includes/Parser.php

   1 <?php
   2 /**
   3  * File for Parser and related classes
   4  *
   5  * @package MediaWiki
   6  * @subpackage Parser
   7  */
   8
   9 /**
  10  * Update this version number when the ParserOutput format
  11  * changes in an incompatible way, so the parser cache
  12  * can automatically discard old data.
  13  */
  14 define( 'MW_PARSER_VERSION', '1.6.1' );
  15
  16 /**
  17  * Variable substitution O(N^2) attack
  18  *
  19  * Without countermeasures, it would be possible to attack the parser by saving
  20  * a page filled with a large number of inclusions of large pages. The size of
  21  * the generated page would be proportional to the square of the input size.
  22  * Hence, we limit the number of inclusions of any given page, thus bringing any
  23  * attack back to O(N).
  24  */
  25
  26 define( 'MAX_INCLUDE_REPEAT', 100 );
  27 define( 'MAX_INCLUDE_SIZE', 1000000 ); // 1 Million
  28
  29 define( 'RLH_FOR_UPDATE', 1 );
  30
  31 # Allowed values for $mOutputType
  32 define( 'OT_HTML', 1 );
  33 define( 'OT_WIKI', 2 );
  34 define( 'OT_MSG' , 3 );
  35
  36 # Flags for setFunctionHook
  37 define( 'SFH_NO_HASH', 1 );
  38
  39 # string parameter for extractTags which will cause it
  40 # to strip HTML comments in addition to regular
  41 # <XML>-style tags. This should not be anything we
  42 # may want to use in wikisyntax
  43 define( 'STRIP_COMMENTS', 'HTMLCommentStrip' );
  44
  45 # Constants needed for external link processing
  46 define( 'HTTP_PROTOCOLS', 'http:\/\/|https:\/\/' );
  47 # Everything except bracket, space, or control characters
  48 define( 'EXT_LINK_URL_CLASS', '[^][<>"\\x00-\\x20\\x7F]' );
  49 # Including space, but excluding newlines
  50 define( 'EXT_LINK_TEXT_CLASS', '[^\]\\x0a\\x0d]' );
  51 define( 'EXT_IMAGE_FNAME_CLASS', '[A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]' );
  52 define( 'EXT_IMAGE_EXTENSIONS', 'gif|png|jpg|jpeg' );
  53 define( 'EXT_LINK_BRACKETED',  '/\[(\b(' . wfUrlProtocols() . ')'.
  54         EXT_LINK_URL_CLASS.'+) *('.EXT_LINK_TEXT_CLASS.'*?)\]/S' );
  55 define( 'EXT_IMAGE_REGEX',
  56         '/^('.HTTP_PROTOCOLS.')'.  # Protocol
  57         '('.EXT_LINK_URL_CLASS.'+)\\/'.  # Hostname and path
  58         '('.EXT_IMAGE_FNAME_CLASS.'+)\\.((?i)'.EXT_IMAGE_EXTENSIONS.')$/S' # Filename
  59 );
  60
  61 // State constants for the definition list colon extraction
  62 define( 'MW_COLON_STATE_TEXT', 0 );
  63 define( 'MW_COLON_STATE_TAG', 1 );
  64 define( 'MW_COLON_STATE_TAGSTART', 2 );
  65 define( 'MW_COLON_STATE_CLOSETAG', 3 );
  66 define( 'MW_COLON_STATE_TAGSLASH', 4 );
  67 define( 'MW_COLON_STATE_COMMENT', 5 );
  68 define( 'MW_COLON_STATE_COMMENTDASH', 6 );
  69 define( 'MW_COLON_STATE_COMMENTDASHDASH', 7 );
  70
  71 /**
  72  * PHP Parser
  73  *
  74  * Processes wiki markup
  75  *
  76  * <pre>
  77  * There are three main entry points into the Parser class:
  78  * parse()
  79  *   produces HTML output
  80  * preSaveTransform().
  81  *   produces altered wiki markup.
  82  * transformMsg()
  83  *   performs brace substitution on MediaWiki messages
  84  *
  85  * Globals used:
  86  *    objects:   $wgLang, $wgContLang
  87  *
  88  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  89  *
  90  * settings:
  91  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  92  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  93  *  $wgLocaltimezone, $wgAllowSpecialInclusion*
  94  *
  95  *  * only within ParserOptions
  96  * </pre>
  97  *
  98  * @package MediaWiki
  99  */
 100 class Parser
 101 {
 102         /**#@+
 103          * @private
 104          */
 105         # Persistent:
 106         var $mTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables;
 107
 108         # Cleared with clearState():
 109         var $mOutput, $mAutonumber, $mDTopen, $mStripState = array();
 110         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 111         var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
 112         var $mTemplates,        // cache of already loaded templates, avoids
 113                                 // multiple SQL queries for the same string
 114             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 115                                 // in this path. Used for loop detection.
 116
 117         # Temporary
 118         # These are variables reset at least once per parse regardless of $clearState
 119         var $mOptions,      // ParserOptions object
 120                 $mTitle,        // Title context, used for self-link rendering and similar things
 121                 $mOutputType,   // Output type, one of the OT_xxx constants
 122                 $mRevisionId;   // ID to display in {{REVISIONID}} tags
 123
 124         /**#@-*/
 125
 126         /**
 127          * Constructor
 128          *
 129          * @public
 130          */
 131         function Parser() {
 132                 $this->mTagHooks = array();
 133                 $this->mFunctionHooks = array();
 134                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 135                 $this->mFirstCall = true;
 136         }
 137
 138         /**
 139          * Do various kinds of initialisation on the first call of the parser
 140          */
 141         function firstCallInit() {
 142                 if ( !$this->mFirstCall ) {
 143                         return;
 144                 }
 145
 146                 wfProfileIn( __METHOD__ );
 147                 global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
 148
 149                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 150
 151                 $this->setFunctionHook( 'ns', array( 'CoreParserFunctions', 'ns' ), SFH_NO_HASH );
 152                 $this->setFunctionHook( 'urlencode', array( 'CoreParserFunctions', 'urlencode' ), SFH_NO_HASH );
 153                 $this->setFunctionHook( 'lcfirst', array( 'CoreParserFunctions', 'lcfirst' ), SFH_NO_HASH );
 154                 $this->setFunctionHook( 'ucfirst', array( 'CoreParserFunctions', 'ucfirst' ), SFH_NO_HASH );
 155                 $this->setFunctionHook( 'lc', array( 'CoreParserFunctions', 'lc' ), SFH_NO_HASH );
 156                 $this->setFunctionHook( 'uc', array( 'CoreParserFunctions', 'uc' ), SFH_NO_HASH );
 157                 $this->setFunctionHook( 'localurl', array( 'CoreParserFunctions', 'localurl' ), SFH_NO_HASH );
 158                 $this->setFunctionHook( 'localurle', array( 'CoreParserFunctions', 'localurle' ), SFH_NO_HASH );
 159                 $this->setFunctionHook( 'fullurl', array( 'CoreParserFunctions', 'fullurl' ), SFH_NO_HASH );
 160                 $this->setFunctionHook( 'fullurle', array( 'CoreParserFunctions', 'fullurle' ), SFH_NO_HASH );
 161                 $this->setFunctionHook( 'formatnum', array( 'CoreParserFunctions', 'formatnum' ), SFH_NO_HASH );
 162                 $this->setFunctionHook( 'grammar', array( 'CoreParserFunctions', 'grammar' ), SFH_NO_HASH );
 163                 $this->setFunctionHook( 'plural', array( 'CoreParserFunctions', 'plural' ), SFH_NO_HASH );
 164                 $this->setFunctionHook( 'numberofpages', array( 'CoreParserFunctions', 'numberofpages' ), SFH_NO_HASH );
 165                 $this->setFunctionHook( 'numberofusers', array( 'CoreParserFunctions', 'numberofusers' ), SFH_NO_HASH );
 166                 $this->setFunctionHook( 'numberofarticles', array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 167                 $this->setFunctionHook( 'numberoffiles', array( 'CoreParserFunctions', 'numberoffiles' ), SFH_NO_HASH );
 168                 $this->setFunctionHook( 'numberofadmins', array( 'CoreParserFunctions', 'numberofadmins' ), SFH_NO_HASH );
 169                 $this->setFunctionHook( 'language', array( 'CoreParserFunctions', 'language' ), SFH_NO_HASH );
 170
 171                 if ( $wgAllowDisplayTitle ) {
 172                         $this->setFunctionHook( 'displaytitle', array( 'CoreParserFunctions', 'displaytitle' ), SFH_NO_HASH );
 173                 }
 174                 if ( $wgAllowSlowParserFunctions ) {
 175                         $this->setFunctionHook( 'pagesinnamespace', array( 'CoreParserFunctions', 'pagesinnamespace' ), SFH_NO_HASH );
 176                 }
 177
 178                 $this->initialiseVariables();
 179
 180                 $this->mFirstCall = false;
 181                 wfProfileOut( __METHOD__ );
 182         }
 183
 184         /**
 185          * Clear Parser state
 186          *
 187          * @private
 188          */
 189         function clearState() {
 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 = array();
 199                 $this->mArgStack = array();
 200                 $this->mInPre = false;
 201                 $this->mInterwikiLinkHolders = array(
 202                         'texts' => array(),
 203                         'titles' => array()
 204                 );
 205                 $this->mLinkHolders = array(
 206                         'namespaces' => array(),
 207                         'dbkeys' => array(),
 208                         'queries' => array(),
 209                         'texts' => array(),
 210                         'titles' => array()
 211                 );
 212                 $this->mRevisionId = null;
 213
 214                 /**
 215                  * Prefix for temporary replacement strings for the multipass parser.
 216                  * \x07 should never appear in input as it's disallowed in XML.
 217                  * Using it at the front also gives us a little extra robustness
 218                  * since it shouldn't match when butted up against identifier-like
 219                  * string constructs.
 220                  */
 221                 $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 222
 223                 # Clear these on every parse, bug 4549
 224                 $this->mTemplates = array();
 225                 $this->mTemplatePath = array();
 226
 227                 $this->mShowToc = true;
 228                 $this->mForceTocPosition = false;
 229
 230                 wfRunHooks( 'ParserClearState', array( &$this ) );
 231         }
 232
 233         /**
 234          * Accessor for mUniqPrefix.
 235          *
 236          * @public
 237          */
 238         function UniqPrefix() {
 239                 return $this->mUniqPrefix;
 240         }
 241
 242         /**
 243          * Convert wikitext to HTML
 244          * Do not call this function recursively.
 245          *
 246          * @private
 247          * @param string $text Text we want to parse
 248          * @param Title &$title A title object
 249          * @param array $options
 250          * @param boolean $linestart
 251          * @param boolean $clearState
 252          * @param int $revid number to pass in {{REVISIONID}}
 253          * @return ParserOutput a ParserOutput
 254          */
 255         function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 256                 /**
 257                  * First pass--just handle <nowiki> sections, pass the rest off
 258                  * to internalParse() which does all the real work.
 259                  */
 260
 261                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 262                 $fname = 'Parser::parse';
 263                 wfProfileIn( $fname );
 264
 265                 if ( $clearState ) {
 266                         $this->clearState();
 267                 }
 268
 269                 $this->mOptions = $options;
 270                 $this->mTitle =& $title;
 271                 $oldRevisionId = $this->mRevisionId;
 272                 if( $revid !== null ) {
 273                         $this->mRevisionId = $revid;
 274                 }
 275                 $this->mOutputType = OT_HTML;
 276
 277                 //$text = $this->strip( $text, $this->mStripState );
 278                 // VOODOO MAGIC FIX! Sometimes the above segfaults in PHP5.
 279                 $x =& $this->mStripState;
 280
 281                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 282                 $text = $this->strip( $text, $x );
 283                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 284
 285                 # Hook to suspend the parser in this state
 286                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$x ) ) ) {
 287                         wfProfileOut( $fname );
 288                         return $text ;
 289                 }
 290
 291                 $text = $this->internalParse( $text );
 292
 293                 $text = $this->unstrip( $text, $this->mStripState );
 294
 295                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 296                 $fixtags = array(
 297                         # french spaces, last one Guillemet-left
 298                         # only if there is something before the space
 299                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 300                         # french spaces, Guillemet-right
 301                         '/(\\302\\253) /' => '\\1&nbsp;',
 302                 );
 303                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 304
 305                 # only once and last
 306                 $text = $this->doBlockLevels( $text, $linestart );
 307
 308                 $this->replaceLinkHolders( $text );
 309
 310                 # the position of the parserConvert() call should not be changed. it
 311                 # assumes that the links are all replaced and the only thing left
 312                 # is the <nowiki> mark.
 313                 # Side-effects: this calls $this->mOutput->setTitleText()
 314                 $text = $wgContLang->parserConvert( $text, $this );
 315
 316                 $text = $this->unstripNoWiki( $text, $this->mStripState );
 317
 318                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 319
 320                 $text = Sanitizer::normalizeCharReferences( $text );
 321
 322                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 323                         $text = Parser::tidy($text);
 324                 } else {
 325                         # attempt to sanitize at least some nesting problems
 326                         # (bug #2702 and quite a few others)
 327                         $tidyregs = array(
 328                                 # ''Something [http://www.cool.com cool''] -->
 329                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 330                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 331                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 332                                 # fix up an anchor inside another anchor, only
 333                                 # at least for a single single nested link (bug 3695)
 334                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 335                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 336                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 337                                 # contains a div, so fix it up here; replace
 338                                 # div with escaped text
 339                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 340                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 341                                 # remove empty italic or bold tag pairs, some
 342                                 # introduced by rules above
 343                                 '/<([bi])><\/\\1>/' => '',
 344                         );
 345
 346                         $text = preg_replace(
 347                                 array_keys( $tidyregs ),
 348                                 array_values( $tidyregs ),
 349                                 $text );
 350                 }
 351
 352                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 353
 354                 $this->mOutput->setText( $text );
 355                 $this->mRevisionId = $oldRevisionId;
 356                 wfProfileOut( $fname );
 357
 358                 return $this->mOutput;
 359         }
 360
 361         /**
 362          * Get a random string
 363          *
 364          * @private
 365          * @static
 366          */
 367         function getRandomString() {
 368                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 369         }
 370
 371         function &getTitle() { return $this->mTitle; }
 372         function getOptions() { return $this->mOptions; }
 373
 374         function getFunctionLang() {
 375                 global $wgLang, $wgContLang;
 376                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 377         }
 378
 379         /**
 380          * Replaces all occurrences of HTML-style comments and the given tags
 381          * in the text with a random marker and returns teh next text. The output
 382          * parameter $matches will be an associative array filled with data in
 383          * the form:
 384          *   'UNIQ-xxxxx' => array(
 385          *     'element',
 386          *     'tag content',
 387          *     array( 'param' => 'x' ),
 388          *     '<element param="x">tag content</element>' ) )
 389          *
 390          * @param $elements list of element names. Comments are always extracted.
 391          * @param $text Source text string.
 392          * @param $uniq_prefix
 393          *
 394          * @private
 395          * @static
 396          */
 397         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 398                 $rand = Parser::getRandomString();
 399                 $n = 1;
 400                 $stripped = '';
 401                 $matches = array();
 402
 403                 $taglist = implode( '|', $elements );
 404                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 405
 406                 while ( '' != $text ) {
 407                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 408                         $stripped .= $p[0];
 409                         if( count( $p ) < 5 ) {
 410                                 break;
 411                         }
 412                         if( count( $p ) > 5 ) {
 413                                 // comment
 414                                 $element    = $p[4];
 415                                 $attributes = '';
 416                                 $close      = '';
 417                                 $inside     = $p[5];
 418                         } else {
 419                                 // tag
 420                                 $element    = $p[1];
 421                                 $attributes = $p[2];
 422                                 $close      = $p[3];
 423                                 $inside     = $p[4];
 424                         }
 425
 426                         $marker = "$uniq_prefix-$element-$rand" . sprintf('%08X', $n++) . '-QINU';
 427                         $stripped .= $marker;
 428
 429                         if ( $close === '/>' ) {
 430                                 // Empty element tag, <tag />
 431                                 $content = null;
 432                                 $text = $inside;
 433                                 $tail = null;
 434                         } else {
 435                                 if( $element == '!--' ) {
 436                                         $end = '/(-->)/';
 437                                 } else {
 438                                         $end = "/(<\\/$element\\s*>)/i";
 439                                 }
 440                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 441                                 $content = $q[0];
 442                                 if( count( $q ) < 3 ) {
 443                                         # No end tag -- let it run out to the end of the text.
 444                                         $tail = '';
 445                                         $text = '';
 446                                 } else {
 447                                         $tail = $q[1];
 448                                         $text = $q[2];
 449                                 }
 450                         }
 451
 452                         $matches[$marker] = array( $element,
 453                                 $content,
 454                                 Sanitizer::decodeTagAttributes( $attributes ),
 455                                 "<$element$attributes$close$content$tail" );
 456                 }
 457                 return $stripped;
 458         }
 459
 460         /**
 461          * Strips and renders nowiki, pre, math, hiero
 462          * If $render is set, performs necessary rendering operations on plugins
 463          * Returns the text, and fills an array with data needed in unstrip()
 464          * If the $state is already a valid strip state, it adds to the state
 465          *
 466          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 467          *  will be stripped in addition to other tags. This is important
 468          *  for section editing, where these comments cause confusion when
 469          *  counting the sections in the wikisource
 470          *
 471          * @param array dontstrip contains tags which should not be stripped;
 472          *  used to prevent stipping of <gallery> when saving (fixes bug 2700)
 473          *
 474          * @private
 475          */
 476         function strip( $text, &$state, $stripcomments = false , $dontstrip = array () ) {
 477                 $render = ($this->mOutputType == OT_HTML);
 478
 479                 # Replace any instances of the placeholders
 480                 $uniq_prefix = $this->mUniqPrefix;
 481                 #$text = str_replace( $uniq_prefix, wfHtmlEscapeFirst( $uniq_prefix ), $text );
 482                 $commentState = array();
 483
 484                 $elements = array_merge(
 485                         array( 'nowiki', 'gallery' ),
 486                         array_keys( $this->mTagHooks ) );
 487                 global $wgRawHtml;
 488                 if( $wgRawHtml ) {
 489                         $elements[] = 'html';
 490                 }
 491                 if( $this->mOptions->getUseTeX() ) {
 492                         $elements[] = 'math';
 493                 }
 494
 495                 # Removing $dontstrip tags from $elements list (currently only 'gallery', fixing bug 2700)
 496                 foreach ( $elements AS $k => $v ) {
 497                         if ( !in_array ( $v , $dontstrip ) ) continue;
 498                         unset ( $elements[$k] );
 499                 }
 500
 501                 $matches = array();
 502                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 503
 504                 foreach( $matches as $marker => $data ) {
 505                         list( $element, $content, $params, $tag ) = $data;
 506                         if( $render ) {
 507                                 $tagName = strtolower( $element );
 508                                 switch( $tagName ) {
 509                                 case '!--':
 510                                         // Comment
 511                                         if( substr( $tag, -3 ) == '-->' ) {
 512                                                 $output = $tag;
 513                                         } else {
 514                                                 // Unclosed comment in input.
 515                                                 // Close it so later stripping can remove it
 516                                                 $output = "$tag-->";
 517                                         }
 518                                         break;
 519                                 case 'html':
 520                                         if( $wgRawHtml ) {
 521                                                 $output = $content;
 522                                                 break;
 523                                         }
 524                                         // Shouldn't happen otherwise. :)
 525                                 case 'nowiki':
 526                                         $output = wfEscapeHTMLTagsOnly( $content );
 527                                         break;
 528                                 case 'math':
 529                                         $output = MathRenderer::renderMath( $content );
 530                                         break;
 531                                 case 'gallery':
 532                                         $output = $this->renderImageGallery( $content, $params );
 533                                         break;
 534                                 default:
 535                                         if( isset( $this->mTagHooks[$tagName] ) ) {
 536                                                 $output = call_user_func_array( $this->mTagHooks[$tagName],
 537                                                         array( $content, $params, $this ) );
 538                                         } else {
 539                                                 throw new MWException( "Invalid call hook $element" );
 540                                         }
 541                                 }
 542                         } else {
 543                                 // Just stripping tags; keep the source
 544                                 $output = $tag;
 545                         }
 546                         if( !$stripcomments && $element == '!--' ) {
 547                                 $commentState[$marker] = $output;
 548                         } else {
 549                                 $state[$element][$marker] = $output;
 550                         }
 551                 }
 552
 553                 # Unstrip comments unless explicitly told otherwise.
 554                 # (The comments are always stripped prior to this point, so as to
 555                 # not invoke any extension tags / parser hooks contained within
 556                 # a comment.)
 557                 if ( !$stripcomments ) {
 558                         // Put them all back and forget them
 559                         $text = strtr( $text, $commentState );
 560                 }
 561
 562                 return $text;
 563         }
 564
 565         /**
 566          * Restores pre, math, and other extensions removed by strip()
 567          *
 568          * always call unstripNoWiki() after this one
 569          * @private
 570          */
 571         function unstrip( $text, &$state ) {
 572                 if ( !is_array( $state ) ) {
 573                         return $text;
 574                 }
 575
 576                 $replacements = array();
 577                 foreach( $state as $tag => $contentDict ) {
 578                         if( $tag != 'nowiki' && $tag != 'html' ) {
 579                                 foreach( $contentDict as $uniq => $content ) {
 580                                         $replacements[$uniq] = $content;
 581                                 }
 582                         }
 583                 }
 584                 $text = strtr( $text, $replacements );
 585
 586                 return $text;
 587         }
 588
 589         /**
 590          * Always call this after unstrip() to preserve the order
 591          *
 592          * @private
 593          */
 594         function unstripNoWiki( $text, &$state ) {
 595                 if ( !is_array( $state ) ) {
 596                         return $text;
 597                 }
 598
 599                 $replacements = array();
 600                 foreach( $state as $tag => $contentDict ) {
 601                         if( $tag == 'nowiki' || $tag == 'html' ) {
 602                                 foreach( $contentDict as $uniq => $content ) {
 603                                         $replacements[$uniq] = $content;
 604                                 }
 605                         }
 606                 }
 607                 $text = strtr( $text, $replacements );
 608
 609                 return $text;
 610         }
 611
 612         /**
 613          * Add an item to the strip state
 614          * Returns the unique tag which must be inserted into the stripped text
 615          * The tag will be replaced with the original text in unstrip()
 616          *
 617          * @private
 618          */
 619         function insertStripItem( $text, &$state ) {
 620                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 621                 if ( !$state ) {
 622                         $state = array();
 623                 }
 624                 $state['item'][$rnd] = $text;
 625                 return $rnd;
 626         }
 627
 628         /**
 629          * Interface with html tidy, used if $wgUseTidy = true.
 630          * If tidy isn't able to correct the markup, the original will be
 631          * returned in all its glory with a warning comment appended.
 632          *
 633          * Either the external tidy program or the in-process tidy extension
 634          * will be used depending on availability. Override the default
 635          * $wgTidyInternal setting to disable the internal if it's not working.
 636          *
 637          * @param string $text Hideous HTML input
 638          * @return string Corrected HTML output
 639          * @public
 640          * @static
 641          */
 642         function tidy( $text ) {
 643                 global $wgTidyInternal;
 644                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 645 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 646 '<head><title>test</title></head><body>'.$text.'</body></html>';
 647                 if( $wgTidyInternal ) {
 648                         $correctedtext = Parser::internalTidy( $wrappedtext );
 649                 } else {
 650                         $correctedtext = Parser::externalTidy( $wrappedtext );
 651                 }
 652                 if( is_null( $correctedtext ) ) {
 653                         wfDebug( "Tidy error detected!\n" );
 654                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 655                 }
 656                 return $correctedtext;
 657         }
 658
 659         /**
 660          * Spawn an external HTML tidy process and get corrected markup back from it.
 661          *
 662          * @private
 663          * @static
 664          */
 665         function externalTidy( $text ) {
 666                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 667                 $fname = 'Parser::externalTidy';
 668                 wfProfileIn( $fname );
 669
 670                 $cleansource = '';
 671                 $opts = ' -utf8';
 672
 673                 $descriptorspec = array(
 674                         0 => array('pipe', 'r'),
 675                         1 => array('pipe', 'w'),
 676                         2 => array('file', '/dev/null', 'a')
 677                 );
 678                 $pipes = array();
 679                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 680                 if (is_resource($process)) {
 681                         // Theoretically, this style of communication could cause a deadlock
 682                         // here. If the stdout buffer fills up, then writes to stdin could
 683                         // block. This doesn't appear to happen with tidy, because tidy only
 684                         // writes to stdout after it's finished reading from stdin. Search
 685                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 686                         fwrite($pipes[0], $text);
 687                         fclose($pipes[0]);
 688                         while (!feof($pipes[1])) {
 689                                 $cleansource .= fgets($pipes[1], 1024);
 690                         }
 691                         fclose($pipes[1]);
 692                         proc_close($process);
 693                 }
 694
 695                 wfProfileOut( $fname );
 696
 697                 if( $cleansource == '' && $text != '') {
 698                         // Some kind of error happened, so we couldn't get the corrected text.
 699                         // Just give up; we'll use the source text and append a warning.
 700                         return null;
 701                 } else {
 702                         return $cleansource;
 703                 }
 704         }
 705
 706         /**
 707          * Use the HTML tidy PECL extension to use the tidy library in-process,
 708          * saving the overhead of spawning a new process. Currently written to
 709          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 710          *
 711          * 'pear install tidy' should be able to compile the extension module.
 712          *
 713          * @private
 714          * @static
 715          */
 716         function internalTidy( $text ) {
 717                 global $wgTidyConf;
 718                 $fname = 'Parser::internalTidy';
 719                 wfProfileIn( $fname );
 720
 721                 tidy_load_config( $wgTidyConf );
 722                 tidy_set_encoding( 'utf8' );
 723                 tidy_parse_string( $text );
 724                 tidy_clean_repair();
 725                 if( tidy_get_status() == 2 ) {
 726                         // 2 is magic number for fatal error
 727                         // http://www.php.net/manual/en/function.tidy-get-status.php
 728                         $cleansource = null;
 729                 } else {
 730                         $cleansource = tidy_get_output();
 731                 }
 732                 wfProfileOut( $fname );
 733                 return $cleansource;
 734         }
 735
 736         /**
 737          * parse the wiki syntax used to render tables
 738          *
 739          * @private
 740          */
 741         function doTableStuff ( $t ) {
 742                 $fname = 'Parser::doTableStuff';
 743                 wfProfileIn( $fname );
 744
 745                 $t = explode ( "\n" , $t ) ;
 746                 $td = array () ; # Is currently a td tag open?
 747                 $ltd = array () ; # Was it TD or TH?
 748                 $tr = array () ; # Is currently a tr tag open?
 749                 $ltr = array () ; # tr attributes
 750                 $has_opened_tr = array(); # Did this table open a <tr> element?
 751                 $indent_level = 0; # indent level of the table
 752                 foreach ( $t AS $k => $x )
 753                 {
 754                         $x = trim ( $x ) ;
 755                         $fc = substr ( $x , 0 , 1 ) ;
 756                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 757                                 $indent_level = strlen( $matches[1] );
 758
 759                                 $attributes = $this->unstripForHTML( $matches[2] );
 760
 761                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 762                                         '<table' . Sanitizer::fixTagAttributes ( $attributes, 'table' ) . '>' ;
 763                                 array_push ( $td , false ) ;
 764                                 array_push ( $ltd , '' ) ;
 765                                 array_push ( $tr , false ) ;
 766                                 array_push ( $ltr , '' ) ;
 767                                 array_push ( $has_opened_tr, false );
 768                         }
 769                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 770                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 771                                 $z = "</table>" . substr ( $x , 2);
 772                                 $l = array_pop ( $ltd ) ;
 773                                 if ( !array_pop ( $has_opened_tr ) ) $z = "<tr><td></td></tr>" . $z ;
 774                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 775                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 776                                 array_pop ( $ltr ) ;
 777                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 778                         }
 779                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 780                                 $x = substr ( $x , 1 ) ;
 781                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 782                                 $z = '' ;
 783                                 $l = array_pop ( $ltd ) ;
 784                                 array_pop ( $has_opened_tr );
 785                                 array_push ( $has_opened_tr , true ) ;
 786                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 787                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 788                                 array_pop ( $ltr ) ;
 789                                 $t[$k] = $z ;
 790                                 array_push ( $tr , false ) ;
 791                                 array_push ( $td , false ) ;
 792                                 array_push ( $ltd , '' ) ;
 793                                 $attributes = $this->unstripForHTML( $x );
 794                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $attributes, 'tr' ) ) ;
 795                         }
 796                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 797                                 # $x is a table row
 798                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 799                                         $fc = '+' ;
 800                                         $x = substr ( $x , 1 ) ;
 801                                 }
 802                                 $after = substr ( $x , 1 ) ;
 803                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 804
 805                                 // Split up multiple cells on the same line.
 806                                 // FIXME: This can result in improper nesting of tags processed
 807                                 // by earlier parser steps, but should avoid splitting up eg
 808                                 // attribute values containing literal "||".
 809                                 $after = wfExplodeMarkup( '||', $after );
 810
 811                                 $t[$k] = '' ;
 812
 813                                 # Loop through each table cell
 814                                 foreach ( $after AS $theline )
 815                                 {
 816                                         $z = '' ;
 817                                         if ( $fc != '+' )
 818                                         {
 819                                                 $tra = array_pop ( $ltr ) ;
 820                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 821                                                 array_push ( $tr , true ) ;
 822                                                 array_push ( $ltr , '' ) ;
 823                                                 array_pop ( $has_opened_tr );
 824                                                 array_push ( $has_opened_tr , true ) ;
 825                                         }
 826
 827                                         $l = array_pop ( $ltd ) ;
 828                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 829                                         if ( $fc == '|' ) $l = 'td' ;
 830                                         else if ( $fc == '!' ) $l = 'th' ;
 831                                         else if ( $fc == '+' ) $l = 'caption' ;
 832                                         else $l = '' ;
 833                                         array_push ( $ltd , $l ) ;
 834
 835                                         # Cell parameters
 836                                         $y = explode ( '|' , $theline , 2 ) ;
 837                                         # Note that a '|' inside an invalid link should not
 838                                         # be mistaken as delimiting cell parameters
 839                                         if ( strpos( $y[0], '[[' ) !== false ) {
 840                                                 $y = array ($theline);
 841                                         }
 842                                         if ( count ( $y ) == 1 )
 843                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 844                                         else {
 845                                                 $attributes = $this->unstripForHTML( $y[0] );
 846                                                 $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($attributes, $l).">{$y[1]}" ;
 847                                         }
 848                                         $t[$k] .= $y ;
 849                                         array_push ( $td , true ) ;
 850                                 }
 851                         }
 852                 }
 853
 854                 # Closing open td, tr && table
 855                 while ( count ( $td ) > 0 )
 856                 {
 857                         $l = array_pop ( $ltd ) ;
 858                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 859                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 860                         if ( !array_pop ( $has_opened_tr ) ) $t[] = "<tr><td></td></tr>" ;
 861                         $t[] = '</table>' ;
 862                 }
 863
 864                 $t = implode ( "\n" , $t ) ;
 865                 # special case: don't return empty table
 866                 if($t == "<table>\n<tr><td></td></tr>\n</table>")
 867                         $t = '';
 868                 wfProfileOut( $fname );
 869                 return $t ;
 870         }
 871
 872         /**
 873          * Helper function for parse() that transforms wiki markup into
 874          * HTML. Only called for $mOutputType == OT_HTML.
 875          *
 876          * @private
 877          */
 878         function internalParse( $text ) {
 879                 $args = array();
 880                 $isMain = true;
 881                 $fname = 'Parser::internalParse';
 882                 wfProfileIn( $fname );
 883
 884                 # Remove <noinclude> tags and <includeonly> sections
 885                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 886                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 887                 $text = preg_replace( '/<includeonly>.*?<\/includeonly>/s', '', $text );
 888
 889                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
 890
 891                 $text = $this->replaceVariables( $text, $args );
 892
 893                 // Tables need to come after variable replacement for things to work
 894                 // properly; putting them before other transformations should keep
 895                 // exciting things like link expansions from showing up in surprising
 896                 // places.
 897                 $text = $this->doTableStuff( $text );
 898
 899                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 900
 901                 $text = $this->stripToc( $text );
 902                 $this->stripNoGallery( $text );
 903                 $text = $this->doHeadings( $text );
 904                 if($this->mOptions->getUseDynamicDates()) {
 905                         $df =& DateFormatter::getInstance();
 906                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 907                 }
 908                 $text = $this->doAllQuotes( $text );
 909                 $text = $this->replaceInternalLinks( $text );
 910                 $text = $this->replaceExternalLinks( $text );
 911
 912                 # replaceInternalLinks may sometimes leave behind
 913                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 914                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 915
 916                 $text = $this->doMagicLinks( $text );
 917                 $text = $this->formatHeadings( $text, $isMain );
 918
 919                 wfProfileOut( $fname );
 920                 return $text;
 921         }
 922
 923         /**
 924          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 925          * magic external links.
 926          *
 927          * @private
 928          */
 929         function &doMagicLinks( &$text ) {
 930                 $text = $this->magicISBN( $text );
 931                 $text = $this->magicRFC( $text, 'RFC ', 'rfcurl' );
 932                 $text = $this->magicRFC( $text, 'PMID ', 'pubmedurl' );
 933                 return $text;
 934         }
 935
 936         /**
 937          * Parse headers and return html
 938          *
 939          * @private
 940          */
 941         function doHeadings( $text ) {
 942                 $fname = 'Parser::doHeadings';
 943                 wfProfileIn( $fname );
 944                 for ( $i = 6; $i >= 1; --$i ) {
 945                         $h = str_repeat( '=', $i );
 946                         $text = preg_replace( "/^{$h}(.+){$h}\\s*$/m",
 947                           "<h{$i}>\\1</h{$i}>\\2", $text );
 948                 }
 949                 wfProfileOut( $fname );
 950                 return $text;
 951         }
 952
 953         /**
 954          * Replace single quotes with HTML markup
 955          * @private
 956          * @return string the altered text
 957          */
 958         function doAllQuotes( $text ) {
 959                 $fname = 'Parser::doAllQuotes';
 960                 wfProfileIn( $fname );
 961                 $outtext = '';
 962                 $lines = explode( "\n", $text );
 963                 foreach ( $lines as $line ) {
 964                         $outtext .= $this->doQuotes ( $line ) . "\n";
 965                 }
 966                 $outtext = substr($outtext, 0,-1);
 967                 wfProfileOut( $fname );
 968                 return $outtext;
 969         }
 970
 971         /**
 972          * Helper function for doAllQuotes()
 973          * @private
 974          */
 975         function doQuotes( $text ) {
 976                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 977                 if ( count( $arr ) == 1 )
 978                         return $text;
 979                 else
 980                 {
 981                         # First, do some preliminary work. This may shift some apostrophes from
 982                         # being mark-up to being text. It also counts the number of occurrences
 983                         # of bold and italics mark-ups.
 984                         $i = 0;
 985                         $numbold = 0;
 986                         $numitalics = 0;
 987                         foreach ( $arr as $r )
 988                         {
 989                                 if ( ( $i % 2 ) == 1 )
 990                                 {
 991                                         # If there are ever four apostrophes, assume the first is supposed to
 992                                         # be text, and the remaining three constitute mark-up for bold text.
 993                                         if ( strlen( $arr[$i] ) == 4 )
 994                                         {
 995                                                 $arr[$i-1] .= "'";
 996                                                 $arr[$i] = "'''";
 997                                         }
 998                                         # If there are more than 5 apostrophes in a row, assume they're all
 999                                         # text except for the last 5.
1000                                         else if ( strlen( $arr[$i] ) > 5 )
1001                                         {
1002                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1003                                                 $arr[$i] = "'''''";
1004                                         }
1005                                         # Count the number of occurrences of bold and italics mark-ups.
1006                                         # We are not counting sequences of five apostrophes.
1007                                         if ( strlen( $arr[$i] ) == 2 ) $numitalics++;  else
1008                                         if ( strlen( $arr[$i] ) == 3 ) $numbold++;     else
1009                                         if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1010                                 }
1011                                 $i++;
1012                         }
1013
1014                         # If there is an odd number of both bold and italics, it is likely
1015                         # that one of the bold ones was meant to be an apostrophe followed
1016                         # by italics. Which one we cannot know for certain, but it is more
1017                         # likely to be one that has a single-letter word before it.
1018                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1019                         {
1020                                 $i = 0;
1021                                 $firstsingleletterword = -1;
1022                                 $firstmultiletterword = -1;
1023                                 $firstspace = -1;
1024                                 foreach ( $arr as $r )
1025                                 {
1026                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1027                                         {
1028                                                 $x1 = substr ($arr[$i-1], -1);
1029                                                 $x2 = substr ($arr[$i-1], -2, 1);
1030                                                 if ($x1 == ' ') {
1031                                                         if ($firstspace == -1) $firstspace = $i;
1032                                                 } else if ($x2 == ' ') {
1033                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1034                                                 } else {
1035                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1036                                                 }
1037                                         }
1038                                         $i++;
1039                                 }
1040
1041                                 # If there is a single-letter word, use it!
1042                                 if ($firstsingleletterword > -1)
1043                                 {
1044                                         $arr [ $firstsingleletterword ] = "''";
1045                                         $arr [ $firstsingleletterword-1 ] .= "'";
1046                                 }
1047                                 # If not, but there's a multi-letter word, use that one.
1048                                 else if ($firstmultiletterword > -1)
1049                                 {
1050                                         $arr [ $firstmultiletterword ] = "''";
1051                                         $arr [ $firstmultiletterword-1 ] .= "'";
1052                                 }
1053                                 # ... otherwise use the first one that has neither.
1054                                 # (notice that it is possible for all three to be -1 if, for example,
1055                                 # there is only one pentuple-apostrophe in the line)
1056                                 else if ($firstspace > -1)
1057                                 {
1058                                         $arr [ $firstspace ] = "''";
1059                                         $arr [ $firstspace-1 ] .= "'";
1060                                 }
1061                         }
1062
1063                         # Now let's actually convert our apostrophic mush to HTML!
1064                         $output = '';
1065                         $buffer = '';
1066                         $state = '';
1067                         $i = 0;
1068                         foreach ($arr as $r)
1069                         {
1070                                 if (($i % 2) == 0)
1071                                 {
1072                                         if ($state == 'both')
1073                                                 $buffer .= $r;
1074                                         else
1075                                                 $output .= $r;
1076                                 }
1077                                 else
1078                                 {
1079                                         if (strlen ($r) == 2)
1080                                         {
1081                                                 if ($state == 'i')
1082                                                 { $output .= '</i>'; $state = ''; }
1083                                                 else if ($state == 'bi')
1084                                                 { $output .= '</i>'; $state = 'b'; }
1085                                                 else if ($state == 'ib')
1086                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1087                                                 else if ($state == 'both')
1088                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1089                                                 else # $state can be 'b' or ''
1090                                                 { $output .= '<i>'; $state .= 'i'; }
1091                                         }
1092                                         else if (strlen ($r) == 3)
1093                                         {
1094                                                 if ($state == 'b')
1095                                                 { $output .= '</b>'; $state = ''; }
1096                                                 else if ($state == 'bi')
1097                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1098                                                 else if ($state == 'ib')
1099                                                 { $output .= '</b>'; $state = 'i'; }
1100                                                 else if ($state == 'both')
1101                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1102                                                 else # $state can be 'i' or ''
1103                                                 { $output .= '<b>'; $state .= 'b'; }
1104                                         }
1105                                         else if (strlen ($r) == 5)
1106                                         {
1107                                                 if ($state == 'b')
1108                                                 { $output .= '</b><i>'; $state = 'i'; }
1109                                                 else if ($state == 'i')
1110                                                 { $output .= '</i><b>'; $state = 'b'; }
1111                                                 else if ($state == 'bi')
1112                                                 { $output .= '</i></b>'; $state = ''; }
1113                                                 else if ($state == 'ib')
1114                                                 { $output .= '</b></i>'; $state = ''; }
1115                                                 else if ($state == 'both')
1116                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1117                                                 else # ($state == '')
1118                                                 { $buffer = ''; $state = 'both'; }
1119                                         }
1120                                 }
1121                                 $i++;
1122                         }
1123                         # Now close all remaining tags.  Notice that the order is important.
1124                         if ($state == 'b' || $state == 'ib')
1125                                 $output .= '</b>';
1126                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1127                                 $output .= '</i>';
1128                         if ($state == 'bi')
1129                                 $output .= '</b>';
1130                         if ($state == 'both')
1131                                 $output .= '<b><i>'.$buffer.'</i></b>';
1132                         return $output;
1133                 }
1134         }
1135
1136         /**
1137          * Replace external links
1138          *
1139          * Note: this is all very hackish and the order of execution matters a lot.
1140          * Make sure to run maintenance/parserTests.php if you change this code.
1141          *
1142          * @private
1143          */
1144         function replaceExternalLinks( $text ) {
1145                 global $wgContLang;
1146                 $fname = 'Parser::replaceExternalLinks';
1147                 wfProfileIn( $fname );
1148
1149                 $sk =& $this->mOptions->getSkin();
1150
1151                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1152
1153                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1154
1155                 $i = 0;
1156                 while ( $i<count( $bits ) ) {
1157                         $url = $bits[$i++];
1158                         $protocol = $bits[$i++];
1159                         $text = $bits[$i++];
1160                         $trail = $bits[$i++];
1161
1162                         # The characters '<' and '>' (which were escaped by
1163                         # removeHTMLtags()) should not be included in
1164                         # URLs, per RFC 2396.
1165                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1166                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1167                                 $url = substr($url, 0, $m2[0][1]);
1168                         }
1169
1170                         # If the link text is an image URL, replace it with an <img> tag
1171                         # This happened by accident in the original parser, but some people used it extensively
1172                         $img = $this->maybeMakeExternalImage( $text );
1173                         if ( $img !== false ) {
1174                                 $text = $img;
1175                         }
1176
1177                         $dtrail = '';
1178
1179                         # Set linktype for CSS - if URL==text, link is essentially free
1180                         $linktype = ($text == $url) ? 'free' : 'text';
1181
1182                         # No link text, e.g. [http://domain.tld/some.link]
1183                         if ( $text == '' ) {
1184                                 # Autonumber if allowed. See bug #5918
1185                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1186                                         $text = '[' . ++$this->mAutonumber . ']';
1187                                         $linktype = 'autonumber';
1188                                 } else {
1189                                         # Otherwise just use the URL
1190                                         $text = htmlspecialchars( $url );
1191                                         $linktype = 'free';
1192                                 }
1193                         } else {
1194                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1195                                 # Check for trail
1196                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1197                         }
1198
1199                         $text = $wgContLang->markNoConversion($text);
1200
1201                         $url = Sanitizer::cleanUrl( $url );
1202
1203                         # Process the trail (i.e. everything after this link up until start of the next link),
1204                         # replacing any non-bracketed links
1205                         $trail = $this->replaceFreeExternalLinks( $trail );
1206
1207                         # Use the encoded URL
1208                         # This means that users can paste URLs directly into the text
1209                         # Funny characters like &ouml; aren't valid in URLs anyway
1210                         # This was changed in August 2004
1211                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1212
1213                         # Register link in the output object.
1214                         # Replace unnecessary URL escape codes with the referenced character
1215                         # This prevents spammers from hiding links from the filters
1216                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1217                         $this->mOutput->addExternalLink( $pasteurized );
1218                 }
1219
1220                 wfProfileOut( $fname );
1221                 return $s;
1222         }
1223
1224         /**
1225          * Replace anything that looks like a URL with a link
1226          * @private
1227          */
1228         function replaceFreeExternalLinks( $text ) {
1229                 global $wgContLang;
1230                 $fname = 'Parser::replaceFreeExternalLinks';
1231                 wfProfileIn( $fname );
1232
1233                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1234                 $s = array_shift( $bits );
1235                 $i = 0;
1236
1237                 $sk =& $this->mOptions->getSkin();
1238
1239                 while ( $i < count( $bits ) ){
1240                         $protocol = $bits[$i++];
1241                         $remainder = $bits[$i++];
1242
1243                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1244                                 # Found some characters after the protocol that look promising
1245                                 $url = $protocol . $m[1];
1246                                 $trail = $m[2];
1247
1248                                 # special case: handle urls as url args:
1249                                 # http://www.example.com/foo?=http://www.example.com/bar
1250                                 if(strlen($trail) == 0 &&
1251                                         isset($bits[$i]) &&
1252                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1253                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1254                                 {
1255                                         # add protocol, arg
1256                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1257                                         $i += 2;
1258                                         $trail = $m[2];
1259                                 }
1260
1261                                 # The characters '<' and '>' (which were escaped by
1262                                 # removeHTMLtags()) should not be included in
1263                                 # URLs, per RFC 2396.
1264                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1265                                         $trail = substr($url, $m2[0][1]) . $trail;
1266                                         $url = substr($url, 0, $m2[0][1]);
1267                                 }
1268
1269                                 # Move trailing punctuation to $trail
1270                                 $sep = ',;\.:!?';
1271                                 # If there is no left bracket, then consider right brackets fair game too
1272                                 if ( strpos( $url, '(' ) === false ) {
1273                                         $sep .= ')';
1274                                 }
1275
1276                                 $numSepChars = strspn( strrev( $url ), $sep );
1277                                 if ( $numSepChars ) {
1278                                         $trail = substr( $url, -$numSepChars ) . $trail;
1279                                         $url = substr( $url, 0, -$numSepChars );
1280                                 }
1281
1282                                 $url = Sanitizer::cleanUrl( $url );
1283
1284                                 # Is this an external image?
1285                                 $text = $this->maybeMakeExternalImage( $url );
1286                                 if ( $text === false ) {
1287                                         # Not an image, make a link
1288                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1289                                         # Register it in the output object...
1290                                         # Replace unnecessary URL escape codes with their equivalent characters
1291                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1292                                         $this->mOutput->addExternalLink( $pasteurized );
1293                                 }
1294                                 $s .= $text . $trail;
1295                         } else {
1296                                 $s .= $protocol . $remainder;
1297                         }
1298                 }
1299                 wfProfileOut( $fname );
1300                 return $s;
1301         }
1302
1303         /**
1304          * Replace unusual URL escape codes with their equivalent characters
1305          * @param string
1306          * @return string
1307          * @static
1308          * @fixme This can merge genuinely required bits in the path or query string,
1309          *        breaking legit URLs. A proper fix would treat the various parts of
1310          *        the URL differently; as a workaround, just use the output for
1311          *        statistical records, not for actual linking/output.
1312          */
1313         static function replaceUnusualEscapes( $url ) {
1314                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1315                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1316         }
1317
1318         /**
1319          * Callback function used in replaceUnusualEscapes().
1320          * Replaces unusual URL escape codes with their equivalent character
1321          * @static
1322          * @private
1323          */
1324         private static function replaceUnusualEscapesCallback( $matches ) {
1325                 $char = urldecode( $matches[0] );
1326                 $ord = ord( $char );
1327                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1328                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1329                         // No, shouldn't be escaped
1330                         return $char;
1331                 } else {
1332                         // Yes, leave it escaped
1333                         return $matches[0];
1334                 }
1335         }
1336
1337         /**
1338          * make an image if it's allowed, either through the global
1339          * option or through the exception
1340          * @private
1341          */
1342         function maybeMakeExternalImage( $url ) {
1343                 $sk =& $this->mOptions->getSkin();
1344                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1345                 $imagesexception = !empty($imagesfrom);
1346                 $text = false;
1347                 if ( $this->mOptions->getAllowExternalImages()
1348                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1349                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1350                                 # Image found
1351                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1352                         }
1353                 }
1354                 return $text;
1355         }
1356
1357         /**
1358          * Process [[ ]] wikilinks
1359          *
1360          * @private
1361          */
1362         function replaceInternalLinks( $s ) {
1363                 global $wgContLang;
1364                 static $fname = 'Parser::replaceInternalLinks' ;
1365
1366                 wfProfileIn( $fname );
1367
1368                 wfProfileIn( $fname.'-setup' );
1369                 static $tc = FALSE;
1370                 # the % is needed to support urlencoded titles as well
1371                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1372
1373                 $sk =& $this->mOptions->getSkin();
1374
1375                 #split the entire text string on occurences of [[
1376                 $a = explode( '[[', ' ' . $s );
1377                 #get the first element (all text up to first [[), and remove the space we added
1378                 $s = array_shift( $a );
1379                 $s = substr( $s, 1 );
1380
1381                 # Match a link having the form [[namespace:link|alternate]]trail
1382                 static $e1 = FALSE;
1383                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1384                 # Match cases where there is no "]]", which might still be images
1385                 static $e1_img = FALSE;
1386                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1387                 # Match the end of a line for a word that's not followed by whitespace,
1388                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1389                 $e2 = wfMsgForContent( 'linkprefix' );
1390
1391                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1392
1393                 if( is_null( $this->mTitle ) ) {
1394                         throw new MWException( 'nooo' );
1395                 }
1396                 $nottalk = !$this->mTitle->isTalkPage();
1397
1398                 if ( $useLinkPrefixExtension ) {
1399                         if ( preg_match( $e2, $s, $m ) ) {
1400                                 $first_prefix = $m[2];
1401                         } else {
1402                                 $first_prefix = false;
1403                         }
1404                 } else {
1405                         $prefix = '';
1406                 }
1407
1408                 $selflink = $this->mTitle->getPrefixedText();
1409                 wfProfileOut( $fname.'-setup' );
1410
1411                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1412                 $useSubpages = $this->areSubpagesAllowed();
1413
1414                 # Loop for each link
1415                 for ($k = 0; isset( $a[$k] ); $k++) {
1416                         $line = $a[$k];
1417                         if ( $useLinkPrefixExtension ) {
1418                                 wfProfileIn( $fname.'-prefixhandling' );
1419                                 if ( preg_match( $e2, $s, $m ) ) {
1420                                         $prefix = $m[2];
1421                                         $s = $m[1];
1422                                 } else {
1423                                         $prefix='';
1424                                 }
1425                                 # first link
1426                                 if($first_prefix) {
1427                                         $prefix = $first_prefix;
1428                                         $first_prefix = false;
1429                                 }
1430                                 wfProfileOut( $fname.'-prefixhandling' );
1431                         }
1432
1433                         $might_be_img = false;
1434
1435                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1436                                 $text = $m[2];
1437                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1438                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1439                                 # the real problem is with the $e1 regex
1440                                 # See bug 1300.
1441                                 #
1442                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1443                                 # and no image is in sight. See bug 2095.
1444                                 #
1445                                 if( $text !== '' &&
1446                                         preg_match( "/^\](.*)/s", $m[3], $n ) &&
1447                                         strpos($text, '[') !== false
1448                                 )
1449                                 {
1450                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1451                                         $m[3] = $n[1];
1452                                 }
1453                                 # fix up urlencoded title texts
1454                                 if(preg_match('/%/', $m[1] ))
1455                                         # Should anchors '#' also be rejected?
1456                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1457                                 $trail = $m[3];
1458                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1459                                 $might_be_img = true;
1460                                 $text = $m[2];
1461                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1462                                 $trail = "";
1463                         } else { # Invalid form; output directly
1464                                 $s .= $prefix . '[[' . $line ;
1465                                 continue;
1466                         }
1467
1468                         # Don't allow internal links to pages containing
1469                         # PROTO: where PROTO is a valid URL protocol; these
1470                         # should be external links.
1471                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1472                                 $s .= $prefix . '[[' . $line ;
1473                                 continue;
1474                         }
1475
1476                         # Make subpage if necessary
1477                         if( $useSubpages ) {
1478                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1479                         } else {
1480                                 $link = $m[1];
1481                         }
1482
1483                         $noforce = (substr($m[1], 0, 1) != ':');
1484                         if (!$noforce) {
1485                                 # Strip off leading ':'
1486                                 $link = substr($link, 1);
1487                         }
1488
1489                         $nt = Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1490                         if( !$nt ) {
1491                                 $s .= $prefix . '[[' . $line;
1492                                 continue;
1493                         }
1494
1495                         #check other language variants of the link
1496                         #if the article does not exist
1497                         if( $checkVariantLink
1498                             && $nt->getArticleID() == 0 ) {
1499                                 $wgContLang->findVariantLink($link, $nt);
1500                         }
1501
1502                         $ns = $nt->getNamespace();
1503                         $iw = $nt->getInterWiki();
1504
1505                         if ($might_be_img) { # if this is actually an invalid link
1506                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1507                                         $found = false;
1508                                         while (isset ($a[$k+1]) ) {
1509                                                 #look at the next 'line' to see if we can close it there
1510                                                 $spliced = array_splice( $a, $k + 1, 1 );
1511                                                 $next_line = array_shift( $spliced );
1512                                                 if( preg_match("/^(.*?]].*?)]](.*)$/sD", $next_line, $m) ) {
1513                                                 # the first ]] closes the inner link, the second the image
1514                                                         $found = true;
1515                                                         $text .= '[[' . $m[1];
1516                                                         $trail = $m[2];
1517                                                         break;
1518                                                 } elseif( preg_match("/^.*?]].*$/sD", $next_line, $m) ) {
1519                                                         #if there's exactly one ]] that's fine, we'll keep looking
1520                                                         $text .= '[[' . $m[0];
1521                                                 } else {
1522                                                         #if $next_line is invalid too, we need look no further
1523                                                         $text .= '[[' . $next_line;
1524                                                         break;
1525                                                 }
1526                                         }
1527                                         if ( !$found ) {
1528                                                 # we couldn't find the end of this imageLink, so output it raw
1529                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1530                                                 $text = $this->replaceInternalLinks($text);
1531                                                 $s .= $prefix . '[[' . $link . '|' . $text;
1532                                                 # note: no $trail, because without an end, there *is* no trail
1533                                                 continue;
1534                                         }
1535                                 } else { #it's not an image, so output it raw
1536                                         $s .= $prefix . '[[' . $link . '|' . $text;
1537                                         # note: no $trail, because without an end, there *is* no trail
1538                                         continue;
1539                                 }
1540                         }
1541
1542                         $wasblank = ( '' == $text );
1543                         if( $wasblank ) $text = $link;
1544
1545
1546                         # Link not escaped by : , create the various objects
1547                         if( $noforce ) {
1548
1549                                 # Interwikis
1550                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1551                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1552                                         $s = rtrim($s . "\n");
1553                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1554                                         continue;
1555                                 }
1556
1557                                 if ( $ns == NS_IMAGE ) {
1558                                         wfProfileIn( "$fname-image" );
1559                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1560                                                 # recursively parse links inside the image caption
1561                                                 # actually, this will parse them in any other parameters, too,
1562                                                 # but it might be hard to fix that, and it doesn't matter ATM
1563                                                 $text = $this->replaceExternalLinks($text);
1564                                                 $text = $this->replaceInternalLinks($text);
1565
1566                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1567                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1568                                                 $this->mOutput->addImage( $nt->getDBkey() );
1569
1570                                                 wfProfileOut( "$fname-image" );
1571                                                 continue;
1572                                         } else {
1573                                                 # We still need to record the image's presence on the page
1574                                                 $this->mOutput->addImage( $nt->getDBkey() );
1575                                         }
1576                                         wfProfileOut( "$fname-image" );
1577
1578                                 }
1579
1580                                 if ( $ns == NS_CATEGORY ) {
1581                                         wfProfileIn( "$fname-category" );
1582                                         $s = rtrim($s . "\n"); # bug 87
1583
1584                                         if ( $wasblank ) {
1585                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1586                                                         $sortkey = $this->mTitle->getText();
1587                                                 } else {
1588                                                         $sortkey = $this->mTitle->getPrefixedText();
1589                                                 }
1590                                         } else {
1591                                                 $sortkey = $text;
1592                                         }
1593                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1594                                         $sortkey = str_replace( "\n", '', $sortkey );
1595                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1596                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1597
1598                                         /**
1599                                          * Strip the whitespace Category links produce, see bug 87
1600                                          * @todo We might want to use trim($tmp, "\n") here.
1601                                          */
1602                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1603
1604                                         wfProfileOut( "$fname-category" );
1605                                         continue;
1606                                 }
1607                         }
1608
1609                         if( ( $nt->getPrefixedText() === $selflink ) &&
1610                             ( $nt->getFragment() === '' ) ) {
1611                                 # Self-links are handled specially; generally de-link and change to bold.
1612                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1613                                 continue;
1614                         }
1615
1616                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1617                         if( $ns == NS_MEDIA ) {
1618                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1619                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1620                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1621                                 $this->mOutput->addImage( $nt->getDBkey() );
1622                                 continue;
1623                         } elseif( $ns == NS_SPECIAL ) {
1624                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1625                                 continue;
1626                         } elseif( $ns == NS_IMAGE ) {
1627                                 $img = new Image( $nt );
1628                                 if( $img->exists() ) {
1629                                         // Force a blue link if the file exists; may be a remote
1630                                         // upload on the shared repository, and we want to see its
1631                                         // auto-generated page.
1632                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1633                                         continue;
1634                                 }
1635                         }
1636                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1637                 }
1638                 wfProfileOut( $fname );
1639                 return $s;
1640         }
1641
1642         /**
1643          * Make a link placeholder. The text returned can be later resolved to a real link with
1644          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1645          * parsing of interwiki links, and secondly to allow all extistence checks and
1646          * article length checks (for stub links) to be bundled into a single query.
1647          *
1648          */
1649         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1650                 if ( ! is_object($nt) ) {
1651                         # Fail gracefully
1652                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1653                 } else {
1654                         # Separate the link trail from the rest of the link
1655                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1656
1657                         if ( $nt->isExternal() ) {
1658                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1659                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1660                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1661                         } else {
1662                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1663                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1664                                 $this->mLinkHolders['queries'][] = $query;
1665                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1666                                 $this->mLinkHolders['titles'][] = $nt;
1667
1668                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1669                         }
1670                 }
1671                 return $retVal;
1672         }
1673
1674         /**
1675          * Render a forced-blue link inline; protect against double expansion of
1676          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1677          * Since this little disaster has to split off the trail text to avoid
1678          * breaking URLs in the following text without breaking trails on the
1679          * wiki links, it's been made into a horrible function.
1680          *
1681          * @param Title $nt
1682          * @param string $text
1683          * @param string $query
1684          * @param string $trail
1685          * @param string $prefix
1686          * @return string HTML-wikitext mix oh yuck
1687          */
1688         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1689                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1690                 $sk =& $this->mOptions->getSkin();
1691                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1692                 return $this->armorLinks( $link ) . $trail;
1693         }
1694
1695         /**
1696          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1697          * going to go through further parsing steps before inline URL expansion.
1698          *
1699          * In particular this is important when using action=render, which causes
1700          * full URLs to be included.
1701          *
1702          * Oh man I hate our multi-layer parser!
1703          *
1704          * @param string more-or-less HTML
1705          * @return string less-or-more HTML with NOPARSE bits
1706          */
1707         function armorLinks( $text ) {
1708                 return preg_replace( "/\b(" . wfUrlProtocols() . ')/',
1709                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1710         }
1711
1712         /**
1713          * Return true if subpage links should be expanded on this page.
1714          * @return bool
1715          */
1716         function areSubpagesAllowed() {
1717                 # Some namespaces don't allow subpages
1718                 global $wgNamespacesWithSubpages;
1719                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1720         }
1721
1722         /**
1723          * Handle link to subpage if necessary
1724          * @param string $target the source of the link
1725          * @param string &$text the link text, modified as necessary
1726          * @return string the full name of the link
1727          * @private
1728          */
1729         function maybeDoSubpageLink($target, &$text) {
1730                 # Valid link forms:
1731                 # Foobar -- normal
1732                 # :Foobar -- override special treatment of prefix (images, language links)
1733                 # /Foobar -- convert to CurrentPage/Foobar
1734                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1735                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1736                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1737
1738                 $fname = 'Parser::maybeDoSubpageLink';
1739                 wfProfileIn( $fname );
1740                 $ret = $target; # default return value is no change
1741
1742                 # Some namespaces don't allow subpages,
1743                 # so only perform processing if subpages are allowed
1744                 if( $this->areSubpagesAllowed() ) {
1745                         # Look at the first character
1746                         if( $target != '' && $target{0} == '/' ) {
1747                                 # / at end means we don't want the slash to be shown
1748                                 if( substr( $target, -1, 1 ) == '/' ) {
1749                                         $target = substr( $target, 1, -1 );
1750                                         $noslash = $target;
1751                                 } else {
1752                                         $noslash = substr( $target, 1 );
1753                                 }
1754
1755                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1756                                 if( '' === $text ) {
1757                                         $text = $target;
1758                                 } # this might be changed for ugliness reasons
1759                         } else {
1760                                 # check for .. subpage backlinks
1761                                 $dotdotcount = 0;
1762                                 $nodotdot = $target;
1763                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1764                                         ++$dotdotcount;
1765                                         $nodotdot = substr( $nodotdot, 3 );
1766                                 }
1767                                 if($dotdotcount > 0) {
1768                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1769                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1770                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1771                                                 # / at the end means don't show full path
1772                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1773                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1774                                                         if( '' === $text ) {
1775                                                                 $text = $nodotdot;
1776                                                         }
1777                                                 }
1778                                                 $nodotdot = trim( $nodotdot );
1779                                                 if( $nodotdot != '' ) {
1780                                                         $ret .= '/' . $nodotdot;
1781                                                 }
1782                                         }
1783                                 }
1784                         }
1785                 }
1786
1787                 wfProfileOut( $fname );
1788                 return $ret;
1789         }
1790
1791         /**#@+
1792          * Used by doBlockLevels()
1793          * @private
1794          */
1795         /* private */ function closeParagraph() {
1796                 $result = '';
1797                 if ( '' != $this->mLastSection ) {
1798                         $result = '</' . $this->mLastSection  . ">\n";
1799                 }
1800                 $this->mInPre = false;
1801                 $this->mLastSection = '';
1802                 return $result;
1803         }
1804         # getCommon() returns the length of the longest common substring
1805         # of both arguments, starting at the beginning of both.
1806         #
1807         /* private */ function getCommon( $st1, $st2 ) {
1808                 $fl = strlen( $st1 );
1809                 $shorter = strlen( $st2 );
1810                 if ( $fl < $shorter ) { $shorter = $fl; }
1811
1812                 for ( $i = 0; $i < $shorter; ++$i ) {
1813                         if ( $st1{$i} != $st2{$i} ) { break; }
1814                 }
1815                 return $i;
1816         }
1817         # These next three functions open, continue, and close the list
1818         # element appropriate to the prefix character passed into them.
1819         #
1820         /* private */ function openList( $char ) {
1821                 $result = $this->closeParagraph();
1822
1823                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1824                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1825                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1826                 else if ( ';' == $char ) {
1827                         $result .= '<dl><dt>';
1828                         $this->mDTopen = true;
1829                 }
1830                 else { $result = '<!-- ERR 1 -->'; }
1831
1832                 return $result;
1833         }
1834
1835         /* private */ function nextItem( $char ) {
1836                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1837                 else if ( ':' == $char || ';' == $char ) {
1838                         $close = '</dd>';
1839                         if ( $this->mDTopen ) { $close = '</dt>'; }
1840                         if ( ';' == $char ) {
1841                                 $this->mDTopen = true;
1842                                 return $close . '<dt>';
1843                         } else {
1844                                 $this->mDTopen = false;
1845                                 return $close . '<dd>';
1846                         }
1847                 }
1848                 return '<!-- ERR 2 -->';
1849         }
1850
1851         /* private */ function closeList( $char ) {
1852                 if ( '*' == $char ) { $text = '</li></ul>'; }
1853                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1854                 else if ( ':' == $char ) {
1855                         if ( $this->mDTopen ) {
1856                                 $this->mDTopen = false;
1857                                 $text = '</dt></dl>';
1858                         } else {
1859                                 $text = '</dd></dl>';
1860                         }
1861                 }
1862                 else {  return '<!-- ERR 3 -->'; }
1863                 return $text."\n";
1864         }
1865         /**#@-*/
1866
1867         /**
1868          * Make lists from lines starting with ':', '*', '#', etc.
1869          *
1870          * @private
1871          * @return string the lists rendered as HTML
1872          */
1873         function doBlockLevels( $text, $linestart ) {
1874                 $fname = 'Parser::doBlockLevels';
1875                 wfProfileIn( $fname );
1876
1877                 # Parsing through the text line by line.  The main thing
1878                 # happening here is handling of block-level elements p, pre,
1879                 # and making lists from lines starting with * # : etc.
1880                 #
1881                 $textLines = explode( "\n", $text );
1882
1883                 $lastPrefix = $output = '';
1884                 $this->mDTopen = $inBlockElem = false;
1885                 $prefixLength = 0;
1886                 $paragraphStack = false;
1887
1888                 if ( !$linestart ) {
1889                         $output .= array_shift( $textLines );
1890                 }
1891                 foreach ( $textLines as $oLine ) {
1892                         $lastPrefixLength = strlen( $lastPrefix );
1893                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1894                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1895                         if ( !$this->mInPre ) {
1896                                 # Multiple prefixes may abut each other for nested lists.
1897                                 $prefixLength = strspn( $oLine, '*#:;' );
1898                                 $pref = substr( $oLine, 0, $prefixLength );
1899
1900                                 # eh?
1901                                 $pref2 = str_replace( ';', ':', $pref );
1902                                 $t = substr( $oLine, $prefixLength );
1903                                 $this->mInPre = !empty($preOpenMatch);
1904                         } else {
1905                                 # Don't interpret any other prefixes in preformatted text
1906                                 $prefixLength = 0;
1907                                 $pref = $pref2 = '';
1908                                 $t = $oLine;
1909                         }
1910
1911                         # List generation
1912                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1913                                 # Same as the last item, so no need to deal with nesting or opening stuff
1914                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1915                                 $paragraphStack = false;
1916
1917                                 if ( substr( $pref, -1 ) == ';') {
1918                                         # The one nasty exception: definition lists work like this:
1919                                         # ; title : definition text
1920                                         # So we check for : in the remainder text to split up the
1921                                         # title and definition, without b0rking links.
1922                                         $term = $t2 = '';
1923                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1924                                                 $t = $t2;
1925                                                 $output .= $term . $this->nextItem( ':' );
1926                                         }
1927                                 }
1928                         } elseif( $prefixLength || $lastPrefixLength ) {
1929                                 # Either open or close a level...
1930                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
1931                                 $paragraphStack = false;
1932
1933                                 while( $commonPrefixLength < $lastPrefixLength ) {
1934                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
1935                                         --$lastPrefixLength;
1936                                 }
1937                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
1938                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
1939                                 }
1940                                 while ( $prefixLength > $commonPrefixLength ) {
1941                                         $char = substr( $pref, $commonPrefixLength, 1 );
1942                                         $output .= $this->openList( $char );
1943
1944                                         if ( ';' == $char ) {
1945                                                 # FIXME: This is dupe of code above
1946                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1947                                                         $t = $t2;
1948                                                         $output .= $term . $this->nextItem( ':' );
1949                                                 }
1950                                         }
1951                                         ++$commonPrefixLength;
1952                                 }
1953                                 $lastPrefix = $pref2;
1954                         }
1955                         if( 0 == $prefixLength ) {
1956                                 wfProfileIn( "$fname-paragraph" );
1957                                 # No prefix (not in list)--go to paragraph mode
1958                                 // XXX: use a stack for nestable elements like span, table and div
1959                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/center|<\\/tr|<\\/td|<\\/th)/iS', $t );
1960                                 $closematch = preg_match(
1961                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
1962                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<center)/iS', $t );
1963                                 if ( $openmatch or $closematch ) {
1964                                         $paragraphStack = false;
1965                                         # TODO bug 5718: paragraph closed
1966                                         $output .= $this->closeParagraph();
1967                                         if ( $preOpenMatch and !$preCloseMatch ) {
1968                                                 $this->mInPre = true;
1969                                         }
1970                                         if ( $closematch ) {
1971                                                 $inBlockElem = false;
1972                                         } else {
1973                                                 $inBlockElem = true;
1974                                         }
1975                                 } else if ( !$inBlockElem && !$this->mInPre ) {
1976                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
1977                                                 // pre
1978                                                 if ($this->mLastSection != 'pre') {
1979                                                         $paragraphStack = false;
1980                                                         $output .= $this->closeParagraph().'<pre>';
1981                                                         $this->mLastSection = 'pre';
1982                                                 }
1983                                                 $t = substr( $t, 1 );
1984                                         } else {
1985                                                 // paragraph
1986                                                 if ( '' == trim($t) ) {
1987                                                         if ( $paragraphStack ) {
1988                                                                 $output .= $paragraphStack.'<br />';
1989                                                                 $paragraphStack = false;
1990                                                                 $this->mLastSection = 'p';
1991                                                         } else {
1992                                                                 if ($this->mLastSection != 'p' ) {
1993                                                                         $output .= $this->closeParagraph();
1994                                                                         $this->mLastSection = '';
1995                                                                         $paragraphStack = '<p>';
1996                                                                 } else {
1997                                                                         $paragraphStack = '</p><p>';
1998                                                                 }
1999                                                         }
2000                                                 } else {
2001                                                         if ( $paragraphStack ) {
2002                                                                 $output .= $paragraphStack;
2003                                                                 $paragraphStack = false;
2004                                                                 $this->mLastSection = 'p';
2005                                                         } else if ($this->mLastSection != 'p') {
2006                                                                 $output .= $this->closeParagraph().'<p>';
2007                                                                 $this->mLastSection = 'p';
2008                                                         }
2009                                                 }
2010                                         }
2011                                 }
2012                                 wfProfileOut( "$fname-paragraph" );
2013                         }
2014                         // somewhere above we forget to get out of pre block (bug 785)
2015                         if($preCloseMatch && $this->mInPre) {
2016                                 $this->mInPre = false;
2017                         }
2018                         if ($paragraphStack === false) {
2019                                 $output .= $t."\n";
2020                         }
2021                 }
2022                 while ( $prefixLength ) {
2023                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2024                         --$prefixLength;
2025                 }
2026                 if ( '' != $this->mLastSection ) {
2027                         $output .= '</' . $this->mLastSection . '>';
2028                         $this->mLastSection = '';
2029                 }
2030
2031                 wfProfileOut( $fname );
2032                 return $output;
2033         }
2034
2035         /**
2036          * Split up a string on ':', ignoring any occurences inside tags
2037          * to prevent illegal overlapping.
2038          * @param string $str the string to split
2039          * @param string &$before set to everything before the ':'
2040          * @param string &$after set to everything after the ':'
2041          * return string the position of the ':', or false if none found
2042          */
2043         function findColonNoLinks($str, &$before, &$after) {
2044                 $fname = 'Parser::findColonNoLinks';
2045                 wfProfileIn( $fname );
2046
2047                 $pos = strpos( $str, ':' );
2048                 if( $pos === false ) {
2049                         // Nothing to find!
2050                         wfProfileOut( $fname );
2051                         return false;
2052                 }
2053
2054                 $lt = strpos( $str, '<' );
2055                 if( $lt === false || $lt > $pos ) {
2056                         // Easy; no tag nesting to worry about
2057                         $before = substr( $str, 0, $pos );
2058                         $after = substr( $str, $pos+1 );
2059                         wfProfileOut( $fname );
2060                         return $pos;
2061                 }
2062
2063                 // Ugly state machine to walk through avoiding tags.
2064                 $state = MW_COLON_STATE_TEXT;
2065                 $stack = 0;
2066                 $len = strlen( $str );
2067                 for( $i = 0; $i < $len; $i++ ) {
2068                         $c = $str{$i};
2069
2070                         switch( $state ) {
2071                         // (Using the number is a performance hack for common cases)
2072                         case 0: // MW_COLON_STATE_TEXT:
2073                                 switch( $c ) {
2074                                 case "<":
2075                                         // Could be either a <start> tag or an </end> tag
2076                                         $state = MW_COLON_STATE_TAGSTART;
2077                                         break;
2078                                 case ":":
2079                                         if( $stack == 0 ) {
2080                                                 // We found it!
2081                                                 $before = substr( $str, 0, $i );
2082                                                 $after = substr( $str, $i + 1 );
2083                                                 wfProfileOut( $fname );
2084                                                 return $i;
2085                                         }
2086                                         // Embedded in a tag; don't break it.
2087                                         break;
2088                                 default:
2089                                         // Skip ahead looking for something interesting
2090                                         $colon = strpos( $str, ':', $i );
2091                                         if( $colon === false ) {
2092                                                 // Nothing else interesting
2093                                                 wfProfileOut( $fname );
2094                                                 return false;
2095                                         }
2096                                         $lt = strpos( $str, '<', $i );
2097                                         if( $stack === 0 ) {
2098                                                 if( $lt === false || $colon < $lt ) {
2099                                                         // We found it!
2100                                                         $before = substr( $str, 0, $colon );
2101                                                         $after = substr( $str, $colon + 1 );
2102                                                         wfProfileOut( $fname );
2103                                                         return $i;
2104                                                 }
2105                                         }
2106                                         if( $lt === false ) {
2107                                                 // Nothing else interesting to find; abort!
2108                                                 // We're nested, but there's no close tags left. Abort!
2109                                                 break 2;
2110                                         }
2111                                         // Skip ahead to next tag start
2112                                         $i = $lt;
2113                                         $state = MW_COLON_STATE_TAGSTART;
2114                                 }
2115                                 break;
2116                         case 1: // MW_COLON_STATE_TAG:
2117                                 // In a <tag>
2118                                 switch( $c ) {
2119                                 case ">":
2120                                         $stack++;
2121                                         $state = MW_COLON_STATE_TEXT;
2122                                         break;
2123                                 case "/":
2124                                         // Slash may be followed by >?
2125                                         $state = MW_COLON_STATE_TAGSLASH;
2126                                         break;
2127                                 default:
2128                                         // ignore
2129                                 }
2130                                 break;
2131                         case 2: // MW_COLON_STATE_TAGSTART:
2132                                 switch( $c ) {
2133                                 case "/":
2134                                         $state = MW_COLON_STATE_CLOSETAG;
2135                                         break;
2136                                 case "!":
2137                                         $state = MW_COLON_STATE_COMMENT;
2138                                         break;
2139                                 case ">":
2140                                         // Illegal early close? This shouldn't happen D:
2141                                         $state = MW_COLON_STATE_TEXT;
2142                                         break;
2143                                 default:
2144                                         $state = MW_COLON_STATE_TAG;
2145                                 }
2146                                 break;
2147                         case 3: // MW_COLON_STATE_CLOSETAG:
2148                                 // In a </tag>
2149                                 if( $c == ">" ) {
2150                                         $stack--;
2151                                         if( $stack < 0 ) {
2152                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2153                                                 wfProfileOut( $fname );
2154                                                 return false;
2155                                         }
2156                                         $state = MW_COLON_STATE_TEXT;
2157                                 }
2158                                 break;
2159                         case MW_COLON_STATE_TAGSLASH:
2160                                 if( $c == ">" ) {
2161                                         // Yes, a self-closed tag <blah/>
2162                                         $state = MW_COLON_STATE_TEXT;
2163                                 } else {
2164                                         // Probably we're jumping the gun, and this is an attribute
2165                                         $state = MW_COLON_STATE_TAG;
2166                                 }
2167                                 break;
2168                         case 5: // MW_COLON_STATE_COMMENT:
2169                                 if( $c == "-" ) {
2170                                         $state = MW_COLON_STATE_COMMENTDASH;
2171                                 }
2172                                 break;
2173                         case MW_COLON_STATE_COMMENTDASH:
2174                                 if( $c == "-" ) {
2175                                         $state = MW_COLON_STATE_COMMENTDASHDASH;
2176                                 } else {
2177                                         $state = MW_COLON_STATE_COMMENT;
2178                                 }
2179                                 break;
2180                         case MW_COLON_STATE_COMMENTDASHDASH:
2181                                 if( $c == ">" ) {
2182                                         $state = MW_COLON_STATE_TEXT;
2183                                 } else {
2184                                         $state = MW_COLON_STATE_COMMENT;
2185                                 }
2186                                 break;
2187                         default:
2188                                 throw new MWException( "State machine error in $fname" );
2189                         }
2190                 }
2191                 if( $stack > 0 ) {
2192                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2193                         return false;
2194                 }
2195                 wfProfileOut( $fname );
2196                 return false;
2197         }
2198
2199         /**
2200          * Return value of a magic variable (like PAGENAME)
2201          *
2202          * @private
2203          */
2204         function getVariableValue( $index ) {
2205                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2206
2207                 /**
2208                  * Some of these require message or data lookups and can be
2209                  * expensive to check many times.
2210                  */
2211                 static $varCache = array();
2212                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) )
2213                         if ( isset( $varCache[$index] ) )
2214                                 return $varCache[$index];
2215
2216                 $ts = time();
2217                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2218
2219                 switch ( $index ) {
2220                         case 'currentmonth':
2221                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
2222                         case 'currentmonthname':
2223                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2224                         case 'currentmonthnamegen':
2225                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2226                         case 'currentmonthabbrev':
2227                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2228                         case 'currentday':
2229                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2230                         case 'currentday2':
2231                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2232                         case 'pagename':
2233                                 return $this->mTitle->getText();
2234                         case 'pagenamee':
2235                                 return $this->mTitle->getPartialURL();
2236                         case 'fullpagename':
2237                                 return $this->mTitle->getPrefixedText();
2238                         case 'fullpagenamee':
2239                                 return $this->mTitle->getPrefixedURL();
2240                         case 'subpagename':
2241                                 return $this->mTitle->getSubpageText();
2242                         case 'subpagenamee':
2243                                 return $this->mTitle->getSubpageUrlForm();
2244                         case 'basepagename':
2245                                 return $this->mTitle->getBaseText();
2246                         case 'basepagenamee':
2247                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2248                         case 'talkpagename':
2249                                 if( $this->mTitle->canTalk() ) {
2250                                         $talkPage = $this->mTitle->getTalkPage();
2251                                         return $talkPage->getPrefixedText();
2252                                 } else {
2253                                         return '';
2254                                 }
2255                         case 'talkpagenamee':
2256                                 if( $this->mTitle->canTalk() ) {
2257                                         $talkPage = $this->mTitle->getTalkPage();
2258                                         return $talkPage->getPrefixedUrl();
2259                                 } else {
2260                                         return '';
2261                                 }
2262                         case 'subjectpagename':
2263                                 $subjPage = $this->mTitle->getSubjectPage();
2264                                 return $subjPage->getPrefixedText();
2265                         case 'subjectpagenamee':
2266                                 $subjPage = $this->mTitle->getSubjectPage();
2267                                 return $subjPage->getPrefixedUrl();
2268                         case 'revisionid':
2269                                 return $this->mRevisionId;
2270                         case 'namespace':
2271                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2272                         case 'namespacee':
2273                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2274                         case 'talkspace':
2275                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2276                         case 'talkspacee':
2277                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2278                         case 'subjectspace':
2279                                 return $this->mTitle->getSubjectNsText();
2280                         case 'subjectspacee':
2281                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2282                         case 'currentdayname':
2283                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2284                         case 'currentyear':
2285                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2286                         case 'currenttime':
2287                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2288                         case 'currentweek':
2289                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2290                                 // int to remove the padding
2291                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2292                         case 'currentdow':
2293                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2294                         case 'numberofarticles':
2295                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
2296                         case 'numberoffiles':
2297                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
2298                         case 'numberofusers':
2299                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfUsers() );
2300                         case 'numberofpages':
2301                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfPages() );
2302                         case 'numberofadmins':
2303                                 return $varCache[$index]  = $wgContLang->formatNum( wfNumberOfAdmins() );
2304                         case 'currenttimestamp':
2305                                 return $varCache[$index] = wfTimestampNow();
2306                         case 'currentversion':
2307                                 global $wgVersion;
2308                                 return $wgVersion;
2309                         case 'sitename':
2310                                 return $wgSitename;
2311                         case 'server':
2312                                 return $wgServer;
2313                         case 'servername':
2314                                 return $wgServerName;
2315                         case 'scriptpath':
2316                                 return $wgScriptPath;
2317                         case 'directionmark':
2318                                 return $wgContLang->getDirMark();
2319                         case 'contentlanguage':
2320                                 global $wgContLanguageCode;
2321                                 return $wgContLanguageCode;
2322                         default:
2323                                 $ret = null;
2324                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2325                                         return $ret;
2326                                 else
2327                                         return null;
2328                 }
2329         }
2330
2331         /**
2332          * initialise the magic variables (like CURRENTMONTHNAME)
2333          *
2334          * @private
2335          */
2336         function initialiseVariables() {
2337                 $fname = 'Parser::initialiseVariables';
2338                 wfProfileIn( $fname );
2339                 $variableIDs = MagicWord::getVariableIDs();
2340
2341                 $this->mVariables = array();
2342                 foreach ( $variableIDs as $id ) {
2343                         $mw =& MagicWord::get( $id );
2344                         $mw->addToArray( $this->mVariables, $id );
2345                 }
2346                 wfProfileOut( $fname );
2347         }
2348
2349         /**
2350          * parse any parentheses in format ((title|part|part))
2351          * and call callbacks to get a replacement text for any found piece
2352          *
2353          * @param string $text The text to parse
2354          * @param array $callbacks rules in form:
2355          *     '{' => array(                            # opening parentheses
2356          *                                      'end' => '}',   # closing parentheses
2357          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2358          *                                                                4 => callback         # replacement callback to call if {{{{..}}}} is found
2359          *                                                                )
2360          *                                      )
2361          * @private
2362          */
2363         function replace_callback ($text, $callbacks) {
2364                 wfProfileIn( __METHOD__ . '-self' );
2365                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2366                 $lastOpeningBrace = -1;                 # last not closed parentheses
2367
2368                 for ($i = 0; $i < strlen($text); $i++) {
2369                         # check for any opening brace
2370                         $rule = null;
2371                         $nextPos = -1;
2372                         foreach ($callbacks as $key => $value) {
2373                                 $pos = strpos ($text, $key, $i);
2374                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)) {
2375                                         $rule = $value;
2376                                         $nextPos = $pos;
2377                                 }
2378                         }
2379
2380                         if ($lastOpeningBrace >= 0) {
2381                                 $pos = strpos ($text, $openingBraceStack[$lastOpeningBrace]['braceEnd'], $i);
2382
2383                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2384                                         $rule = null;
2385                                         $nextPos = $pos;
2386                                 }
2387
2388                                 $pos = strpos ($text, '|', $i);
2389
2390                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2391                                         $rule = null;
2392                                         $nextPos = $pos;
2393                                 }
2394                         }
2395
2396                         if ($nextPos == -1)
2397                                 break;
2398
2399                         $i = $nextPos;
2400
2401                         # found openning brace, lets add it to parentheses stack
2402                         if (null != $rule) {
2403                                 $piece = array('brace' => $text[$i],
2404                                                            'braceEnd' => $rule['end'],
2405                                                            'count' => 1,
2406                                                            'title' => '',
2407                                                            'parts' => null);
2408
2409                                 # count openning brace characters
2410                                 while ($i+1 < strlen($text) && $text[$i+1] == $piece['brace']) {
2411                                         $piece['count']++;
2412                                         $i++;
2413                                 }
2414
2415                                 $piece['startAt'] = $i+1;
2416                                 $piece['partStart'] = $i+1;
2417
2418                                 # we need to add to stack only if openning brace count is enough for any given rule
2419                                 foreach ($rule['cb'] as $cnt => $fn) {
2420                                         if ($piece['count'] >= $cnt) {
2421                                                 $lastOpeningBrace ++;
2422                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2423                                                 break;
2424                                         }
2425                                 }
2426
2427                                 continue;
2428                         }
2429                         else if ($lastOpeningBrace >= 0) {
2430                                 # first check if it is a closing brace
2431                                 if ($openingBraceStack[$lastOpeningBrace]['braceEnd'] == $text[$i]) {
2432                                         # lets check if it is enough characters for closing brace
2433                                         $count = 1;
2434                                         while ($i+$count < strlen($text) && $text[$i+$count] == $text[$i])
2435                                                 $count++;
2436
2437                                         # if there are more closing parentheses than opening ones, we parse less
2438                                         if ($openingBraceStack[$lastOpeningBrace]['count'] < $count)
2439                                                 $count = $openingBraceStack[$lastOpeningBrace]['count'];
2440
2441                                         # check for maximum matching characters (if there are 5 closing characters, we will probably need only 3 - depending on the rules)
2442                                         $matchingCount = 0;
2443                                         $matchingCallback = null;
2444                                         foreach ($callbacks[$openingBraceStack[$lastOpeningBrace]['brace']]['cb'] as $cnt => $fn) {
2445                                                 if ($count >= $cnt && $matchingCount < $cnt) {
2446                                                         $matchingCount = $cnt;
2447                                                         $matchingCallback = $fn;
2448                                                 }
2449                                         }
2450
2451                                         if ($matchingCount == 0) {
2452                                                 $i += $count - 1;
2453                                                 continue;
2454                                         }
2455
2456                                         # lets set a title or last part (if '|' was found)
2457                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts'])
2458                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2459                                         else
2460                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2461
2462                                         $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2463                                         $pieceEnd = $i + $matchingCount;
2464
2465                                         if( is_callable( $matchingCallback ) ) {
2466                                                 $cbArgs = array (
2467                                                                                  'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2468                                                                                  'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2469                                                                                  'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2470                                                                                  'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == "\n")),
2471                                                                                  );
2472                                                 # finally we can call a user callback and replace piece of text
2473                                                 wfProfileOut( __METHOD__ . '-self' );
2474                                                 $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2475                                                 wfProfileIn( __METHOD__ . '-self' );
2476                                                 $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2477                                                 $i = $pieceStart + strlen($replaceWith) - 1;
2478                                         }
2479                                         else {
2480                                                 # null value for callback means that parentheses should be parsed, but not replaced
2481                                                 $i += $matchingCount - 1;
2482                                         }
2483
2484                                         # reset last openning parentheses, but keep it in case there are unused characters
2485                                         $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2486                                                                    'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2487                                                                    'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2488                                                                    'title' => '',
2489                                                                    'parts' => null,
2490                                                                    'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2491                                         $openingBraceStack[$lastOpeningBrace--] = null;
2492
2493                                         if ($matchingCount < $piece['count']) {
2494                                                 $piece['count'] -= $matchingCount;
2495                                                 $piece['startAt'] -= $matchingCount;
2496                                                 $piece['partStart'] = $piece['startAt'];
2497                                                 # do we still qualify for any callback with remaining count?
2498                                                 foreach ($callbacks[$piece['brace']]['cb'] as $cnt => $fn) {
2499                                                         if ($piece['count'] >= $cnt) {
2500                                                                 $lastOpeningBrace ++;
2501                                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2502                                                                 break;
2503                                                         }
2504                                                 }
2505                                         }
2506                                         continue;
2507                                 }
2508
2509                                 # lets set a title if it is a first separator, or next part otherwise
2510                                 if ($text[$i] == '|') {
2511                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2512                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2513                                                 $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2514                                         }
2515                                         else
2516                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2517
2518                                         $openingBraceStack[$lastOpeningBrace]['partStart'] = $i + 1;
2519                                 }
2520                         }
2521                 }
2522
2523                 wfProfileOut( __METHOD__ . '-self' );
2524                 return $text;
2525         }
2526
2527         /**
2528          * Replace magic variables, templates, and template arguments
2529          * with the appropriate text. Templates are substituted recursively,
2530          * taking care to avoid infinite loops.
2531          *
2532          * Note that the substitution depends on value of $mOutputType:
2533          *  OT_WIKI: only {{subst:}} templates
2534          *  OT_MSG: only magic variables
2535          *  OT_HTML: all templates and magic variables
2536          *
2537          * @param string $tex The text to transform
2538          * @param array $args Key-value pairs representing template parameters to substitute
2539          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2540          * @private
2541          */
2542         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2543                 # Prevent too big inclusions
2544                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
2545                         return $text;
2546                 }
2547
2548                 $fname = 'Parser::replaceVariables';
2549                 wfProfileIn( $fname );
2550
2551                 # This function is called recursively. To keep track of arguments we need a stack:
2552                 array_push( $this->mArgStack, $args );
2553
2554                 $braceCallbacks = array();
2555                 if ( !$argsOnly ) {
2556                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2557                 }
2558                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
2559                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2560                 }
2561                 $callbacks = array();
2562                 $callbacks['{'] = array('end' => '}', 'cb' => $braceCallbacks);
2563                 $callbacks['['] = array('end' => ']', 'cb' => array(2=>null));
2564                 $text = $this->replace_callback ($text, $callbacks);
2565
2566                 array_pop( $this->mArgStack );
2567
2568                 wfProfileOut( $fname );
2569                 return $text;
2570         }
2571
2572         /**
2573          * Replace magic variables
2574          * @private
2575          */
2576         function variableSubstitution( $matches ) {
2577                 $fname = 'Parser::variableSubstitution';
2578                 $varname = $matches[1];
2579                 wfProfileIn( $fname );
2580                 $skip = false;
2581                 if ( $this->mOutputType == OT_WIKI ) {
2582                         # Do only magic variables prefixed by SUBST
2583                         $mwSubst =& MagicWord::get( 'subst' );
2584                         if (!$mwSubst->matchStartAndRemove( $varname ))
2585                                 $skip = true;
2586                         # Note that if we don't substitute the variable below,
2587                         # we don't remove the {{subst:}} magic word, in case
2588                         # it is a template rather than a magic variable.
2589                 }
2590                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2591                         $id = $this->mVariables[$varname];
2592                         $text = $this->getVariableValue( $id );
2593                         $this->mOutput->mContainsOldMagic = true;
2594                 } else {
2595                         $text = $matches[0];
2596                 }
2597                 wfProfileOut( $fname );
2598                 return $text;
2599         }
2600
2601         # Split template arguments
2602         function getTemplateArgs( $argsString ) {
2603                 if ( $argsString === '' ) {
2604                         return array();
2605                 }
2606
2607                 $args = explode( '|', substr( $argsString, 1 ) );
2608
2609                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2610                 # merged with the next arg because the '|' character between belongs
2611                 # to the link syntax and not the template parameter syntax.
2612                 $argc = count($args);
2613
2614                 for ( $i = 0; $i < $argc-1; $i++ ) {
2615                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2616                                 $args[$i] .= '|'.$args[$i+1];
2617                                 array_splice($args, $i+1, 1);
2618                                 $i--;
2619                                 $argc--;
2620                         }
2621                 }
2622
2623                 return $args;
2624         }
2625
2626         /**
2627          * Return the text of a template, after recursively
2628          * replacing any variables or templates within the template.
2629          *
2630          * @param array $piece The parts of the template
2631          *  $piece['text']: matched text
2632          *  $piece['title']: the title, i.e. the part before the |
2633          *  $piece['parts']: the parameter array
2634          * @return string the text of the template
2635          * @private
2636          */
2637         function braceSubstitution( $piece ) {
2638                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $action;
2639                 $fname = 'Parser::braceSubstitution';
2640                 wfProfileIn( $fname );
2641
2642                 # Flags
2643                 $found = false;             # $text has been filled
2644                 $nowiki = false;            # wiki markup in $text should be escaped
2645                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2646                 $noargs = false;            # Don't replace triple-brace arguments in $text
2647                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2648                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2649                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2650
2651                 # Title object, where $text came from
2652                 $title = NULL;
2653
2654                 $linestart = '';
2655
2656                 # $part1 is the bit before the first |, and must contain only title characters
2657                 # $args is a list of arguments, starting from index 0, not including $part1
2658
2659                 $part1 = $piece['title'];
2660                 # If the third subpattern matched anything, it will start with |
2661
2662                 if (null == $piece['parts']) {
2663                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2664                         if ($replaceWith != $piece['text']) {
2665                                 $text = $replaceWith;
2666                                 $found = true;
2667                                 $noparse = true;
2668                                 $noargs = true;
2669                         }
2670                 }
2671
2672                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2673                 $argc = count( $args );
2674
2675                 # SUBST
2676                 if ( !$found ) {
2677                         $mwSubst =& MagicWord::get( 'subst' );
2678                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2679                                 # One of two possibilities is true:
2680                                 # 1) Found SUBST but not in the PST phase
2681                                 # 2) Didn't find SUBST and in the PST phase
2682                                 # In either case, return without further processing
2683                                 $text = $piece['text'];
2684                                 $found = true;
2685                                 $noparse = true;
2686                                 $noargs = true;
2687                         }
2688                 }
2689
2690                 # MSG, MSGNW, INT and RAW
2691                 if ( !$found ) {
2692                         # Check for MSGNW:
2693                         $mwMsgnw =& MagicWord::get( 'msgnw' );
2694                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2695                                 $nowiki = true;
2696                         } else {
2697                                 # Remove obsolete MSG:
2698                                 $mwMsg =& MagicWord::get( 'msg' );
2699                                 $mwMsg->matchStartAndRemove( $part1 );
2700                         }
2701
2702                         # Check for RAW:
2703                         $mwRaw =& MagicWord::get( 'raw' );
2704                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2705                                 $forceRawInterwiki = true;
2706                         }
2707
2708                         # Check if it is an internal message
2709                         $mwInt =& MagicWord::get( 'int' );
2710                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2711                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2712                                         $text = $linestart . wfMsgReal( $part1, $args, true );
2713                                         $found = true;
2714                                 }
2715                         }
2716                 }
2717
2718                 # Parser functions
2719                 if ( !$found ) {
2720                         wfProfileIn( __METHOD__ . '-pfunc' );
2721
2722                         $colonPos = strpos( $part1, ':' );
2723                         if ( $colonPos !== false ) {
2724                                 # Case sensitive functions
2725                                 $function = substr( $part1, 0, $colonPos );
2726                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2727                                         $function = $this->mFunctionSynonyms[1][$function];
2728                                 } else {
2729                                         # Case insensitive functions
2730                                         $function = strtolower( $function );
2731                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2732                                                 $function = $this->mFunctionSynonyms[0][$function];
2733                                         } else {
2734                                                 $function = false;
2735                                         }
2736                                 }
2737                                 if ( $function ) {
2738                                         $funcArgs = array_map( 'trim', $args );
2739                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
2740                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
2741                                         $found = true;
2742
2743                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
2744                                         //$noargs = true;
2745                                         //$noparse = true;
2746
2747                                         if ( is_array( $result ) ) {
2748                                                 if ( isset( $result[0] ) ) {
2749                                                         $text = $linestart . $result[0];
2750                                                         unset( $result[0] );
2751                                                 }
2752
2753                                                 // Extract flags into the local scope
2754                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
2755                                                 extract( $result );
2756                                         } else {
2757                                                 $text = $linestart . $result;
2758                                         }
2759                                 }
2760                         }
2761                         wfProfileOut( __METHOD__ . '-pfunc' );
2762                 }
2763
2764                 # Template table test
2765
2766                 # Did we encounter this template already? If yes, it is in the cache
2767                 # and we need to check for loops.
2768                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
2769                         $found = true;
2770
2771                         # Infinite loop test
2772                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2773                                 $noparse = true;
2774                                 $noargs = true;
2775                                 $found = true;
2776                                 $text = $linestart .
2777                                         '{{' . $part1 . '}}' .
2778                                         '<!-- WARNING: template loop detected -->';
2779                                 wfDebug( "$fname: template loop broken at '$part1'\n" );
2780                         } else {
2781                                 # set $text to cached message.
2782                                 $text = $linestart . $this->mTemplates[$piece['title']];
2783                         }
2784                 }
2785
2786                 # Load from database
2787                 $lastPathLevel = $this->mTemplatePath;
2788                 if ( !$found ) {
2789                         wfProfileIn( __METHOD__ . '-loadtpl' );
2790                         $ns = NS_TEMPLATE;
2791                         # declaring $subpage directly in the function call
2792                         # does not work correctly with references and breaks
2793                         # {{/subpage}}-style inclusions
2794                         $subpage = '';
2795                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2796                         if ($subpage !== '') {
2797                                 $ns = $this->mTitle->getNamespace();
2798                         }
2799                         $title = Title::newFromText( $part1, $ns );
2800
2801
2802                         if ( !is_null( $title ) ) {
2803                                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
2804                                 # Check for language variants if the template is not found
2805                                 if($checkVariantLink && $title->getArticleID() == 0){
2806                                         $wgContLang->findVariantLink($part1, $title);
2807                                 }
2808
2809                                 if ( !$title->isExternal() ) {
2810                                         # Check for excessive inclusion
2811                                         $dbk = $title->getPrefixedDBkey();
2812                                         if ( $this->incrementIncludeCount( $dbk ) ) {
2813                                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->mOutputType != OT_WIKI ) {
2814                                                         $text = SpecialPage::capturePath( $title );
2815                                                         if ( is_string( $text ) ) {
2816                                                                 $found = true;
2817                                                                 $noparse = true;
2818                                                                 $noargs = true;
2819                                                                 $isHTML = true;
2820                                                                 $this->disableCache();
2821                                                         }
2822                                                 } else {
2823                                                         $articleContent = $this->fetchTemplate( $title );
2824                                                         if ( $articleContent !== false ) {
2825                                                                 $found = true;
2826                                                                 $text = $articleContent;
2827                                                                 $replaceHeadings = true;
2828                                                         }
2829                                                 }
2830                                         }
2831
2832                                         # If the title is valid but undisplayable, make a link to it
2833                                         if ( $this->mOutputType == OT_HTML && !$found ) {
2834                                                 $text = '[['.$title->getPrefixedText().']]';
2835                                                 $found = true;
2836                                         }
2837                                 } elseif ( $title->isTrans() ) {
2838                                         // Interwiki transclusion
2839                                         if ( $this->mOutputType == OT_HTML && !$forceRawInterwiki ) {
2840                                                 $text = $this->interwikiTransclude( $title, 'render' );
2841                                                 $isHTML = true;
2842                                                 $noparse = true;
2843                                         } else {
2844                                                 $text = $this->interwikiTransclude( $title, 'raw' );
2845                                                 $replaceHeadings = true;
2846                                         }
2847                                         $found = true;
2848                                 }
2849
2850                                 # Template cache array insertion
2851                                 # Use the original $piece['title'] not the mangled $part1, so that
2852                                 # modifiers such as RAW: produce separate cache entries
2853                                 if( $found ) {
2854                                         if( $isHTML ) {
2855                                                 // A special page; don't store it in the template cache.
2856                                         } else {
2857                                                 $this->mTemplates[$piece['title']] = $text;
2858                                         }
2859                                         $text = $linestart . $text;
2860                                 }
2861                         }
2862                         wfProfileOut( __METHOD__ . '-loadtpl' );
2863                 }
2864
2865                 # Recursive parsing, escaping and link table handling
2866                 # Only for HTML output
2867                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2868                         $text = wfEscapeWikiText( $text );
2869                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found ) {
2870                         if ( $noargs ) {
2871                                 $assocArgs = array();
2872                         } else {
2873                                 # Clean up argument array
2874                                 $assocArgs = array();
2875                                 $index = 1;
2876                                 foreach( $args as $arg ) {
2877                                         $eqpos = strpos( $arg, '=' );
2878                                         if ( $eqpos === false ) {
2879                                                 $assocArgs[$index++] = $arg;
2880                                         } else {
2881                                                 $name = trim( substr( $arg, 0, $eqpos ) );
2882                                                 $value = trim( substr( $arg, $eqpos+1 ) );
2883                                                 if ( $value === false ) {
2884                                                         $value = '';
2885                                                 }
2886                                                 if ( $name !== false ) {
2887                                                         $assocArgs[$name] = $value;
2888                                                 }
2889                                         }
2890                                 }
2891
2892                                 # Add a new element to the templace recursion path
2893                                 $this->mTemplatePath[$part1] = 1;
2894                         }
2895
2896                         if ( !$noparse ) {
2897                                 # If there are any <onlyinclude> tags, only include them
2898                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
2899                                         preg_match_all( '/<onlyinclude>(.*?)\n?<\/onlyinclude>/s', $text, $m );
2900                                         $text = '';
2901                                         foreach ($m[1] as $piece)
2902                                                 $text .= $piece;
2903                                 }
2904                                 # Remove <noinclude> sections and <includeonly> tags
2905                                 $text = preg_replace( '/<noinclude>.*?<\/noinclude>/s', '', $text );
2906                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
2907
2908                                 if( $this->mOutputType == OT_HTML ) {
2909                                         # Strip <nowiki>, <pre>, etc.
2910                                         $text = $this->strip( $text, $this->mStripState );
2911                                         $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
2912                                 }
2913                                 $text = $this->replaceVariables( $text, $assocArgs );
2914
2915                                 # If the template begins with a table or block-level
2916                                 # element, it should be treated as beginning a new line.
2917                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) {
2918                                         $text = "\n" . $text;
2919                                 }
2920                         } elseif ( !$noargs ) {
2921                                 # $noparse and !$noargs
2922                                 # Just replace the arguments, not any double-brace items
2923                                 # This is used for rendered interwiki transclusion
2924                                 $text = $this->replaceVariables( $text, $assocArgs, true );
2925                         }
2926                 }
2927                 # Prune lower levels off the recursion check path
2928                 $this->mTemplatePath = $lastPathLevel;
2929
2930                 if ( !$found ) {
2931                         wfProfileOut( $fname );
2932                         return $piece['text'];
2933                 } else {
2934                         wfProfileIn( __METHOD__ . '-placeholders' );
2935                         if ( $isHTML ) {
2936                                 # Replace raw HTML by a placeholder
2937                                 # Add a blank line preceding, to prevent it from mucking up
2938                                 # immediately preceding headings
2939                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
2940                         } else {
2941                                 # replace ==section headers==
2942                                 # XXX this needs to go away once we have a better parser.
2943                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
2944                                         if( !is_null( $title ) )
2945                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
2946                                         else
2947                                                 $encodedname = base64_encode("");
2948                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
2949                                                 PREG_SPLIT_DELIM_CAPTURE);
2950                                         $text = '';
2951                                         $nsec = 0;
2952                                         for( $i = 0; $i < count($m); $i += 2 ) {
2953                                                 $text .= $m[$i];
2954                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
2955                                                 $hl = $m[$i + 1];
2956                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
2957                                                         $text .= $hl;
2958                                                         continue;
2959                                                 }
2960                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
2961                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
2962                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
2963
2964                                                 $nsec++;
2965                                         }
2966                                 }
2967                         }
2968                         wfProfileOut( __METHOD__ . '-placeholders' );
2969                 }
2970
2971                 # Prune lower levels off the recursion check path
2972                 $this->mTemplatePath = $lastPathLevel;
2973
2974                 if ( !$found ) {
2975                         wfProfileOut( $fname );
2976                         return $piece['text'];
2977                 } else {
2978                         wfProfileOut( $fname );
2979                         return $text;
2980                 }
2981         }
2982
2983         /**
2984          * Fetch the unparsed text of a template and register a reference to it.
2985          */
2986         function fetchTemplate( $title ) {
2987                 $text = false;
2988                 // Loop to fetch the article, with up to 1 redirect
2989                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
2990                         $rev = Revision::newFromTitle( $title );
2991                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
2992                         if ( !$rev ) {
2993                                 break;
2994                         }
2995                         $text = $rev->getText();
2996                         if ( $text === false ) {
2997                                 break;
2998                         }
2999                         // Redirect?
3000                         $title = Title::newFromRedirect( $text );
3001                 }
3002                 return $text;
3003         }
3004
3005         /**
3006          * Transclude an interwiki link.
3007          */
3008         function interwikiTransclude( $title, $action ) {
3009                 global $wgEnableScaryTranscluding, $wgCanonicalNamespaceNames;
3010
3011                 if (!$wgEnableScaryTranscluding)
3012                         return wfMsg('scarytranscludedisabled');
3013
3014                 // The namespace will actually only be 0 or 10, depending on whether there was a leading :
3015                 // But we'll handle it generally anyway
3016                 if ( $title->getNamespace() ) {
3017                         // Use the canonical namespace, which should work anywhere
3018                         $articleName = $wgCanonicalNamespaceNames[$title->getNamespace()] . ':' . $title->getDBkey();
3019                 } else {
3020                         $articleName = $title->getDBkey();
3021                 }
3022
3023                 $url = str_replace('$1', urlencode($articleName), Title::getInterwikiLink($title->getInterwiki()));
3024                 $url .= "?action=$action";
3025                 if (strlen($url) > 255)
3026                         return wfMsg('scarytranscludetoolong');
3027                 return $this->fetchScaryTemplateMaybeFromCache($url);
3028         }
3029
3030         function fetchScaryTemplateMaybeFromCache($url) {
3031                 global $wgTranscludeCacheExpiry;
3032                 $dbr =& wfGetDB(DB_SLAVE);
3033                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3034                                 array('tc_url' => $url));
3035                 if ($obj) {
3036                         $time = $obj->tc_time;
3037                         $text = $obj->tc_contents;
3038                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3039                                 return $text;
3040                         }
3041                 }
3042
3043                 $text = Http::get($url);
3044                 if (!$text)
3045                         return wfMsg('scarytranscludefailed', $url);
3046
3047                 $dbw =& wfGetDB(DB_MASTER);
3048                 $dbw->replace('transcache', array('tc_url'), array(
3049                         'tc_url' => $url,
3050                         'tc_time' => time(),
3051                         'tc_contents' => $text));
3052                 return $text;
3053         }
3054
3055
3056         /**
3057          * Triple brace replacement -- used for template arguments
3058          * @private
3059          */
3060         function argSubstitution( $matches ) {
3061                 $arg = trim( $matches['title'] );
3062                 $text = $matches['text'];
3063                 $inputArgs = end( $this->mArgStack );
3064
3065                 if ( array_key_exists( $arg, $inputArgs ) ) {
3066                         $text = $inputArgs[$arg];
3067                 } else if ($this->mOutputType == OT_HTML && null != $matches['parts'] && count($matches['parts']) > 0) {
3068                         $text = $matches['parts'][0];
3069                 }
3070
3071                 return $text;
3072         }
3073
3074         /**
3075          * Returns true if the function is allowed to include this entity
3076          * @private
3077          */
3078         function incrementIncludeCount( $dbk ) {
3079                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
3080                         $this->mIncludeCount[$dbk] = 0;
3081                 }
3082                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
3083                         return true;
3084                 } else {
3085                         return false;
3086                 }
3087         }
3088
3089         /**
3090          * Detect __NOGALLERY__ magic word and set a placeholder
3091          */
3092         function stripNoGallery( &$text ) {
3093                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3094                 # do not add TOC
3095                 $mw = MagicWord::get( 'nogallery' );
3096                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3097         }
3098
3099         /**
3100          * Detect __TOC__ magic word and set a placeholder
3101          */
3102         function stripToc( $text ) {
3103                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3104                 # do not add TOC
3105                 $mw = MagicWord::get( 'notoc' );
3106                 if( $mw->matchAndRemove( $text ) ) {
3107                         $this->mShowToc = false;
3108                 }
3109
3110                 $mw = MagicWord::get( 'toc' );
3111                 if( $mw->match( $text ) ) {
3112                         $this->mShowToc = true;
3113                         $this->mForceTocPosition = true;
3114
3115                         // Set a placeholder. At the end we'll fill it in with the TOC.
3116                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3117
3118                         // Only keep the first one.
3119                         $text = $mw->replace( '', $text );
3120                 }
3121                 return $text;
3122         }
3123
3124         /**
3125          * This function accomplishes several tasks:
3126          * 1) Auto-number headings if that option is enabled
3127          * 2) Add an [edit] link to sections for logged in users who have enabled the option
3128          * 3) Add a Table of contents on the top for users who have enabled the option
3129          * 4) Auto-anchor headings
3130          *
3131          * It loops through all headlines, collects the necessary data, then splits up the
3132          * string and re-inserts the newly formatted headlines.
3133          *
3134          * @param string $text
3135          * @param boolean $isMain
3136          * @private
3137          */
3138         function formatHeadings( $text, $isMain=true ) {
3139                 global $wgMaxTocLevel, $wgContLang;
3140
3141                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3142                 if( !$this->mTitle->userCanEdit() ) {
3143                         $showEditLink = 0;
3144                 } else {
3145                         $showEditLink = $this->mOptions->getEditSection();
3146                 }
3147
3148                 # Inhibit editsection links if requested in the page
3149                 $esw =& MagicWord::get( 'noeditsection' );
3150                 if( $esw->matchAndRemove( $text ) ) {
3151                         $showEditLink = 0;
3152                 }
3153
3154                 # Get all headlines for numbering them and adding funky stuff like [edit]
3155                 # links - this is for later, but we need the number of headlines right now
3156                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
3157
3158                 # if there are fewer than 4 headlines in the article, do not show TOC
3159                 # unless it's been explicitly enabled.
3160                 $enoughToc = $this->mShowToc &&
3161                         (($numMatches >= 4) || $this->mForceTocPosition);
3162
3163                 # Allow user to stipulate that a page should have a "new section"
3164                 # link added via __NEWSECTIONLINK__
3165                 $mw =& MagicWord::get( 'newsectionlink' );
3166                 if( $mw->matchAndRemove( $text ) )
3167                         $this->mOutput->setNewSection( true );
3168
3169                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3170                 # override above conditions and always show TOC above first header
3171                 $mw =& MagicWord::get( 'forcetoc' );
3172                 if ($mw->matchAndRemove( $text ) ) {
3173                         $this->mShowToc = true;
3174                         $enoughToc = true;
3175                 }
3176
3177                 # Never ever show TOC if no headers
3178                 if( $numMatches < 1 ) {
3179                         $enoughToc = false;
3180                 }
3181
3182                 # We need this to perform operations on the HTML
3183                 $sk =& $this->mOptions->getSkin();
3184
3185                 # headline counter
3186                 $headlineCount = 0;
3187                 $sectionCount = 0; # headlineCount excluding template sections
3188
3189                 # Ugh .. the TOC should have neat indentation levels which can be
3190                 # passed to the skin functions. These are determined here
3191                 $toc = '';
3192                 $full = '';
3193                 $head = array();
3194                 $sublevelCount = array();
3195                 $levelCount = array();
3196                 $toclevel = 0;
3197                 $level = 0;
3198                 $prevlevel = 0;
3199                 $toclevel = 0;
3200                 $prevtoclevel = 0;
3201
3202                 foreach( $matches[3] as $headline ) {
3203                         $istemplate = 0;
3204                         $templatetitle = '';
3205                         $templatesection = 0;
3206                         $numbering = '';
3207
3208                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3209                                 $istemplate = 1;
3210                                 $templatetitle = base64_decode($mat[1]);
3211                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3212                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3213                         }
3214
3215                         if( $toclevel ) {
3216                                 $prevlevel = $level;
3217                                 $prevtoclevel = $toclevel;
3218                         }
3219                         $level = $matches[1][$headlineCount];
3220
3221                         if( $doNumberHeadings || $enoughToc ) {
3222
3223                                 if ( $level > $prevlevel ) {
3224                                         # Increase TOC level
3225                                         $toclevel++;
3226                                         $sublevelCount[$toclevel] = 0;
3227                                         if( $toclevel<$wgMaxTocLevel ) {
3228                                                 $toc .= $sk->tocIndent();
3229                                         }
3230                                 }
3231                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3232                                         # Decrease TOC level, find level to jump to
3233
3234                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3235                                                 # Can only go down to level 1
3236                                                 $toclevel = 1;
3237                                         } else {
3238                                                 for ($i = $toclevel; $i > 0; $i--) {
3239                                                         if ( $levelCount[$i] == $level ) {
3240                                                                 # Found last matching level
3241                                                                 $toclevel = $i;
3242                                                                 break;
3243                                                         }
3244                                                         elseif ( $levelCount[$i] < $level ) {
3245                                                                 # Found first matching level below current level
3246                                                                 $toclevel = $i + 1;
3247                                                                 break;
3248                                                         }
3249                                                 }
3250                                         }
3251                                         if( $toclevel<$wgMaxTocLevel ) {
3252                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3253                                         }
3254                                 }
3255                                 else {
3256                                         # No change in level, end TOC line
3257                                         if( $toclevel<$wgMaxTocLevel ) {
3258                                                 $toc .= $sk->tocLineEnd();
3259                                         }
3260                                 }
3261
3262                                 $levelCount[$toclevel] = $level;
3263
3264                                 # count number of headlines for each level
3265                                 @$sublevelCount[$toclevel]++;
3266                                 $dot = 0;
3267                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3268                                         if( !empty( $sublevelCount[$i] ) ) {
3269                                                 if( $dot ) {
3270                                                         $numbering .= '.';
3271                                                 }
3272                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3273                                                 $dot = 1;
3274                                         }
3275                                 }
3276                         }
3277
3278                         # The canonized header is a version of the header text safe to use for links
3279                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3280                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
3281                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
3282
3283                         # Remove link placeholders by the link text.
3284                         #     <!--LINK number-->
3285                         # turns into
3286                         #     link text with suffix
3287                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3288                                                             "\$this->mLinkHolders['texts'][\$1]",
3289                                                             $canonized_headline );
3290                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3291                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3292                                                             $canonized_headline );
3293
3294                         # strip out HTML
3295                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3296                         $tocline = trim( $canonized_headline );
3297                         # Save headline for section edit hint before it's escaped
3298                         $headline_hint = trim( $canonized_headline );
3299                         $canonized_headline = Sanitizer::escapeId( $tocline );
3300                         $refers[$headlineCount] = $canonized_headline;
3301
3302                         # count how many in assoc. array so we can track dupes in anchors
3303                         @$refers[$canonized_headline]++;
3304                         $refcount[$headlineCount]=$refers[$canonized_headline];
3305
3306                         # Don't number the heading if it is the only one (looks silly)
3307                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3308                                 # the two are different if the line contains a link
3309                                 $headline=$numbering . ' ' . $headline;
3310                         }
3311
3312                         # Create the anchor for linking from the TOC to the section
3313                         $anchor = $canonized_headline;
3314                         if($refcount[$headlineCount] > 1 ) {
3315                                 $anchor .= '_' . $refcount[$headlineCount];
3316                         }
3317                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3318                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3319                         }
3320                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3321                                 if ( empty( $head[$headlineCount] ) ) {
3322                                         $head[$headlineCount] = '';
3323                                 }
3324                                 if( $istemplate )
3325                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
3326                                 else
3327                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3328                         }
3329
3330                         # give headline the correct <h#> tag
3331                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
3332
3333                         $headlineCount++;
3334                         if( !$istemplate )
3335                                 $sectionCount++;
3336                 }
3337
3338                 if( $enoughToc ) {
3339                         if( $toclevel<$wgMaxTocLevel ) {
3340                                 $toc .= $sk->tocUnindent( $toclevel - 1 );
3341                         }
3342                         $toc = $sk->tocList( $toc );
3343                 }
3344
3345                 # split up and insert constructed headlines
3346
3347                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3348                 $i = 0;
3349
3350                 foreach( $blocks as $block ) {
3351                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3352                                 # This is the [edit] link that appears for the top block of text when
3353                                 # section editing is enabled
3354
3355                                 # Disabled because it broke block formatting
3356                                 # For example, a bullet point in the top line
3357                                 # $full .= $sk->editSectionLink(0);
3358                         }
3359                         $full .= $block;
3360                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3361                                 # Top anchor now in skin
3362                                 $full = $full.$toc;
3363                         }
3364
3365                         if( !empty( $head[$i] ) ) {
3366                                 $full .= $head[$i];
3367                         }
3368                         $i++;
3369                 }
3370                 if( $this->mForceTocPosition ) {
3371                         return str_replace( '<!--MWTOC-->', $toc, $full );
3372                 } else {
3373                         return $full;
3374                 }
3375         }
3376
3377         /**
3378          * Return an HTML link for the "ISBN 123456" text
3379          * @private
3380          */
3381         function magicISBN( $text ) {
3382                 $fname = 'Parser::magicISBN';
3383                 wfProfileIn( $fname );
3384
3385                 $a = split( 'ISBN ', ' '.$text );
3386                 if ( count ( $a ) < 2 ) {
3387                         wfProfileOut( $fname );
3388                         return $text;
3389                 }
3390                 $text = substr( array_shift( $a ), 1);
3391                 $valid = '0123456789-Xx';
3392
3393                 foreach ( $a as $x ) {
3394                         # hack: don't replace inside thumbnail title/alt
3395                         # attributes
3396                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3397                                 $text .= "ISBN $x";
3398                                 continue;
3399                         }
3400
3401                         $isbn = $blank = '' ;
3402                         while ( $x !== '' && ' ' == $x{0} ) {
3403                                 $blank .= ' ';
3404                                 $x = substr( $x, 1 );
3405                         }
3406                         if ( $x == '' ) { # blank isbn
3407                                 $text .= "ISBN $blank";
3408                                 continue;
3409                         }
3410                         while ( strstr( $valid, $x{0} ) != false ) {
3411                                 $isbn .= $x{0};
3412                                 $x = substr( $x, 1 );
3413                         }
3414                         $num = str_replace( '-', '', $isbn );
3415                         $num = str_replace( ' ', '', $num );
3416                         $num = str_replace( 'x', 'X', $num );
3417
3418                         if ( '' == $num ) {
3419                                 $text .= "ISBN $blank$x";
3420                         } else {
3421                                 $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
3422                                 $text .= '<a href="' .
3423                                         $titleObj->escapeLocalUrl( 'isbn='.$num ) .
3424                                         "\" class=\"internal\">ISBN $isbn</a>";
3425                                 $text .= $x;
3426                         }
3427                 }
3428                 wfProfileOut( $fname );
3429                 return $text;
3430         }
3431
3432         /**
3433          * Return an HTML link for the "RFC 1234" text
3434          *
3435          * @private
3436          * @param string $text     Text to be processed
3437          * @param string $keyword  Magic keyword to use (default RFC)
3438          * @param string $urlmsg   Interface message to use (default rfcurl)
3439          * @return string
3440          */
3441         function magicRFC( $text, $keyword='RFC ', $urlmsg='rfcurl'  ) {
3442
3443                 $valid = '0123456789';
3444                 $internal = false;
3445
3446                 $a = split( $keyword, ' '.$text );
3447                 if ( count ( $a ) < 2 ) {
3448                         return $text;
3449                 }
3450                 $text = substr( array_shift( $a ), 1);
3451
3452                 /* Check if keyword is preceed by [[.
3453                  * This test is made here cause of the array_shift above
3454                  * that prevent the test to be done in the foreach.
3455                  */
3456                 if ( substr( $text, -2 ) == '[[' ) {
3457                         $internal = true;
3458                 }
3459
3460                 foreach ( $a as $x ) {
3461                         /* token might be empty if we have RFC RFC 1234 */
3462                         if ( $x=='' ) {
3463                                 $text.=$keyword;
3464                                 continue;
3465                                 }
3466
3467                         # hack: don't replace inside thumbnail title/alt
3468                         # attributes
3469                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3470                                 $text .= $keyword . $x;
3471                                 continue;
3472                         }
3473
3474                         $id = $blank = '' ;
3475
3476                         /** remove and save whitespaces in $blank */
3477                         while ( $x{0} == ' ' ) {
3478                                 $blank .= ' ';
3479                                 $x = substr( $x, 1 );
3480                         }
3481
3482                         /** remove and save the rfc number in $id */
3483                         while ( strstr( $valid, $x{0} ) != false ) {
3484                                 $id .= $x{0};
3485                                 $x = substr( $x, 1 );
3486                         }
3487
3488                         if ( $id == '' ) {
3489                                 /* call back stripped spaces*/
3490                                 $text .= $keyword.$blank.$x;
3491                         } elseif( $internal ) {
3492                                 /* normal link */
3493                                 $text .= $keyword.$id.$x;
3494                         } else {
3495                                 /* build the external link*/
3496                                 $url = wfMsg( $urlmsg, $id);
3497                                 $sk =& $this->mOptions->getSkin();
3498                                 $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
3499                                 $text .= "<a href=\"{$url}\"{$la}>{$keyword}{$id}</a>{$x}";
3500                         }
3501
3502                         /* Check if the next RFC keyword is preceed by [[ */
3503                         $internal = ( substr($x,-2) == '[[' );
3504                 }
3505                 return $text;
3506         }
3507
3508         /**
3509          * Transform wiki markup when saving a page by doing \r\n -> \n
3510          * conversion, substitting signatures, {{subst:}} templates, etc.
3511          *
3512          * @param string $text the text to transform
3513          * @param Title &$title the Title object for the current article
3514          * @param User &$user the User object describing the current user
3515          * @param ParserOptions $options parsing options
3516          * @param bool $clearState whether to clear the parser state first
3517          * @return string the altered wiki markup
3518          * @public
3519          */
3520         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
3521                 $this->mOptions = $options;
3522                 $this->mTitle =& $title;
3523                 $this->mOutputType = OT_WIKI;
3524
3525                 if ( $clearState ) {
3526                         $this->clearState();
3527                 }
3528
3529                 $stripState = false;
3530                 $pairs = array(
3531                         "\r\n" => "\n",
3532                 );
3533                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3534                 $text = $this->strip( $text, $stripState, true, array( 'gallery' ) );
3535                 $text = $this->pstPass2( $text, $stripState, $user );
3536                 $text = $this->unstrip( $text, $stripState );
3537                 $text = $this->unstripNoWiki( $text, $stripState );
3538                 return $text;
3539         }
3540
3541         /**
3542          * Pre-save transform helper function
3543          * @private
3544          */
3545         function pstPass2( $text, &$stripState, &$user ) {
3546                 global $wgContLang, $wgLocaltimezone;
3547
3548                 /* Note: This is the timestamp saved as hardcoded wikitext to
3549                  * the database, we use $wgContLang here in order to give
3550                  * everyone the same signature and use the default one rather
3551                  * than the one selected in each user's preferences.
3552                  */
3553                 if ( isset( $wgLocaltimezone ) ) {
3554                         $oldtz = getenv( 'TZ' );
3555                         putenv( 'TZ='.$wgLocaltimezone );
3556                 }
3557                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3558                   ' (' . date( 'T' ) . ')';
3559                 if ( isset( $wgLocaltimezone ) ) {
3560                         putenv( 'TZ='.$oldtz );
3561                 }
3562
3563                 # Variable replacement
3564                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3565                 $text = $this->replaceVariables( $text );
3566
3567                 # Strip out <nowiki> etc. added via replaceVariables
3568                 $text = $this->strip( $text, $stripState, false, array( 'gallery' ) );
3569
3570                 # Signatures
3571                 $sigText = $this->getUserSig( $user );
3572                 $text = strtr( $text, array(
3573                         '~~~~~' => $d,
3574                         '~~~~' => "$sigText $d",
3575                         '~~~' => $sigText
3576                 ) );
3577
3578                 # Context links: [[|name]] and [[name (context)|]]
3579                 #
3580                 global $wgLegalTitleChars;
3581                 $tc = "[$wgLegalTitleChars]";
3582                 $np = str_replace( array( '(', ')' ), array( '', '' ), $tc ); # No parens
3583
3584                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3585                 $conpat = "/^({$np}+) \\(({$tc}+)\\)$/";
3586
3587                 $p1 = "/\[\[({$np}+) \\(({$np}+)\\)\\|]]/";             # [[page (context)|]]
3588                 $p2 = "/\[\[\\|({$tc}+)]]/";                                    # [[|page]]
3589                 $p3 = "/\[\[(:*$namespacechar+):({$np}+)\\|]]/";                # [[namespace:page|]] and [[:namespace:page|]]
3590                 $p4 = "/\[\[(:*$namespacechar+):({$np}+) \\(({$np}+)\\)\\|]]/"; # [[ns:page (cont)|]] and [[:ns:page (cont)|]]
3591                 $context = '';
3592                 $t = $this->mTitle->getText();
3593                 if ( preg_match( $conpat, $t, $m ) ) {
3594                         $context = $m[2];
3595                 }
3596                 $text = preg_replace( $p4, '[[\\1:\\2 (\\3)|\\2]]', $text );
3597                 $text = preg_replace( $p1, '[[\\1 (\\2)|\\1]]', $text );
3598                 $text = preg_replace( $p3, '[[\\1:\\2|\\2]]', $text );
3599
3600                 if ( '' == $context ) {
3601                         $text = preg_replace( $p2, '[[\\1]]', $text );
3602                 } else {
3603                         $text = preg_replace( $p2, "[[\\1 ({$context})|\\1]]", $text );
3604                 }
3605
3606                 # Trim trailing whitespace
3607                 # __END__ tag allows for trailing
3608                 # whitespace to be deliberately included
3609                 $text = rtrim( $text );
3610                 $mw =& MagicWord::get( 'end' );
3611                 $mw->matchAndRemove( $text );
3612
3613                 return $text;
3614         }
3615
3616         /**
3617          * Fetch the user's signature text, if any, and normalize to
3618          * validated, ready-to-insert wikitext.
3619          *
3620          * @param User $user
3621          * @return string
3622          * @private
3623          */
3624         function getUserSig( &$user ) {
3625                 $username = $user->getName();
3626                 $nickname = $user->getOption( 'nickname' );
3627                 $nickname = $nickname === '' ? $username : $nickname;
3628
3629                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3630                         # Sig. might contain markup; validate this
3631                         if( $this->validateSig( $nickname ) !== false ) {
3632                                 # Validated; clean up (if needed) and return it
3633                                 return $this->cleanSig( $nickname, true );
3634                         } else {
3635                                 # Failed to validate; fall back to the default
3636                                 $nickname = $username;
3637                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3638                         }
3639                 }
3640
3641                 // Make sure nickname doesnt get a sig in a sig
3642                 $nickname = $this->cleanSigInSig( $nickname );
3643
3644                 # If we're still here, make it a link to the user page
3645                 $userpage = $user->getUserPage();
3646                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3647         }
3648
3649         /**
3650          * Check that the user's signature contains no bad XML
3651          *
3652          * @param string $text
3653          * @return mixed An expanded string, or false if invalid.
3654          */
3655         function validateSig( $text ) {
3656                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3657         }
3658
3659         /**
3660          * Clean up signature text
3661          *
3662          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3663          * 2) Substitute all transclusions
3664          *
3665          * @param string $text
3666          * @param $parsing Whether we're cleaning (preferences save) or parsing
3667          * @return string Signature text
3668          */
3669         function cleanSig( $text, $parsing = false ) {
3670                 global $wgTitle;
3671                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
3672
3673                 $substWord = MagicWord::get( 'subst' );
3674                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3675                 $substText = '{{' . $substWord->getSynonym( 0 );
3676
3677                 $text = preg_replace( $substRegex, $substText, $text );
3678                 $text = $this->cleanSigInSig( $text );
3679                 $text = $this->replaceVariables( $text );
3680
3681                 $this->clearState();
3682                 return $text;
3683         }
3684
3685         /**
3686          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3687          * @param string $text
3688          * @return string Signature text with /~{3,5}/ removed
3689          */
3690         function cleanSigInSig( $text ) {
3691                 $text = preg_replace( '/~{3,5}/', '', $text );
3692                 return $text;
3693         }
3694
3695         /**
3696          * Set up some variables which are usually set up in parse()
3697          * so that an external function can call some class members with confidence
3698          * @public
3699          */
3700         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3701                 $this->mTitle =& $title;
3702                 $this->mOptions = $options;
3703                 $this->mOutputType = $outputType;
3704                 if ( $clearState ) {
3705                         $this->clearState();
3706                 }
3707         }
3708
3709         /**
3710          * Transform a MediaWiki message by replacing magic variables.
3711          *
3712          * @param string $text the text to transform
3713          * @param ParserOptions $options  options
3714          * @return string the text with variables substituted
3715          * @public
3716          */
3717         function transformMsg( $text, $options ) {
3718                 global $wgTitle;
3719                 static $executing = false;
3720
3721                 $fname = "Parser::transformMsg";
3722
3723                 # Guard against infinite recursion
3724                 if ( $executing ) {
3725                         return $text;
3726                 }
3727                 $executing = true;
3728
3729                 wfProfileIn($fname);
3730
3731                 $this->mTitle = $wgTitle;
3732                 $this->mOptions = $options;
3733                 $this->mOutputType = OT_MSG;
3734                 $this->clearState();
3735                 $text = $this->replaceVariables( $text );
3736
3737                 $executing = false;
3738                 wfProfileOut($fname);
3739                 return $text;
3740         }
3741
3742         /**
3743          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3744          * The callback should have the following form:
3745          *    function myParserHook( $text, $params, &$parser ) { ... }
3746          *
3747          * Transform and return $text. Use $parser for any required context, e.g. use
3748          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3749          *
3750          * @public
3751          *
3752          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3753          * @param mixed $callback The callback function (and object) to use for the tag
3754          *
3755          * @return The old value of the mTagHooks array associated with the hook
3756          */
3757         function setHook( $tag, $callback ) {
3758                 $tag = strtolower( $tag );
3759                 $oldVal = @$this->mTagHooks[$tag];
3760                 $this->mTagHooks[$tag] = $callback;
3761
3762                 return $oldVal;
3763         }
3764
3765         /**
3766          * Create a function, e.g. {{sum:1|2|3}}
3767          * The callback function should have the form:
3768          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3769          *
3770          * The callback may either return the text result of the function, or an array with the text
3771          * in element 0, and a number of flags in the other elements. The names of the flags are
3772          * specified in the keys. Valid flags are:
3773          *   found                     The text returned is valid, stop processing the template. This
3774          *                             is on by default.
3775          *   nowiki                    Wiki markup in the return value should be escaped
3776          *   noparse                   Unsafe HTML tags should not be stripped, etc.
3777          *   noargs                    Don't replace triple-brace arguments in the return value
3778          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3779          *
3780          * @public
3781          *
3782          * @param mixed $id The magic word ID
3783          * @param mixed $callback The callback function (and object) to use
3784          * @param integer $flags a combination of the following flags:
3785          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
3786          *
3787          * @return The old callback function for this name, if any
3788          */
3789         function setFunctionHook( $id, $callback, $flags = 0 ) {
3790                 $oldVal = @$this->mFunctionHooks[$id];
3791                 $this->mFunctionHooks[$id] = $callback;
3792
3793                 # Add to function cache
3794                 $mw = MagicWord::get( $id );
3795                 if ( !$mw ) {
3796                         throw new MWException( 'The calling convention to Parser::setFunctionHook() has changed, ' .
3797                                 'it is now required to pass a MagicWord ID as the first parameter.' );
3798                 }
3799
3800                 $synonyms = $mw->getSynonyms();
3801                 $sensitive = intval( $mw->isCaseSensitive() );
3802
3803                 foreach ( $synonyms as $syn ) {
3804                         # Case
3805                         if ( !$sensitive ) {
3806                                 $syn = strtolower( $syn );
3807                         }
3808                         # Add leading hash
3809                         if ( !( $flags & SFH_NO_HASH ) ) {
3810                                 $syn = '#' . $syn;
3811                         }
3812                         # Remove trailing colon
3813                         if ( substr( $syn, -1, 1 ) == ':' ) {
3814                                 $syn = substr( $syn, 0, -1 );
3815                         }
3816                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
3817                 }
3818                 return $oldVal;
3819         }
3820
3821         /**
3822          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3823          * Placeholders created in Skin::makeLinkObj()
3824          * Returns an array of links found, indexed by PDBK:
3825          *  0 - broken
3826          *  1 - normal link
3827          *  2 - stub
3828          * $options is a bit field, RLH_FOR_UPDATE to select for update
3829          */
3830         function replaceLinkHolders( &$text, $options = 0 ) {
3831                 global $wgUser;
3832                 global $wgOutputReplace;
3833
3834                 $fname = 'Parser::replaceLinkHolders';
3835                 wfProfileIn( $fname );
3836
3837                 $pdbks = array();
3838                 $colours = array();
3839                 $sk =& $this->mOptions->getSkin();
3840                 $linkCache =& LinkCache::singleton();
3841
3842                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3843                         wfProfileIn( $fname.'-check' );
3844                         $dbr =& wfGetDB( DB_SLAVE );
3845                         $page = $dbr->tableName( 'page' );
3846                         $threshold = $wgUser->getOption('stubthreshold');
3847
3848                         # Sort by namespace
3849                         asort( $this->mLinkHolders['namespaces'] );
3850
3851                         # Generate query
3852                         $query = false;
3853                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3854                                 # Make title object
3855                                 $title = $this->mLinkHolders['titles'][$key];
3856
3857                                 # Skip invalid entries.
3858                                 # Result will be ugly, but prevents crash.
3859                                 if ( is_null( $title ) ) {
3860                                         continue;
3861                                 }
3862                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3863
3864                                 # Check if it's a static known link, e.g. interwiki
3865                                 if ( $title->isAlwaysKnown() ) {
3866                                         $colours[$pdbk] = 1;
3867                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3868                                         $colours[$pdbk] = 1;
3869                                         $this->mOutput->addLink( $title, $id );
3870                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3871                                         $colours[$pdbk] = 0;
3872                                 } else {
3873                                         # Not in the link cache, add it to the query
3874                                         if ( !isset( $current ) ) {
3875                                                 $current = $ns;
3876                                                 $query =  "SELECT page_id, page_namespace, page_title";
3877                                                 if ( $threshold > 0 ) {
3878                                                         $query .= ', page_len, page_is_redirect';
3879                                                 }
3880                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
3881                                         } elseif ( $current != $ns ) {
3882                                                 $current = $ns;
3883                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
3884                                         } else {
3885                                                 $query .= ', ';
3886                                         }
3887
3888                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3889                                 }
3890                         }
3891                         if ( $query ) {
3892                                 $query .= '))';
3893                                 if ( $options & RLH_FOR_UPDATE ) {
3894                                         $query .= ' FOR UPDATE';
3895                                 }
3896
3897                                 $res = $dbr->query( $query, $fname );
3898
3899                                 # Fetch data and form into an associative array
3900                                 # non-existent = broken
3901                                 # 1 = known
3902                                 # 2 = stub
3903                                 while ( $s = $dbr->fetchObject($res) ) {
3904                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
3905                                         $pdbk = $title->getPrefixedDBkey();
3906                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
3907                                         $this->mOutput->addLink( $title, $s->page_id );
3908
3909                                         if ( $threshold >  0 ) {
3910                                                 $size = $s->page_len;
3911                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
3912                                                         $colours[$pdbk] = 1;
3913                                                 } else {
3914                                                         $colours[$pdbk] = 2;
3915                                                 }
3916                                         } else {
3917                                                 $colours[$pdbk] = 1;
3918                                         }
3919                                 }
3920                         }
3921                         wfProfileOut( $fname.'-check' );
3922
3923                         # Construct search and replace arrays
3924                         wfProfileIn( $fname.'-construct' );
3925                         $wgOutputReplace = array();
3926                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3927                                 $pdbk = $pdbks[$key];
3928                                 $searchkey = "<!--LINK $key-->";
3929                                 $title = $this->mLinkHolders['titles'][$key];
3930                                 if ( empty( $colours[$pdbk] ) ) {
3931                                         $linkCache->addBadLinkObj( $title );
3932                                         $colours[$pdbk] = 0;
3933                                         $this->mOutput->addLink( $title, 0 );
3934                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
3935                                                                         $this->mLinkHolders['texts'][$key],
3936                                                                         $this->mLinkHolders['queries'][$key] );
3937                                 } elseif ( $colours[$pdbk] == 1 ) {
3938                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
3939                                                                         $this->mLinkHolders['texts'][$key],
3940                                                                         $this->mLinkHolders['queries'][$key] );
3941                                 } elseif ( $colours[$pdbk] == 2 ) {
3942                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
3943                                                                         $this->mLinkHolders['texts'][$key],
3944                                                                         $this->mLinkHolders['queries'][$key] );
3945                                 }
3946                         }
3947                         wfProfileOut( $fname.'-construct' );
3948
3949                         # Do the thing
3950                         wfProfileIn( $fname.'-replace' );
3951
3952                         $text = preg_replace_callback(
3953                                 '/(<!--LINK .*?-->)/',
3954                                 "wfOutputReplaceMatches",
3955                                 $text);
3956
3957                         wfProfileOut( $fname.'-replace' );
3958                 }
3959
3960                 # Now process interwiki link holders
3961                 # This is quite a bit simpler than internal links
3962                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
3963                         wfProfileIn( $fname.'-interwiki' );
3964                         # Make interwiki link HTML
3965                         $wgOutputReplace = array();
3966                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
3967                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
3968                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
3969                         }
3970
3971                         $text = preg_replace_callback(
3972                                 '/<!--IWLINK (.*?)-->/',
3973                                 "wfOutputReplaceMatches",
3974                                 $text );
3975                         wfProfileOut( $fname.'-interwiki' );
3976                 }
3977
3978                 wfProfileOut( $fname );
3979                 return $colours;
3980         }
3981
3982         /**
3983          * Replace <!--LINK--> link placeholders with plain text of links
3984          * (not HTML-formatted).
3985          * @param string $text
3986          * @return string
3987          */
3988         function replaceLinkHoldersText( $text ) {
3989                 $fname = 'Parser::replaceLinkHoldersText';
3990                 wfProfileIn( $fname );
3991
3992                 $text = preg_replace_callback(
3993                         '/<!--(LINK|IWLINK) (.*?)-->/',
3994                         array( &$this, 'replaceLinkHoldersTextCallback' ),
3995                         $text );
3996
3997                 wfProfileOut( $fname );
3998                 return $text;
3999         }
4000
4001         /**
4002          * @param array $matches
4003          * @return string
4004          * @private
4005          */
4006         function replaceLinkHoldersTextCallback( $matches ) {
4007                 $type = $matches[1];
4008                 $key  = $matches[2];
4009                 if( $type == 'LINK' ) {
4010                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4011                                 return $this->mLinkHolders['texts'][$key];
4012                         }
4013                 } elseif( $type == 'IWLINK' ) {
4014                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4015                                 return $this->mInterwikiLinkHolders['texts'][$key];
4016                         }
4017                 }
4018                 return $matches[0];
4019         }
4020
4021         /**
4022          * Tag hook handler for 'pre'.
4023          */
4024         function renderPreTag( $text, $attribs, $parser ) {
4025                 // Backwards-compatibility hack
4026                 $content = preg_replace( '!<nowiki>(.*?)</nowiki>!is', '\\1', $text );
4027
4028                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4029                 return wfOpenElement( 'pre', $attribs ) .
4030                         wfEscapeHTMLTagsOnly( $content ) .
4031                         '</pre>';
4032         }
4033
4034         /**
4035          * Renders an image gallery from a text with one line per image.
4036          * text labels may be given by using |-style alternative text. E.g.
4037          *   Image:one.jpg|The number "1"
4038          *   Image:tree.jpg|A tree
4039          * given as text will return the HTML of a gallery with two images,
4040          * labeled 'The number "1"' and
4041          * 'A tree'.
4042          */
4043         function renderImageGallery( $text, $params ) {
4044                 $ig = new ImageGallery();
4045                 $ig->setShowBytes( false );
4046                 $ig->setShowFilename( false );
4047                 $ig->setParsing();
4048                 $ig->useSkin( $this->mOptions->getSkin() );
4049
4050                 if( isset( $params['caption'] ) )
4051                         $ig->setCaption( $params['caption'] );
4052
4053                 $lines = explode( "\n", $text );
4054                 foreach ( $lines as $line ) {
4055                         # match lines like these:
4056                         # Image:someimage.jpg|This is some image
4057                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4058                         # Skip empty lines
4059                         if ( count( $matches ) == 0 ) {
4060                                 continue;
4061                         }
4062                         $tp = Title::newFromText( $matches[1] );
4063                         $nt =& $tp;
4064                         if( is_null( $nt ) ) {
4065                                 # Bogus title. Ignore these so we don't bomb out later.
4066                                 continue;
4067                         }
4068                         if ( isset( $matches[3] ) ) {
4069                                 $label = $matches[3];
4070                         } else {
4071                                 $label = '';
4072                         }
4073
4074                         $pout = $this->parse( $label,
4075                                 $this->mTitle,
4076                                 $this->mOptions,
4077                                 false, // Strip whitespace...?
4078                                 false  // Don't clear state!
4079                         );
4080                         $html = $pout->getText();
4081
4082                         $ig->add( new Image( $nt ), $html );
4083
4084                         # Only add real images (bug #5586)
4085                         if ( $nt->getNamespace() == NS_IMAGE ) {
4086                                 $this->mOutput->addImage( $nt->getDBkey() );
4087                         }
4088                 }
4089                 return $ig->toHTML();
4090         }
4091
4092         /**
4093          * Parse image options text and use it to make an image
4094          */
4095         function makeImage( &$nt, $options ) {
4096                 global $wgUseImageResize;
4097
4098                 $align = '';
4099
4100                 # Check if the options text is of the form "options|alt text"
4101                 # Options are:
4102                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4103                 #  * left               no resizing, just left align. label is used for alt= only
4104                 #  * right              same, but right aligned
4105                 #  * none               same, but not aligned
4106                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4107                 #  * center             center the image
4108                 #  * framed             Keep original image size, no magnify-button.
4109
4110                 $part = explode( '|', $options);
4111
4112                 $mwThumb  =& MagicWord::get( 'img_thumbnail' );
4113                 $mwManualThumb =& MagicWord::get( 'img_manualthumb' );
4114                 $mwLeft   =& MagicWord::get( 'img_left' );
4115                 $mwRight  =& MagicWord::get( 'img_right' );
4116                 $mwNone   =& MagicWord::get( 'img_none' );
4117                 $mwWidth  =& MagicWord::get( 'img_width' );
4118                 $mwCenter =& MagicWord::get( 'img_center' );
4119                 $mwFramed =& MagicWord::get( 'img_framed' );
4120                 $caption = '';
4121
4122                 $width = $height = $framed = $thumb = false;
4123                 $manual_thumb = '' ;
4124
4125                 foreach( $part as $key => $val ) {
4126                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
4127                                 $thumb=true;
4128                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
4129                                 # use manually specified thumbnail
4130                                 $thumb=true;
4131                                 $manual_thumb = $match;
4132                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
4133                                 # remember to set an alignment, don't render immediately
4134                                 $align = 'right';
4135                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
4136                                 # remember to set an alignment, don't render immediately
4137                                 $align = 'left';
4138                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
4139                                 # remember to set an alignment, don't render immediately
4140                                 $align = 'center';
4141                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
4142                                 # remember to set an alignment, don't render immediately
4143                                 $align = 'none';
4144                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
4145                                 wfDebug( "img_width match: $match\n" );
4146                                 # $match is the image width in pixels
4147                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
4148                                         $width = intval( $m[1] );
4149                                         $height = intval( $m[2] );
4150                                 } else {
4151                                         $width = intval($match);
4152                                 }
4153                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
4154                                 $framed=true;
4155                         } else {
4156                                 $caption = $val;
4157                         }
4158                 }
4159                 # Strip bad stuff out of the alt text
4160                 $alt = $this->replaceLinkHoldersText( $caption );
4161
4162                 # make sure there are no placeholders in thumbnail attributes
4163                 # that are later expanded to html- so expand them now and
4164                 # remove the tags
4165                 $alt = $this->unstrip($alt, $this->mStripState);
4166                 $alt = Sanitizer::stripAllTags( $alt );
4167
4168                 # Linker does the rest
4169                 $sk =& $this->mOptions->getSkin();
4170                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
4171         }
4172
4173         /**
4174          * Set a flag in the output object indicating that the content is dynamic and
4175          * shouldn't be cached.
4176          */
4177         function disableCache() {
4178                 wfDebug( "Parser output marked as uncacheable.\n" );
4179                 $this->mOutput->mCacheTime = -1;
4180         }
4181
4182         /**#@+
4183          * Callback from the Sanitizer for expanding items found in HTML attribute
4184          * values, so they can be safely tested and escaped.
4185          * @param string $text
4186          * @param array $args
4187          * @return string
4188          * @private
4189          */
4190         function attributeStripCallback( &$text, $args ) {
4191                 $text = $this->replaceVariables( $text, $args );
4192                 $text = $this->unstripForHTML( $text );
4193                 return $text;
4194         }
4195
4196         function unstripForHTML( $text ) {
4197                 $text = $this->unstrip( $text, $this->mStripState );
4198                 $text = $this->unstripNoWiki( $text, $this->mStripState );
4199                 return $text;
4200         }
4201         /**#@-*/
4202
4203         /**#@+
4204          * Accessor/mutator
4205          */
4206         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4207         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4208         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4209         /**#@-*/
4210
4211         /**#@+
4212          * Accessor
4213          */
4214         function getTags() { return array_keys( $this->mTagHooks ); }
4215         /**#@-*/
4216
4217
4218         /**
4219          * Break wikitext input into sections, and either pull or replace
4220          * some particular section's text.
4221          *
4222          * External callers should use the getSection and replaceSection methods.
4223          *
4224          * @param $text Page wikitext
4225          * @param $section Numbered section. 0 pulls the text before the first
4226          *                 heading; other numbers will pull the given section
4227          *                 along with its lower-level subsections.
4228          * @param $mode One of "get" or "replace"
4229          * @param $newtext Replacement text for section data.
4230          * @return string for "get", the extracted section text.
4231          *                for "replace", the whole page with the section replaced.
4232          */
4233         private function extractSections( $text, $section, $mode, $newtext='' ) {
4234                 # strip NOWIKI etc. to avoid confusion (true-parameter causes HTML
4235                 # comments to be stripped as well)
4236                 $striparray = array();
4237
4238                 $oldOutputType = $this->mOutputType;
4239                 $oldOptions = $this->mOptions;
4240                 $this->mOptions = new ParserOptions();
4241                 $this->mOutputType = OT_WIKI;
4242
4243                 $striptext = $this->strip( $text, $striparray, true );
4244
4245                 $this->mOutputType = $oldOutputType;
4246                 $this->mOptions = $oldOptions;
4247
4248                 # now that we can be sure that no pseudo-sections are in the source,
4249                 # split it up by section
4250                 $uniq = preg_quote( $this->uniqPrefix(), '/' );
4251                 $comment = "(?:$uniq-!--.*?QINU)";
4252                 $secs = preg_split(
4253                 /*
4254                         "/
4255                         ^(
4256                         (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4257                         (?:
4258                                 (=+) # Should this be limited to 6?
4259                                 .+?  # Section title...
4260                                 \\2  # Ending = count must match start
4261                         |
4262                                 ^
4263                                 <h([1-6])\b.*?>
4264                                 .*?
4265                                 <\/h\\3\s*>
4266                         )
4267                         (?:$comment|<\/?noinclude>|\s+)* # Trailing whitespace ok
4268                         )$
4269                         /mix",
4270                 */
4271                         "/
4272                         (
4273                                 ^
4274                                 (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4275                                 (=+) # Should this be limited to 6?
4276                                 .+?  # Section title...
4277                                 \\2  # Ending = count must match start
4278                                 (?:$comment|<\/?noinclude>|[ \\t]+)* # Trailing whitespace ok
4279                                 $
4280                         |
4281                                 <h([1-6])\b.*?>
4282                                 .*?
4283                                 <\/h\\3\s*>
4284                         )
4285                         /mix",
4286                         $striptext, -1,
4287                         PREG_SPLIT_DELIM_CAPTURE);
4288
4289                 if( $mode == "get" ) {
4290                         if( $section == 0 ) {
4291                                 // "Section 0" returns the content before any other section.
4292                                 $rv = $secs[0];
4293                         } else {
4294                                 $rv = "";
4295                         }
4296                 } elseif( $mode == "replace" ) {
4297                         if( $section == 0 ) {
4298                                 $rv = $newtext . "\n\n";
4299                                 $remainder = true;
4300                         } else {
4301                                 $rv = $secs[0];
4302                                 $remainder = false;
4303                         }
4304                 }
4305                 $count = 0;
4306                 $sectionLevel = 0;
4307                 for( $index = 1; $index < count( $secs ); ) {
4308                         $headerLine = $secs[$index++];
4309                         if( $secs[$index] ) {
4310                                 // A wiki header
4311                                 $headerLevel = strlen( $secs[$index++] );
4312                         } else {
4313                                 // An HTML header
4314                                 $index++;
4315                                 $headerLevel = intval( $secs[$index++] );
4316                         }
4317                         $content = $secs[$index++];
4318
4319                         $count++;
4320                         if( $mode == "get" ) {
4321                                 if( $count == $section ) {
4322                                         $rv = $headerLine . $content;
4323                                         $sectionLevel = $headerLevel;
4324                                 } elseif( $count > $section ) {
4325                                         if( $sectionLevel && $headerLevel > $sectionLevel ) {
4326                                                 $rv .= $headerLine . $content;
4327                                         } else {
4328                                                 // Broke out to a higher-level section
4329                                                 break;
4330                                         }
4331                                 }
4332                         } elseif( $mode == "replace" ) {
4333                                 if( $count < $section ) {
4334                                         $rv .= $headerLine . $content;
4335                                 } elseif( $count == $section ) {
4336                                         $rv .= $newtext . "\n\n";
4337                                         $sectionLevel = $headerLevel;
4338                                 } elseif( $count > $section ) {
4339                                         if( $headerLevel <= $sectionLevel ) {
4340                                                 // Passed the section's sub-parts.
4341                                                 $remainder = true;
4342                                         }
4343                                         if( $remainder ) {
4344                                                 $rv .= $headerLine . $content;
4345                                         }
4346                                 }
4347                         }
4348                 }
4349                 # reinsert stripped tags
4350                 $rv = $this->unstrip( $rv, $striparray );
4351                 $rv = $this->unstripNoWiki( $rv, $striparray );
4352                 $rv = trim( $rv );
4353                 return $rv;
4354         }
4355
4356         /**
4357          * This function returns the text of a section, specified by a number ($section).
4358          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4359          * the first section before any such heading (section 0).
4360          *
4361          * If a section contains subsections, these are also returned.
4362          *
4363          * @param $text String: text to look in
4364          * @param $section Integer: section number
4365          * @return string text of the requested section
4366          */
4367         function getSection( $text, $section ) {
4368                 return $this->extractSections( $text, $section, "get" );
4369         }
4370
4371         function replaceSection( $oldtext, $section, $text ) {
4372                 return $this->extractSections( $oldtext, $section, "replace", $text );
4373         }
4374
4375 }
4376
4377 /**
4378  * @todo document
4379  * @package MediaWiki
4380  */
4381 class ParserOutput
4382 {
4383         var $mText,             # The output text
4384                 $mLanguageLinks,    # List of the full text of language links, in the order they appear
4385                 $mCategories,       # Map of category names to sort keys
4386                 $mContainsOldMagic, # Boolean variable indicating if the input contained variables like {{CURRENTDAY}}
4387                 $mCacheTime,        # Time when this object was generated, or -1 for uncacheable. Used in ParserCache.
4388                 $mVersion,          # Compatibility check
4389                 $mTitleText,        # title text of the chosen language variant
4390                 $mLinks,            # 2-D map of NS/DBK to ID for the links in the document. ID=zero for broken.
4391                 $mTemplates,        # 2-D map of NS/DBK to ID for the template references. ID=zero for broken.
4392                 $mImages,           # DB keys of the images used, in the array key only
4393                 $mExternalLinks,    # External link URLs, in the key only
4394                 $mHTMLtitle,            # Display HTML title
4395                 $mSubtitle,                     # Additional subtitle
4396                 $mNewSection,           # Show a new section link?
4397                 $mNoGallery;            # No gallery on category page? (__NOGALLERY__)
4398
4399         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
4400                 $containsOldMagic = false, $titletext = '' )
4401         {
4402                 $this->mText = $text;
4403                 $this->mLanguageLinks = $languageLinks;
4404                 $this->mCategories = $categoryLinks;
4405                 $this->mContainsOldMagic = $containsOldMagic;
4406                 $this->mCacheTime = '';
4407                 $this->mVersion = MW_PARSER_VERSION;
4408                 $this->mTitleText = $titletext;
4409                 $this->mLinks = array();
4410                 $this->mTemplates = array();
4411                 $this->mImages = array();
4412                 $this->mExternalLinks = array();
4413                 $this->mHTMLtitle = "" ;
4414                 $this->mSubtitle = "" ;
4415                 $this->mNewSection = false;
4416                 $this->mNoGallery = false;
4417         }
4418
4419         function getText()                   { return $this->mText; }
4420         function &getLanguageLinks()          { return $this->mLanguageLinks; }
4421         function getCategoryLinks()          { return array_keys( $this->mCategories ); }
4422         function &getCategories()            { return $this->mCategories; }
4423         function getCacheTime()              { return $this->mCacheTime; }
4424         function getTitleText()              { return $this->mTitleText; }
4425         function &getLinks()                 { return $this->mLinks; }
4426         function &getTemplates()             { return $this->mTemplates; }
4427         function &getImages()                { return $this->mImages; }
4428         function &getExternalLinks()         { return $this->mExternalLinks; }
4429         function getNoGallery()              { return $this->mNoGallery; }
4430         function getSubtitle()               { return $this->mSubtitle; }
4431
4432         function containsOldMagic()          { return $this->mContainsOldMagic; }
4433         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
4434         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
4435         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategories, $cl ); }
4436         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
4437         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
4438         function setTitleText( $t )          { return wfSetVar($this->mTitleText, $t); }
4439         function setSubtitle( $st )          { return wfSetVar( $this->mSubtitle, $st ); }
4440
4441         function addCategory( $c, $sort )    { $this->mCategories[$c] = $sort; }
4442         function addImage( $name )           { $this->mImages[$name] = 1; }
4443         function addLanguageLink( $t )       { $this->mLanguageLinks[] = $t; }
4444         function addExternalLink( $url )     { $this->mExternalLinks[$url] = 1; }
4445
4446         function setNewSection( $value ) {
4447                 $this->mNewSection = (bool)$value;
4448         }
4449         function getNewSection() {
4450                 return (bool)$this->mNewSection;
4451         }
4452
4453         function addLink( $title, $id ) {
4454                 $ns = $title->getNamespace();
4455                 $dbk = $title->getDBkey();
4456                 if ( !isset( $this->mLinks[$ns] ) ) {
4457                         $this->mLinks[$ns] = array();
4458                 }
4459                 $this->mLinks[$ns][$dbk] = $id;
4460         }
4461
4462         function addTemplate( $title, $id ) {
4463                 $ns = $title->getNamespace();
4464                 $dbk = $title->getDBkey();
4465                 if ( !isset( $this->mTemplates[$ns] ) ) {
4466                         $this->mTemplates[$ns] = array();
4467                 }
4468                 $this->mTemplates[$ns][$dbk] = $id;
4469         }
4470
4471         /**
4472          * Return true if this cached output object predates the global or
4473          * per-article cache invalidation timestamps, or if it comes from
4474          * an incompatible older version.
4475          *
4476          * @param string $touched the affected article's last touched timestamp
4477          * @return bool
4478          * @public
4479          */
4480         function expired( $touched ) {
4481                 global $wgCacheEpoch;
4482                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
4483                        $this->getCacheTime() < $touched ||
4484                        $this->getCacheTime() <= $wgCacheEpoch ||
4485                        !isset( $this->mVersion ) ||
4486                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
4487         }
4488 }
4489
4490 /**
4491  * Set options of the Parser
4492  * @todo document
4493  * @package MediaWiki
4494  */
4495 class ParserOptions
4496 {
4497         # All variables are private
4498         var $mUseTeX;                    # Use texvc to expand <math> tags
4499         var $mUseDynamicDates;           # Use DateFormatter to format dates
4500         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
4501         var $mAllowExternalImages;       # Allow external images inline
4502         var $mAllowExternalImagesFrom;   # If not, any exception?
4503         var $mSkin;                      # Reference to the preferred skin
4504         var $mDateFormat;                # Date format index
4505         var $mEditSection;               # Create "edit section" links
4506         var $mNumberHeadings;            # Automatically number headings
4507         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
4508         var $mTidy;                      # Ask for tidy cleanup
4509         var $mInterfaceMessage;          # Which lang to call for PLURAL and GRAMMAR
4510
4511         var $mUser;                      # Stored user object, just used to initialise the skin
4512
4513         function getUseTeX()                        { return $this->mUseTeX; }
4514         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
4515         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
4516         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
4517         function getAllowExternalImagesFrom()       { return $this->mAllowExternalImagesFrom; }
4518         function getDateFormat()                    { return $this->mDateFormat; }
4519         function getEditSection()                   { return $this->mEditSection; }
4520         function getNumberHeadings()                { return $this->mNumberHeadings; }
4521         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
4522         function getTidy()                          { return $this->mTidy; }
4523         function getInterfaceMessage()              { return $this->mInterfaceMessage; }
4524
4525         function &getSkin() {
4526                 if ( !isset( $this->mSkin ) ) {
4527                         $this->mSkin = $this->mUser->getSkin();
4528                 }
4529                 return $this->mSkin;
4530         }
4531
4532         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
4533         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
4534         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
4535         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
4536         function setAllowExternalImagesFrom( $x )   { return wfSetVar( $this->mAllowExternalImagesFrom, $x ); }
4537         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
4538         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
4539         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
4540         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
4541         function setTidy( $x )                      { return wfSetVar( $this->mTidy, $x); }
4542         function setSkin( &$x ) { $this->mSkin =& $x; }
4543         function setInterfaceMessage( $x )          { return wfSetVar( $this->mInterfaceMessage, $x); }
4544
4545         function ParserOptions( $user = null ) {
4546                 $this->initialiseFromUser( $user );
4547         }
4548
4549         /**
4550          * Get parser options
4551          * @static
4552          */
4553         static function newFromUser( $user ) {
4554                 return new ParserOptions( $user );
4555         }
4556
4557         /** Get user options */
4558         function initialiseFromUser( $userInput ) {
4559                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages;
4560                 global $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion;
4561                 $fname = 'ParserOptions::initialiseFromUser';
4562                 wfProfileIn( $fname );
4563                 if ( !$userInput ) {
4564                         global $wgUser;
4565                         if ( isset( $wgUser ) ) {
4566                                 $user = $wgUser;
4567                         } else {
4568                                 $user = new User;
4569                                 $user->setLoaded( true );
4570                         }
4571                 } else {
4572                         $user =& $userInput;
4573                 }
4574
4575                 $this->mUser = $user;
4576
4577                 $this->mUseTeX = $wgUseTeX;
4578                 $this->mUseDynamicDates = $wgUseDynamicDates;
4579                 $this->mInterwikiMagic = $wgInterwikiMagic;
4580                 $this->mAllowExternalImages = $wgAllowExternalImages;
4581                 $this->mAllowExternalImagesFrom = $wgAllowExternalImagesFrom;
4582                 $this->mSkin = null; # Deferred
4583                 $this->mDateFormat = $user->getOption( 'date' );
4584                 $this->mEditSection = true;
4585                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
4586                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
4587                 $this->mTidy = false;
4588                 $this->mInterfaceMessage = false;
4589                 wfProfileOut( $fname );
4590         }
4591 }
4592
4593 /**
4594  * Callback function used by Parser::replaceLinkHolders()
4595  * to substitute link placeholders.
4596  */
4597 function &wfOutputReplaceMatches( $matches ) {
4598         global $wgOutputReplace;
4599         return $wgOutputReplace[$matches[1]];
4600 }
4601
4602 /**
4603  * Return the total number of articles
4604  */
4605 function wfNumberOfArticles() {
4606         global $wgNumberOfArticles;
4607
4608         wfLoadSiteStats();
4609         return $wgNumberOfArticles;
4610 }
4611
4612 /**
4613  * Return the number of files
4614  */
4615 function wfNumberOfFiles() {
4616         $fname = 'wfNumberOfFiles';
4617
4618         wfProfileIn( $fname );
4619         $dbr =& wfGetDB( DB_SLAVE );
4620         $numImages = $dbr->selectField('site_stats', 'ss_images', array(), $fname );
4621         wfProfileOut( $fname );
4622
4623         return $numImages;
4624 }
4625
4626 /**
4627  * Return the number of user accounts
4628  * @return integer
4629  */
4630 function wfNumberOfUsers() {
4631         wfProfileIn( 'wfNumberOfUsers' );
4632         $dbr =& wfGetDB( DB_SLAVE );
4633         $count = $dbr->selectField( 'site_stats', 'ss_users', array(), 'wfNumberOfUsers' );
4634         wfProfileOut( 'wfNumberOfUsers' );
4635         return (int)$count;
4636 }
4637
4638 /**
4639  * Return the total number of pages
4640  * @return integer
4641  */
4642 function wfNumberOfPages() {
4643         wfProfileIn( 'wfNumberOfPages' );
4644         $dbr =& wfGetDB( DB_SLAVE );
4645         $count = $dbr->selectField( 'site_stats', 'ss_total_pages', array(), 'wfNumberOfPages' );
4646         wfProfileOut( 'wfNumberOfPages' );
4647         return (int)$count;
4648 }
4649
4650 /**
4651  * Return the total number of admins
4652  *
4653  * @return integer
4654  */
4655 function wfNumberOfAdmins() {
4656         static $admins = -1;
4657         wfProfileIn( 'wfNumberOfAdmins' );
4658         if( $admins == -1 ) {
4659                 $dbr =& wfGetDB( DB_SLAVE );
4660                 $admins = $dbr->selectField( 'user_groups', 'COUNT(*)', array( 'ug_group' => 'sysop' ), 'wfNumberOfAdmins' );
4661         }
4662         wfProfileOut( 'wfNumberOfAdmins' );
4663         return (int)$admins;
4664 }
4665
4666 /**
4667  * Count the number of pages in a particular namespace
4668  *
4669  * @param $ns Namespace
4670  * @return integer
4671  */
4672 function wfPagesInNs( $ns ) {
4673         static $pageCount = array();
4674         wfProfileIn( 'wfPagesInNs' );
4675         if( !isset( $pageCount[$ns] ) ) {
4676                 $dbr =& wfGetDB( DB_SLAVE );
4677                 $pageCount[$ns] = $dbr->selectField( 'page', 'COUNT(*)', array( 'page_namespace' => $ns ), 'wfPagesInNs' );
4678         }
4679         wfProfileOut( 'wfPagesInNs' );
4680         return (int)$pageCount[$ns];
4681 }
4682
4683 /**
4684  * Get various statistics from the database
4685  * @private
4686  */
4687 function wfLoadSiteStats() {
4688         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
4689         $fname = 'wfLoadSiteStats';
4690
4691         if ( -1 != $wgNumberOfArticles ) return;
4692         $dbr =& wfGetDB( DB_SLAVE );
4693         $s = $dbr->selectRow( 'site_stats',
4694                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
4695                 array( 'ss_row_id' => 1 ), $fname
4696         );
4697
4698         if ( $s === false ) {
4699                 return;
4700         } else {
4701                 $wgTotalViews = $s->ss_total_views;
4702                 $wgTotalEdits = $s->ss_total_edits;
4703                 $wgNumberOfArticles = $s->ss_good_articles;
4704         }
4705 }
4706
4707 /**
4708  * Escape html tags
4709  * Basically replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
4710  *
4711  * @param $in String: text that might contain HTML tags.
4712  * @return string Escaped string
4713  */
4714 function wfEscapeHTMLTagsOnly( $in ) {
4715         return str_replace(
4716                 array( '"', '>', '<' ),
4717                 array( '&quot;', '&gt;', '&lt;' ),
4718                 $in );
4719 }
4720
4721 ?>