
4970                 $headline = StringUtils::delimiterReplace( '<', '>', '', $headline );
4971                 $headline = trim( $headline );
4972                 $sectionanchor = '#' . urlencode( str_replace( ' ', '_', $headline ) );
4973                 $replacearray = array(
4974                         '%3A' => ':',
4975                         '%' => '.'
4976                 );
4977                 return str_replace(
4978                         array_keys( $replacearray ),
4979                         array_values( $replacearray ),
4980                         $sectionanchor );
4981         }
4982
4983         /**
4984          * Strips a text string of wikitext for use in a section anchor
4985          *
4986          * Accepts a text string and then removes all wikitext from the
4987          * string and leaves only the resultant text (i.e. the result of
4988          * [[User:WikiSysop|Sysop]] would be "Sysop" and the result of
4989          * [[User:WikiSysop]] would be "User:WikiSysop") - this is intended
4990          * to create valid section anchors by mimicing the output of the
4991          * parser when headings are parsed.
4992          *
4993          * @param $text string Text string to be stripped of wikitext
4994          * for use in a Section anchor
4995          * @return Filtered text string
4996          */
4997         public function stripSectionName( $text ) {
4998                 # Strip internal link markup
4999                 $text = preg_replace('/\[\[:?([^[|]+)\|([^[]+)\]\]/','$2',$text);
5000                 $text = preg_replace('/\[\[:?([^[]+)\|?\]\]/','$1',$text);
5001
5002                 # Strip external link markup (FIXME: Not Tolerant to blank link text
5003                 # I.E. [http://www.mediawiki.org] will render as [1] or something depending
5004                 # on how many empty links there are on the page - need to figure that out.
5005                 $text = preg_replace('/\[(?:' . wfUrlProtocols() . ')([^ ]+?) ([^[]+)\]/','$2',$text);
5006
5007                 # Parse wikitext quotes (italics & bold)
5008                 $text = $this->doQuotes($text);
5009
5010                 # Strip HTML tags
5011                 $text = StringUtils::delimiterReplace( '<', '>', '', $text );
5012                 return $text;
5013         }
5014
5015         function srvus( $text ) {
5016                 return $this->testSrvus( $text, $this->mOutputType );
5017         }
5018
5019         /**
5020          * strip/replaceVariables/unstrip for preprocessor regression testing
5021          */
5022         function testSrvus( $text, $title, $options, $outputType = self::OT_HTML ) {
5023                 $this->clearState();
5024                 if ( ! ( $title instanceof Title ) ) {
5025                         $title = Title::newFromText( $title );
5026                 }
5027                 $this->mTitle = $title;
5028                 $this->mOptions = $options;
5029                 $this->setOutputType( $outputType );
5030                 $text = $this->replaceVariables( $text );
5031                 $text = $this->mStripState->unstripBoth( $text );
5032                 $text = Sanitizer::removeHTMLtags( $text );
5033                 return $text;
5034         }
5035
5036         function testPst( $text, $title, $options ) {
5037                 global $wgUser;
5038                 if ( ! ( $title instanceof Title ) ) {
5039                         $title = Title::newFromText( $title );
5040                 }
5041                 return $this->preSaveTransform( $text, $title, $wgUser, $options );
5042         }
5043
5044         function testPreprocess( $text, $title, $options ) {
5045                 if ( ! ( $title instanceof Title ) ) {
5046                         $title = Title::newFromText( $title );
5047                 }
5048                 return $this->testSrvus( $text, $title, $options, self::OT_PREPROCESS );
5049         }
5050
5051         function markerSkipCallback( $s, $callback ) {
5052                 $i = 0;
5053                 $out = '';
5054                 while ( $i < strlen( $s ) ) {
5055                         $markerStart = strpos( $s, $this->mUniqPrefix, $i );
5056                         if ( $markerStart === false ) {
5057                                 $out .= call_user_func( $callback, substr( $s, $i ) );
5058                                 break;
5059                         } else {
5060                                 $out .= call_user_func( $callback, substr( $s, $i, $markerStart - $i ) );
5061                                 $markerEnd = strpos( $s, self::MARKER_SUFFIX, $markerStart );
5062                                 if ( $markerEnd === false ) {
5063                                         $out .= substr( $s, $markerStart );
5064                                         break;
5065                                 } else {
5066                                         $markerEnd += strlen( self::MARKER_SUFFIX );
5067                                         $out .= substr( $s, $markerStart, $markerEnd - $markerStart );
5068                                         $i = $markerEnd;
5069                                 }
5070                         }
5071                 }
5072                 return $out;
5073         }
5074
5075         function serialiseHalfParsedText( $text ) {
5076                 $data = array();
5077                 $data['text'] = $text;
5078
5079                 // First, find all strip markers, and store their
5080                 //  data in an array.
5081                 $stripState = new StripState;
5082                 $pos = 0;
5083                 while( ( $start_pos = strpos( $text, $this->mUniqPrefix, $pos ) ) && ( $end_pos = strpos( $text, self::MARKER_SUFFIX, $pos ) ) ) {
5084                         $end_pos += strlen( self::MARKER_SUFFIX );
5085                         $marker = substr( $text, $start_pos, $end_pos-$start_pos );
5086
5087                         if ( !empty( $this->mStripState->general->data[$marker] ) ) {
5088                                 $replaceArray = $stripState->general;
5089                                 $stripText = $this->mStripState->general->data[$marker];
5090                         } elseif ( !empty( $this->mStripState->nowiki->data[$marker] ) ) {
5091                                 $replaceArray = $stripState->nowiki;
5092                                 $stripText = $this->mStripState->nowiki->data[$marker];
5093                         } else {
5094                                 throw new MWException( "Hanging strip marker: '$marker'." );
5095                         }
5096
5097                         $replaceArray->setPair( $marker, $stripText );
5098                         $pos = $end_pos;
5099                 }
5100                 $data['stripstate'] = $stripState;
5101
5102                 // Now, find all of our links, and store THEIR
5103                 //  data in an array! :)
5104                 $links = array( 'internal' => array(), 'interwiki' => array() );
5105                 $pos = 0;
5106
5107                 // Internal links
5108                 while( ( $start_pos = strpos( $text, '<!--LINK ', $pos ) ) ) {
5109                         list( $ns, $trail ) = explode( ':', substr( $text, $start_pos + strlen( '<!--LINK ' ) ), 2 );
5110
5111                         $ns = trim($ns);
5112                         if (empty( $links['internal'][$ns] )) {
5113                                 $links['internal'][$ns] = array();
5114                         }
5115
5116                         $key = trim( substr( $trail, 0, strpos( $trail, '-->' ) ) );
5117                         $links['internal'][$ns][] = $this->mLinkHolders->internals[$ns][$key];
5118                         $pos = $start_pos + strlen( "<!--LINK $ns:$key-->" );
5119                 }
5120
5121                 $pos = 0;
5122
5123                 // Interwiki links
5124                 while( ( $start_pos = strpos( $text, '<!--IWLINK ', $pos ) ) ) {
5125                         $data = substr( $text, $start_pos );
5126                         $key = trim( substr( $data, 0, strpos( $data, '-->' ) ) );
5127                         $links['interwiki'][] = $this->mLinkHolders->interwiki[$key];
5128                         $pos = $start_pos + strlen( "<!--IWLINK $key-->" );
5129                 }
5130
5131                 $data['linkholder'] = $links;
5132
5133                 return $data;
5134         }
5135
5136         function unserialiseHalfParsedText( $data, $intPrefix = null /* Unique identifying prefix */ ) {
5137                 if (!$intPrefix)
5138                         $intPrefix = $this->getRandomString();
5139
5140                 // First, extract the strip state.
5141                 $stripState = $data['stripstate'];
5142                 $this->mStripState->general->merge( $stripState->general );
5143                 $this->mStripState->nowiki->merge( $stripState->nowiki );
5144
5145                 // Now, extract the text, and renumber links
5146                 $text = $data['text'];
5147                 $links = $data['linkholder'];
5148
5149                 // Internal...
5150                 foreach( $links['internal'] as $ns => $nsLinks ) {
5151                         foreach( $nsLinks as $key => $entry ) {
5152                                 $newKey = $intPrefix . '-' . $key;
5153                                 $this->mLinkHolders->internals[$ns][$newKey] = $entry;
5154
5155                                 $text = str_replace( "<!--LINK $ns:$key-->", "<!--LINK $ns:$newKey-->", $text );
5156                         }
5157                 }
5158
5159                 // Interwiki...
5160                 foreach( $links['interwiki'] as $key => $entry ) {
5161                         $newKey = "$intPrefix-$key";
5162                         $this->mLinkHolders->interwikis[$newKey] = $entry;
5163
5164                         $text = str_replace( "<!--IWLINK $key-->", "<!--IWLINK $newKey-->", $text );
5165                 }
5166
5167                 // Should be good to go.
5168                 return $text;
5169         }
5170 }
5171
5172 /**
5173  * @todo document, briefly.
5174  * @ingroup Parser
5175  */
5176 class StripState {
5177         var $general, $nowiki;
5178
5179         function __construct() {
5180                 $this->general = new ReplacementArray;
5181                 $this->nowiki = new ReplacementArray;
5182         }
5183
5184         function unstripGeneral( $text ) {
5185                 wfProfileIn( __METHOD__ );
5186                 do {
5187                         $oldText = $text;
5188                         $text = $this->general->replace( $text );
5189                 } while ( $text !== $oldText );
5190                 wfProfileOut( __METHOD__ );
5191                 return $text;
5192         }
5193
5194         function unstripNoWiki( $text ) {
5195                 wfProfileIn( __METHOD__ );
5196                 do {
5197                         $oldText = $text;
5198                         $text = $this->nowiki->replace( $text );
5199                 } while ( $text !== $oldText );
5200                 wfProfileOut( __METHOD__ );
5201                 return $text;
5202         }
5203
5204         function unstripBoth( $text ) {
5205                 wfProfileIn( __METHOD__ );
5206                 do {
5207                         $oldText = $text;
5208                         $text = $this->general->replace( $text );
5209                         $text = $this->nowiki->replace( $text );
5210                 } while ( $text !== $oldText );
5211                 wfProfileOut( __METHOD__ );
5212                 return $text;
5213         }
5214 }
5215
5216 /**
5217  * @todo document, briefly.
5218  * @ingroup Parser
5219  */
5220 class OnlyIncludeReplacer {
5221         var $output = '';
5222
5223         function replace( $matches ) {
5224                 if ( substr( $matches[1], -1 ) === "\n" ) {
5225                         $this->output .= substr( $matches[1], 0, -1 );
5226                 } else {
5227                         $this->output .= $matches[1];
5228                 }
5229         }
5230 }