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                 $useSubpages = $this->areSubpagesAllowed();
1412
1413                 # Loop for each link
1414                 for ($k = 0; isset( $a[$k] ); $k++) {
1415                         $line = $a[$k];
1416                         if ( $useLinkPrefixExtension ) {
1417                                 wfProfileIn( $fname.'-prefixhandling' );
1418                                 if ( preg_match( $e2, $s, $m ) ) {
1419                                         $prefix = $m[2];
1420                                         $s = $m[1];
1421                                 } else {
1422                                         $prefix='';
1423                                 }
1424                                 # first link
1425                                 if($first_prefix) {
1426                                         $prefix = $first_prefix;
1427                                         $first_prefix = false;
1428                                 }
1429                                 wfProfileOut( $fname.'-prefixhandling' );
1430                         }
1431
1432                         $might_be_img = false;
1433
1434                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1435                                 $text = $m[2];
1436                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1437                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1438                                 # the real problem is with the $e1 regex
1439                                 # See bug 1300.
1440                                 #
1441                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1442                                 # and no image is in sight. See bug 2095.
1443                                 #
1444                                 if( $text !== '' &&
1445                                         preg_match( "/^\](.*)/s", $m[3], $n ) &&
1446                                         strpos($text, '[') !== false
1447                                 )
1448                                 {
1449                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1450                                         $m[3] = $n[1];
1451                                 }
1452                                 # fix up urlencoded title texts
1453                                 if(preg_match('/%/', $m[1] ))
1454                                         # Should anchors '#' also be rejected?
1455                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1456                                 $trail = $m[3];
1457                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1458                                 $might_be_img = true;
1459                                 $text = $m[2];
1460                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1461                                 $trail = "";
1462                         } else { # Invalid form; output directly
1463                                 $s .= $prefix . '[[' . $line ;
1464                                 continue;
1465                         }
1466
1467                         # Don't allow internal links to pages containing
1468                         # PROTO: where PROTO is a valid URL protocol; these
1469                         # should be external links.
1470                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1471                                 $s .= $prefix . '[[' . $line ;
1472                                 continue;
1473                         }
1474
1475                         # Make subpage if necessary
1476                         if( $useSubpages ) {
1477                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1478                         } else {
1479                                 $link = $m[1];
1480                         }
1481
1482                         $noforce = (substr($m[1], 0, 1) != ':');
1483                         if (!$noforce) {
1484                                 # Strip off leading ':'
1485                                 $link = substr($link, 1);
1486                         }
1487
1488                         $nt = Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1489                         if( !$nt ) {
1490                                 $s .= $prefix . '[[' . $line;
1491                                 continue;
1492                         }
1493
1494                         $ns = $nt->getNamespace();
1495                         $iw = $nt->getInterWiki();
1496
1497                         if ($might_be_img) { # if this is actually an invalid link
1498                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1499                                         $found = false;
1500                                         while (isset ($a[$k+1]) ) {
1501                                                 #look at the next 'line' to see if we can close it there
1502                                                 $spliced = array_splice( $a, $k + 1, 1 );
1503                                                 $next_line = array_shift( $spliced );
1504                                                 if( preg_match("/^(.*?]].*?)]](.*)$/sD", $next_line, $m) ) {
1505                                                 # the first ]] closes the inner link, the second the image
1506                                                         $found = true;
1507                                                         $text .= '[[' . $m[1];
1508                                                         $trail = $m[2];
1509                                                         break;
1510                                                 } elseif( preg_match("/^.*?]].*$/sD", $next_line, $m) ) {
1511                                                         #if there's exactly one ]] that's fine, we'll keep looking
1512                                                         $text .= '[[' . $m[0];
1513                                                 } else {
1514                                                         #if $next_line is invalid too, we need look no further
1515                                                         $text .= '[[' . $next_line;
1516                                                         break;
1517                                                 }
1518                                         }
1519                                         if ( !$found ) {
1520                                                 # we couldn't find the end of this imageLink, so output it raw
1521                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1522                                                 $text = $this->replaceInternalLinks($text);
1523                                                 $s .= $prefix . '[[' . $link . '|' . $text;
1524                                                 # note: no $trail, because without an end, there *is* no trail
1525                                                 continue;
1526                                         }
1527                                 } else { #it's not an image, so output it raw
1528                                         $s .= $prefix . '[[' . $link . '|' . $text;
1529                                         # note: no $trail, because without an end, there *is* no trail
1530                                         continue;
1531                                 }
1532                         }
1533
1534                         $wasblank = ( '' == $text );
1535                         if( $wasblank ) $text = $link;
1536
1537
1538                         # Link not escaped by : , create the various objects
1539                         if( $noforce ) {
1540
1541                                 # Interwikis
1542                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1543                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1544                                         $s = rtrim($s . "\n");
1545                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1546                                         continue;
1547                                 }
1548
1549                                 if ( $ns == NS_IMAGE ) {
1550                                         wfProfileIn( "$fname-image" );
1551                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1552                                                 # recursively parse links inside the image caption
1553                                                 # actually, this will parse them in any other parameters, too,
1554                                                 # but it might be hard to fix that, and it doesn't matter ATM
1555                                                 $text = $this->replaceExternalLinks($text);
1556                                                 $text = $this->replaceInternalLinks($text);
1557
1558                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1559                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1560                                                 $this->mOutput->addImage( $nt->getDBkey() );
1561
1562                                                 wfProfileOut( "$fname-image" );
1563                                                 continue;
1564                                         } else {
1565                                                 # We still need to record the image's presence on the page
1566                                                 $this->mOutput->addImage( $nt->getDBkey() );
1567                                         }
1568                                         wfProfileOut( "$fname-image" );
1569
1570                                 }
1571
1572                                 if ( $ns == NS_CATEGORY ) {
1573                                         wfProfileIn( "$fname-category" );
1574                                         $s = rtrim($s . "\n"); # bug 87
1575
1576                                         if ( $wasblank ) {
1577                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1578                                                         $sortkey = $this->mTitle->getText();
1579                                                 } else {
1580                                                         $sortkey = $this->mTitle->getPrefixedText();
1581                                                 }
1582                                         } else {
1583                                                 $sortkey = $text;
1584                                         }
1585                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1586                                         $sortkey = str_replace( "\n", '', $sortkey );
1587                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1588                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1589
1590                                         /**
1591                                          * Strip the whitespace Category links produce, see bug 87
1592                                          * @todo We might want to use trim($tmp, "\n") here.
1593                                          */
1594                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1595
1596                                         wfProfileOut( "$fname-category" );
1597                                         continue;
1598                                 }
1599                         }
1600
1601                         if( ( $nt->getPrefixedText() === $selflink ) &&
1602                             ( $nt->getFragment() === '' ) ) {
1603                                 # Self-links are handled specially; generally de-link and change to bold.
1604                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1605                                 continue;
1606                         }
1607
1608                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1609                         if( $ns == NS_MEDIA ) {
1610                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1611                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1612                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1613                                 $this->mOutput->addImage( $nt->getDBkey() );
1614                                 continue;
1615                         } elseif( $ns == NS_SPECIAL ) {
1616                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1617                                 continue;
1618                         } elseif( $ns == NS_IMAGE ) {
1619                                 $img = new Image( $nt );
1620                                 if( $img->exists() ) {
1621                                         // Force a blue link if the file exists; may be a remote
1622                                         // upload on the shared repository, and we want to see its
1623                                         // auto-generated page.
1624                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1625                                         continue;
1626                                 }
1627                         }
1628                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1629                 }
1630                 wfProfileOut( $fname );
1631                 return $s;
1632         }
1633
1634         /**
1635          * Make a link placeholder. The text returned can be later resolved to a real link with
1636          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1637          * parsing of interwiki links, and secondly to allow all extistence checks and
1638          * article length checks (for stub links) to be bundled into a single query.
1639          *
1640          */
1641         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1642                 if ( ! is_object($nt) ) {
1643                         # Fail gracefully
1644                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1645                 } else {
1646                         # Separate the link trail from the rest of the link
1647                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1648
1649                         if ( $nt->isExternal() ) {
1650                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1651                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1652                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1653                         } else {
1654                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1655                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1656                                 $this->mLinkHolders['queries'][] = $query;
1657                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1658                                 $this->mLinkHolders['titles'][] = $nt;
1659
1660                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1661                         }
1662                 }
1663                 return $retVal;
1664         }
1665
1666         /**
1667          * Render a forced-blue link inline; protect against double expansion of
1668          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1669          * Since this little disaster has to split off the trail text to avoid
1670          * breaking URLs in the following text without breaking trails on the
1671          * wiki links, it's been made into a horrible function.
1672          *
1673          * @param Title $nt
1674          * @param string $text
1675          * @param string $query
1676          * @param string $trail
1677          * @param string $prefix
1678          * @return string HTML-wikitext mix oh yuck
1679          */
1680         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1681                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1682                 $sk =& $this->mOptions->getSkin();
1683                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1684                 return $this->armorLinks( $link ) . $trail;
1685         }
1686
1687         /**
1688          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1689          * going to go through further parsing steps before inline URL expansion.
1690          *
1691          * In particular this is important when using action=render, which causes
1692          * full URLs to be included.
1693          *
1694          * Oh man I hate our multi-layer parser!
1695          *
1696          * @param string more-or-less HTML
1697          * @return string less-or-more HTML with NOPARSE bits
1698          */
1699         function armorLinks( $text ) {
1700                 return preg_replace( "/\b(" . wfUrlProtocols() . ')/',
1701                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1702         }
1703
1704         /**
1705          * Return true if subpage links should be expanded on this page.
1706          * @return bool
1707          */
1708         function areSubpagesAllowed() {
1709                 # Some namespaces don't allow subpages
1710                 global $wgNamespacesWithSubpages;
1711                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1712         }
1713
1714         /**
1715          * Handle link to subpage if necessary
1716          * @param string $target the source of the link
1717          * @param string &$text the link text, modified as necessary
1718          * @return string the full name of the link
1719          * @private
1720          */
1721         function maybeDoSubpageLink($target, &$text) {
1722                 # Valid link forms:
1723                 # Foobar -- normal
1724                 # :Foobar -- override special treatment of prefix (images, language links)
1725                 # /Foobar -- convert to CurrentPage/Foobar
1726                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1727                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1728                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1729
1730                 $fname = 'Parser::maybeDoSubpageLink';
1731                 wfProfileIn( $fname );
1732                 $ret = $target; # default return value is no change
1733
1734                 # Some namespaces don't allow subpages,
1735                 # so only perform processing if subpages are allowed
1736                 if( $this->areSubpagesAllowed() ) {
1737                         # Look at the first character
1738                         if( $target != '' && $target{0} == '/' ) {
1739                                 # / at end means we don't want the slash to be shown
1740                                 if( substr( $target, -1, 1 ) == '/' ) {
1741                                         $target = substr( $target, 1, -1 );
1742                                         $noslash = $target;
1743                                 } else {
1744                                         $noslash = substr( $target, 1 );
1745                                 }
1746
1747                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1748                                 if( '' === $text ) {
1749                                         $text = $target;
1750                                 } # this might be changed for ugliness reasons
1751                         } else {
1752                                 # check for .. subpage backlinks
1753                                 $dotdotcount = 0;
1754                                 $nodotdot = $target;
1755                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1756                                         ++$dotdotcount;
1757                                         $nodotdot = substr( $nodotdot, 3 );
1758                                 }
1759                                 if($dotdotcount > 0) {
1760                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1761                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1762                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1763                                                 # / at the end means don't show full path
1764                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1765                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1766                                                         if( '' === $text ) {
1767                                                                 $text = $nodotdot;
1768                                                         }
1769                                                 }
1770                                                 $nodotdot = trim( $nodotdot );
1771                                                 if( $nodotdot != '' ) {
1772                                                         $ret .= '/' . $nodotdot;
1773                                                 }
1774                                         }
1775                                 }
1776                         }
1777                 }
1778
1779                 wfProfileOut( $fname );
1780                 return $ret;
1781         }
1782
1783         /**#@+
1784          * Used by doBlockLevels()
1785          * @private
1786          */
1787         /* private */ function closeParagraph() {
1788                 $result = '';
1789                 if ( '' != $this->mLastSection ) {
1790                         $result = '</' . $this->mLastSection  . ">\n";
1791                 }
1792                 $this->mInPre = false;
1793                 $this->mLastSection = '';
1794                 return $result;
1795         }
1796         # getCommon() returns the length of the longest common substring
1797         # of both arguments, starting at the beginning of both.
1798         #
1799         /* private */ function getCommon( $st1, $st2 ) {
1800                 $fl = strlen( $st1 );
1801                 $shorter = strlen( $st2 );
1802                 if ( $fl < $shorter ) { $shorter = $fl; }
1803
1804                 for ( $i = 0; $i < $shorter; ++$i ) {
1805                         if ( $st1{$i} != $st2{$i} ) { break; }
1806                 }
1807                 return $i;
1808         }
1809         # These next three functions open, continue, and close the list
1810         # element appropriate to the prefix character passed into them.
1811         #
1812         /* private */ function openList( $char ) {
1813                 $result = $this->closeParagraph();
1814
1815                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1816                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1817                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1818                 else if ( ';' == $char ) {
1819                         $result .= '<dl><dt>';
1820                         $this->mDTopen = true;
1821                 }
1822                 else { $result = '<!-- ERR 1 -->'; }
1823
1824                 return $result;
1825         }
1826
1827         /* private */ function nextItem( $char ) {
1828                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1829                 else if ( ':' == $char || ';' == $char ) {
1830                         $close = '</dd>';
1831                         if ( $this->mDTopen ) { $close = '</dt>'; }
1832                         if ( ';' == $char ) {
1833                                 $this->mDTopen = true;
1834                                 return $close . '<dt>';
1835                         } else {
1836                                 $this->mDTopen = false;
1837                                 return $close . '<dd>';
1838                         }
1839                 }
1840                 return '<!-- ERR 2 -->';
1841         }
1842
1843         /* private */ function closeList( $char ) {
1844                 if ( '*' == $char ) { $text = '</li></ul>'; }
1845                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1846                 else if ( ':' == $char ) {
1847                         if ( $this->mDTopen ) {
1848                                 $this->mDTopen = false;
1849                                 $text = '</dt></dl>';
1850                         } else {
1851                                 $text = '</dd></dl>';
1852                         }
1853                 }
1854                 else {  return '<!-- ERR 3 -->'; }
1855                 return $text."\n";
1856         }
1857         /**#@-*/
1858
1859         /**
1860          * Make lists from lines starting with ':', '*', '#', etc.
1861          *
1862          * @private
1863          * @return string the lists rendered as HTML
1864          */
1865         function doBlockLevels( $text, $linestart ) {
1866                 $fname = 'Parser::doBlockLevels';
1867                 wfProfileIn( $fname );
1868
1869                 # Parsing through the text line by line.  The main thing
1870                 # happening here is handling of block-level elements p, pre,
1871                 # and making lists from lines starting with * # : etc.
1872                 #
1873                 $textLines = explode( "\n", $text );
1874
1875                 $lastPrefix = $output = '';
1876                 $this->mDTopen = $inBlockElem = false;
1877                 $prefixLength = 0;
1878                 $paragraphStack = false;
1879
1880                 if ( !$linestart ) {
1881                         $output .= array_shift( $textLines );
1882                 }
1883                 foreach ( $textLines as $oLine ) {
1884                         $lastPrefixLength = strlen( $lastPrefix );
1885                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1886                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1887                         if ( !$this->mInPre ) {
1888                                 # Multiple prefixes may abut each other for nested lists.
1889                                 $prefixLength = strspn( $oLine, '*#:;' );
1890                                 $pref = substr( $oLine, 0, $prefixLength );
1891
1892                                 # eh?
1893                                 $pref2 = str_replace( ';', ':', $pref );
1894                                 $t = substr( $oLine, $prefixLength );
1895                                 $this->mInPre = !empty($preOpenMatch);
1896                         } else {
1897                                 # Don't interpret any other prefixes in preformatted text
1898                                 $prefixLength = 0;
1899                                 $pref = $pref2 = '';
1900                                 $t = $oLine;
1901                         }
1902
1903                         # List generation
1904                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1905                                 # Same as the last item, so no need to deal with nesting or opening stuff
1906                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1907                                 $paragraphStack = false;
1908
1909                                 if ( substr( $pref, -1 ) == ';') {
1910                                         # The one nasty exception: definition lists work like this:
1911                                         # ; title : definition text
1912                                         # So we check for : in the remainder text to split up the
1913                                         # title and definition, without b0rking links.
1914                                         $term = $t2 = '';
1915                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1916                                                 $t = $t2;
1917                                                 $output .= $term . $this->nextItem( ':' );
1918                                         }
1919                                 }
1920                         } elseif( $prefixLength || $lastPrefixLength ) {
1921                                 # Either open or close a level...
1922                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
1923                                 $paragraphStack = false;
1924
1925                                 while( $commonPrefixLength < $lastPrefixLength ) {
1926                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
1927                                         --$lastPrefixLength;
1928                                 }
1929                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
1930                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
1931                                 }
1932                                 while ( $prefixLength > $commonPrefixLength ) {
1933                                         $char = substr( $pref, $commonPrefixLength, 1 );
1934                                         $output .= $this->openList( $char );
1935
1936                                         if ( ';' == $char ) {
1937                                                 # FIXME: This is dupe of code above
1938                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1939                                                         $t = $t2;
1940                                                         $output .= $term . $this->nextItem( ':' );
1941                                                 }
1942                                         }
1943                                         ++$commonPrefixLength;
1944                                 }
1945                                 $lastPrefix = $pref2;
1946                         }
1947                         if( 0 == $prefixLength ) {
1948                                 wfProfileIn( "$fname-paragraph" );
1949                                 # No prefix (not in list)--go to paragraph mode
1950                                 // XXX: use a stack for nestable elements like span, table and div
1951                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/center|<\\/tr|<\\/td|<\\/th)/iS', $t );
1952                                 $closematch = preg_match(
1953                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
1954                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<center)/iS', $t );
1955                                 if ( $openmatch or $closematch ) {
1956                                         $paragraphStack = false;
1957                                         # TODO bug 5718: paragraph closed
1958                                         $output .= $this->closeParagraph();
1959                                         if ( $preOpenMatch and !$preCloseMatch ) {
1960                                                 $this->mInPre = true;
1961                                         }
1962                                         if ( $closematch ) {
1963                                                 $inBlockElem = false;
1964                                         } else {
1965                                                 $inBlockElem = true;
1966                                         }
1967                                 } else if ( !$inBlockElem && !$this->mInPre ) {
1968                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
1969                                                 // pre
1970                                                 if ($this->mLastSection != 'pre') {
1971                                                         $paragraphStack = false;
1972                                                         $output .= $this->closeParagraph().'<pre>';
1973                                                         $this->mLastSection = 'pre';
1974                                                 }
1975                                                 $t = substr( $t, 1 );
1976                                         } else {
1977                                                 // paragraph
1978                                                 if ( '' == trim($t) ) {
1979                                                         if ( $paragraphStack ) {
1980                                                                 $output .= $paragraphStack.'<br />';
1981                                                                 $paragraphStack = false;
1982                                                                 $this->mLastSection = 'p';
1983                                                         } else {
1984                                                                 if ($this->mLastSection != 'p' ) {
1985                                                                         $output .= $this->closeParagraph();
1986                                                                         $this->mLastSection = '';
1987                                                                         $paragraphStack = '<p>';
1988                                                                 } else {
1989                                                                         $paragraphStack = '</p><p>';
1990                                                                 }
1991                                                         }
1992                                                 } else {
1993                                                         if ( $paragraphStack ) {
1994                                                                 $output .= $paragraphStack;
1995                                                                 $paragraphStack = false;
1996                                                                 $this->mLastSection = 'p';
1997                                                         } else if ($this->mLastSection != 'p') {
1998                                                                 $output .= $this->closeParagraph().'<p>';
1999                                                                 $this->mLastSection = 'p';
2000                                                         }
2001                                                 }
2002                                         }
2003                                 }
2004                                 wfProfileOut( "$fname-paragraph" );
2005                         }
2006                         // somewhere above we forget to get out of pre block (bug 785)
2007                         if($preCloseMatch && $this->mInPre) {
2008                                 $this->mInPre = false;
2009                         }
2010                         if ($paragraphStack === false) {
2011                                 $output .= $t."\n";
2012                         }
2013                 }
2014                 while ( $prefixLength ) {
2015                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2016                         --$prefixLength;
2017                 }
2018                 if ( '' != $this->mLastSection ) {
2019                         $output .= '</' . $this->mLastSection . '>';
2020                         $this->mLastSection = '';
2021                 }
2022
2023                 wfProfileOut( $fname );
2024                 return $output;
2025         }
2026
2027         /**
2028          * Split up a string on ':', ignoring any occurences inside tags
2029          * to prevent illegal overlapping.
2030          * @param string $str the string to split
2031          * @param string &$before set to everything before the ':'
2032          * @param string &$after set to everything after the ':'
2033          * return string the position of the ':', or false if none found
2034          */
2035         function findColonNoLinks($str, &$before, &$after) {
2036                 $fname = 'Parser::findColonNoLinks';
2037                 wfProfileIn( $fname );
2038
2039                 $pos = strpos( $str, ':' );
2040                 if( $pos === false ) {
2041                         // Nothing to find!
2042                         wfProfileOut( $fname );
2043                         return false;
2044                 }
2045
2046                 $lt = strpos( $str, '<' );
2047                 if( $lt === false || $lt > $pos ) {
2048                         // Easy; no tag nesting to worry about
2049                         $before = substr( $str, 0, $pos );
2050                         $after = substr( $str, $pos+1 );
2051                         wfProfileOut( $fname );
2052                         return $pos;
2053                 }
2054
2055                 // Ugly state machine to walk through avoiding tags.
2056                 $state = MW_COLON_STATE_TEXT;
2057                 $stack = 0;
2058                 $len = strlen( $str );
2059                 for( $i = 0; $i < $len; $i++ ) {
2060                         $c = $str{$i};
2061
2062                         switch( $state ) {
2063                         // (Using the number is a performance hack for common cases)
2064                         case 0: // MW_COLON_STATE_TEXT:
2065                                 switch( $c ) {
2066                                 case "<":
2067                                         // Could be either a <start> tag or an </end> tag
2068                                         $state = MW_COLON_STATE_TAGSTART;
2069                                         break;
2070                                 case ":":
2071                                         if( $stack == 0 ) {
2072                                                 // We found it!
2073                                                 $before = substr( $str, 0, $i );
2074                                                 $after = substr( $str, $i + 1 );
2075                                                 wfProfileOut( $fname );
2076                                                 return $i;
2077                                         }
2078                                         // Embedded in a tag; don't break it.
2079                                         break;
2080                                 default:
2081                                         // Skip ahead looking for something interesting
2082                                         $colon = strpos( $str, ':', $i );
2083                                         if( $colon === false ) {
2084                                                 // Nothing else interesting
2085                                                 wfProfileOut( $fname );
2086                                                 return false;
2087                                         }
2088                                         $lt = strpos( $str, '<', $i );
2089                                         if( $stack === 0 ) {
2090                                                 if( $lt === false || $colon < $lt ) {
2091                                                         // We found it!
2092                                                         $before = substr( $str, 0, $colon );
2093                                                         $after = substr( $str, $colon + 1 );
2094                                                         wfProfileOut( $fname );
2095                                                         return $i;
2096                                                 }
2097                                         }
2098                                         if( $lt === false ) {
2099                                                 // Nothing else interesting to find; abort!
2100                                                 // We're nested, but there's no close tags left. Abort!
2101                                                 break 2;
2102                                         }
2103                                         // Skip ahead to next tag start
2104                                         $i = $lt;
2105                                         $state = MW_COLON_STATE_TAGSTART;
2106                                 }
2107                                 break;
2108                         case 1: // MW_COLON_STATE_TAG:
2109                                 // In a <tag>
2110                                 switch( $c ) {
2111                                 case ">":
2112                                         $stack++;
2113                                         $state = MW_COLON_STATE_TEXT;
2114                                         break;
2115                                 case "/":
2116                                         // Slash may be followed by >?
2117                                         $state = MW_COLON_STATE_TAGSLASH;
2118                                         break;
2119                                 default:
2120                                         // ignore
2121                                 }
2122                                 break;
2123                         case 2: // MW_COLON_STATE_TAGSTART:
2124                                 switch( $c ) {
2125                                 case "/":
2126                                         $state = MW_COLON_STATE_CLOSETAG;
2127                                         break;
2128                                 case "!":
2129                                         $state = MW_COLON_STATE_COMMENT;
2130                                         break;
2131                                 case ">":
2132                                         // Illegal early close? This shouldn't happen D:
2133                                         $state = MW_COLON_STATE_TEXT;
2134                                         break;
2135                                 default:
2136                                         $state = MW_COLON_STATE_TAG;
2137                                 }
2138                                 break;
2139                         case 3: // MW_COLON_STATE_CLOSETAG:
2140                                 // In a </tag>
2141                                 if( $c == ">" ) {
2142                                         $stack--;
2143                                         if( $stack < 0 ) {
2144                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2145                                                 wfProfileOut( $fname );
2146                                                 return false;
2147                                         }
2148                                         $state = MW_COLON_STATE_TEXT;
2149                                 }
2150                                 break;
2151                         case MW_COLON_STATE_TAGSLASH:
2152                                 if( $c == ">" ) {
2153                                         // Yes, a self-closed tag <blah/>
2154                                         $state = MW_COLON_STATE_TEXT;
2155                                 } else {
2156                                         // Probably we're jumping the gun, and this is an attribute
2157                                         $state = MW_COLON_STATE_TAG;
2158                                 }
2159                                 break;
2160                         case 5: // MW_COLON_STATE_COMMENT:
2161                                 if( $c == "-" ) {
2162                                         $state = MW_COLON_STATE_COMMENTDASH;
2163                                 }
2164                                 break;
2165                         case MW_COLON_STATE_COMMENTDASH:
2166                                 if( $c == "-" ) {
2167                                         $state = MW_COLON_STATE_COMMENTDASHDASH;
2168                                 } else {
2169                                         $state = MW_COLON_STATE_COMMENT;
2170                                 }
2171                                 break;
2172                         case MW_COLON_STATE_COMMENTDASHDASH:
2173                                 if( $c == ">" ) {
2174                                         $state = MW_COLON_STATE_TEXT;
2175                                 } else {
2176                                         $state = MW_COLON_STATE_COMMENT;
2177                                 }
2178                                 break;
2179                         default:
2180                                 throw new MWException( "State machine error in $fname" );
2181                         }
2182                 }
2183                 if( $stack > 0 ) {
2184                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2185                         return false;
2186                 }
2187                 wfProfileOut( $fname );
2188                 return false;
2189         }
2190
2191         /**
2192          * Return value of a magic variable (like PAGENAME)
2193          *
2194          * @private
2195          */
2196         function getVariableValue( $index ) {
2197                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2198
2199                 /**
2200                  * Some of these require message or data lookups and can be
2201                  * expensive to check many times.
2202                  */
2203                 static $varCache = array();
2204                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) )
2205                         if ( isset( $varCache[$index] ) )
2206                                 return $varCache[$index];
2207
2208                 $ts = time();
2209                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2210
2211                 switch ( $index ) {
2212                         case 'currentmonth':
2213                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
2214                         case 'currentmonthname':
2215                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2216                         case 'currentmonthnamegen':
2217                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2218                         case 'currentmonthabbrev':
2219                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2220                         case 'currentday':
2221                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2222                         case 'currentday2':
2223                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2224                         case 'pagename':
2225                                 return $this->mTitle->getText();
2226                         case 'pagenamee':
2227                                 return $this->mTitle->getPartialURL();
2228                         case 'fullpagename':
2229                                 return $this->mTitle->getPrefixedText();
2230                         case 'fullpagenamee':
2231                                 return $this->mTitle->getPrefixedURL();
2232                         case 'subpagename':
2233                                 return $this->mTitle->getSubpageText();
2234                         case 'subpagenamee':
2235                                 return $this->mTitle->getSubpageUrlForm();
2236                         case 'basepagename':
2237                                 return $this->mTitle->getBaseText();
2238                         case 'basepagenamee':
2239                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2240                         case 'talkpagename':
2241                                 if( $this->mTitle->canTalk() ) {
2242                                         $talkPage = $this->mTitle->getTalkPage();
2243                                         return $talkPage->getPrefixedText();
2244                                 } else {
2245                                         return '';
2246                                 }
2247                         case 'talkpagenamee':
2248                                 if( $this->mTitle->canTalk() ) {
2249                                         $talkPage = $this->mTitle->getTalkPage();
2250                                         return $talkPage->getPrefixedUrl();
2251                                 } else {
2252                                         return '';
2253                                 }
2254                         case 'subjectpagename':
2255                                 $subjPage = $this->mTitle->getSubjectPage();
2256                                 return $subjPage->getPrefixedText();
2257                         case 'subjectpagenamee':
2258                                 $subjPage = $this->mTitle->getSubjectPage();
2259                                 return $subjPage->getPrefixedUrl();
2260                         case 'revisionid':
2261                                 return $this->mRevisionId;
2262                         case 'namespace':
2263                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2264                         case 'namespacee':
2265                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2266                         case 'talkspace':
2267                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2268                         case 'talkspacee':
2269                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2270                         case 'subjectspace':
2271                                 return $this->mTitle->getSubjectNsText();
2272                         case 'subjectspacee':
2273                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2274                         case 'currentdayname':
2275                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2276                         case 'currentyear':
2277                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2278                         case 'currenttime':
2279                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2280                         case 'currentweek':
2281                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2282                                 // int to remove the padding
2283                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2284                         case 'currentdow':
2285                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2286                         case 'numberofarticles':
2287                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
2288                         case 'numberoffiles':
2289                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
2290                         case 'numberofusers':
2291                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfUsers() );
2292                         case 'numberofpages':
2293                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfPages() );
2294                         case 'numberofadmins':
2295                                 return $varCache[$index]  = $wgContLang->formatNum( wfNumberOfAdmins() );
2296                         case 'currenttimestamp':
2297                                 return $varCache[$index] = wfTimestampNow();
2298                         case 'currentversion':
2299                                 global $wgVersion;
2300                                 return $wgVersion;
2301                         case 'sitename':
2302                                 return $wgSitename;
2303                         case 'server':
2304                                 return $wgServer;
2305                         case 'servername':
2306                                 return $wgServerName;
2307                         case 'scriptpath':
2308                                 return $wgScriptPath;
2309                         case 'directionmark':
2310                                 return $wgContLang->getDirMark();
2311                         case 'contentlanguage':
2312                                 global $wgContLanguageCode;
2313                                 return $wgContLanguageCode;
2314                         default:
2315                                 $ret = null;
2316                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2317                                         return $ret;
2318                                 else
2319                                         return null;
2320                 }
2321         }
2322
2323         /**
2324          * initialise the magic variables (like CURRENTMONTHNAME)
2325          *
2326          * @private
2327          */
2328         function initialiseVariables() {
2329                 $fname = 'Parser::initialiseVariables';
2330                 wfProfileIn( $fname );
2331                 $variableIDs = MagicWord::getVariableIDs();
2332
2333                 $this->mVariables = array();
2334                 foreach ( $variableIDs as $id ) {
2335                         $mw =& MagicWord::get( $id );
2336                         $mw->addToArray( $this->mVariables, $id );
2337                 }
2338                 wfProfileOut( $fname );
2339         }
2340
2341         /**
2342          * parse any parentheses in format ((title|part|part))
2343          * and call callbacks to get a replacement text for any found piece
2344          *
2345          * @param string $text The text to parse
2346          * @param array $callbacks rules in form:
2347          *     '{' => array(                            # opening parentheses
2348          *                                      'end' => '}',   # closing parentheses
2349          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2350          *                                                                4 => callback         # replacement callback to call if {{{{..}}}} is found
2351          *                                                                )
2352          *                                      )
2353          * @private
2354          */
2355         function replace_callback ($text, $callbacks) {
2356                 wfProfileIn( __METHOD__ . '-self' );
2357                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2358                 $lastOpeningBrace = -1;                 # last not closed parentheses
2359
2360                 for ($i = 0; $i < strlen($text); $i++) {
2361                         # check for any opening brace
2362                         $rule = null;
2363                         $nextPos = -1;
2364                         foreach ($callbacks as $key => $value) {
2365                                 $pos = strpos ($text, $key, $i);
2366                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)) {
2367                                         $rule = $value;
2368                                         $nextPos = $pos;
2369                                 }
2370                         }
2371
2372                         if ($lastOpeningBrace >= 0) {
2373                                 $pos = strpos ($text, $openingBraceStack[$lastOpeningBrace]['braceEnd'], $i);
2374
2375                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2376                                         $rule = null;
2377                                         $nextPos = $pos;
2378                                 }
2379
2380                                 $pos = strpos ($text, '|', $i);
2381
2382                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2383                                         $rule = null;
2384                                         $nextPos = $pos;
2385                                 }
2386                         }
2387
2388                         if ($nextPos == -1)
2389                                 break;
2390
2391                         $i = $nextPos;
2392
2393                         # found openning brace, lets add it to parentheses stack
2394                         if (null != $rule) {
2395                                 $piece = array('brace' => $text[$i],
2396                                                            'braceEnd' => $rule['end'],
2397                                                            'count' => 1,
2398                                                            'title' => '',
2399                                                            'parts' => null);
2400
2401                                 # count openning brace characters
2402                                 while ($i+1 < strlen($text) && $text[$i+1] == $piece['brace']) {
2403                                         $piece['count']++;
2404                                         $i++;
2405                                 }
2406
2407                                 $piece['startAt'] = $i+1;
2408                                 $piece['partStart'] = $i+1;
2409
2410                                 # we need to add to stack only if openning brace count is enough for any given rule
2411                                 foreach ($rule['cb'] as $cnt => $fn) {
2412                                         if ($piece['count'] >= $cnt) {
2413                                                 $lastOpeningBrace ++;
2414                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2415                                                 break;
2416                                         }
2417                                 }
2418
2419                                 continue;
2420                         }
2421                         else if ($lastOpeningBrace >= 0) {
2422                                 # first check if it is a closing brace
2423                                 if ($openingBraceStack[$lastOpeningBrace]['braceEnd'] == $text[$i]) {
2424                                         # lets check if it is enough characters for closing brace
2425                                         $count = 1;
2426                                         while ($i+$count < strlen($text) && $text[$i+$count] == $text[$i])
2427                                                 $count++;
2428
2429                                         # if there are more closing parentheses than opening ones, we parse less
2430                                         if ($openingBraceStack[$lastOpeningBrace]['count'] < $count)
2431                                                 $count = $openingBraceStack[$lastOpeningBrace]['count'];
2432
2433                                         # check for maximum matching characters (if there are 5 closing characters, we will probably need only 3 - depending on the rules)
2434                                         $matchingCount = 0;
2435                                         $matchingCallback = null;
2436                                         foreach ($callbacks[$openingBraceStack[$lastOpeningBrace]['brace']]['cb'] as $cnt => $fn) {
2437                                                 if ($count >= $cnt && $matchingCount < $cnt) {
2438                                                         $matchingCount = $cnt;
2439                                                         $matchingCallback = $fn;
2440                                                 }
2441                                         }
2442
2443                                         if ($matchingCount == 0) {
2444                                                 $i += $count - 1;
2445                                                 continue;
2446                                         }
2447
2448                                         # lets set a title or last part (if '|' was found)
2449                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts'])
2450                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2451                                         else
2452                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2453
2454                                         $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2455                                         $pieceEnd = $i + $matchingCount;
2456
2457                                         if( is_callable( $matchingCallback ) ) {
2458                                                 $cbArgs = array (
2459                                                                                  'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2460                                                                                  'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2461                                                                                  'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2462                                                                                  'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == "\n")),
2463                                                                                  );
2464                                                 # finally we can call a user callback and replace piece of text
2465                                                 wfProfileOut( __METHOD__ . '-self' );
2466                                                 $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2467                                                 wfProfileIn( __METHOD__ . '-self' );
2468                                                 $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2469                                                 $i = $pieceStart + strlen($replaceWith) - 1;
2470                                         }
2471                                         else {
2472                                                 # null value for callback means that parentheses should be parsed, but not replaced
2473                                                 $i += $matchingCount - 1;
2474                                         }
2475
2476                                         # reset last openning parentheses, but keep it in case there are unused characters
2477                                         $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2478                                                                    'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2479                                                                    'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2480                                                                    'title' => '',
2481                                                                    'parts' => null,
2482                                                                    'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2483                                         $openingBraceStack[$lastOpeningBrace--] = null;
2484
2485                                         if ($matchingCount < $piece['count']) {
2486                                                 $piece['count'] -= $matchingCount;
2487                                                 $piece['startAt'] -= $matchingCount;
2488                                                 $piece['partStart'] = $piece['startAt'];
2489                                                 # do we still qualify for any callback with remaining count?
2490                                                 foreach ($callbacks[$piece['brace']]['cb'] as $cnt => $fn) {
2491                                                         if ($piece['count'] >= $cnt) {
2492                                                                 $lastOpeningBrace ++;
2493                                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2494                                                                 break;
2495                                                         }
2496                                                 }
2497                                         }
2498                                         continue;
2499                                 }
2500
2501                                 # lets set a title if it is a first separator, or next part otherwise
2502                                 if ($text[$i] == '|') {
2503                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2504                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2505                                                 $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2506                                         }
2507                                         else
2508                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2509
2510                                         $openingBraceStack[$lastOpeningBrace]['partStart'] = $i + 1;
2511                                 }
2512                         }
2513                 }
2514
2515                 wfProfileOut( __METHOD__ . '-self' );
2516                 return $text;
2517         }
2518
2519         /**
2520          * Replace magic variables, templates, and template arguments
2521          * with the appropriate text. Templates are substituted recursively,
2522          * taking care to avoid infinite loops.
2523          *
2524          * Note that the substitution depends on value of $mOutputType:
2525          *  OT_WIKI: only {{subst:}} templates
2526          *  OT_MSG: only magic variables
2527          *  OT_HTML: all templates and magic variables
2528          *
2529          * @param string $tex The text to transform
2530          * @param array $args Key-value pairs representing template parameters to substitute
2531          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2532          * @private
2533          */
2534         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2535                 # Prevent too big inclusions
2536                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
2537                         return $text;
2538                 }
2539
2540                 $fname = 'Parser::replaceVariables';
2541                 wfProfileIn( $fname );
2542
2543                 # This function is called recursively. To keep track of arguments we need a stack:
2544                 array_push( $this->mArgStack, $args );
2545
2546                 $braceCallbacks = array();
2547                 if ( !$argsOnly ) {
2548                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2549                 }
2550                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
2551                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2552                 }
2553                 $callbacks = array();
2554                 $callbacks['{'] = array('end' => '}', 'cb' => $braceCallbacks);
2555                 $callbacks['['] = array('end' => ']', 'cb' => array(2=>null));
2556                 $text = $this->replace_callback ($text, $callbacks);
2557
2558                 array_pop( $this->mArgStack );
2559
2560                 wfProfileOut( $fname );
2561                 return $text;
2562         }
2563
2564         /**
2565          * Replace magic variables
2566          * @private
2567          */
2568         function variableSubstitution( $matches ) {
2569                 $fname = 'Parser::variableSubstitution';
2570                 $varname = $matches[1];
2571                 wfProfileIn( $fname );
2572                 $skip = false;
2573                 if ( $this->mOutputType == OT_WIKI ) {
2574                         # Do only magic variables prefixed by SUBST
2575                         $mwSubst =& MagicWord::get( 'subst' );
2576                         if (!$mwSubst->matchStartAndRemove( $varname ))
2577                                 $skip = true;
2578                         # Note that if we don't substitute the variable below,
2579                         # we don't remove the {{subst:}} magic word, in case
2580                         # it is a template rather than a magic variable.
2581                 }
2582                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2583                         $id = $this->mVariables[$varname];
2584                         $text = $this->getVariableValue( $id );
2585                         $this->mOutput->mContainsOldMagic = true;
2586                 } else {
2587                         $text = $matches[0];
2588                 }
2589                 wfProfileOut( $fname );
2590                 return $text;
2591         }
2592
2593         # Split template arguments
2594         function getTemplateArgs( $argsString ) {
2595                 if ( $argsString === '' ) {
2596                         return array();
2597                 }
2598
2599                 $args = explode( '|', substr( $argsString, 1 ) );
2600
2601                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2602                 # merged with the next arg because the '|' character between belongs
2603                 # to the link syntax and not the template parameter syntax.
2604                 $argc = count($args);
2605
2606                 for ( $i = 0; $i < $argc-1; $i++ ) {
2607                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2608                                 $args[$i] .= '|'.$args[$i+1];
2609                                 array_splice($args, $i+1, 1);
2610                                 $i--;
2611                                 $argc--;
2612                         }
2613                 }
2614
2615                 return $args;
2616         }
2617
2618         /**
2619          * Return the text of a template, after recursively
2620          * replacing any variables or templates within the template.
2621          *
2622          * @param array $piece The parts of the template
2623          *  $piece['text']: matched text
2624          *  $piece['title']: the title, i.e. the part before the |
2625          *  $piece['parts']: the parameter array
2626          * @return string the text of the template
2627          * @private
2628          */
2629         function braceSubstitution( $piece ) {
2630                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $action;
2631                 $fname = 'Parser::braceSubstitution';
2632                 wfProfileIn( $fname );
2633
2634                 # Flags
2635                 $found = false;             # $text has been filled
2636                 $nowiki = false;            # wiki markup in $text should be escaped
2637                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2638                 $noargs = false;            # Don't replace triple-brace arguments in $text
2639                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2640                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2641                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2642
2643                 # Title object, where $text came from
2644                 $title = NULL;
2645
2646                 $linestart = '';
2647
2648                 # $part1 is the bit before the first |, and must contain only title characters
2649                 # $args is a list of arguments, starting from index 0, not including $part1
2650
2651                 $part1 = $piece['title'];
2652                 # If the third subpattern matched anything, it will start with |
2653
2654                 if (null == $piece['parts']) {
2655                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2656                         if ($replaceWith != $piece['text']) {
2657                                 $text = $replaceWith;
2658                                 $found = true;
2659                                 $noparse = true;
2660                                 $noargs = true;
2661                         }
2662                 }
2663
2664                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2665                 $argc = count( $args );
2666
2667                 # SUBST
2668                 if ( !$found ) {
2669                         $mwSubst =& MagicWord::get( 'subst' );
2670                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2671                                 # One of two possibilities is true:
2672                                 # 1) Found SUBST but not in the PST phase
2673                                 # 2) Didn't find SUBST and in the PST phase
2674                                 # In either case, return without further processing
2675                                 $text = $piece['text'];
2676                                 $found = true;
2677                                 $noparse = true;
2678                                 $noargs = true;
2679                         }
2680                 }
2681
2682                 # MSG, MSGNW, INT and RAW
2683                 if ( !$found ) {
2684                         # Check for MSGNW:
2685                         $mwMsgnw =& MagicWord::get( 'msgnw' );
2686                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2687                                 $nowiki = true;
2688                         } else {
2689                                 # Remove obsolete MSG:
2690                                 $mwMsg =& MagicWord::get( 'msg' );
2691                                 $mwMsg->matchStartAndRemove( $part1 );
2692                         }
2693
2694                         # Check for RAW:
2695                         $mwRaw =& MagicWord::get( 'raw' );
2696                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2697                                 $forceRawInterwiki = true;
2698                         }
2699
2700                         # Check if it is an internal message
2701                         $mwInt =& MagicWord::get( 'int' );
2702                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2703                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2704                                         $text = $linestart . wfMsgReal( $part1, $args, true );
2705                                         $found = true;
2706                                 }
2707                         }
2708                 }
2709
2710                 # Parser functions
2711                 if ( !$found ) {
2712                         wfProfileIn( __METHOD__ . '-pfunc' );
2713
2714                         $colonPos = strpos( $part1, ':' );
2715                         if ( $colonPos !== false ) {
2716                                 # Case sensitive functions
2717                                 $function = substr( $part1, 0, $colonPos );
2718                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2719                                         $function = $this->mFunctionSynonyms[1][$function];
2720                                 } else {
2721                                         # Case insensitive functions
2722                                         $function = strtolower( $function );
2723                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2724                                                 $function = $this->mFunctionSynonyms[0][$function];
2725                                         } else {
2726                                                 $function = false;
2727                                         }
2728                                 }
2729                                 if ( $function ) {
2730                                         $funcArgs = array_map( 'trim', $args );
2731                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
2732                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
2733                                         $found = true;
2734
2735                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
2736                                         //$noargs = true;
2737                                         //$noparse = true;
2738
2739                                         if ( is_array( $result ) ) {
2740                                                 if ( isset( $result[0] ) ) {
2741                                                         $text = $linestart . $result[0];
2742                                                         unset( $result[0] );
2743                                                 }
2744
2745                                                 // Extract flags into the local scope
2746                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
2747                                                 extract( $result );
2748                                         } else {
2749                                                 $text = $linestart . $result;
2750                                         }
2751                                 }
2752                         }
2753                         wfProfileOut( __METHOD__ . '-pfunc' );
2754                 }
2755
2756                 # Template table test
2757
2758                 # Did we encounter this template already? If yes, it is in the cache
2759                 # and we need to check for loops.
2760                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
2761                         $found = true;
2762
2763                         # Infinite loop test
2764                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2765                                 $noparse = true;
2766                                 $noargs = true;
2767                                 $found = true;
2768                                 $text = $linestart .
2769                                         '{{' . $part1 . '}}' .
2770                                         '<!-- WARNING: template loop detected -->';
2771                                 wfDebug( "$fname: template loop broken at '$part1'\n" );
2772                         } else {
2773                                 # set $text to cached message.
2774                                 $text = $linestart . $this->mTemplates[$piece['title']];
2775                         }
2776                 }
2777
2778                 # Load from database
2779                 $lastPathLevel = $this->mTemplatePath;
2780                 if ( !$found ) {
2781                         wfProfileIn( __METHOD__ . '-loadtpl' );
2782                         $ns = NS_TEMPLATE;
2783                         # declaring $subpage directly in the function call
2784                         # does not work correctly with references and breaks
2785                         # {{/subpage}}-style inclusions
2786                         $subpage = '';
2787                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2788                         if ($subpage !== '') {
2789                                 $ns = $this->mTitle->getNamespace();
2790                         }
2791                         $title = Title::newFromText( $part1, $ns );
2792
2793
2794                         if ( !is_null( $title ) ) {
2795                                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
2796                                 # Check for language variants if the template is not found
2797                                 if($checkVariantLink && $title->getArticleID() == 0){
2798                                         $wgContLang->findVariantLink($part1, $title);
2799                                 }
2800
2801                                 if ( !$title->isExternal() ) {
2802                                         # Check for excessive inclusion
2803                                         $dbk = $title->getPrefixedDBkey();
2804                                         if ( $this->incrementIncludeCount( $dbk ) ) {
2805                                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->mOutputType != OT_WIKI ) {
2806                                                         $text = SpecialPage::capturePath( $title );
2807                                                         if ( is_string( $text ) ) {
2808                                                                 $found = true;
2809                                                                 $noparse = true;
2810                                                                 $noargs = true;
2811                                                                 $isHTML = true;
2812                                                                 $this->disableCache();
2813                                                         }
2814                                                 } else {
2815                                                         $articleContent = $this->fetchTemplate( $title );
2816                                                         if ( $articleContent !== false ) {
2817                                                                 $found = true;
2818                                                                 $text = $articleContent;
2819                                                                 $replaceHeadings = true;
2820                                                         }
2821                                                 }
2822                                         }
2823
2824                                         # If the title is valid but undisplayable, make a link to it
2825                                         if ( $this->mOutputType == OT_HTML && !$found ) {
2826                                                 $text = '[['.$title->getPrefixedText().']]';
2827                                                 $found = true;
2828                                         }
2829                                 } elseif ( $title->isTrans() ) {
2830                                         // Interwiki transclusion
2831                                         if ( $this->mOutputType == OT_HTML && !$forceRawInterwiki ) {
2832                                                 $text = $this->interwikiTransclude( $title, 'render' );
2833                                                 $isHTML = true;
2834                                                 $noparse = true;
2835                                         } else {
2836                                                 $text = $this->interwikiTransclude( $title, 'raw' );
2837                                                 $replaceHeadings = true;
2838                                         }
2839                                         $found = true;
2840                                 }
2841
2842                                 # Template cache array insertion
2843                                 # Use the original $piece['title'] not the mangled $part1, so that
2844                                 # modifiers such as RAW: produce separate cache entries
2845                                 if( $found ) {
2846                                         if( $isHTML ) {
2847                                                 // A special page; don't store it in the template cache.
2848                                         } else {
2849                                                 $this->mTemplates[$piece['title']] = $text;
2850                                         }
2851                                         $text = $linestart . $text;
2852                                 }
2853                         }
2854                         wfProfileOut( __METHOD__ . '-loadtpl' );
2855                 }
2856
2857                 # Recursive parsing, escaping and link table handling
2858                 # Only for HTML output
2859                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2860                         $text = wfEscapeWikiText( $text );
2861                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found ) {
2862                         if ( $noargs ) {
2863                                 $assocArgs = array();
2864                         } else {
2865                                 # Clean up argument array
2866                                 $assocArgs = array();
2867                                 $index = 1;
2868                                 foreach( $args as $arg ) {
2869                                         $eqpos = strpos( $arg, '=' );
2870                                         if ( $eqpos === false ) {
2871                                                 $assocArgs[$index++] = $arg;
2872                                         } else {
2873                                                 $name = trim( substr( $arg, 0, $eqpos ) );
2874                                                 $value = trim( substr( $arg, $eqpos+1 ) );
2875                                                 if ( $value === false ) {
2876                                                         $value = '';
2877                                                 }
2878                                                 if ( $name !== false ) {
2879                                                         $assocArgs[$name] = $value;
2880                                                 }
2881                                         }
2882                                 }
2883
2884                                 # Add a new element to the templace recursion path
2885                                 $this->mTemplatePath[$part1] = 1;
2886                         }
2887
2888                         if ( !$noparse ) {
2889                                 # If there are any <onlyinclude> tags, only include them
2890                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
2891                                         preg_match_all( '/<onlyinclude>(.*?)\n?<\/onlyinclude>/s', $text, $m );
2892                                         $text = '';
2893                                         foreach ($m[1] as $piece)
2894                                                 $text .= $piece;
2895                                 }
2896                                 # Remove <noinclude> sections and <includeonly> tags
2897                                 $text = preg_replace( '/<noinclude>.*?<\/noinclude>/s', '', $text );
2898                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
2899
2900                                 if( $this->mOutputType == OT_HTML ) {
2901                                         # Strip <nowiki>, <pre>, etc.
2902                                         $text = $this->strip( $text, $this->mStripState );
2903                                         $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
2904                                 }
2905                                 $text = $this->replaceVariables( $text, $assocArgs );
2906
2907                                 # If the template begins with a table or block-level
2908                                 # element, it should be treated as beginning a new line.
2909                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) {
2910                                         $text = "\n" . $text;
2911                                 }
2912                         } elseif ( !$noargs ) {
2913                                 # $noparse and !$noargs
2914                                 # Just replace the arguments, not any double-brace items
2915                                 # This is used for rendered interwiki transclusion
2916                                 $text = $this->replaceVariables( $text, $assocArgs, true );
2917                         }
2918                 }
2919                 # Prune lower levels off the recursion check path
2920                 $this->mTemplatePath = $lastPathLevel;
2921
2922                 if ( !$found ) {
2923                         wfProfileOut( $fname );
2924                         return $piece['text'];
2925                 } else {
2926                         wfProfileIn( __METHOD__ . '-placeholders' );
2927                         if ( $isHTML ) {
2928                                 # Replace raw HTML by a placeholder
2929                                 # Add a blank line preceding, to prevent it from mucking up
2930                                 # immediately preceding headings
2931                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
2932                         } else {
2933                                 # replace ==section headers==
2934                                 # XXX this needs to go away once we have a better parser.
2935                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
2936                                         if( !is_null( $title ) )
2937                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
2938                                         else
2939                                                 $encodedname = base64_encode("");
2940                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
2941                                                 PREG_SPLIT_DELIM_CAPTURE);
2942                                         $text = '';
2943                                         $nsec = 0;
2944                                         for( $i = 0; $i < count($m); $i += 2 ) {
2945                                                 $text .= $m[$i];
2946                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
2947                                                 $hl = $m[$i + 1];
2948                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
2949                                                         $text .= $hl;
2950                                                         continue;
2951                                                 }
2952                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
2953                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
2954                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
2955
2956                                                 $nsec++;
2957                                         }
2958                                 }
2959                         }
2960                         wfProfileOut( __METHOD__ . '-placeholders' );
2961                 }
2962
2963                 # Prune lower levels off the recursion check path
2964                 $this->mTemplatePath = $lastPathLevel;
2965
2966                 if ( !$found ) {
2967                         wfProfileOut( $fname );
2968                         return $piece['text'];
2969                 } else {
2970                         wfProfileOut( $fname );
2971                         return $text;
2972                 }
2973         }
2974
2975         /**
2976          * Fetch the unparsed text of a template and register a reference to it.
2977          */
2978         function fetchTemplate( $title ) {
2979                 $text = false;
2980                 // Loop to fetch the article, with up to 1 redirect
2981                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
2982                         $rev = Revision::newFromTitle( $title );
2983                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
2984                         if ( !$rev ) {
2985                                 break;
2986                         }
2987                         $text = $rev->getText();
2988                         if ( $text === false ) {
2989                                 break;
2990                         }
2991                         // Redirect?
2992                         $title = Title::newFromRedirect( $text );
2993                 }
2994                 return $text;
2995         }
2996
2997         /**
2998          * Transclude an interwiki link.
2999          */
3000         function interwikiTransclude( $title, $action ) {
3001                 global $wgEnableScaryTranscluding, $wgCanonicalNamespaceNames;
3002
3003                 if (!$wgEnableScaryTranscluding)
3004                         return wfMsg('scarytranscludedisabled');
3005
3006                 // The namespace will actually only be 0 or 10, depending on whether there was a leading :
3007                 // But we'll handle it generally anyway
3008                 if ( $title->getNamespace() ) {
3009                         // Use the canonical namespace, which should work anywhere
3010                         $articleName = $wgCanonicalNamespaceNames[$title->getNamespace()] . ':' . $title->getDBkey();
3011                 } else {
3012                         $articleName = $title->getDBkey();
3013                 }
3014
3015                 $url = str_replace('$1', urlencode($articleName), Title::getInterwikiLink($title->getInterwiki()));
3016                 $url .= "?action=$action";
3017                 if (strlen($url) > 255)
3018                         return wfMsg('scarytranscludetoolong');
3019                 return $this->fetchScaryTemplateMaybeFromCache($url);
3020         }
3021
3022         function fetchScaryTemplateMaybeFromCache($url) {
3023                 global $wgTranscludeCacheExpiry;
3024                 $dbr =& wfGetDB(DB_SLAVE);
3025                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3026                                 array('tc_url' => $url));
3027                 if ($obj) {
3028                         $time = $obj->tc_time;
3029                         $text = $obj->tc_contents;
3030                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3031                                 return $text;
3032                         }
3033                 }
3034
3035                 $text = Http::get($url);
3036                 if (!$text)
3037                         return wfMsg('scarytranscludefailed', $url);
3038
3039                 $dbw =& wfGetDB(DB_MASTER);
3040                 $dbw->replace('transcache', array('tc_url'), array(
3041                         'tc_url' => $url,
3042                         'tc_time' => time(),
3043                         'tc_contents' => $text));
3044                 return $text;
3045         }
3046
3047
3048         /**
3049          * Triple brace replacement -- used for template arguments
3050          * @private
3051          */
3052         function argSubstitution( $matches ) {
3053                 $arg = trim( $matches['title'] );
3054                 $text = $matches['text'];
3055                 $inputArgs = end( $this->mArgStack );
3056
3057                 if ( array_key_exists( $arg, $inputArgs ) ) {
3058                         $text = $inputArgs[$arg];
3059                 } else if ($this->mOutputType == OT_HTML && null != $matches['parts'] && count($matches['parts']) > 0) {
3060                         $text = $matches['parts'][0];
3061                 }
3062
3063                 return $text;
3064         }
3065
3066         /**
3067          * Returns true if the function is allowed to include this entity
3068          * @private
3069          */
3070         function incrementIncludeCount( $dbk ) {
3071                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
3072                         $this->mIncludeCount[$dbk] = 0;
3073                 }
3074                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
3075                         return true;
3076                 } else {
3077                         return false;
3078                 }
3079         }
3080
3081         /**
3082          * Detect __NOGALLERY__ magic word and set a placeholder
3083          */
3084         function stripNoGallery( &$text ) {
3085                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3086                 # do not add TOC
3087                 $mw = MagicWord::get( 'nogallery' );
3088                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3089         }
3090
3091         /**
3092          * Detect __TOC__ magic word and set a placeholder
3093          */
3094         function stripToc( $text ) {
3095                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3096                 # do not add TOC
3097                 $mw = MagicWord::get( 'notoc' );
3098                 if( $mw->matchAndRemove( $text ) ) {
3099                         $this->mShowToc = false;
3100                 }
3101
3102                 $mw = MagicWord::get( 'toc' );
3103                 if( $mw->match( $text ) ) {
3104                         $this->mShowToc = true;
3105                         $this->mForceTocPosition = true;
3106
3107                         // Set a placeholder. At the end we'll fill it in with the TOC.
3108                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3109
3110                         // Only keep the first one.
3111                         $text = $mw->replace( '', $text );
3112                 }
3113                 return $text;
3114         }
3115
3116         /**
3117          * This function accomplishes several tasks:
3118          * 1) Auto-number headings if that option is enabled
3119          * 2) Add an [edit] link to sections for logged in users who have enabled the option
3120          * 3) Add a Table of contents on the top for users who have enabled the option
3121          * 4) Auto-anchor headings
3122          *
3123          * It loops through all headlines, collects the necessary data, then splits up the
3124          * string and re-inserts the newly formatted headlines.
3125          *
3126          * @param string $text
3127          * @param boolean $isMain
3128          * @private
3129          */
3130         function formatHeadings( $text, $isMain=true ) {
3131                 global $wgMaxTocLevel, $wgContLang;
3132
3133                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3134                 if( !$this->mTitle->userCanEdit() ) {
3135                         $showEditLink = 0;
3136                 } else {
3137                         $showEditLink = $this->mOptions->getEditSection();
3138                 }
3139
3140                 # Inhibit editsection links if requested in the page
3141                 $esw =& MagicWord::get( 'noeditsection' );
3142                 if( $esw->matchAndRemove( $text ) ) {
3143                         $showEditLink = 0;
3144                 }
3145
3146                 # Get all headlines for numbering them and adding funky stuff like [edit]
3147                 # links - this is for later, but we need the number of headlines right now
3148                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
3149
3150                 # if there are fewer than 4 headlines in the article, do not show TOC
3151                 # unless it's been explicitly enabled.
3152                 $enoughToc = $this->mShowToc &&
3153                         (($numMatches >= 4) || $this->mForceTocPosition);
3154
3155                 # Allow user to stipulate that a page should have a "new section"
3156                 # link added via __NEWSECTIONLINK__
3157                 $mw =& MagicWord::get( 'newsectionlink' );
3158                 if( $mw->matchAndRemove( $text ) )
3159                         $this->mOutput->setNewSection( true );
3160
3161                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3162                 # override above conditions and always show TOC above first header
3163                 $mw =& MagicWord::get( 'forcetoc' );
3164                 if ($mw->matchAndRemove( $text ) ) {
3165                         $this->mShowToc = true;
3166                         $enoughToc = true;
3167                 }
3168
3169                 # Never ever show TOC if no headers
3170                 if( $numMatches < 1 ) {
3171                         $enoughToc = false;
3172                 }
3173
3174                 # We need this to perform operations on the HTML
3175                 $sk =& $this->mOptions->getSkin();
3176
3177                 # headline counter
3178                 $headlineCount = 0;
3179                 $sectionCount = 0; # headlineCount excluding template sections
3180
3181                 # Ugh .. the TOC should have neat indentation levels which can be
3182                 # passed to the skin functions. These are determined here
3183                 $toc = '';
3184                 $full = '';
3185                 $head = array();
3186                 $sublevelCount = array();
3187                 $levelCount = array();
3188                 $toclevel = 0;
3189                 $level = 0;
3190                 $prevlevel = 0;
3191                 $toclevel = 0;
3192                 $prevtoclevel = 0;
3193
3194                 foreach( $matches[3] as $headline ) {
3195                         $istemplate = 0;
3196                         $templatetitle = '';
3197                         $templatesection = 0;
3198                         $numbering = '';
3199
3200                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3201                                 $istemplate = 1;
3202                                 $templatetitle = base64_decode($mat[1]);
3203                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3204                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3205                         }
3206
3207                         if( $toclevel ) {
3208                                 $prevlevel = $level;
3209                                 $prevtoclevel = $toclevel;
3210                         }
3211                         $level = $matches[1][$headlineCount];
3212
3213                         if( $doNumberHeadings || $enoughToc ) {
3214
3215                                 if ( $level > $prevlevel ) {
3216                                         # Increase TOC level
3217                                         $toclevel++;
3218                                         $sublevelCount[$toclevel] = 0;
3219                                         if( $toclevel<$wgMaxTocLevel ) {
3220                                                 $toc .= $sk->tocIndent();
3221                                         }
3222                                 }
3223                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3224                                         # Decrease TOC level, find level to jump to
3225
3226                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3227                                                 # Can only go down to level 1
3228                                                 $toclevel = 1;
3229                                         } else {
3230                                                 for ($i = $toclevel; $i > 0; $i--) {
3231                                                         if ( $levelCount[$i] == $level ) {
3232                                                                 # Found last matching level
3233                                                                 $toclevel = $i;
3234                                                                 break;
3235                                                         }
3236                                                         elseif ( $levelCount[$i] < $level ) {
3237                                                                 # Found first matching level below current level
3238                                                                 $toclevel = $i + 1;
3239                                                                 break;
3240                                                         }
3241                                                 }
3242                                         }
3243                                         if( $toclevel<$wgMaxTocLevel ) {
3244                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3245                                         }
3246                                 }
3247                                 else {
3248                                         # No change in level, end TOC line
3249                                         if( $toclevel<$wgMaxTocLevel ) {
3250                                                 $toc .= $sk->tocLineEnd();
3251                                         }
3252                                 }
3253
3254                                 $levelCount[$toclevel] = $level;
3255
3256                                 # count number of headlines for each level
3257                                 @$sublevelCount[$toclevel]++;
3258                                 $dot = 0;
3259                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3260                                         if( !empty( $sublevelCount[$i] ) ) {
3261                                                 if( $dot ) {
3262                                                         $numbering .= '.';
3263                                                 }
3264                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3265                                                 $dot = 1;
3266                                         }
3267                                 }
3268                         }
3269
3270                         # The canonized header is a version of the header text safe to use for links
3271                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3272                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
3273                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
3274
3275                         # Remove link placeholders by the link text.
3276                         #     <!--LINK number-->
3277                         # turns into
3278                         #     link text with suffix
3279                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3280                                                             "\$this->mLinkHolders['texts'][\$1]",
3281                                                             $canonized_headline );
3282                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3283                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3284                                                             $canonized_headline );
3285
3286                         # strip out HTML
3287                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3288                         $tocline = trim( $canonized_headline );
3289                         # Save headline for section edit hint before it's escaped
3290                         $headline_hint = trim( $canonized_headline );
3291                         $canonized_headline = Sanitizer::escapeId( $tocline );
3292                         $refers[$headlineCount] = $canonized_headline;
3293
3294                         # count how many in assoc. array so we can track dupes in anchors
3295                         @$refers[$canonized_headline]++;
3296                         $refcount[$headlineCount]=$refers[$canonized_headline];
3297
3298                         # Don't number the heading if it is the only one (looks silly)
3299                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3300                                 # the two are different if the line contains a link
3301                                 $headline=$numbering . ' ' . $headline;
3302                         }
3303
3304                         # Create the anchor for linking from the TOC to the section
3305                         $anchor = $canonized_headline;
3306                         if($refcount[$headlineCount] > 1 ) {
3307                                 $anchor .= '_' . $refcount[$headlineCount];
3308                         }
3309                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3310                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3311                         }
3312                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3313                                 if ( empty( $head[$headlineCount] ) ) {
3314                                         $head[$headlineCount] = '';
3315                                 }
3316                                 if( $istemplate )
3317                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
3318                                 else
3319                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3320                         }
3321
3322                         # give headline the correct <h#> tag
3323                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
3324
3325                         $headlineCount++;
3326                         if( !$istemplate )
3327                                 $sectionCount++;
3328                 }
3329
3330                 if( $enoughToc ) {
3331                         if( $toclevel<$wgMaxTocLevel ) {
3332                                 $toc .= $sk->tocUnindent( $toclevel - 1 );
3333                         }
3334                         $toc = $sk->tocList( $toc );
3335                 }
3336
3337                 # split up and insert constructed headlines
3338
3339                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3340                 $i = 0;
3341
3342                 foreach( $blocks as $block ) {
3343                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3344                                 # This is the [edit] link that appears for the top block of text when
3345                                 # section editing is enabled
3346
3347                                 # Disabled because it broke block formatting
3348                                 # For example, a bullet point in the top line
3349                                 # $full .= $sk->editSectionLink(0);
3350                         }
3351                         $full .= $block;
3352                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3353                                 # Top anchor now in skin
3354                                 $full = $full.$toc;
3355                         }
3356
3357                         if( !empty( $head[$i] ) ) {
3358                                 $full .= $head[$i];
3359                         }
3360                         $i++;
3361                 }
3362                 if( $this->mForceTocPosition ) {
3363                         return str_replace( '<!--MWTOC-->', $toc, $full );
3364                 } else {
3365                         return $full;
3366                 }
3367         }
3368
3369         /**
3370          * Return an HTML link for the "ISBN 123456" text
3371          * @private
3372          */
3373         function magicISBN( $text ) {
3374                 $fname = 'Parser::magicISBN';
3375                 wfProfileIn( $fname );
3376
3377                 $a = split( 'ISBN ', ' '.$text );
3378                 if ( count ( $a ) < 2 ) {
3379                         wfProfileOut( $fname );
3380                         return $text;
3381                 }
3382                 $text = substr( array_shift( $a ), 1);
3383                 $valid = '0123456789-Xx';
3384
3385                 foreach ( $a as $x ) {
3386                         # hack: don't replace inside thumbnail title/alt
3387                         # attributes
3388                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3389                                 $text .= "ISBN $x";
3390                                 continue;
3391                         }
3392
3393                         $isbn = $blank = '' ;
3394                         while ( $x !== '' && ' ' == $x{0} ) {
3395                                 $blank .= ' ';
3396                                 $x = substr( $x, 1 );
3397                         }
3398                         if ( $x == '' ) { # blank isbn
3399                                 $text .= "ISBN $blank";
3400                                 continue;
3401                         }
3402                         while ( strstr( $valid, $x{0} ) != false ) {
3403                                 $isbn .= $x{0};
3404                                 $x = substr( $x, 1 );
3405                         }
3406                         $num = str_replace( '-', '', $isbn );
3407                         $num = str_replace( ' ', '', $num );
3408                         $num = str_replace( 'x', 'X', $num );
3409
3410                         if ( '' == $num ) {
3411                                 $text .= "ISBN $blank$x";
3412                         } else {
3413                                 $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
3414                                 $text .= '<a href="' .
3415                                         $titleObj->escapeLocalUrl( 'isbn='.$num ) .
3416                                         "\" class=\"internal\">ISBN $isbn</a>";
3417                                 $text .= $x;
3418                         }
3419                 }
3420                 wfProfileOut( $fname );
3421                 return $text;
3422         }
3423
3424         /**
3425          * Return an HTML link for the "RFC 1234" text
3426          *
3427          * @private
3428          * @param string $text     Text to be processed
3429          * @param string $keyword  Magic keyword to use (default RFC)
3430          * @param string $urlmsg   Interface message to use (default rfcurl)
3431          * @return string
3432          */
3433         function magicRFC( $text, $keyword='RFC ', $urlmsg='rfcurl'  ) {
3434
3435                 $valid = '0123456789';
3436                 $internal = false;
3437
3438                 $a = split( $keyword, ' '.$text );
3439                 if ( count ( $a ) < 2 ) {
3440                         return $text;
3441                 }
3442                 $text = substr( array_shift( $a ), 1);
3443
3444                 /* Check if keyword is preceed by [[.
3445                  * This test is made here cause of the array_shift above
3446                  * that prevent the test to be done in the foreach.
3447                  */
3448                 if ( substr( $text, -2 ) == '[[' ) {
3449                         $internal = true;
3450                 }
3451
3452                 foreach ( $a as $x ) {
3453                         /* token might be empty if we have RFC RFC 1234 */
3454                         if ( $x=='' ) {
3455                                 $text.=$keyword;
3456                                 continue;
3457                                 }
3458
3459                         # hack: don't replace inside thumbnail title/alt
3460                         # attributes
3461                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3462                                 $text .= $keyword . $x;
3463                                 continue;
3464                         }
3465
3466                         $id = $blank = '' ;
3467
3468                         /** remove and save whitespaces in $blank */
3469                         while ( $x{0} == ' ' ) {
3470                                 $blank .= ' ';
3471                                 $x = substr( $x, 1 );
3472                         }
3473
3474                         /** remove and save the rfc number in $id */
3475                         while ( strstr( $valid, $x{0} ) != false ) {
3476                                 $id .= $x{0};
3477                                 $x = substr( $x, 1 );
3478                         }
3479
3480                         if ( $id == '' ) {
3481                                 /* call back stripped spaces*/
3482                                 $text .= $keyword.$blank.$x;
3483                         } elseif( $internal ) {
3484                                 /* normal link */
3485                                 $text .= $keyword.$id.$x;
3486                         } else {
3487                                 /* build the external link*/
3488                                 $url = wfMsg( $urlmsg, $id);
3489                                 $sk =& $this->mOptions->getSkin();
3490                                 $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
3491                                 $text .= "<a href=\"{$url}\"{$la}>{$keyword}{$id}</a>{$x}";
3492                         }
3493
3494                         /* Check if the next RFC keyword is preceed by [[ */
3495                         $internal = ( substr($x,-2) == '[[' );
3496                 }
3497                 return $text;
3498         }
3499
3500         /**
3501          * Transform wiki markup when saving a page by doing \r\n -> \n
3502          * conversion, substitting signatures, {{subst:}} templates, etc.
3503          *
3504          * @param string $text the text to transform
3505          * @param Title &$title the Title object for the current article
3506          * @param User &$user the User object describing the current user
3507          * @param ParserOptions $options parsing options
3508          * @param bool $clearState whether to clear the parser state first
3509          * @return string the altered wiki markup
3510          * @public
3511          */
3512         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
3513                 $this->mOptions = $options;
3514                 $this->mTitle =& $title;
3515                 $this->mOutputType = OT_WIKI;
3516
3517                 if ( $clearState ) {
3518                         $this->clearState();
3519                 }
3520
3521                 $stripState = false;
3522                 $pairs = array(
3523                         "\r\n" => "\n",
3524                 );
3525                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3526                 $text = $this->strip( $text, $stripState, true, array( 'gallery' ) );
3527                 $text = $this->pstPass2( $text, $stripState, $user );
3528                 $text = $this->unstrip( $text, $stripState );
3529                 $text = $this->unstripNoWiki( $text, $stripState );
3530                 return $text;
3531         }
3532
3533         /**
3534          * Pre-save transform helper function
3535          * @private
3536          */
3537         function pstPass2( $text, &$stripState, &$user ) {
3538                 global $wgContLang, $wgLocaltimezone;
3539
3540                 /* Note: This is the timestamp saved as hardcoded wikitext to
3541                  * the database, we use $wgContLang here in order to give
3542                  * everyone the same signature and use the default one rather
3543                  * than the one selected in each user's preferences.
3544                  */
3545                 if ( isset( $wgLocaltimezone ) ) {
3546                         $oldtz = getenv( 'TZ' );
3547                         putenv( 'TZ='.$wgLocaltimezone );
3548                 }
3549                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3550                   ' (' . date( 'T' ) . ')';
3551                 if ( isset( $wgLocaltimezone ) ) {
3552                         putenv( 'TZ='.$oldtz );
3553                 }
3554
3555                 # Variable replacement
3556                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3557                 $text = $this->replaceVariables( $text );
3558
3559                 # Strip out <nowiki> etc. added via replaceVariables
3560                 $text = $this->strip( $text, $stripState, false, array( 'gallery' ) );
3561
3562                 # Signatures
3563                 $sigText = $this->getUserSig( $user );
3564                 $text = strtr( $text, array(
3565                         '~~~~~' => $d,
3566                         '~~~~' => "$sigText $d",
3567                         '~~~' => $sigText
3568                 ) );
3569
3570                 # Context links: [[|name]] and [[name (context)|]]
3571                 #
3572                 global $wgLegalTitleChars;
3573                 $tc = "[$wgLegalTitleChars]";
3574                 $np = str_replace( array( '(', ')' ), array( '', '' ), $tc ); # No parens
3575
3576                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3577                 $conpat = "/^({$np}+) \\(({$tc}+)\\)$/";
3578
3579                 $p1 = "/\[\[({$np}+) \\(({$np}+)\\)\\|]]/";             # [[page (context)|]]
3580                 $p2 = "/\[\[\\|({$tc}+)]]/";                                    # [[|page]]
3581                 $p3 = "/\[\[(:*$namespacechar+):({$np}+)\\|]]/";                # [[namespace:page|]] and [[:namespace:page|]]
3582                 $p4 = "/\[\[(:*$namespacechar+):({$np}+) \\(({$np}+)\\)\\|]]/"; # [[ns:page (cont)|]] and [[:ns:page (cont)|]]
3583                 $context = '';
3584                 $t = $this->mTitle->getText();
3585                 if ( preg_match( $conpat, $t, $m ) ) {
3586                         $context = $m[2];
3587                 }
3588                 $text = preg_replace( $p4, '[[\\1:\\2 (\\3)|\\2]]', $text );
3589                 $text = preg_replace( $p1, '[[\\1 (\\2)|\\1]]', $text );
3590                 $text = preg_replace( $p3, '[[\\1:\\2|\\2]]', $text );
3591
3592                 if ( '' == $context ) {
3593                         $text = preg_replace( $p2, '[[\\1]]', $text );
3594                 } else {
3595                         $text = preg_replace( $p2, "[[\\1 ({$context})|\\1]]", $text );
3596                 }
3597
3598                 # Trim trailing whitespace
3599                 # __END__ tag allows for trailing
3600                 # whitespace to be deliberately included
3601                 $text = rtrim( $text );
3602                 $mw =& MagicWord::get( 'end' );
3603                 $mw->matchAndRemove( $text );
3604
3605                 return $text;
3606         }
3607
3608         /**
3609          * Fetch the user's signature text, if any, and normalize to
3610          * validated, ready-to-insert wikitext.
3611          *
3612          * @param User $user
3613          * @return string
3614          * @private
3615          */
3616         function getUserSig( &$user ) {
3617                 $username = $user->getName();
3618                 $nickname = $user->getOption( 'nickname' );
3619                 $nickname = $nickname === '' ? $username : $nickname;
3620
3621                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3622                         # Sig. might contain markup; validate this
3623                         if( $this->validateSig( $nickname ) !== false ) {
3624                                 # Validated; clean up (if needed) and return it
3625                                 return $this->cleanSig( $nickname, true );
3626                         } else {
3627                                 # Failed to validate; fall back to the default
3628                                 $nickname = $username;
3629                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3630                         }
3631                 }
3632
3633                 // Make sure nickname doesnt get a sig in a sig
3634                 $nickname = $this->cleanSigInSig( $nickname );
3635
3636                 # If we're still here, make it a link to the user page
3637                 $userpage = $user->getUserPage();
3638                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3639         }
3640
3641         /**
3642          * Check that the user's signature contains no bad XML
3643          *
3644          * @param string $text
3645          * @return mixed An expanded string, or false if invalid.
3646          */
3647         function validateSig( $text ) {
3648                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3649         }
3650
3651         /**
3652          * Clean up signature text
3653          *
3654          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3655          * 2) Substitute all transclusions
3656          *
3657          * @param string $text
3658          * @param $parsing Whether we're cleaning (preferences save) or parsing
3659          * @return string Signature text
3660          */
3661         function cleanSig( $text, $parsing = false ) {
3662                 global $wgTitle;
3663                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
3664
3665                 $substWord = MagicWord::get( 'subst' );
3666                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3667                 $substText = '{{' . $substWord->getSynonym( 0 );
3668
3669                 $text = preg_replace( $substRegex, $substText, $text );
3670                 $text = $this->cleanSigInSig( $text );
3671                 $text = $this->replaceVariables( $text );
3672
3673                 $this->clearState();
3674                 return $text;
3675         }
3676
3677         /**
3678          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3679          * @param string $text
3680          * @return string Signature text with /~{3,5}/ removed
3681          */
3682         function cleanSigInSig( $text ) {
3683                 $text = preg_replace( '/~{3,5}/', '', $text );
3684                 return $text;
3685         }
3686
3687         /**
3688          * Set up some variables which are usually set up in parse()
3689          * so that an external function can call some class members with confidence
3690          * @public
3691          */
3692         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3693                 $this->mTitle =& $title;
3694                 $this->mOptions = $options;
3695                 $this->mOutputType = $outputType;
3696                 if ( $clearState ) {
3697                         $this->clearState();
3698                 }
3699         }
3700
3701         /**
3702          * Transform a MediaWiki message by replacing magic variables.
3703          *
3704          * @param string $text the text to transform
3705          * @param ParserOptions $options  options
3706          * @return string the text with variables substituted
3707          * @public
3708          */
3709         function transformMsg( $text, $options ) {
3710                 global $wgTitle;
3711                 static $executing = false;
3712
3713                 $fname = "Parser::transformMsg";
3714
3715                 # Guard against infinite recursion
3716                 if ( $executing ) {
3717                         return $text;
3718                 }
3719                 $executing = true;
3720
3721                 wfProfileIn($fname);
3722
3723                 $this->mTitle = $wgTitle;
3724                 $this->mOptions = $options;
3725                 $this->mOutputType = OT_MSG;
3726                 $this->clearState();
3727                 $text = $this->replaceVariables( $text );
3728
3729                 $executing = false;
3730                 wfProfileOut($fname);
3731                 return $text;
3732         }
3733
3734         /**
3735          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3736          * The callback should have the following form:
3737          *    function myParserHook( $text, $params, &$parser ) { ... }
3738          *
3739          * Transform and return $text. Use $parser for any required context, e.g. use
3740          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3741          *
3742          * @public
3743          *
3744          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3745          * @param mixed $callback The callback function (and object) to use for the tag
3746          *
3747          * @return The old value of the mTagHooks array associated with the hook
3748          */
3749         function setHook( $tag, $callback ) {
3750                 $tag = strtolower( $tag );
3751                 $oldVal = @$this->mTagHooks[$tag];
3752                 $this->mTagHooks[$tag] = $callback;
3753
3754                 return $oldVal;
3755         }
3756
3757         /**
3758          * Create a function, e.g. {{sum:1|2|3}}
3759          * The callback function should have the form:
3760          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3761          *
3762          * The callback may either return the text result of the function, or an array with the text
3763          * in element 0, and a number of flags in the other elements. The names of the flags are
3764          * specified in the keys. Valid flags are:
3765          *   found                     The text returned is valid, stop processing the template. This
3766          *                             is on by default.
3767          *   nowiki                    Wiki markup in the return value should be escaped
3768          *   noparse                   Unsafe HTML tags should not be stripped, etc.
3769          *   noargs                    Don't replace triple-brace arguments in the return value
3770          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3771          *
3772          * @public
3773          *
3774          * @param mixed $id The magic word ID
3775          * @param mixed $callback The callback function (and object) to use
3776          * @param integer $flags a combination of the following flags:
3777          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
3778          *
3779          * @return The old callback function for this name, if any
3780          */
3781         function setFunctionHook( $id, $callback, $flags = 0 ) {
3782                 $oldVal = @$this->mFunctionHooks[$id];
3783                 $this->mFunctionHooks[$id] = $callback;
3784
3785                 # Add to function cache
3786                 $mw = MagicWord::get( $id );
3787                 if ( !$mw ) {
3788                         throw new MWException( 'The calling convention to Parser::setFunctionHook() has changed, ' .
3789                                 'it is now required to pass a MagicWord ID as the first parameter.' );
3790                 }
3791
3792                 $synonyms = $mw->getSynonyms();
3793                 $sensitive = intval( $mw->isCaseSensitive() );
3794
3795                 foreach ( $synonyms as $syn ) {
3796                         # Case
3797                         if ( !$sensitive ) {
3798                                 $syn = strtolower( $syn );
3799                         }
3800                         # Add leading hash
3801                         if ( !( $flags & SFH_NO_HASH ) ) {
3802                                 $syn = '#' . $syn;
3803                         }
3804                         # Remove trailing colon
3805                         if ( substr( $syn, -1, 1 ) == ':' ) {
3806                                 $syn = substr( $syn, 0, -1 );
3807                         }
3808                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
3809                 }
3810                 return $oldVal;
3811         }
3812
3813         /**
3814          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3815          * Placeholders created in Skin::makeLinkObj()
3816          * Returns an array of links found, indexed by PDBK:
3817          *  0 - broken
3818          *  1 - normal link
3819          *  2 - stub
3820          * $options is a bit field, RLH_FOR_UPDATE to select for update
3821          */
3822         function replaceLinkHolders( &$text, $options = 0 ) {
3823                 global $wgUser;
3824                 global $wgOutputReplace;
3825                 global $wgContLang, $wgLanguageCode;
3826
3827                 $fname = 'Parser::replaceLinkHolders';
3828                 wfProfileIn( $fname );
3829
3830                 $pdbks = array();
3831                 $colours = array();
3832                 $sk =& $this->mOptions->getSkin();
3833                 $linkCache =& LinkCache::singleton();
3834
3835                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3836                         wfProfileIn( $fname.'-check' );
3837                         $dbr =& wfGetDB( DB_SLAVE );
3838                         $page = $dbr->tableName( 'page' );
3839                         $threshold = $wgUser->getOption('stubthreshold');
3840
3841                         # Sort by namespace
3842                         asort( $this->mLinkHolders['namespaces'] );
3843
3844                         # Generate query
3845                         $query = false;
3846                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3847                                 # Make title object
3848                                 $title = $this->mLinkHolders['titles'][$key];
3849
3850                                 # Skip invalid entries.
3851                                 # Result will be ugly, but prevents crash.
3852                                 if ( is_null( $title ) ) {
3853                                         continue;
3854                                 }
3855                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3856
3857                                 # Check if it's a static known link, e.g. interwiki
3858                                 if ( $title->isAlwaysKnown() ) {
3859                                         $colours[$pdbk] = 1;
3860                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3861                                         $colours[$pdbk] = 1;
3862                                         $this->mOutput->addLink( $title, $id );
3863                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3864                                         $colours[$pdbk] = 0;
3865                                 } else {
3866                                         # Not in the link cache, add it to the query
3867                                         if ( !isset( $current ) ) {
3868                                                 $current = $ns;
3869                                                 $query =  "SELECT page_id, page_namespace, page_title";
3870                                                 if ( $threshold > 0 ) {
3871                                                         $query .= ', page_len, page_is_redirect';
3872                                                 }
3873                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
3874                                         } elseif ( $current != $ns ) {
3875                                                 $current = $ns;
3876                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
3877                                         } else {
3878                                                 $query .= ', ';
3879                                         }
3880
3881                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3882                                 }
3883                         }
3884                         if ( $query ) {
3885                                 $query .= '))';
3886                                 if ( $options & RLH_FOR_UPDATE ) {
3887                                         $query .= ' FOR UPDATE';
3888                                 }
3889
3890                                 $res = $dbr->query( $query, $fname );
3891
3892                                 # Fetch data and form into an associative array
3893                                 # non-existent = broken
3894                                 # 1 = known
3895                                 # 2 = stub
3896                                 while ( $s = $dbr->fetchObject($res) ) {
3897                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
3898                                         $pdbk = $title->getPrefixedDBkey();
3899                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
3900                                         $this->mOutput->addLink( $title, $s->page_id );
3901
3902                                         if ( $threshold >  0 ) {
3903                                                 $size = $s->page_len;
3904                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
3905                                                         $colours[$pdbk] = 1;
3906                                                 } else {
3907                                                         $colours[$pdbk] = 2;
3908                                                 }
3909                                         } else {
3910                                                 $colours[$pdbk] = 1;
3911                                         }
3912                                 }
3913                         }
3914                         wfProfileOut( $fname.'-check' );
3915
3916                         # Do a second query for links in different language variants (if needed)
3917                         if(sizeof($wgContLang->getVariants())>1){
3918                                 $linkBatch = new LinkBatch();
3919
3920                                 // Add variants of links to link batch
3921                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3922                                         $title = $this->mLinkHolders['titles'][$key];
3923                                         if ( is_null( $title ) )
3924                                                 continue;
3925
3926                                         $pdbk = $title->getPrefixedDBkey();
3927
3928                                         // generate all variants of the link title text
3929                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($title->getText());
3930
3931                                         // if link was not found (in first query), add all variants to query
3932                                         if ( !isset($colours[$pdbk]) ){
3933                                                 foreach($allTextVariants as $textVariant){
3934                                                         $linkBatch->addObj( Title::makeTitleSafe( $ns, $textVariant ) );
3935                                                 }
3936                                         }
3937                                         // if link was found add only variant with fixed title
3938                                         else if($colours[$pdbk] == 1){
3939                                                 $fixedCode = $wgLanguageCode.'-fixed';
3940                                                 if( isset($allTextVariants[$fixedCode]) ){
3941                                                         $linkBatch->addObj( Title::makeTitleSafe( $ns, $allTextVariants[$fixedCode]  ) );
3942                                                 }
3943                                         }
3944                                 }
3945
3946                                 # fetch link variants into cache
3947                                 $linkBatch->execute();
3948
3949                                 # check if links are found in some of the variants
3950                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3951                                         $title = $this->mLinkHolders['titles'][$key];
3952                                         if ( is_null( $title ) )
3953                                                 continue;
3954
3955                                         $pdbk = $title->getPrefixedDBkey();
3956                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($title->getText());
3957
3958                                         // If link has already been found, check only fixed variant
3959                                         if(isset($colours[$pdbk]) && $colours[$pdbk] == 1){
3960                                                 $fixedCode = $wgLanguageCode.'-fixed';
3961
3962                                                 if( isset($allTextVariants[$fixedCode]) ){
3963                                                         $allTextVariants = array($fixedCode => $allTextVariants[$fixedCode]);
3964                                                 }
3965                                         }
3966
3967                                         // process the link variants
3968                                         if ( !isset($colours[$pdbk]) || $colours[$pdbk] == 1 ){
3969                                                 foreach($allTextVariants as $variantCode => $textVariant){
3970                                                         $variantTitle=Title::makeTitleSafe( $ns, $textVariant );
3971                                                         if(is_null($variantTitle)) continue;
3972
3973                                                         $varpdbk = $variantTitle->getPrefixedDBkey();
3974
3975                                                         if($linkCache->getGoodLinkID( $varpdbk ) != 0){
3976
3977                                                                 // found link in some of the variants, replace the link holder data
3978                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
3979                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
3980
3981                                                                 if($wgContLang->getPreferredVariant() == $wgLanguageCode)
3982                                                                         $this->mLinkHolders['texts'][$key] = $this->mLinkHolders['texts'][$key];
3983                                                                 else
3984                                                                         $this->mLinkHolders['texts'][$key] = $variantTitle->getText();
3985
3986                                                                 $pdbks[$key] = $varpdbk;
3987                                                                 $colours[$varpdbk] = 1;
3988
3989                                                                 break;
3990                                                         }
3991                                                 }
3992                                         }
3993                                 }
3994                         }
3995
3996                         # Construct search and replace arrays
3997                         wfProfileIn( $fname.'-construct' );
3998                         $wgOutputReplace = array();
3999                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4000                                 $pdbk = $pdbks[$key];
4001                                 $searchkey = "<!--LINK $key-->";
4002                                 $title = $this->mLinkHolders['titles'][$key];
4003                                 if ( empty( $colours[$pdbk] ) ) {
4004                                         $linkCache->addBadLinkObj( $title );
4005                                         $colours[$pdbk] = 0;
4006                                         $this->mOutput->addLink( $title, 0 );
4007                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
4008                                                                         $this->mLinkHolders['texts'][$key],
4009                                                                         $this->mLinkHolders['queries'][$key] );
4010                                 } elseif ( $colours[$pdbk] == 1 ) {
4011                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
4012                                                                         $this->mLinkHolders['texts'][$key],
4013                                                                         $this->mLinkHolders['queries'][$key] );
4014                                 } elseif ( $colours[$pdbk] == 2 ) {
4015                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
4016                                                                         $this->mLinkHolders['texts'][$key],
4017                                                                         $this->mLinkHolders['queries'][$key] );
4018                                 }
4019                         }
4020                         wfProfileOut( $fname.'-construct' );
4021
4022                         # Do the thing
4023                         wfProfileIn( $fname.'-replace' );
4024
4025                         $text = preg_replace_callback(
4026                                 '/(<!--LINK .*?-->)/',
4027                                 "wfOutputReplaceMatches",
4028                                 $text);
4029
4030                         wfProfileOut( $fname.'-replace' );
4031                 }
4032
4033                 # Now process interwiki link holders
4034                 # This is quite a bit simpler than internal links
4035                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4036                         wfProfileIn( $fname.'-interwiki' );
4037                         # Make interwiki link HTML
4038                         $wgOutputReplace = array();
4039                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4040                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4041                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
4042                         }
4043
4044                         $text = preg_replace_callback(
4045                                 '/<!--IWLINK (.*?)-->/',
4046                                 "wfOutputReplaceMatches",
4047                                 $text );
4048                         wfProfileOut( $fname.'-interwiki' );
4049                 }
4050
4051                 wfProfileOut( $fname );
4052                 return $colours;
4053         }
4054
4055         /**
4056          * Replace <!--LINK--> link placeholders with plain text of links
4057          * (not HTML-formatted).
4058          * @param string $text
4059          * @return string
4060          */
4061         function replaceLinkHoldersText( $text ) {
4062                 $fname = 'Parser::replaceLinkHoldersText';
4063                 wfProfileIn( $fname );
4064
4065                 $text = preg_replace_callback(
4066                         '/<!--(LINK|IWLINK) (.*?)-->/',
4067                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4068                         $text );
4069
4070                 wfProfileOut( $fname );
4071                 return $text;
4072         }
4073
4074         /**
4075          * @param array $matches
4076          * @return string
4077          * @private
4078          */
4079         function replaceLinkHoldersTextCallback( $matches ) {
4080                 $type = $matches[1];
4081                 $key  = $matches[2];
4082                 if( $type == 'LINK' ) {
4083                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4084                                 return $this->mLinkHolders['texts'][$key];
4085                         }
4086                 } elseif( $type == 'IWLINK' ) {
4087                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4088                                 return $this->mInterwikiLinkHolders['texts'][$key];
4089                         }
4090                 }
4091                 return $matches[0];
4092         }
4093
4094         /**
4095          * Tag hook handler for 'pre'.
4096          */
4097         function renderPreTag( $text, $attribs, $parser ) {
4098                 // Backwards-compatibility hack
4099                 $content = preg_replace( '!<nowiki>(.*?)</nowiki>!is', '\\1', $text );
4100
4101                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4102                 return wfOpenElement( 'pre', $attribs ) .
4103                         wfEscapeHTMLTagsOnly( $content ) .
4104                         '</pre>';
4105         }
4106
4107         /**
4108          * Renders an image gallery from a text with one line per image.
4109          * text labels may be given by using |-style alternative text. E.g.
4110          *   Image:one.jpg|The number "1"
4111          *   Image:tree.jpg|A tree
4112          * given as text will return the HTML of a gallery with two images,
4113          * labeled 'The number "1"' and
4114          * 'A tree'.
4115          */
4116         function renderImageGallery( $text, $params ) {
4117                 $ig = new ImageGallery();
4118                 $ig->setShowBytes( false );
4119                 $ig->setShowFilename( false );
4120                 $ig->setParsing();
4121                 $ig->useSkin( $this->mOptions->getSkin() );
4122
4123                 if( isset( $params['caption'] ) )
4124                         $ig->setCaption( $params['caption'] );
4125
4126                 $lines = explode( "\n", $text );
4127                 foreach ( $lines as $line ) {
4128                         # match lines like these:
4129                         # Image:someimage.jpg|This is some image
4130                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4131                         # Skip empty lines
4132                         if ( count( $matches ) == 0 ) {
4133                                 continue;
4134                         }
4135                         $tp = Title::newFromText( $matches[1] );
4136                         $nt =& $tp;
4137                         if( is_null( $nt ) ) {
4138                                 # Bogus title. Ignore these so we don't bomb out later.
4139                                 continue;
4140                         }
4141                         if ( isset( $matches[3] ) ) {
4142                                 $label = $matches[3];
4143                         } else {
4144                                 $label = '';
4145                         }
4146
4147                         $pout = $this->parse( $label,
4148                                 $this->mTitle,
4149                                 $this->mOptions,
4150                                 false, // Strip whitespace...?
4151                                 false  // Don't clear state!
4152                         );
4153                         $html = $pout->getText();
4154
4155                         $ig->add( new Image( $nt ), $html );
4156
4157                         # Only add real images (bug #5586)
4158                         if ( $nt->getNamespace() == NS_IMAGE ) {
4159                                 $this->mOutput->addImage( $nt->getDBkey() );
4160                         }
4161                 }
4162                 return $ig->toHTML();
4163         }
4164
4165         /**
4166          * Parse image options text and use it to make an image
4167          */
4168         function makeImage( &$nt, $options ) {
4169                 global $wgUseImageResize;
4170
4171                 $align = '';
4172
4173                 # Check if the options text is of the form "options|alt text"
4174                 # Options are:
4175                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4176                 #  * left               no resizing, just left align. label is used for alt= only
4177                 #  * right              same, but right aligned
4178                 #  * none               same, but not aligned
4179                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4180                 #  * center             center the image
4181                 #  * framed             Keep original image size, no magnify-button.
4182
4183                 $part = explode( '|', $options);
4184
4185                 $mwThumb  =& MagicWord::get( 'img_thumbnail' );
4186                 $mwManualThumb =& MagicWord::get( 'img_manualthumb' );
4187                 $mwLeft   =& MagicWord::get( 'img_left' );
4188                 $mwRight  =& MagicWord::get( 'img_right' );
4189                 $mwNone   =& MagicWord::get( 'img_none' );
4190                 $mwWidth  =& MagicWord::get( 'img_width' );
4191                 $mwCenter =& MagicWord::get( 'img_center' );
4192                 $mwFramed =& MagicWord::get( 'img_framed' );
4193                 $caption = '';
4194
4195                 $width = $height = $framed = $thumb = false;
4196                 $manual_thumb = '' ;
4197
4198                 foreach( $part as $key => $val ) {
4199                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
4200                                 $thumb=true;
4201                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
4202                                 # use manually specified thumbnail
4203                                 $thumb=true;
4204                                 $manual_thumb = $match;
4205                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
4206                                 # remember to set an alignment, don't render immediately
4207                                 $align = 'right';
4208                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
4209                                 # remember to set an alignment, don't render immediately
4210                                 $align = 'left';
4211                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
4212                                 # remember to set an alignment, don't render immediately
4213                                 $align = 'center';
4214                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
4215                                 # remember to set an alignment, don't render immediately
4216                                 $align = 'none';
4217                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
4218                                 wfDebug( "img_width match: $match\n" );
4219                                 # $match is the image width in pixels
4220                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
4221                                         $width = intval( $m[1] );
4222                                         $height = intval( $m[2] );
4223                                 } else {
4224                                         $width = intval($match);
4225                                 }
4226                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
4227                                 $framed=true;
4228                         } else {
4229                                 $caption = $val;
4230                         }
4231                 }
4232                 # Strip bad stuff out of the alt text
4233                 $alt = $this->replaceLinkHoldersText( $caption );
4234
4235                 # make sure there are no placeholders in thumbnail attributes
4236                 # that are later expanded to html- so expand them now and
4237                 # remove the tags
4238                 $alt = $this->unstrip($alt, $this->mStripState);
4239                 $alt = Sanitizer::stripAllTags( $alt );
4240
4241                 # Linker does the rest
4242                 $sk =& $this->mOptions->getSkin();
4243                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
4244         }
4245
4246         /**
4247          * Set a flag in the output object indicating that the content is dynamic and
4248          * shouldn't be cached.
4249          */
4250         function disableCache() {
4251                 wfDebug( "Parser output marked as uncacheable.\n" );
4252                 $this->mOutput->mCacheTime = -1;
4253         }
4254
4255         /**#@+
4256          * Callback from the Sanitizer for expanding items found in HTML attribute
4257          * values, so they can be safely tested and escaped.
4258          * @param string $text
4259          * @param array $args
4260          * @return string
4261          * @private
4262          */
4263         function attributeStripCallback( &$text, $args ) {
4264                 $text = $this->replaceVariables( $text, $args );
4265                 $text = $this->unstripForHTML( $text );
4266                 return $text;
4267         }
4268
4269         function unstripForHTML( $text ) {
4270                 $text = $this->unstrip( $text, $this->mStripState );
4271                 $text = $this->unstripNoWiki( $text, $this->mStripState );
4272                 return $text;
4273         }
4274         /**#@-*/
4275
4276         /**#@+
4277          * Accessor/mutator
4278          */
4279         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4280         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4281         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4282         /**#@-*/
4283
4284         /**#@+
4285          * Accessor
4286          */
4287         function getTags() { return array_keys( $this->mTagHooks ); }
4288         /**#@-*/
4289
4290
4291         /**
4292          * Break wikitext input into sections, and either pull or replace
4293          * some particular section's text.
4294          *
4295          * External callers should use the getSection and replaceSection methods.
4296          *
4297          * @param $text Page wikitext
4298          * @param $section Numbered section. 0 pulls the text before the first
4299          *                 heading; other numbers will pull the given section
4300          *                 along with its lower-level subsections.
4301          * @param $mode One of "get" or "replace"
4302          * @param $newtext Replacement text for section data.
4303          * @return string for "get", the extracted section text.
4304          *                for "replace", the whole page with the section replaced.
4305          */
4306         private function extractSections( $text, $section, $mode, $newtext='' ) {
4307                 # strip NOWIKI etc. to avoid confusion (true-parameter causes HTML
4308                 # comments to be stripped as well)
4309                 $striparray = array();
4310
4311                 $oldOutputType = $this->mOutputType;
4312                 $oldOptions = $this->mOptions;
4313                 $this->mOptions = new ParserOptions();
4314                 $this->mOutputType = OT_WIKI;
4315
4316                 $striptext = $this->strip( $text, $striparray, true );
4317
4318                 $this->mOutputType = $oldOutputType;
4319                 $this->mOptions = $oldOptions;
4320
4321                 # now that we can be sure that no pseudo-sections are in the source,
4322                 # split it up by section
4323                 $uniq = preg_quote( $this->uniqPrefix(), '/' );
4324                 $comment = "(?:$uniq-!--.*?QINU)";
4325                 $secs = preg_split(
4326                 /*
4327                         "/
4328                         ^(
4329                         (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4330                         (?:
4331                                 (=+) # Should this be limited to 6?
4332                                 .+?  # Section title...
4333                                 \\2  # Ending = count must match start
4334                         |
4335                                 ^
4336                                 <h([1-6])\b.*?>
4337                                 .*?
4338                                 <\/h\\3\s*>
4339                         )
4340                         (?:$comment|<\/?noinclude>|\s+)* # Trailing whitespace ok
4341                         )$
4342                         /mix",
4343                 */
4344                         "/
4345                         (
4346                                 ^
4347                                 (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4348                                 (=+) # Should this be limited to 6?
4349                                 .+?  # Section title...
4350                                 \\2  # Ending = count must match start
4351                                 (?:$comment|<\/?noinclude>|[ \\t]+)* # Trailing whitespace ok
4352                                 $
4353                         |
4354                                 <h([1-6])\b.*?>
4355                                 .*?
4356                                 <\/h\\3\s*>
4357                         )
4358                         /mix",
4359                         $striptext, -1,
4360                         PREG_SPLIT_DELIM_CAPTURE);
4361
4362                 if( $mode == "get" ) {
4363                         if( $section == 0 ) {
4364                                 // "Section 0" returns the content before any other section.
4365                                 $rv = $secs[0];
4366                         } else {
4367                                 $rv = "";
4368                         }
4369                 } elseif( $mode == "replace" ) {
4370                         if( $section == 0 ) {
4371                                 $rv = $newtext . "\n\n";
4372                                 $remainder = true;
4373                         } else {
4374                                 $rv = $secs[0];
4375                                 $remainder = false;
4376                         }
4377                 }
4378                 $count = 0;
4379                 $sectionLevel = 0;
4380                 for( $index = 1; $index < count( $secs ); ) {
4381                         $headerLine = $secs[$index++];
4382                         if( $secs[$index] ) {
4383                                 // A wiki header
4384                                 $headerLevel = strlen( $secs[$index++] );
4385                         } else {
4386                                 // An HTML header
4387                                 $index++;
4388                                 $headerLevel = intval( $secs[$index++] );
4389                         }
4390                         $content = $secs[$index++];
4391
4392                         $count++;
4393                         if( $mode == "get" ) {
4394                                 if( $count == $section ) {
4395                                         $rv = $headerLine . $content;
4396                                         $sectionLevel = $headerLevel;
4397                                 } elseif( $count > $section ) {
4398                                         if( $sectionLevel && $headerLevel > $sectionLevel ) {
4399                                                 $rv .= $headerLine . $content;
4400                                         } else {
4401                                                 // Broke out to a higher-level section
4402                                                 break;
4403                                         }
4404                                 }
4405                         } elseif( $mode == "replace" ) {
4406                                 if( $count < $section ) {
4407                                         $rv .= $headerLine . $content;
4408                                 } elseif( $count == $section ) {
4409                                         $rv .= $newtext . "\n\n";
4410                                         $sectionLevel = $headerLevel;
4411                                 } elseif( $count > $section ) {
4412                                         if( $headerLevel <= $sectionLevel ) {
4413                                                 // Passed the section's sub-parts.
4414                                                 $remainder = true;
4415                                         }
4416                                         if( $remainder ) {
4417                                                 $rv .= $headerLine . $content;
4418                                         }
4419                                 }
4420                         }
4421                 }
4422                 # reinsert stripped tags
4423                 $rv = $this->unstrip( $rv, $striparray );
4424                 $rv = $this->unstripNoWiki( $rv, $striparray );
4425                 $rv = trim( $rv );
4426                 return $rv;
4427         }
4428
4429         /**
4430          * This function returns the text of a section, specified by a number ($section).
4431          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4432          * the first section before any such heading (section 0).
4433          *
4434          * If a section contains subsections, these are also returned.
4435          *
4436          * @param $text String: text to look in
4437          * @param $section Integer: section number
4438          * @return string text of the requested section
4439          */
4440         function getSection( $text, $section ) {
4441                 return $this->extractSections( $text, $section, "get" );
4442         }
4443
4444         function replaceSection( $oldtext, $section, $text ) {
4445                 return $this->extractSections( $oldtext, $section, "replace", $text );
4446         }
4447
4448 }
4449
4450 /**
4451  * @todo document
4452  * @package MediaWiki
4453  */
4454 class ParserOutput
4455 {
4456         var $mText,             # The output text
4457                 $mLanguageLinks,    # List of the full text of language links, in the order they appear
4458                 $mCategories,       # Map of category names to sort keys
4459                 $mContainsOldMagic, # Boolean variable indicating if the input contained variables like {{CURRENTDAY}}
4460                 $mCacheTime,        # Time when this object was generated, or -1 for uncacheable. Used in ParserCache.
4461                 $mVersion,          # Compatibility check
4462                 $mTitleText,        # title text of the chosen language variant
4463                 $mLinks,            # 2-D map of NS/DBK to ID for the links in the document. ID=zero for broken.
4464                 $mTemplates,        # 2-D map of NS/DBK to ID for the template references. ID=zero for broken.
4465                 $mImages,           # DB keys of the images used, in the array key only
4466                 $mExternalLinks,    # External link URLs, in the key only
4467                 $mHTMLtitle,            # Display HTML title
4468                 $mSubtitle,                     # Additional subtitle
4469                 $mNewSection,           # Show a new section link?
4470                 $mNoGallery;            # No gallery on category page? (__NOGALLERY__)
4471
4472         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
4473                 $containsOldMagic = false, $titletext = '' )
4474         {
4475                 $this->mText = $text;
4476                 $this->mLanguageLinks = $languageLinks;
4477                 $this->mCategories = $categoryLinks;
4478                 $this->mContainsOldMagic = $containsOldMagic;
4479                 $this->mCacheTime = '';
4480                 $this->mVersion = MW_PARSER_VERSION;
4481                 $this->mTitleText = $titletext;
4482                 $this->mLinks = array();
4483                 $this->mTemplates = array();
4484                 $this->mImages = array();
4485                 $this->mExternalLinks = array();
4486                 $this->mHTMLtitle = "" ;
4487                 $this->mSubtitle = "" ;
4488                 $this->mNewSection = false;
4489                 $this->mNoGallery = false;
4490         }
4491
4492         function getText()                   { return $this->mText; }
4493         function &getLanguageLinks()          { return $this->mLanguageLinks; }
4494         function getCategoryLinks()          { return array_keys( $this->mCategories ); }
4495         function &getCategories()            { return $this->mCategories; }
4496         function getCacheTime()              { return $this->mCacheTime; }
4497         function getTitleText()              { return $this->mTitleText; }
4498         function &getLinks()                 { return $this->mLinks; }
4499         function &getTemplates()             { return $this->mTemplates; }
4500         function &getImages()                { return $this->mImages; }
4501         function &getExternalLinks()         { return $this->mExternalLinks; }
4502         function getNoGallery()              { return $this->mNoGallery; }
4503
4504         function containsOldMagic()          { return $this->mContainsOldMagic; }
4505         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
4506         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
4507         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategories, $cl ); }
4508         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
4509         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
4510         function setTitleText( $t )          { return wfSetVar ($this->mTitleText, $t); }
4511
4512         function addCategory( $c, $sort )    { $this->mCategories[$c] = $sort; }
4513         function addImage( $name )           { $this->mImages[$name] = 1; }
4514         function addLanguageLink( $t )       { $this->mLanguageLinks[] = $t; }
4515         function addExternalLink( $url )     { $this->mExternalLinks[$url] = 1; }
4516
4517         function setNewSection( $value ) {
4518                 $this->mNewSection = (bool)$value;
4519         }
4520         function getNewSection() {
4521                 return (bool)$this->mNewSection;
4522         }
4523
4524         function addLink( $title, $id ) {
4525                 $ns = $title->getNamespace();
4526                 $dbk = $title->getDBkey();
4527                 if ( !isset( $this->mLinks[$ns] ) ) {
4528                         $this->mLinks[$ns] = array();
4529                 }
4530                 $this->mLinks[$ns][$dbk] = $id;
4531         }
4532
4533         function addTemplate( $title, $id ) {
4534                 $ns = $title->getNamespace();
4535                 $dbk = $title->getDBkey();
4536                 if ( !isset( $this->mTemplates[$ns] ) ) {
4537                         $this->mTemplates[$ns] = array();
4538                 }
4539                 $this->mTemplates[$ns][$dbk] = $id;
4540         }
4541
4542         /**
4543          * Return true if this cached output object predates the global or
4544          * per-article cache invalidation timestamps, or if it comes from
4545          * an incompatible older version.
4546          *
4547          * @param string $touched the affected article's last touched timestamp
4548          * @return bool
4549          * @public
4550          */
4551         function expired( $touched ) {
4552                 global $wgCacheEpoch;
4553                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
4554                        $this->getCacheTime() < $touched ||
4555                        $this->getCacheTime() <= $wgCacheEpoch ||
4556                        !isset( $this->mVersion ) ||
4557                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
4558         }
4559 }
4560
4561 /**
4562  * Set options of the Parser
4563  * @todo document
4564  * @package MediaWiki
4565  */
4566 class ParserOptions
4567 {
4568         # All variables are private
4569         var $mUseTeX;                    # Use texvc to expand <math> tags
4570         var $mUseDynamicDates;           # Use DateFormatter to format dates
4571         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
4572         var $mAllowExternalImages;       # Allow external images inline
4573         var $mAllowExternalImagesFrom;   # If not, any exception?
4574         var $mSkin;                      # Reference to the preferred skin
4575         var $mDateFormat;                # Date format index
4576         var $mEditSection;               # Create "edit section" links
4577         var $mNumberHeadings;            # Automatically number headings
4578         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
4579         var $mTidy;                      # Ask for tidy cleanup
4580         var $mInterfaceMessage;          # Which lang to call for PLURAL and GRAMMAR
4581
4582         var $mUser;                      # Stored user object, just used to initialise the skin
4583
4584         function getUseTeX()                        { return $this->mUseTeX; }
4585         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
4586         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
4587         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
4588         function getAllowExternalImagesFrom()       { return $this->mAllowExternalImagesFrom; }
4589         function getDateFormat()                    { return $this->mDateFormat; }
4590         function getEditSection()                   { return $this->mEditSection; }
4591         function getNumberHeadings()                { return $this->mNumberHeadings; }
4592         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
4593         function getTidy()                          { return $this->mTidy; }
4594         function getInterfaceMessage()              { return $this->mInterfaceMessage; }
4595
4596         function &getSkin() {
4597                 if ( !isset( $this->mSkin ) ) {
4598                         $this->mSkin = $this->mUser->getSkin();
4599                 }
4600                 return $this->mSkin;
4601         }
4602
4603         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
4604         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
4605         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
4606         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
4607         function setAllowExternalImagesFrom( $x )   { return wfSetVar( $this->mAllowExternalImagesFrom, $x ); }
4608         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
4609         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
4610         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
4611         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
4612         function setTidy( $x )                      { return wfSetVar( $this->mTidy, $x); }
4613         function setSkin( &$x ) { $this->mSkin =& $x; }
4614         function setInterfaceMessage( $x )          { return wfSetVar( $this->mInterfaceMessage, $x); }
4615
4616         function ParserOptions( $user = null ) {
4617                 $this->initialiseFromUser( $user );
4618         }
4619
4620         /**
4621          * Get parser options
4622          * @static
4623          */
4624         static function newFromUser( $user ) {
4625                 return new ParserOptions( $user );
4626         }
4627
4628         /** Get user options */
4629         function initialiseFromUser( $userInput ) {
4630                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages;
4631                 global $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion;
4632                 $fname = 'ParserOptions::initialiseFromUser';
4633                 wfProfileIn( $fname );
4634                 if ( !$userInput ) {
4635                         global $wgUser;
4636                         if ( isset( $wgUser ) ) {
4637                                 $user = $wgUser;
4638                         } else {
4639                                 $user = new User;
4640                                 $user->setLoaded( true );
4641                         }
4642                 } else {
4643                         $user =& $userInput;
4644                 }
4645
4646                 $this->mUser = $user;
4647
4648                 $this->mUseTeX = $wgUseTeX;
4649                 $this->mUseDynamicDates = $wgUseDynamicDates;
4650                 $this->mInterwikiMagic = $wgInterwikiMagic;
4651                 $this->mAllowExternalImages = $wgAllowExternalImages;
4652                 $this->mAllowExternalImagesFrom = $wgAllowExternalImagesFrom;
4653                 $this->mSkin = null; # Deferred
4654                 $this->mDateFormat = $user->getOption( 'date' );
4655                 $this->mEditSection = true;
4656                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
4657                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
4658                 $this->mTidy = false;
4659                 $this->mInterfaceMessage = false;
4660                 wfProfileOut( $fname );
4661         }
4662 }
4663
4664 /**
4665  * Callback function used by Parser::replaceLinkHolders()
4666  * to substitute link placeholders.
4667  */
4668 function &wfOutputReplaceMatches( $matches ) {
4669         global $wgOutputReplace;
4670         return $wgOutputReplace[$matches[1]];
4671 }
4672
4673 /**
4674  * Return the total number of articles
4675  */
4676 function wfNumberOfArticles() {
4677         global $wgNumberOfArticles;
4678
4679         wfLoadSiteStats();
4680         return $wgNumberOfArticles;
4681 }
4682
4683 /**
4684  * Return the number of files
4685  */
4686 function wfNumberOfFiles() {
4687         $fname = 'wfNumberOfFiles';
4688
4689         wfProfileIn( $fname );
4690         $dbr =& wfGetDB( DB_SLAVE );
4691         $numImages = $dbr->selectField('site_stats', 'ss_images', array(), $fname );
4692         wfProfileOut( $fname );
4693
4694         return $numImages;
4695 }
4696
4697 /**
4698  * Return the number of user accounts
4699  * @return integer
4700  */
4701 function wfNumberOfUsers() {
4702         wfProfileIn( 'wfNumberOfUsers' );
4703         $dbr =& wfGetDB( DB_SLAVE );
4704         $count = $dbr->selectField( 'site_stats', 'ss_users', array(), 'wfNumberOfUsers' );
4705         wfProfileOut( 'wfNumberOfUsers' );
4706         return (int)$count;
4707 }
4708
4709 /**
4710  * Return the total number of pages
4711  * @return integer
4712  */
4713 function wfNumberOfPages() {
4714         wfProfileIn( 'wfNumberOfPages' );
4715         $dbr =& wfGetDB( DB_SLAVE );
4716         $count = $dbr->selectField( 'site_stats', 'ss_total_pages', array(), 'wfNumberOfPages' );
4717         wfProfileOut( 'wfNumberOfPages' );
4718         return (int)$count;
4719 }
4720
4721 /**
4722  * Return the total number of admins
4723  *
4724  * @return integer
4725  */
4726 function wfNumberOfAdmins() {
4727         static $admins = -1;
4728         wfProfileIn( 'wfNumberOfAdmins' );
4729         if( $admins == -1 ) {
4730                 $dbr =& wfGetDB( DB_SLAVE );
4731                 $admins = $dbr->selectField( 'user_groups', 'COUNT(*)', array( 'ug_group' => 'sysop' ), 'wfNumberOfAdmins' );
4732         }
4733         wfProfileOut( 'wfNumberOfAdmins' );
4734         return (int)$admins;
4735 }
4736
4737 /**
4738  * Count the number of pages in a particular namespace
4739  *
4740  * @param $ns Namespace
4741  * @return integer
4742  */
4743 function wfPagesInNs( $ns ) {
4744         static $pageCount = array();
4745         wfProfileIn( 'wfPagesInNs' );
4746         if( !isset( $pageCount[$ns] ) ) {
4747                 $dbr =& wfGetDB( DB_SLAVE );
4748                 $pageCount[$ns] = $dbr->selectField( 'page', 'COUNT(*)', array( 'page_namespace' => $ns ), 'wfPagesInNs' );
4749         }
4750         wfProfileOut( 'wfPagesInNs' );
4751         return (int)$pageCount[$ns];
4752 }
4753
4754 /**
4755  * Get various statistics from the database
4756  * @private
4757  */
4758 function wfLoadSiteStats() {
4759         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
4760         $fname = 'wfLoadSiteStats';
4761
4762         if ( -1 != $wgNumberOfArticles ) return;
4763         $dbr =& wfGetDB( DB_SLAVE );
4764         $s = $dbr->selectRow( 'site_stats',
4765                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
4766                 array( 'ss_row_id' => 1 ), $fname
4767         );
4768
4769         if ( $s === false ) {
4770                 return;
4771         } else {
4772                 $wgTotalViews = $s->ss_total_views;
4773                 $wgTotalEdits = $s->ss_total_edits;
4774                 $wgNumberOfArticles = $s->ss_good_articles;
4775         }
4776 }
4777
4778 /**
4779  * Escape html tags
4780  * Basically replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
4781  *
4782  * @param $in String: text that might contain HTML tags.
4783  * @return string Escaped string
4784  */
4785 function wfEscapeHTMLTagsOnly( $in ) {
4786         return str_replace(
4787                 array( '"', '>', '<' ),
4788                 array( '&quot;', '&gt;', '&lt;' ),
4789                 $in );
4790 }
4791
4792 ?>