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