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( MAG_NS, array( 'CoreParserFunctions', 'ns' ), SFH_NO_HASH );
 152                 $this->setFunctionHook( MAG_URLENCODE, array( 'CoreParserFunctions', 'urlencode' ), SFH_NO_HASH );
 153                 $this->setFunctionHook( MAG_LCFIRST, array( 'CoreParserFunctions', 'lcfirst' ), SFH_NO_HASH );
 154                 $this->setFunctionHook( MAG_UCFIRST, array( 'CoreParserFunctions', 'ucfirst' ), SFH_NO_HASH );
 155                 $this->setFunctionHook( MAG_LC, array( 'CoreParserFunctions', 'lc' ), SFH_NO_HASH );
 156                 $this->setFunctionHook( MAG_UC, array( 'CoreParserFunctions', 'uc' ), SFH_NO_HASH );
 157                 $this->setFunctionHook( MAG_LOCALURL, array( 'CoreParserFunctions', 'localurl' ), SFH_NO_HASH );
 158                 $this->setFunctionHook( MAG_LOCALURLE, array( 'CoreParserFunctions', 'localurle' ), SFH_NO_HASH );
 159                 $this->setFunctionHook( MAG_FULLURL, array( 'CoreParserFunctions', 'fullurl' ), SFH_NO_HASH );
 160                 $this->setFunctionHook( MAG_FULLURLE, array( 'CoreParserFunctions', 'fullurle' ), SFH_NO_HASH );
 161                 $this->setFunctionHook( MAG_FORMATNUM, array( 'CoreParserFunctions', 'formatnum' ), SFH_NO_HASH );
 162                 $this->setFunctionHook( MAG_GRAMMAR, array( 'CoreParserFunctions', 'grammar' ), SFH_NO_HASH );
 163                 $this->setFunctionHook( MAG_PLURAL, array( 'CoreParserFunctions', 'plural' ), SFH_NO_HASH );
 164                 $this->setFunctionHook( MAG_NUMBEROFPAGES, array( 'CoreParserFunctions', 'numberofpages' ), SFH_NO_HASH );
 165                 $this->setFunctionHook( MAG_NUMBEROFUSERS, array( 'CoreParserFunctions', 'numberofusers' ), SFH_NO_HASH );
 166                 $this->setFunctionHook( MAG_NUMBEROFARTICLES, array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 167                 $this->setFunctionHook( MAG_NUMBEROFFILES, array( 'CoreParserFunctions', 'numberoffiles' ), SFH_NO_HASH );
 168                 $this->setFunctionHook( MAG_NUMBEROFADMINS, array( 'CoreParserFunctions', 'numberofadmins' ), SFH_NO_HASH );
 169                 $this->setFunctionHook( MAG_LANGUAGE, array( 'CoreParserFunctions', 'language' ), SFH_NO_HASH );
 170
 171                 if ( $wgAllowDisplayTitle ) {
 172                         $this->setFunctionHook( MAG_DISPLAYTITLE, array( 'CoreParserFunctions', 'displaytitle' ), SFH_NO_HASH );
 173                 }
 174                 if ( $wgAllowSlowParserFunctions ) {
 175                         $this->setFunctionHook( MAG_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                         # Normalize any HTML entities in input. They will be
1202                         # re-escaped by makeExternalLink().
1203                         $url = Sanitizer::decodeCharReferences( $url );
1204
1205                         # Escape any control characters introduced by the above step
1206                         $url = preg_replace( '/[\][<>"\\x00-\\x20\\x7F]/e', "urlencode('\\0')", $url );
1207
1208                         # Process the trail (i.e. everything after this link up until start of the next link),
1209                         # replacing any non-bracketed links
1210                         $trail = $this->replaceFreeExternalLinks( $trail );
1211
1212                         # Use the encoded URL
1213                         # This means that users can paste URLs directly into the text
1214                         # Funny characters like &ouml; aren't valid in URLs anyway
1215                         # This was changed in August 2004
1216                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1217
1218                         # Register link in the output object.
1219                         # Replace unnecessary URL escape codes with the referenced character
1220                         # This prevents spammers from hiding links from the filters
1221                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1222                         $this->mOutput->addExternalLink( $pasteurized );
1223                 }
1224
1225                 wfProfileOut( $fname );
1226                 return $s;
1227         }
1228
1229         /**
1230          * Replace anything that looks like a URL with a link
1231          * @private
1232          */
1233         function replaceFreeExternalLinks( $text ) {
1234                 global $wgContLang;
1235                 $fname = 'Parser::replaceFreeExternalLinks';
1236                 wfProfileIn( $fname );
1237
1238                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1239                 $s = array_shift( $bits );
1240                 $i = 0;
1241
1242                 $sk =& $this->mOptions->getSkin();
1243
1244                 while ( $i < count( $bits ) ){
1245                         $protocol = $bits[$i++];
1246                         $remainder = $bits[$i++];
1247
1248                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1249                                 # Found some characters after the protocol that look promising
1250                                 $url = $protocol . $m[1];
1251                                 $trail = $m[2];
1252
1253                                 # special case: handle urls as url args:
1254                                 # http://www.example.com/foo?=http://www.example.com/bar
1255                                 if(strlen($trail) == 0 &&
1256                                         isset($bits[$i]) &&
1257                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1258                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1259                                 {
1260                                         # add protocol, arg
1261                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1262                                         $i += 2;
1263                                         $trail = $m[2];
1264                                 }
1265
1266                                 # The characters '<' and '>' (which were escaped by
1267                                 # removeHTMLtags()) should not be included in
1268                                 # URLs, per RFC 2396.
1269                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1270                                         $trail = substr($url, $m2[0][1]) . $trail;
1271                                         $url = substr($url, 0, $m2[0][1]);
1272                                 }
1273
1274                                 # Move trailing punctuation to $trail
1275                                 $sep = ',;\.:!?';
1276                                 # If there is no left bracket, then consider right brackets fair game too
1277                                 if ( strpos( $url, '(' ) === false ) {
1278                                         $sep .= ')';
1279                                 }
1280
1281                                 $numSepChars = strspn( strrev( $url ), $sep );
1282                                 if ( $numSepChars ) {
1283                                         $trail = substr( $url, -$numSepChars ) . $trail;
1284                                         $url = substr( $url, 0, -$numSepChars );
1285                                 }
1286
1287                                 # Normalize any HTML entities in input. They will be
1288                                 # re-escaped by makeExternalLink() or maybeMakeExternalImage()
1289                                 $url = Sanitizer::decodeCharReferences( $url );
1290
1291                                 # Escape any control characters introduced by the above step
1292                                 $url = preg_replace( '/[\][<>"\\x00-\\x20\\x7F]/e', "urlencode('\\0')", $url );
1293
1294                                 # Is this an external image?
1295                                 $text = $this->maybeMakeExternalImage( $url );
1296                                 if ( $text === false ) {
1297                                         # Not an image, make a link
1298                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1299                                         # Register it in the output object...
1300                                         # Replace unnecessary URL escape codes with their equivalent characters
1301                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1302                                         $this->mOutput->addExternalLink( $pasteurized );
1303                                 }
1304                                 $s .= $text . $trail;
1305                         } else {
1306                                 $s .= $protocol . $remainder;
1307                         }
1308                 }
1309                 wfProfileOut( $fname );
1310                 return $s;
1311         }
1312
1313         /**
1314          * Replace unusual URL escape codes with their equivalent characters
1315          * @param string
1316          * @return string
1317          * @static
1318          * @fixme This can merge genuinely required bits in the path or query string,
1319          *        breaking legit URLs. A proper fix would treat the various parts of
1320          *        the URL differently; as a workaround, just use the output for
1321          *        statistical records, not for actual linking/output.
1322          */
1323         static function replaceUnusualEscapes( $url ) {
1324                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1325                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1326         }
1327
1328         /**
1329          * Callback function used in replaceUnusualEscapes().
1330          * Replaces unusual URL escape codes with their equivalent character
1331          * @static
1332          * @private
1333          */
1334         private static function replaceUnusualEscapesCallback( $matches ) {
1335                 $char = urldecode( $matches[0] );
1336                 $ord = ord( $char );
1337                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1338                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1339                         // No, shouldn't be escaped
1340                         return $char;
1341                 } else {
1342                         // Yes, leave it escaped
1343                         return $matches[0];
1344                 }
1345         }
1346
1347         /**
1348          * make an image if it's allowed, either through the global
1349          * option or through the exception
1350          * @private
1351          */
1352         function maybeMakeExternalImage( $url ) {
1353                 $sk =& $this->mOptions->getSkin();
1354                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1355                 $imagesexception = !empty($imagesfrom);
1356                 $text = false;
1357                 if ( $this->mOptions->getAllowExternalImages()
1358                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1359                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1360                                 # Image found
1361                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1362                         }
1363                 }
1364                 return $text;
1365         }
1366
1367         /**
1368          * Process [[ ]] wikilinks
1369          *
1370          * @private
1371          */
1372         function replaceInternalLinks( $s ) {
1373                 global $wgContLang;
1374                 static $fname = 'Parser::replaceInternalLinks' ;
1375
1376                 wfProfileIn( $fname );
1377
1378                 wfProfileIn( $fname.'-setup' );
1379                 static $tc = FALSE;
1380                 # the % is needed to support urlencoded titles as well
1381                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1382
1383                 $sk =& $this->mOptions->getSkin();
1384
1385                 #split the entire text string on occurences of [[
1386                 $a = explode( '[[', ' ' . $s );
1387                 #get the first element (all text up to first [[), and remove the space we added
1388                 $s = array_shift( $a );
1389                 $s = substr( $s, 1 );
1390
1391                 # Match a link having the form [[namespace:link|alternate]]trail
1392                 static $e1 = FALSE;
1393                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1394                 # Match cases where there is no "]]", which might still be images
1395                 static $e1_img = FALSE;
1396                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1397                 # Match the end of a line for a word that's not followed by whitespace,
1398                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1399                 $e2 = wfMsgForContent( 'linkprefix' );
1400
1401                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1402
1403                 if( is_null( $this->mTitle ) ) {
1404                         throw new MWException( 'nooo' );
1405                 }
1406                 $nottalk = !$this->mTitle->isTalkPage();
1407
1408                 if ( $useLinkPrefixExtension ) {
1409                         if ( preg_match( $e2, $s, $m ) ) {
1410                                 $first_prefix = $m[2];
1411                         } else {
1412                                 $first_prefix = false;
1413                         }
1414                 } else {
1415                         $prefix = '';
1416                 }
1417
1418                 $selflink = $this->mTitle->getPrefixedText();
1419                 wfProfileOut( $fname.'-setup' );
1420
1421                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1422                 $useSubpages = $this->areSubpagesAllowed();
1423
1424                 # Loop for each link
1425                 for ($k = 0; isset( $a[$k] ); $k++) {
1426                         $line = $a[$k];
1427                         if ( $useLinkPrefixExtension ) {
1428                                 wfProfileIn( $fname.'-prefixhandling' );
1429                                 if ( preg_match( $e2, $s, $m ) ) {
1430                                         $prefix = $m[2];
1431                                         $s = $m[1];
1432                                 } else {
1433                                         $prefix='';
1434                                 }
1435                                 # first link
1436                                 if($first_prefix) {
1437                                         $prefix = $first_prefix;
1438                                         $first_prefix = false;
1439                                 }
1440                                 wfProfileOut( $fname.'-prefixhandling' );
1441                         }
1442
1443                         $might_be_img = false;
1444
1445                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1446                                 $text = $m[2];
1447                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1448                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1449                                 # the real problem is with the $e1 regex
1450                                 # See bug 1300.
1451                                 #
1452                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1453                                 # and no image is in sight. See bug 2095.
1454                                 #
1455                                 if( $text !== '' &&
1456                                         preg_match( "/^\](.*)/s", $m[3], $n ) &&
1457                                         strpos($text, '[') !== false
1458                                 )
1459                                 {
1460                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1461                                         $m[3] = $n[1];
1462                                 }
1463                                 # fix up urlencoded title texts
1464                                 if(preg_match('/%/', $m[1] ))
1465                                         # Should anchors '#' also be rejected?
1466                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1467                                 $trail = $m[3];
1468                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1469                                 $might_be_img = true;
1470                                 $text = $m[2];
1471                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1472                                 $trail = "";
1473                         } else { # Invalid form; output directly
1474                                 $s .= $prefix . '[[' . $line ;
1475                                 continue;
1476                         }
1477
1478                         # Don't allow internal links to pages containing
1479                         # PROTO: where PROTO is a valid URL protocol; these
1480                         # should be external links.
1481                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1482                                 $s .= $prefix . '[[' . $line ;
1483                                 continue;
1484                         }
1485
1486                         # Make subpage if necessary
1487                         if( $useSubpages ) {
1488                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1489                         } else {
1490                                 $link = $m[1];
1491                         }
1492
1493                         $noforce = (substr($m[1], 0, 1) != ':');
1494                         if (!$noforce) {
1495                                 # Strip off leading ':'
1496                                 $link = substr($link, 1);
1497                         }
1498
1499                         $nt = Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1500                         if( !$nt ) {
1501                                 $s .= $prefix . '[[' . $line;
1502                                 continue;
1503                         }
1504
1505                         #check other language variants of the link
1506                         #if the article does not exist
1507                         if( $checkVariantLink
1508                             && $nt->getArticleID() == 0 ) {
1509                                 $wgContLang->findVariantLink($link, $nt);
1510                         }
1511
1512                         $ns = $nt->getNamespace();
1513                         $iw = $nt->getInterWiki();
1514
1515                         if ($might_be_img) { # if this is actually an invalid link
1516                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1517                                         $found = false;
1518                                         while (isset ($a[$k+1]) ) {
1519                                                 #look at the next 'line' to see if we can close it there
1520                                                 $spliced = array_splice( $a, $k + 1, 1 );
1521                                                 $next_line = array_shift( $spliced );
1522                                                 if( preg_match("/^(.*?]].*?)]](.*)$/sD", $next_line, $m) ) {
1523                                                 # the first ]] closes the inner link, the second the image
1524                                                         $found = true;
1525                                                         $text .= '[[' . $m[1];
1526                                                         $trail = $m[2];
1527                                                         break;
1528                                                 } elseif( preg_match("/^.*?]].*$/sD", $next_line, $m) ) {
1529                                                         #if there's exactly one ]] that's fine, we'll keep looking
1530                                                         $text .= '[[' . $m[0];
1531                                                 } else {
1532                                                         #if $next_line is invalid too, we need look no further
1533                                                         $text .= '[[' . $next_line;
1534                                                         break;
1535                                                 }
1536                                         }
1537                                         if ( !$found ) {
1538                                                 # we couldn't find the end of this imageLink, so output it raw
1539                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1540                                                 $text = $this->replaceInternalLinks($text);
1541                                                 $s .= $prefix . '[[' . $link . '|' . $text;
1542                                                 # note: no $trail, because without an end, there *is* no trail
1543                                                 continue;
1544                                         }
1545                                 } else { #it's not an image, so output it raw
1546                                         $s .= $prefix . '[[' . $link . '|' . $text;
1547                                         # note: no $trail, because without an end, there *is* no trail
1548                                         continue;
1549                                 }
1550                         }
1551
1552                         $wasblank = ( '' == $text );
1553                         if( $wasblank ) $text = $link;
1554
1555
1556                         # Link not escaped by : , create the various objects
1557                         if( $noforce ) {
1558
1559                                 # Interwikis
1560                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1561                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1562                                         $s = rtrim($s . "\n");
1563                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1564                                         continue;
1565                                 }
1566
1567                                 if ( $ns == NS_IMAGE ) {
1568                                         wfProfileIn( "$fname-image" );
1569                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1570                                                 # recursively parse links inside the image caption
1571                                                 # actually, this will parse them in any other parameters, too,
1572                                                 # but it might be hard to fix that, and it doesn't matter ATM
1573                                                 $text = $this->replaceExternalLinks($text);
1574                                                 $text = $this->replaceInternalLinks($text);
1575
1576                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1577                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1578                                                 $this->mOutput->addImage( $nt->getDBkey() );
1579
1580                                                 wfProfileOut( "$fname-image" );
1581                                                 continue;
1582                                         } else {
1583                                                 # We still need to record the image's presence on the page
1584                                                 $this->mOutput->addImage( $nt->getDBkey() );
1585                                         }
1586                                         wfProfileOut( "$fname-image" );
1587
1588                                 }
1589
1590                                 if ( $ns == NS_CATEGORY ) {
1591                                         wfProfileIn( "$fname-category" );
1592                                         $s = rtrim($s . "\n"); # bug 87
1593
1594                                         if ( $wasblank ) {
1595                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1596                                                         $sortkey = $this->mTitle->getText();
1597                                                 } else {
1598                                                         $sortkey = $this->mTitle->getPrefixedText();
1599                                                 }
1600                                         } else {
1601                                                 $sortkey = $text;
1602                                         }
1603                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1604                                         $sortkey = str_replace( "\n", '', $sortkey );
1605                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1606                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1607
1608                                         /**
1609                                          * Strip the whitespace Category links produce, see bug 87
1610                                          * @todo We might want to use trim($tmp, "\n") here.
1611                                          */
1612                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1613
1614                                         wfProfileOut( "$fname-category" );
1615                                         continue;
1616                                 }
1617                         }
1618
1619                         if( ( $nt->getPrefixedText() === $selflink ) &&
1620                             ( $nt->getFragment() === '' ) ) {
1621                                 # Self-links are handled specially; generally de-link and change to bold.
1622                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1623                                 continue;
1624                         }
1625
1626                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1627                         if( $ns == NS_MEDIA ) {
1628                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1629                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1630                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1631                                 $this->mOutput->addImage( $nt->getDBkey() );
1632                                 continue;
1633                         } elseif( $ns == NS_SPECIAL ) {
1634                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1635                                 continue;
1636                         } elseif( $ns == NS_IMAGE ) {
1637                                 $img = Image::newFromTitle( $nt );
1638                                 if( $img->exists() ) {
1639                                         // Force a blue link if the file exists; may be a remote
1640                                         // upload on the shared repository, and we want to see its
1641                                         // auto-generated page.
1642                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1643                                         continue;
1644                                 }
1645                         }
1646                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1647                 }
1648                 wfProfileOut( $fname );
1649                 return $s;
1650         }
1651
1652         /**
1653          * Make a link placeholder. The text returned can be later resolved to a real link with
1654          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1655          * parsing of interwiki links, and secondly to allow all extistence checks and
1656          * article length checks (for stub links) to be bundled into a single query.
1657          *
1658          */
1659         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1660                 if ( ! is_object($nt) ) {
1661                         # Fail gracefully
1662                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1663                 } else {
1664                         # Separate the link trail from the rest of the link
1665                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1666
1667                         if ( $nt->isExternal() ) {
1668                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1669                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1670                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1671                         } else {
1672                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1673                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1674                                 $this->mLinkHolders['queries'][] = $query;
1675                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1676                                 $this->mLinkHolders['titles'][] = $nt;
1677
1678                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1679                         }
1680                 }
1681                 return $retVal;
1682         }
1683
1684         /**
1685          * Render a forced-blue link inline; protect against double expansion of
1686          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1687          * Since this little disaster has to split off the trail text to avoid
1688          * breaking URLs in the following text without breaking trails on the
1689          * wiki links, it's been made into a horrible function.
1690          *
1691          * @param Title $nt
1692          * @param string $text
1693          * @param string $query
1694          * @param string $trail
1695          * @param string $prefix
1696          * @return string HTML-wikitext mix oh yuck
1697          */
1698         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1699                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1700                 $sk =& $this->mOptions->getSkin();
1701                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1702                 return $this->armorLinks( $link ) . $trail;
1703         }
1704
1705         /**
1706          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1707          * going to go through further parsing steps before inline URL expansion.
1708          *
1709          * In particular this is important when using action=render, which causes
1710          * full URLs to be included.
1711          *
1712          * Oh man I hate our multi-layer parser!
1713          *
1714          * @param string more-or-less HTML
1715          * @return string less-or-more HTML with NOPARSE bits
1716          */
1717         function armorLinks( $text ) {
1718                 return preg_replace( "/\b(" . wfUrlProtocols() . ')/',
1719                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1720         }
1721
1722         /**
1723          * Return true if subpage links should be expanded on this page.
1724          * @return bool
1725          */
1726         function areSubpagesAllowed() {
1727                 # Some namespaces don't allow subpages
1728                 global $wgNamespacesWithSubpages;
1729                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1730         }
1731
1732         /**
1733          * Handle link to subpage if necessary
1734          * @param string $target the source of the link
1735          * @param string &$text the link text, modified as necessary
1736          * @return string the full name of the link
1737          * @private
1738          */
1739         function maybeDoSubpageLink($target, &$text) {
1740                 # Valid link forms:
1741                 # Foobar -- normal
1742                 # :Foobar -- override special treatment of prefix (images, language links)
1743                 # /Foobar -- convert to CurrentPage/Foobar
1744                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1745                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1746                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1747
1748                 $fname = 'Parser::maybeDoSubpageLink';
1749                 wfProfileIn( $fname );
1750                 $ret = $target; # default return value is no change
1751
1752                 # Some namespaces don't allow subpages,
1753                 # so only perform processing if subpages are allowed
1754                 if( $this->areSubpagesAllowed() ) {
1755                         # Look at the first character
1756                         if( $target != '' && $target{0} == '/' ) {
1757                                 # / at end means we don't want the slash to be shown
1758                                 if( substr( $target, -1, 1 ) == '/' ) {
1759                                         $target = substr( $target, 1, -1 );
1760                                         $noslash = $target;
1761                                 } else {
1762                                         $noslash = substr( $target, 1 );
1763                                 }
1764
1765                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1766                                 if( '' === $text ) {
1767                                         $text = $target;
1768                                 } # this might be changed for ugliness reasons
1769                         } else {
1770                                 # check for .. subpage backlinks
1771                                 $dotdotcount = 0;
1772                                 $nodotdot = $target;
1773                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1774                                         ++$dotdotcount;
1775                                         $nodotdot = substr( $nodotdot, 3 );
1776                                 }
1777                                 if($dotdotcount > 0) {
1778                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1779                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1780                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1781                                                 # / at the end means don't show full path
1782                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1783                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1784                                                         if( '' === $text ) {
1785                                                                 $text = $nodotdot;
1786                                                         }
1787                                                 }
1788                                                 $nodotdot = trim( $nodotdot );
1789                                                 if( $nodotdot != '' ) {
1790                                                         $ret .= '/' . $nodotdot;
1791                                                 }
1792                                         }
1793                                 }
1794                         }
1795                 }
1796
1797                 wfProfileOut( $fname );
1798                 return $ret;
1799         }
1800
1801         /**#@+
1802          * Used by doBlockLevels()
1803          * @private
1804          */
1805         /* private */ function closeParagraph() {
1806                 $result = '';
1807                 if ( '' != $this->mLastSection ) {
1808                         $result = '</' . $this->mLastSection  . ">\n";
1809                 }
1810                 $this->mInPre = false;
1811                 $this->mLastSection = '';
1812                 return $result;
1813         }
1814         # getCommon() returns the length of the longest common substring
1815         # of both arguments, starting at the beginning of both.
1816         #
1817         /* private */ function getCommon( $st1, $st2 ) {
1818                 $fl = strlen( $st1 );
1819                 $shorter = strlen( $st2 );
1820                 if ( $fl < $shorter ) { $shorter = $fl; }
1821
1822                 for ( $i = 0; $i < $shorter; ++$i ) {
1823                         if ( $st1{$i} != $st2{$i} ) { break; }
1824                 }
1825                 return $i;
1826         }
1827         # These next three functions open, continue, and close the list
1828         # element appropriate to the prefix character passed into them.
1829         #
1830         /* private */ function openList( $char ) {
1831                 $result = $this->closeParagraph();
1832
1833                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1834                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1835                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1836                 else if ( ';' == $char ) {
1837                         $result .= '<dl><dt>';
1838                         $this->mDTopen = true;
1839                 }
1840                 else { $result = '<!-- ERR 1 -->'; }
1841
1842                 return $result;
1843         }
1844
1845         /* private */ function nextItem( $char ) {
1846                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1847                 else if ( ':' == $char || ';' == $char ) {
1848                         $close = '</dd>';
1849                         if ( $this->mDTopen ) { $close = '</dt>'; }
1850                         if ( ';' == $char ) {
1851                                 $this->mDTopen = true;
1852                                 return $close . '<dt>';
1853                         } else {
1854                                 $this->mDTopen = false;
1855                                 return $close . '<dd>';
1856                         }
1857                 }
1858                 return '<!-- ERR 2 -->';
1859         }
1860
1861         /* private */ function closeList( $char ) {
1862                 if ( '*' == $char ) { $text = '</li></ul>'; }
1863                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1864                 else if ( ':' == $char ) {
1865                         if ( $this->mDTopen ) {
1866                                 $this->mDTopen = false;
1867                                 $text = '</dt></dl>';
1868                         } else {
1869                                 $text = '</dd></dl>';
1870                         }
1871                 }
1872                 else {  return '<!-- ERR 3 -->'; }
1873                 return $text."\n";
1874         }
1875         /**#@-*/
1876
1877         /**
1878          * Make lists from lines starting with ':', '*', '#', etc.
1879          *
1880          * @private
1881          * @return string the lists rendered as HTML
1882          */
1883         function doBlockLevels( $text, $linestart ) {
1884                 $fname = 'Parser::doBlockLevels';
1885                 wfProfileIn( $fname );
1886
1887                 # Parsing through the text line by line.  The main thing
1888                 # happening here is handling of block-level elements p, pre,
1889                 # and making lists from lines starting with * # : etc.
1890                 #
1891                 $textLines = explode( "\n", $text );
1892
1893                 $lastPrefix = $output = '';
1894                 $this->mDTopen = $inBlockElem = false;
1895                 $prefixLength = 0;
1896                 $paragraphStack = false;
1897
1898                 if ( !$linestart ) {
1899                         $output .= array_shift( $textLines );
1900                 }
1901                 foreach ( $textLines as $oLine ) {
1902                         $lastPrefixLength = strlen( $lastPrefix );
1903                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1904                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1905                         if ( !$this->mInPre ) {
1906                                 # Multiple prefixes may abut each other for nested lists.
1907                                 $prefixLength = strspn( $oLine, '*#:;' );
1908                                 $pref = substr( $oLine, 0, $prefixLength );
1909
1910                                 # eh?
1911                                 $pref2 = str_replace( ';', ':', $pref );
1912                                 $t = substr( $oLine, $prefixLength );
1913                                 $this->mInPre = !empty($preOpenMatch);
1914                         } else {
1915                                 # Don't interpret any other prefixes in preformatted text
1916                                 $prefixLength = 0;
1917                                 $pref = $pref2 = '';
1918                                 $t = $oLine;
1919                         }
1920
1921                         # List generation
1922                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1923                                 # Same as the last item, so no need to deal with nesting or opening stuff
1924                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1925                                 $paragraphStack = false;
1926
1927                                 if ( substr( $pref, -1 ) == ';') {
1928                                         # The one nasty exception: definition lists work like this:
1929                                         # ; title : definition text
1930                                         # So we check for : in the remainder text to split up the
1931                                         # title and definition, without b0rking links.
1932                                         $term = $t2 = '';
1933                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1934                                                 $t = $t2;
1935                                                 $output .= $term . $this->nextItem( ':' );
1936                                         }
1937                                 }
1938                         } elseif( $prefixLength || $lastPrefixLength ) {
1939                                 # Either open or close a level...
1940                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
1941                                 $paragraphStack = false;
1942
1943                                 while( $commonPrefixLength < $lastPrefixLength ) {
1944                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
1945                                         --$lastPrefixLength;
1946                                 }
1947                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
1948                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
1949                                 }
1950                                 while ( $prefixLength > $commonPrefixLength ) {
1951                                         $char = substr( $pref, $commonPrefixLength, 1 );
1952                                         $output .= $this->openList( $char );
1953
1954                                         if ( ';' == $char ) {
1955                                                 # FIXME: This is dupe of code above
1956                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1957                                                         $t = $t2;
1958                                                         $output .= $term . $this->nextItem( ':' );
1959                                                 }
1960                                         }
1961                                         ++$commonPrefixLength;
1962                                 }
1963                                 $lastPrefix = $pref2;
1964                         }
1965                         if( 0 == $prefixLength ) {
1966                                 wfProfileIn( "$fname-paragraph" );
1967                                 # No prefix (not in list)--go to paragraph mode
1968                                 // XXX: use a stack for nestable elements like span, table and div
1969                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/center|<\\/tr|<\\/td|<\\/th)/iS', $t );
1970                                 $closematch = preg_match(
1971                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
1972                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<center)/iS', $t );
1973                                 if ( $openmatch or $closematch ) {
1974                                         $paragraphStack = false;
1975                                         # TODO bug 5718: paragraph closed
1976                                         $output .= $this->closeParagraph();
1977                                         if ( $preOpenMatch and !$preCloseMatch ) {
1978                                                 $this->mInPre = true;
1979                                         }
1980                                         if ( $closematch ) {
1981                                                 $inBlockElem = false;
1982                                         } else {
1983                                                 $inBlockElem = true;
1984                                         }
1985                                 } else if ( !$inBlockElem && !$this->mInPre ) {
1986                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
1987                                                 // pre
1988                                                 if ($this->mLastSection != 'pre') {
1989                                                         $paragraphStack = false;
1990                                                         $output .= $this->closeParagraph().'<pre>';
1991                                                         $this->mLastSection = 'pre';
1992                                                 }
1993                                                 $t = substr( $t, 1 );
1994                                         } else {
1995                                                 // paragraph
1996                                                 if ( '' == trim($t) ) {
1997                                                         if ( $paragraphStack ) {
1998                                                                 $output .= $paragraphStack.'<br />';
1999                                                                 $paragraphStack = false;
2000                                                                 $this->mLastSection = 'p';
2001                                                         } else {
2002                                                                 if ($this->mLastSection != 'p' ) {
2003                                                                         $output .= $this->closeParagraph();
2004                                                                         $this->mLastSection = '';
2005                                                                         $paragraphStack = '<p>';
2006                                                                 } else {
2007                                                                         $paragraphStack = '</p><p>';
2008                                                                 }
2009                                                         }
2010                                                 } else {
2011                                                         if ( $paragraphStack ) {
2012                                                                 $output .= $paragraphStack;
2013                                                                 $paragraphStack = false;
2014                                                                 $this->mLastSection = 'p';
2015                                                         } else if ($this->mLastSection != 'p') {
2016                                                                 $output .= $this->closeParagraph().'<p>';
2017                                                                 $this->mLastSection = 'p';
2018                                                         }
2019                                                 }
2020                                         }
2021                                 }
2022                                 wfProfileOut( "$fname-paragraph" );
2023                         }
2024                         // somewhere above we forget to get out of pre block (bug 785)
2025                         if($preCloseMatch && $this->mInPre) {
2026                                 $this->mInPre = false;
2027                         }
2028                         if ($paragraphStack === false) {
2029                                 $output .= $t."\n";
2030                         }
2031                 }
2032                 while ( $prefixLength ) {
2033                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2034                         --$prefixLength;
2035                 }
2036                 if ( '' != $this->mLastSection ) {
2037                         $output .= '</' . $this->mLastSection . '>';
2038                         $this->mLastSection = '';
2039                 }
2040
2041                 wfProfileOut( $fname );
2042                 return $output;
2043         }
2044
2045         /**
2046          * Split up a string on ':', ignoring any occurences inside tags
2047          * to prevent illegal overlapping.
2048          * @param string $str the string to split
2049          * @param string &$before set to everything before the ':'
2050          * @param string &$after set to everything after the ':'
2051          * return string the position of the ':', or false if none found
2052          */
2053         function findColonNoLinks($str, &$before, &$after) {
2054                 $fname = 'Parser::findColonNoLinks';
2055                 wfProfileIn( $fname );
2056
2057                 $pos = strpos( $str, ':' );
2058                 if( $pos === false ) {
2059                         // Nothing to find!
2060                         wfProfileOut( $fname );
2061                         return false;
2062                 }
2063
2064                 $lt = strpos( $str, '<' );
2065                 if( $lt === false || $lt > $pos ) {
2066                         // Easy; no tag nesting to worry about
2067                         $before = substr( $str, 0, $pos );
2068                         $after = substr( $str, $pos+1 );
2069                         wfProfileOut( $fname );
2070                         return $pos;
2071                 }
2072
2073                 // Ugly state machine to walk through avoiding tags.
2074                 $state = MW_COLON_STATE_TEXT;
2075                 $stack = 0;
2076                 $len = strlen( $str );
2077                 for( $i = 0; $i < $len; $i++ ) {
2078                         $c = $str{$i};
2079
2080                         switch( $state ) {
2081                         // (Using the number is a performance hack for common cases)
2082                         case 0: // MW_COLON_STATE_TEXT:
2083                                 switch( $c ) {
2084                                 case "<":
2085                                         // Could be either a <start> tag or an </end> tag
2086                                         $state = MW_COLON_STATE_TAGSTART;
2087                                         break;
2088                                 case ":":
2089                                         if( $stack == 0 ) {
2090                                                 // We found it!
2091                                                 $before = substr( $str, 0, $i );
2092                                                 $after = substr( $str, $i + 1 );
2093                                                 wfProfileOut( $fname );
2094                                                 return $i;
2095                                         }
2096                                         // Embedded in a tag; don't break it.
2097                                         break;
2098                                 default:
2099                                         // Skip ahead looking for something interesting
2100                                         $colon = strpos( $str, ':', $i );
2101                                         if( $colon === false ) {
2102                                                 // Nothing else interesting
2103                                                 wfProfileOut( $fname );
2104                                                 return false;
2105                                         }
2106                                         $lt = strpos( $str, '<', $i );
2107                                         if( $stack === 0 ) {
2108                                                 if( $lt === false || $colon < $lt ) {
2109                                                         // We found it!
2110                                                         $before = substr( $str, 0, $colon );
2111                                                         $after = substr( $str, $colon + 1 );
2112                                                         wfProfileOut( $fname );
2113                                                         return $i;
2114                                                 }
2115                                         }
2116                                         if( $lt === false ) {
2117                                                 // Nothing else interesting to find; abort!
2118                                                 // We're nested, but there's no close tags left. Abort!
2119                                                 break 2;
2120                                         }
2121                                         // Skip ahead to next tag start
2122                                         $i = $lt;
2123                                         $state = MW_COLON_STATE_TAGSTART;
2124                                 }
2125                                 break;
2126                         case 1: // MW_COLON_STATE_TAG:
2127                                 // In a <tag>
2128                                 switch( $c ) {
2129                                 case ">":
2130                                         $stack++;
2131                                         $state = MW_COLON_STATE_TEXT;
2132                                         break;
2133                                 case "/":
2134                                         // Slash may be followed by >?
2135                                         $state = MW_COLON_STATE_TAGSLASH;
2136                                         break;
2137                                 default:
2138                                         // ignore
2139                                 }
2140                                 break;
2141                         case 2: // MW_COLON_STATE_TAGSTART:
2142                                 switch( $c ) {
2143                                 case "/":
2144                                         $state = MW_COLON_STATE_CLOSETAG;
2145                                         break;
2146                                 case "!":
2147                                         $state = MW_COLON_STATE_COMMENT;
2148                                         break;
2149                                 case ">":
2150                                         // Illegal early close? This shouldn't happen D:
2151                                         $state = MW_COLON_STATE_TEXT;
2152                                         break;
2153                                 default:
2154                                         $state = MW_COLON_STATE_TAG;
2155                                 }
2156                                 break;
2157                         case 3: // MW_COLON_STATE_CLOSETAG:
2158                                 // In a </tag>
2159                                 if( $c == ">" ) {
2160                                         $stack--;
2161                                         if( $stack < 0 ) {
2162                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2163                                                 wfProfileOut( $fname );
2164                                                 return false;
2165                                         }
2166                                         $state = MW_COLON_STATE_TEXT;
2167                                 }
2168                                 break;
2169                         case MW_COLON_STATE_TAGSLASH:
2170                                 if( $c == ">" ) {
2171                                         // Yes, a self-closed tag <blah/>
2172                                         $state = MW_COLON_STATE_TEXT;
2173                                 } else {
2174                                         // Probably we're jumping the gun, and this is an attribute
2175                                         $state = MW_COLON_STATE_TAG;
2176                                 }
2177                                 break;
2178                         case 5: // MW_COLON_STATE_COMMENT:
2179                                 if( $c == "-" ) {
2180                                         $state = MW_COLON_STATE_COMMENTDASH;
2181                                 }
2182                                 break;
2183                         case MW_COLON_STATE_COMMENTDASH:
2184                                 if( $c == "-" ) {
2185                                         $state = MW_COLON_STATE_COMMENTDASHDASH;
2186                                 } else {
2187                                         $state = MW_COLON_STATE_COMMENT;
2188                                 }
2189                                 break;
2190                         case MW_COLON_STATE_COMMENTDASHDASH:
2191                                 if( $c == ">" ) {
2192                                         $state = MW_COLON_STATE_TEXT;
2193                                 } else {
2194                                         $state = MW_COLON_STATE_COMMENT;
2195                                 }
2196                                 break;
2197                         default:
2198                                 throw new MWException( "State machine error in $fname" );
2199                         }
2200                 }
2201                 if( $stack > 0 ) {
2202                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2203                         return false;
2204                 }
2205                 wfProfileOut( $fname );
2206                 return false;
2207         }
2208
2209         /**
2210          * Return value of a magic variable (like PAGENAME)
2211          *
2212          * @private
2213          */
2214         function getVariableValue( $index ) {
2215                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2216
2217                 /**
2218                  * Some of these require message or data lookups and can be
2219                  * expensive to check many times.
2220                  */
2221                 static $varCache = array();
2222                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) )
2223                         if ( isset( $varCache[$index] ) )
2224                                 return $varCache[$index];
2225
2226                 $ts = time();
2227                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2228
2229                 switch ( $index ) {
2230                         case MAG_CURRENTMONTH:
2231                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
2232                         case MAG_CURRENTMONTHNAME:
2233                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2234                         case MAG_CURRENTMONTHNAMEGEN:
2235                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2236                         case MAG_CURRENTMONTHABBREV:
2237                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2238                         case MAG_CURRENTDAY:
2239                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2240                         case MAG_CURRENTDAY2:
2241                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2242                         case MAG_PAGENAME:
2243                                 return $this->mTitle->getText();
2244                         case MAG_PAGENAMEE:
2245                                 return $this->mTitle->getPartialURL();
2246                         case MAG_FULLPAGENAME:
2247                                 return $this->mTitle->getPrefixedText();
2248                         case MAG_FULLPAGENAMEE:
2249                                 return $this->mTitle->getPrefixedURL();
2250                         case MAG_SUBPAGENAME:
2251                                 return $this->mTitle->getSubpageText();
2252                         case MAG_SUBPAGENAMEE:
2253                                 return $this->mTitle->getSubpageUrlForm();
2254                         case MAG_BASEPAGENAME:
2255                                 return $this->mTitle->getBaseText();
2256                         case MAG_BASEPAGENAMEE:
2257                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2258                         case MAG_TALKPAGENAME:
2259                                 if( $this->mTitle->canTalk() ) {
2260                                         $talkPage = $this->mTitle->getTalkPage();
2261                                         return $talkPage->getPrefixedText();
2262                                 } else {
2263                                         return '';
2264                                 }
2265                         case MAG_TALKPAGENAMEE:
2266                                 if( $this->mTitle->canTalk() ) {
2267                                         $talkPage = $this->mTitle->getTalkPage();
2268                                         return $talkPage->getPrefixedUrl();
2269                                 } else {
2270                                         return '';
2271                                 }
2272                         case MAG_SUBJECTPAGENAME:
2273                                 $subjPage = $this->mTitle->getSubjectPage();
2274                                 return $subjPage->getPrefixedText();
2275                         case MAG_SUBJECTPAGENAMEE:
2276                                 $subjPage = $this->mTitle->getSubjectPage();
2277                                 return $subjPage->getPrefixedUrl();
2278                         case MAG_REVISIONID:
2279                                 return $this->mRevisionId;
2280                         case MAG_NAMESPACE:
2281                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2282                         case MAG_NAMESPACEE:
2283                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2284                         case MAG_TALKSPACE:
2285                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2286                         case MAG_TALKSPACEE:
2287                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2288                         case MAG_SUBJECTSPACE:
2289                                 return $this->mTitle->getSubjectNsText();
2290                         case MAG_SUBJECTSPACEE:
2291                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2292                         case MAG_CURRENTDAYNAME:
2293                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2294                         case MAG_CURRENTYEAR:
2295                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2296                         case MAG_CURRENTTIME:
2297                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2298                         case MAG_CURRENTWEEK:
2299                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2300                                 // int to remove the padding
2301                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2302                         case MAG_CURRENTDOW:
2303                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2304                         case MAG_NUMBEROFARTICLES:
2305                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
2306                         case MAG_NUMBEROFFILES:
2307                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
2308                         case MAG_NUMBEROFUSERS:
2309                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfUsers() );
2310                         case MAG_NUMBEROFPAGES:
2311                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfPages() );
2312                         case MAG_NUMBEROFADMINS:
2313                                 return $varCache[$index]  = $wgContLang->formatNum( wfNumberOfAdmins() );
2314                         case MAG_CURRENTTIMESTAMP:
2315                                 return $varCache[$index] = wfTimestampNow();
2316                         case MAG_CURRENTVERSION:
2317                                 global $wgVersion;
2318                                 return $wgVersion;
2319                         case MAG_SITENAME:
2320                                 return $wgSitename;
2321                         case MAG_SERVER:
2322                                 return $wgServer;
2323                         case MAG_SERVERNAME:
2324                                 return $wgServerName;
2325                         case MAG_SCRIPTPATH:
2326                                 return $wgScriptPath;
2327                         case MAG_DIRECTIONMARK:
2328                                 return $wgContLang->getDirMark();
2329                         case MAG_CONTENTLANGUAGE:
2330                                 global $wgContLanguageCode;
2331                                 return $wgContLanguageCode;
2332                         default:
2333                                 $ret = null;
2334                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2335                                         return $ret;
2336                                 else
2337                                         return null;
2338                 }
2339         }
2340
2341         /**
2342          * initialise the magic variables (like CURRENTMONTHNAME)
2343          *
2344          * @private
2345          */
2346         function initialiseVariables() {
2347                 $fname = 'Parser::initialiseVariables';
2348                 wfProfileIn( $fname );
2349                 global $wgVariableIDs;
2350
2351                 $this->mVariables = array();
2352                 foreach ( $wgVariableIDs as $id ) {
2353                         $mw =& MagicWord::get( $id );
2354                         $mw->addToArray( $this->mVariables, $id );
2355                 }
2356                 wfProfileOut( $fname );
2357         }
2358
2359         /**
2360          * parse any parentheses in format ((title|part|part))
2361          * and call callbacks to get a replacement text for any found piece
2362          *
2363          * @param string $text The text to parse
2364          * @param array $callbacks rules in form:
2365          *     '{' => array(                            # opening parentheses
2366          *                                      'end' => '}',   # closing parentheses
2367          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2368          *                                                                4 => callback         # replacement callback to call if {{{{..}}}} is found
2369          *                                                                )
2370          *                                      )
2371          * @private
2372          */
2373         function replace_callback ($text, $callbacks) {
2374                 wfProfileIn( __METHOD__ . '-self' );
2375                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2376                 $lastOpeningBrace = -1;                 # last not closed parentheses
2377
2378                 for ($i = 0; $i < strlen($text); $i++) {
2379                         # check for any opening brace
2380                         $rule = null;
2381                         $nextPos = -1;
2382                         foreach ($callbacks as $key => $value) {
2383                                 $pos = strpos ($text, $key, $i);
2384                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)) {
2385                                         $rule = $value;
2386                                         $nextPos = $pos;
2387                                 }
2388                         }
2389
2390                         if ($lastOpeningBrace >= 0) {
2391                                 $pos = strpos ($text, $openingBraceStack[$lastOpeningBrace]['braceEnd'], $i);
2392
2393                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2394                                         $rule = null;
2395                                         $nextPos = $pos;
2396                                 }
2397
2398                                 $pos = strpos ($text, '|', $i);
2399
2400                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2401                                         $rule = null;
2402                                         $nextPos = $pos;
2403                                 }
2404                         }
2405
2406                         if ($nextPos == -1)
2407                                 break;
2408
2409                         $i = $nextPos;
2410
2411                         # found openning brace, lets add it to parentheses stack
2412                         if (null != $rule) {
2413                                 $piece = array('brace' => $text[$i],
2414                                                            'braceEnd' => $rule['end'],
2415                                                            'count' => 1,
2416                                                            'title' => '',
2417                                                            'parts' => null);
2418
2419                                 # count openning brace characters
2420                                 while ($i+1 < strlen($text) && $text[$i+1] == $piece['brace']) {
2421                                         $piece['count']++;
2422                                         $i++;
2423                                 }
2424
2425                                 $piece['startAt'] = $i+1;
2426                                 $piece['partStart'] = $i+1;
2427
2428                                 # we need to add to stack only if openning brace count is enough for any given rule
2429                                 foreach ($rule['cb'] as $cnt => $fn) {
2430                                         if ($piece['count'] >= $cnt) {
2431                                                 $lastOpeningBrace ++;
2432                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2433                                                 break;
2434                                         }
2435                                 }
2436
2437                                 continue;
2438                         }
2439                         else if ($lastOpeningBrace >= 0) {
2440                                 # first check if it is a closing brace
2441                                 if ($openingBraceStack[$lastOpeningBrace]['braceEnd'] == $text[$i]) {
2442                                         # lets check if it is enough characters for closing brace
2443                                         $count = 1;
2444                                         while ($i+$count < strlen($text) && $text[$i+$count] == $text[$i])
2445                                                 $count++;
2446
2447                                         # if there are more closing parentheses than opening ones, we parse less
2448                                         if ($openingBraceStack[$lastOpeningBrace]['count'] < $count)
2449                                                 $count = $openingBraceStack[$lastOpeningBrace]['count'];
2450
2451                                         # check for maximum matching characters (if there are 5 closing characters, we will probably need only 3 - depending on the rules)
2452                                         $matchingCount = 0;
2453                                         $matchingCallback = null;
2454                                         foreach ($callbacks[$openingBraceStack[$lastOpeningBrace]['brace']]['cb'] as $cnt => $fn) {
2455                                                 if ($count >= $cnt && $matchingCount < $cnt) {
2456                                                         $matchingCount = $cnt;
2457                                                         $matchingCallback = $fn;
2458                                                 }
2459                                         }
2460
2461                                         if ($matchingCount == 0) {
2462                                                 $i += $count - 1;
2463                                                 continue;
2464                                         }
2465
2466                                         # lets set a title or last part (if '|' was found)
2467                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts'])
2468                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2469                                         else
2470                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2471
2472                                         $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2473                                         $pieceEnd = $i + $matchingCount;
2474
2475                                         if( is_callable( $matchingCallback ) ) {
2476                                                 $cbArgs = array (
2477                                                                                  'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2478                                                                                  'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2479                                                                                  'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2480                                                                                  'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == "\n")),
2481                                                                                  );
2482                                                 # finally we can call a user callback and replace piece of text
2483                                                 wfProfileOut( __METHOD__ . '-self' );
2484                                                 $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2485                                                 wfProfileIn( __METHOD__ . '-self' );
2486                                                 $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2487                                                 $i = $pieceStart + strlen($replaceWith) - 1;
2488                                         }
2489                                         else {
2490                                                 # null value for callback means that parentheses should be parsed, but not replaced
2491                                                 $i += $matchingCount - 1;
2492                                         }
2493
2494                                         # reset last openning parentheses, but keep it in case there are unused characters
2495                                         $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2496                                                                    'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2497                                                                    'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2498                                                                    'title' => '',
2499                                                                    'parts' => null,
2500                                                                    'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2501                                         $openingBraceStack[$lastOpeningBrace--] = null;
2502
2503                                         if ($matchingCount < $piece['count']) {
2504                                                 $piece['count'] -= $matchingCount;
2505                                                 $piece['startAt'] -= $matchingCount;
2506                                                 $piece['partStart'] = $piece['startAt'];
2507                                                 # do we still qualify for any callback with remaining count?
2508                                                 foreach ($callbacks[$piece['brace']]['cb'] as $cnt => $fn) {
2509                                                         if ($piece['count'] >= $cnt) {
2510                                                                 $lastOpeningBrace ++;
2511                                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2512                                                                 break;
2513                                                         }
2514                                                 }
2515                                         }
2516                                         continue;
2517                                 }
2518
2519                                 # lets set a title if it is a first separator, or next part otherwise
2520                                 if ($text[$i] == '|') {
2521                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2522                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2523                                                 $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2524                                         }
2525                                         else
2526                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2527
2528                                         $openingBraceStack[$lastOpeningBrace]['partStart'] = $i + 1;
2529                                 }
2530                         }
2531                 }
2532
2533                 wfProfileOut( __METHOD__ . '-self' );
2534                 return $text;
2535         }
2536
2537         /**
2538          * Replace magic variables, templates, and template arguments
2539          * with the appropriate text. Templates are substituted recursively,
2540          * taking care to avoid infinite loops.
2541          *
2542          * Note that the substitution depends on value of $mOutputType:
2543          *  OT_WIKI: only {{subst:}} templates
2544          *  OT_MSG: only magic variables
2545          *  OT_HTML: all templates and magic variables
2546          *
2547          * @param string $tex The text to transform
2548          * @param array $args Key-value pairs representing template parameters to substitute
2549          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2550          * @private
2551          */
2552         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2553                 # Prevent too big inclusions
2554                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
2555                         return $text;
2556                 }
2557
2558                 $fname = 'Parser::replaceVariables';
2559                 wfProfileIn( $fname );
2560
2561                 # This function is called recursively. To keep track of arguments we need a stack:
2562                 array_push( $this->mArgStack, $args );
2563
2564                 $braceCallbacks = array();
2565                 if ( !$argsOnly ) {
2566                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2567                 }
2568                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
2569                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2570                 }
2571                 $callbacks = array();
2572                 $callbacks['{'] = array('end' => '}', 'cb' => $braceCallbacks);
2573                 $callbacks['['] = array('end' => ']', 'cb' => array(2=>null));
2574                 $text = $this->replace_callback ($text, $callbacks);
2575
2576                 array_pop( $this->mArgStack );
2577
2578                 wfProfileOut( $fname );
2579                 return $text;
2580         }
2581
2582         /**
2583          * Replace magic variables
2584          * @private
2585          */
2586         function variableSubstitution( $matches ) {
2587                 $fname = 'Parser::variableSubstitution';
2588                 $varname = $matches[1];
2589                 wfProfileIn( $fname );
2590                 $skip = false;
2591                 if ( $this->mOutputType == OT_WIKI ) {
2592                         # Do only magic variables prefixed by SUBST
2593                         $mwSubst =& MagicWord::get( MAG_SUBST );
2594                         if (!$mwSubst->matchStartAndRemove( $varname ))
2595                                 $skip = true;
2596                         # Note that if we don't substitute the variable below,
2597                         # we don't remove the {{subst:}} magic word, in case
2598                         # it is a template rather than a magic variable.
2599                 }
2600                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2601                         $id = $this->mVariables[$varname];
2602                         $text = $this->getVariableValue( $id );
2603                         $this->mOutput->mContainsOldMagic = true;
2604                 } else {
2605                         $text = $matches[0];
2606                 }
2607                 wfProfileOut( $fname );
2608                 return $text;
2609         }
2610
2611         # Split template arguments
2612         function getTemplateArgs( $argsString ) {
2613                 if ( $argsString === '' ) {
2614                         return array();
2615                 }
2616
2617                 $args = explode( '|', substr( $argsString, 1 ) );
2618
2619                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2620                 # merged with the next arg because the '|' character between belongs
2621                 # to the link syntax and not the template parameter syntax.
2622                 $argc = count($args);
2623
2624                 for ( $i = 0; $i < $argc-1; $i++ ) {
2625                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2626                                 $args[$i] .= '|'.$args[$i+1];
2627                                 array_splice($args, $i+1, 1);
2628                                 $i--;
2629                                 $argc--;
2630                         }
2631                 }
2632
2633                 return $args;
2634         }
2635
2636         /**
2637          * Return the text of a template, after recursively
2638          * replacing any variables or templates within the template.
2639          *
2640          * @param array $piece The parts of the template
2641          *  $piece['text']: matched text
2642          *  $piece['title']: the title, i.e. the part before the |
2643          *  $piece['parts']: the parameter array
2644          * @return string the text of the template
2645          * @private
2646          */
2647         function braceSubstitution( $piece ) {
2648                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $action;
2649                 $fname = 'Parser::braceSubstitution';
2650                 wfProfileIn( $fname );
2651
2652                 # Flags
2653                 $found = false;             # $text has been filled
2654                 $nowiki = false;            # wiki markup in $text should be escaped
2655                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2656                 $noargs = false;            # Don't replace triple-brace arguments in $text
2657                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2658                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2659                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2660
2661                 # Title object, where $text came from
2662                 $title = NULL;
2663
2664                 $linestart = '';
2665
2666                 # $part1 is the bit before the first |, and must contain only title characters
2667                 # $args is a list of arguments, starting from index 0, not including $part1
2668
2669                 $part1 = $piece['title'];
2670                 # If the third subpattern matched anything, it will start with |
2671
2672                 if (null == $piece['parts']) {
2673                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2674                         if ($replaceWith != $piece['text']) {
2675                                 $text = $replaceWith;
2676                                 $found = true;
2677                                 $noparse = true;
2678                                 $noargs = true;
2679                         }
2680                 }
2681
2682                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2683                 $argc = count( $args );
2684
2685                 # SUBST
2686                 if ( !$found ) {
2687                         $mwSubst =& MagicWord::get( MAG_SUBST );
2688                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2689                                 # One of two possibilities is true:
2690                                 # 1) Found SUBST but not in the PST phase
2691                                 # 2) Didn't find SUBST and in the PST phase
2692                                 # In either case, return without further processing
2693                                 $text = $piece['text'];
2694                                 $found = true;
2695                                 $noparse = true;
2696                                 $noargs = true;
2697                         }
2698                 }
2699
2700                 # MSG, MSGNW, INT and RAW
2701                 if ( !$found ) {
2702                         # Check for MSGNW:
2703                         $mwMsgnw =& MagicWord::get( MAG_MSGNW );
2704                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2705                                 $nowiki = true;
2706                         } else {
2707                                 # Remove obsolete MSG:
2708                                 $mwMsg =& MagicWord::get( MAG_MSG );
2709                                 $mwMsg->matchStartAndRemove( $part1 );
2710                         }
2711
2712                         # Check for RAW:
2713                         $mwRaw =& MagicWord::get( MAG_RAW );
2714                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2715                                 $forceRawInterwiki = true;
2716                         }
2717
2718                         # Check if it is an internal message
2719                         $mwInt =& MagicWord::get( MAG_INT );
2720                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2721                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2722                                         $text = $linestart . wfMsgReal( $part1, $args, true );
2723                                         $found = true;
2724                                 }
2725                         }
2726                 }
2727
2728                 # Parser functions
2729                 if ( !$found ) {
2730                         wfProfileIn( __METHOD__ . '-pfunc' );
2731
2732                         $colonPos = strpos( $part1, ':' );
2733                         if ( $colonPos !== false ) {
2734                                 # Case sensitive functions
2735                                 $function = substr( $part1, 0, $colonPos );
2736                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2737                                         $function = $this->mFunctionSynonyms[1][$function];
2738                                 } else {
2739                                         # Case insensitive functions
2740                                         $function = strtolower( $function );
2741                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2742                                                 $function = $this->mFunctionSynonyms[0][$function];
2743                                         } else {
2744                                                 $function = false;
2745                                         }
2746                                 }
2747                                 if ( $function ) {
2748                                         $funcArgs = array_map( 'trim', $args );
2749                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
2750                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
2751                                         $found = true;
2752
2753                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
2754                                         //$noargs = true;
2755                                         //$noparse = true;
2756
2757                                         if ( is_array( $result ) ) {
2758                                                 if ( isset( $result[0] ) ) {
2759                                                         $text = $linestart . $result[0];
2760                                                         unset( $result[0] );
2761                                                 }
2762
2763                                                 // Extract flags into the local scope
2764                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
2765                                                 extract( $result );
2766                                         } else {
2767                                                 $text = $linestart . $result;
2768                                         }
2769                                 }
2770                         }
2771                         wfProfileOut( __METHOD__ . '-pfunc' );
2772                 }
2773
2774                 # Template table test
2775
2776                 # Did we encounter this template already? If yes, it is in the cache
2777                 # and we need to check for loops.
2778                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
2779                         $found = true;
2780
2781                         # Infinite loop test
2782                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2783                                 $noparse = true;
2784                                 $noargs = true;
2785                                 $found = true;
2786                                 $text = $linestart .
2787                                         '{{' . $part1 . '}}' .
2788                                         '<!-- WARNING: template loop detected -->';
2789                                 wfDebug( "$fname: template loop broken at '$part1'\n" );
2790                         } else {
2791                                 # set $text to cached message.
2792                                 $text = $linestart . $this->mTemplates[$piece['title']];
2793                         }
2794                 }
2795
2796                 # Load from database
2797                 $lastPathLevel = $this->mTemplatePath;
2798                 if ( !$found ) {
2799                         wfProfileIn( __METHOD__ . '-loadtpl' );
2800                         $ns = NS_TEMPLATE;
2801                         # declaring $subpage directly in the function call
2802                         # does not work correctly with references and breaks
2803                         # {{/subpage}}-style inclusions
2804                         $subpage = '';
2805                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2806                         if ($subpage !== '') {
2807                                 $ns = $this->mTitle->getNamespace();
2808                         }
2809                         $title = Title::newFromText( $part1, $ns );
2810
2811
2812                         if ( !is_null( $title ) ) {
2813                                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
2814                                 # Check for language variants if the template is not found
2815                                 if($checkVariantLink && $title->getArticleID() == 0){
2816                                         $wgContLang->findVariantLink($part1, $title);
2817                                 }
2818
2819                                 if ( !$title->isExternal() ) {
2820                                         # Check for excessive inclusion
2821                                         $dbk = $title->getPrefixedDBkey();
2822                                         if ( $this->incrementIncludeCount( $dbk ) ) {
2823                                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->mOutputType != OT_WIKI ) {
2824                                                         $text = SpecialPage::capturePath( $title );
2825                                                         if ( is_string( $text ) ) {
2826                                                                 $found = true;
2827                                                                 $noparse = true;
2828                                                                 $noargs = true;
2829                                                                 $isHTML = true;
2830                                                                 $this->disableCache();
2831                                                         }
2832                                                 } else {
2833                                                         $articleContent = $this->fetchTemplate( $title );
2834                                                         if ( $articleContent !== false ) {
2835                                                                 $found = true;
2836                                                                 $text = $articleContent;
2837                                                                 $replaceHeadings = true;
2838                                                         }
2839                                                 }
2840                                         }
2841
2842                                         # If the title is valid but undisplayable, make a link to it
2843                                         if ( $this->mOutputType == OT_HTML && !$found ) {
2844                                                 $text = '[['.$title->getPrefixedText().']]';
2845                                                 $found = true;
2846                                         }
2847                                 } elseif ( $title->isTrans() ) {
2848                                         // Interwiki transclusion
2849                                         if ( $this->mOutputType == OT_HTML && !$forceRawInterwiki ) {
2850                                                 $text = $this->interwikiTransclude( $title, 'render' );
2851                                                 $isHTML = true;
2852                                                 $noparse = true;
2853                                         } else {
2854                                                 $text = $this->interwikiTransclude( $title, 'raw' );
2855                                                 $replaceHeadings = true;
2856                                         }
2857                                         $found = true;
2858                                 }
2859
2860                                 # Template cache array insertion
2861                                 # Use the original $piece['title'] not the mangled $part1, so that
2862                                 # modifiers such as RAW: produce separate cache entries
2863                                 if( $found ) {
2864                                         if( $isHTML ) {
2865                                                 // A special page; don't store it in the template cache.
2866                                         } else {
2867                                                 $this->mTemplates[$piece['title']] = $text;
2868                                         }
2869                                         $text = $linestart . $text;
2870                                 }
2871                         }
2872                         wfProfileOut( __METHOD__ . '-loadtpl' );
2873                 }
2874
2875                 # Recursive parsing, escaping and link table handling
2876                 # Only for HTML output
2877                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2878                         $text = wfEscapeWikiText( $text );
2879                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found ) {
2880                         if ( $noargs ) {
2881                                 $assocArgs = array();
2882                         } else {
2883                                 # Clean up argument array
2884                                 $assocArgs = array();
2885                                 $index = 1;
2886                                 foreach( $args as $arg ) {
2887                                         $eqpos = strpos( $arg, '=' );
2888                                         if ( $eqpos === false ) {
2889                                                 $assocArgs[$index++] = $arg;
2890                                         } else {
2891                                                 $name = trim( substr( $arg, 0, $eqpos ) );
2892                                                 $value = trim( substr( $arg, $eqpos+1 ) );
2893                                                 if ( $value === false ) {
2894                                                         $value = '';
2895                                                 }
2896                                                 if ( $name !== false ) {
2897                                                         $assocArgs[$name] = $value;
2898                                                 }
2899                                         }
2900                                 }
2901
2902                                 # Add a new element to the templace recursion path
2903                                 $this->mTemplatePath[$part1] = 1;
2904                         }
2905
2906                         if ( !$noparse ) {
2907                                 # If there are any <onlyinclude> tags, only include them
2908                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
2909                                         preg_match_all( '/<onlyinclude>(.*?)\n?<\/onlyinclude>/s', $text, $m );
2910                                         $text = '';
2911                                         foreach ($m[1] as $piece)
2912                                                 $text .= $piece;
2913                                 }
2914                                 # Remove <noinclude> sections and <includeonly> tags
2915                                 $text = preg_replace( '/<noinclude>.*?<\/noinclude>/s', '', $text );
2916                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
2917
2918                                 if( $this->mOutputType == OT_HTML ) {
2919                                         # Strip <nowiki>, <pre>, etc.
2920                                         $text = $this->strip( $text, $this->mStripState );
2921                                         $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
2922                                 }
2923                                 $text = $this->replaceVariables( $text, $assocArgs );
2924
2925                                 # If the template begins with a table or block-level
2926                                 # element, it should be treated as beginning a new line.
2927                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) {
2928                                         $text = "\n" . $text;
2929                                 }
2930                         } elseif ( !$noargs ) {
2931                                 # $noparse and !$noargs
2932                                 # Just replace the arguments, not any double-brace items
2933                                 # This is used for rendered interwiki transclusion
2934                                 $text = $this->replaceVariables( $text, $assocArgs, true );
2935                         }
2936                 }
2937                 # Prune lower levels off the recursion check path
2938                 $this->mTemplatePath = $lastPathLevel;
2939
2940                 if ( !$found ) {
2941                         wfProfileOut( $fname );
2942                         return $piece['text'];
2943                 } else {
2944                         wfProfileIn( __METHOD__ . '-placeholders' );
2945                         if ( $isHTML ) {
2946                                 # Replace raw HTML by a placeholder
2947                                 # Add a blank line preceding, to prevent it from mucking up
2948                                 # immediately preceding headings
2949                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
2950                         } else {
2951                                 # replace ==section headers==
2952                                 # XXX this needs to go away once we have a better parser.
2953                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
2954                                         if( !is_null( $title ) )
2955                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
2956                                         else
2957                                                 $encodedname = base64_encode("");
2958                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
2959                                                 PREG_SPLIT_DELIM_CAPTURE);
2960                                         $text = '';
2961                                         $nsec = 0;
2962                                         for( $i = 0; $i < count($m); $i += 2 ) {
2963                                                 $text .= $m[$i];
2964                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
2965                                                 $hl = $m[$i + 1];
2966                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
2967                                                         $text .= $hl;
2968                                                         continue;
2969                                                 }
2970                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
2971                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
2972                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
2973
2974                                                 $nsec++;
2975                                         }
2976                                 }
2977                         }
2978                         wfProfileOut( __METHOD__ . '-placeholders' );
2979                 }
2980
2981                 # Prune lower levels off the recursion check path
2982                 $this->mTemplatePath = $lastPathLevel;
2983
2984                 if ( !$found ) {
2985                         wfProfileOut( $fname );
2986                         return $piece['text'];
2987                 } else {
2988                         wfProfileOut( $fname );
2989                         return $text;
2990                 }
2991         }
2992
2993         /**
2994          * Fetch the unparsed text of a template and register a reference to it.
2995          */
2996         function fetchTemplate( $title ) {
2997                 $text = false;
2998                 // Loop to fetch the article, with up to 1 redirect
2999                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3000                         $rev = Revision::newFromTitle( $title );
3001                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
3002                         if ( !$rev ) {
3003                                 break;
3004                         }
3005                         $text = $rev->getText();
3006                         if ( $text === false ) {
3007                                 break;
3008                         }
3009                         // Redirect?
3010                         $title = Title::newFromRedirect( $text );
3011                 }
3012                 return $text;
3013         }
3014
3015         /**
3016          * Transclude an interwiki link.
3017          */
3018         function interwikiTransclude( $title, $action ) {
3019                 global $wgEnableScaryTranscluding, $wgCanonicalNamespaceNames;
3020
3021                 if (!$wgEnableScaryTranscluding)
3022                         return wfMsg('scarytranscludedisabled');
3023
3024                 // The namespace will actually only be 0 or 10, depending on whether there was a leading :
3025                 // But we'll handle it generally anyway
3026                 if ( $title->getNamespace() ) {
3027                         // Use the canonical namespace, which should work anywhere
3028                         $articleName = $wgCanonicalNamespaceNames[$title->getNamespace()] . ':' . $title->getDBkey();
3029                 } else {
3030                         $articleName = $title->getDBkey();
3031                 }
3032
3033                 $url = str_replace('$1', urlencode($articleName), Title::getInterwikiLink($title->getInterwiki()));
3034                 $url .= "?action=$action";
3035                 if (strlen($url) > 255)
3036                         return wfMsg('scarytranscludetoolong');
3037                 return $this->fetchScaryTemplateMaybeFromCache($url);
3038         }
3039
3040         function fetchScaryTemplateMaybeFromCache($url) {
3041                 global $wgTranscludeCacheExpiry;
3042                 $dbr =& wfGetDB(DB_SLAVE);
3043                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3044                                 array('tc_url' => $url));
3045                 if ($obj) {
3046                         $time = $obj->tc_time;
3047                         $text = $obj->tc_contents;
3048                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3049                                 return $text;
3050                         }
3051                 }
3052
3053                 $text = Http::get($url);
3054                 if (!$text)
3055                         return wfMsg('scarytranscludefailed', $url);
3056
3057                 $dbw =& wfGetDB(DB_MASTER);
3058                 $dbw->replace('transcache', array('tc_url'), array(
3059                         'tc_url' => $url,
3060                         'tc_time' => time(),
3061                         'tc_contents' => $text));
3062                 return $text;
3063         }
3064
3065
3066         /**
3067          * Triple brace replacement -- used for template arguments
3068          * @private
3069          */
3070         function argSubstitution( $matches ) {
3071                 $arg = trim( $matches['title'] );
3072                 $text = $matches['text'];
3073                 $inputArgs = end( $this->mArgStack );
3074
3075                 if ( array_key_exists( $arg, $inputArgs ) ) {
3076                         $text = $inputArgs[$arg];
3077                 } else if ($this->mOutputType == OT_HTML && null != $matches['parts'] && count($matches['parts']) > 0) {
3078                         $text = $matches['parts'][0];
3079                 }
3080
3081                 return $text;
3082         }
3083
3084         /**
3085          * Returns true if the function is allowed to include this entity
3086          * @private
3087          */
3088         function incrementIncludeCount( $dbk ) {
3089                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
3090                         $this->mIncludeCount[$dbk] = 0;
3091                 }
3092                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
3093                         return true;
3094                 } else {
3095                         return false;
3096                 }
3097         }
3098
3099         /**
3100          * Detect __NOGALLERY__ magic word and set a placeholder
3101          */
3102         function stripNoGallery( &$text ) {
3103                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3104                 # do not add TOC
3105                 $mw = MagicWord::get( MAG_NOGALLERY );
3106                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3107         }
3108
3109         /**
3110          * Detect __TOC__ magic word and set a placeholder
3111          */
3112         function stripToc( $text ) {
3113                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3114                 # do not add TOC
3115                 $mw = MagicWord::get( MAG_NOTOC );
3116                 if( $mw->matchAndRemove( $text ) ) {
3117                         $this->mShowToc = false;
3118                 }
3119
3120                 $mw = MagicWord::get( MAG_TOC );
3121                 if( $mw->match( $text ) ) {
3122                         $this->mShowToc = true;
3123                         $this->mForceTocPosition = true;
3124
3125                         // Set a placeholder. At the end we'll fill it in with the TOC.
3126                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3127
3128                         // Only keep the first one.
3129                         $text = $mw->replace( '', $text );
3130                 }
3131                 return $text;
3132         }
3133
3134         /**
3135          * This function accomplishes several tasks:
3136          * 1) Auto-number headings if that option is enabled
3137          * 2) Add an [edit] link to sections for logged in users who have enabled the option
3138          * 3) Add a Table of contents on the top for users who have enabled the option
3139          * 4) Auto-anchor headings
3140          *
3141          * It loops through all headlines, collects the necessary data, then splits up the
3142          * string and re-inserts the newly formatted headlines.
3143          *
3144          * @param string $text
3145          * @param boolean $isMain
3146          * @private
3147          */
3148         function formatHeadings( $text, $isMain=true ) {
3149                 global $wgMaxTocLevel, $wgContLang;
3150
3151                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3152                 if( !$this->mTitle->userCanEdit() ) {
3153                         $showEditLink = 0;
3154                 } else {
3155                         $showEditLink = $this->mOptions->getEditSection();
3156                 }
3157
3158                 # Inhibit editsection links if requested in the page
3159                 $esw =& MagicWord::get( MAG_NOEDITSECTION );
3160                 if( $esw->matchAndRemove( $text ) ) {
3161                         $showEditLink = 0;
3162                 }
3163
3164                 # Get all headlines for numbering them and adding funky stuff like [edit]
3165                 # links - this is for later, but we need the number of headlines right now
3166                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
3167
3168                 # if there are fewer than 4 headlines in the article, do not show TOC
3169                 # unless it's been explicitly enabled.
3170                 $enoughToc = $this->mShowToc &&
3171                         (($numMatches >= 4) || $this->mForceTocPosition);
3172
3173                 # Allow user to stipulate that a page should have a "new section"
3174                 # link added via __NEWSECTIONLINK__
3175                 $mw =& MagicWord::get( MAG_NEWSECTIONLINK );
3176                 if( $mw->matchAndRemove( $text ) )
3177                         $this->mOutput->setNewSection( true );
3178
3179                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3180                 # override above conditions and always show TOC above first header
3181                 $mw =& MagicWord::get( MAG_FORCETOC );
3182                 if ($mw->matchAndRemove( $text ) ) {
3183                         $this->mShowToc = true;
3184                         $enoughToc = true;
3185                 }
3186
3187                 # Never ever show TOC if no headers
3188                 if( $numMatches < 1 ) {
3189                         $enoughToc = false;
3190                 }
3191
3192                 # We need this to perform operations on the HTML
3193                 $sk =& $this->mOptions->getSkin();
3194
3195                 # headline counter
3196                 $headlineCount = 0;
3197                 $sectionCount = 0; # headlineCount excluding template sections
3198
3199                 # Ugh .. the TOC should have neat indentation levels which can be
3200                 # passed to the skin functions. These are determined here
3201                 $toc = '';
3202                 $full = '';
3203                 $head = array();
3204                 $sublevelCount = array();
3205                 $levelCount = array();
3206                 $toclevel = 0;
3207                 $level = 0;
3208                 $prevlevel = 0;
3209                 $toclevel = 0;
3210                 $prevtoclevel = 0;
3211
3212                 foreach( $matches[3] as $headline ) {
3213                         $istemplate = 0;
3214                         $templatetitle = '';
3215                         $templatesection = 0;
3216                         $numbering = '';
3217
3218                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3219                                 $istemplate = 1;
3220                                 $templatetitle = base64_decode($mat[1]);
3221                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3222                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3223                         }
3224
3225                         if( $toclevel ) {
3226                                 $prevlevel = $level;
3227                                 $prevtoclevel = $toclevel;
3228                         }
3229                         $level = $matches[1][$headlineCount];
3230
3231                         if( $doNumberHeadings || $enoughToc ) {
3232
3233                                 if ( $level > $prevlevel ) {
3234                                         # Increase TOC level
3235                                         $toclevel++;
3236                                         $sublevelCount[$toclevel] = 0;
3237                                         if( $toclevel<$wgMaxTocLevel ) {
3238                                                 $toc .= $sk->tocIndent();
3239                                         }
3240                                 }
3241                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3242                                         # Decrease TOC level, find level to jump to
3243
3244                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3245                                                 # Can only go down to level 1
3246                                                 $toclevel = 1;
3247                                         } else {
3248                                                 for ($i = $toclevel; $i > 0; $i--) {
3249                                                         if ( $levelCount[$i] == $level ) {
3250                                                                 # Found last matching level
3251                                                                 $toclevel = $i;
3252                                                                 break;
3253                                                         }
3254                                                         elseif ( $levelCount[$i] < $level ) {
3255                                                                 # Found first matching level below current level
3256                                                                 $toclevel = $i + 1;
3257                                                                 break;
3258                                                         }
3259                                                 }
3260                                         }
3261                                         if( $toclevel<$wgMaxTocLevel ) {
3262                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3263                                         }
3264                                 }
3265                                 else {
3266                                         # No change in level, end TOC line
3267                                         if( $toclevel<$wgMaxTocLevel ) {
3268                                                 $toc .= $sk->tocLineEnd();
3269                                         }
3270                                 }
3271
3272                                 $levelCount[$toclevel] = $level;
3273
3274                                 # count number of headlines for each level
3275                                 @$sublevelCount[$toclevel]++;
3276                                 $dot = 0;
3277                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3278                                         if( !empty( $sublevelCount[$i] ) ) {
3279                                                 if( $dot ) {
3280                                                         $numbering .= '.';
3281                                                 }
3282                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3283                                                 $dot = 1;
3284                                         }
3285                                 }
3286                         }
3287
3288                         # The canonized header is a version of the header text safe to use for links
3289                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3290                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
3291                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
3292
3293                         # Remove link placeholders by the link text.
3294                         #     <!--LINK number-->
3295                         # turns into
3296                         #     link text with suffix
3297                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3298                                                             "\$this->mLinkHolders['texts'][\$1]",
3299                                                             $canonized_headline );
3300                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3301                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3302                                                             $canonized_headline );
3303
3304                         # strip out HTML
3305                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3306                         $tocline = trim( $canonized_headline );
3307                         # Save headline for section edit hint before it's escaped
3308                         $headline_hint = trim( $canonized_headline );
3309                         $canonized_headline = Sanitizer::escapeId( $tocline );
3310                         $refers[$headlineCount] = $canonized_headline;
3311
3312                         # count how many in assoc. array so we can track dupes in anchors
3313                         @$refers[$canonized_headline]++;
3314                         $refcount[$headlineCount]=$refers[$canonized_headline];
3315
3316                         # Don't number the heading if it is the only one (looks silly)
3317                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3318                                 # the two are different if the line contains a link
3319                                 $headline=$numbering . ' ' . $headline;
3320                         }
3321
3322                         # Create the anchor for linking from the TOC to the section
3323                         $anchor = $canonized_headline;
3324                         if($refcount[$headlineCount] > 1 ) {
3325                                 $anchor .= '_' . $refcount[$headlineCount];
3326                         }
3327                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3328                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3329                         }
3330                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3331                                 if ( empty( $head[$headlineCount] ) ) {
3332                                         $head[$headlineCount] = '';
3333                                 }
3334                                 if( $istemplate )
3335                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
3336                                 else
3337                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3338                         }
3339
3340                         # give headline the correct <h#> tag
3341                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
3342
3343                         $headlineCount++;
3344                         if( !$istemplate )
3345                                 $sectionCount++;
3346                 }
3347
3348                 if( $enoughToc ) {
3349                         if( $toclevel<$wgMaxTocLevel ) {
3350                                 $toc .= $sk->tocUnindent( $toclevel - 1 );
3351                         }
3352                         $toc = $sk->tocList( $toc );
3353                 }
3354
3355                 # split up and insert constructed headlines
3356
3357                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3358                 $i = 0;
3359
3360                 foreach( $blocks as $block ) {
3361                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3362                                 # This is the [edit] link that appears for the top block of text when
3363                                 # section editing is enabled
3364
3365                                 # Disabled because it broke block formatting
3366                                 # For example, a bullet point in the top line
3367                                 # $full .= $sk->editSectionLink(0);
3368                         }
3369                         $full .= $block;
3370                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3371                                 # Top anchor now in skin
3372                                 $full = $full.$toc;
3373                         }
3374
3375                         if( !empty( $head[$i] ) ) {
3376                                 $full .= $head[$i];
3377                         }
3378                         $i++;
3379                 }
3380                 if( $this->mForceTocPosition ) {
3381                         return str_replace( '<!--MWTOC-->', $toc, $full );
3382                 } else {
3383                         return $full;
3384                 }
3385         }
3386
3387         /**
3388          * Return an HTML link for the "ISBN 123456" text
3389          * @private
3390          */
3391         function magicISBN( $text ) {
3392                 $fname = 'Parser::magicISBN';
3393                 wfProfileIn( $fname );
3394
3395                 $a = split( 'ISBN ', ' '.$text );
3396                 if ( count ( $a ) < 2 ) {
3397                         wfProfileOut( $fname );
3398                         return $text;
3399                 }
3400                 $text = substr( array_shift( $a ), 1);
3401                 $valid = '0123456789-Xx';
3402
3403                 foreach ( $a as $x ) {
3404                         # hack: don't replace inside thumbnail title/alt
3405                         # attributes
3406                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3407                                 $text .= "ISBN $x";
3408                                 continue;
3409                         }
3410
3411                         $isbn = $blank = '' ;
3412                         while ( $x !== '' && ' ' == $x{0} ) {
3413                                 $blank .= ' ';
3414                                 $x = substr( $x, 1 );
3415                         }
3416                         if ( $x == '' ) { # blank isbn
3417                                 $text .= "ISBN $blank";
3418                                 continue;
3419                         }
3420                         while ( strstr( $valid, $x{0} ) != false ) {
3421                                 $isbn .= $x{0};
3422                                 $x = substr( $x, 1 );
3423                         }
3424                         $num = str_replace( '-', '', $isbn );
3425                         $num = str_replace( ' ', '', $num );
3426                         $num = str_replace( 'x', 'X', $num );
3427
3428                         if ( '' == $num ) {
3429                                 $text .= "ISBN $blank$x";
3430                         } else {
3431                                 $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
3432                                 $text .= '<a href="' .
3433                                         $titleObj->escapeLocalUrl( 'isbn='.$num ) .
3434                                         "\" class=\"internal\">ISBN $isbn</a>";
3435                                 $text .= $x;
3436                         }
3437                 }
3438                 wfProfileOut( $fname );
3439                 return $text;
3440         }
3441
3442         /**
3443          * Return an HTML link for the "RFC 1234" text
3444          *
3445          * @private
3446          * @param string $text     Text to be processed
3447          * @param string $keyword  Magic keyword to use (default RFC)
3448          * @param string $urlmsg   Interface message to use (default rfcurl)
3449          * @return string
3450          */
3451         function magicRFC( $text, $keyword='RFC ', $urlmsg='rfcurl'  ) {
3452
3453                 $valid = '0123456789';
3454                 $internal = false;
3455
3456                 $a = split( $keyword, ' '.$text );
3457                 if ( count ( $a ) < 2 ) {
3458                         return $text;
3459                 }
3460                 $text = substr( array_shift( $a ), 1);
3461
3462                 /* Check if keyword is preceed by [[.
3463                  * This test is made here cause of the array_shift above
3464                  * that prevent the test to be done in the foreach.
3465                  */
3466                 if ( substr( $text, -2 ) == '[[' ) {
3467                         $internal = true;
3468                 }
3469
3470                 foreach ( $a as $x ) {
3471                         /* token might be empty if we have RFC RFC 1234 */
3472                         if ( $x=='' ) {
3473                                 $text.=$keyword;
3474                                 continue;
3475                                 }
3476
3477                         # hack: don't replace inside thumbnail title/alt
3478                         # attributes
3479                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3480                                 $text .= $keyword . $x;
3481                                 continue;
3482                         }
3483
3484                         $id = $blank = '' ;
3485
3486                         /** remove and save whitespaces in $blank */
3487                         while ( $x{0} == ' ' ) {
3488                                 $blank .= ' ';
3489                                 $x = substr( $x, 1 );
3490                         }
3491
3492                         /** remove and save the rfc number in $id */
3493                         while ( strstr( $valid, $x{0} ) != false ) {
3494                                 $id .= $x{0};
3495                                 $x = substr( $x, 1 );
3496                         }
3497
3498                         if ( $id == '' ) {
3499                                 /* call back stripped spaces*/
3500                                 $text .= $keyword.$blank.$x;
3501                         } elseif( $internal ) {
3502                                 /* normal link */
3503                                 $text .= $keyword.$id.$x;
3504                         } else {
3505                                 /* build the external link*/
3506                                 $url = wfMsg( $urlmsg, $id);
3507                                 $sk =& $this->mOptions->getSkin();
3508                                 $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
3509                                 $text .= "<a href=\"{$url}\"{$la}>{$keyword}{$id}</a>{$x}";
3510                         }
3511
3512                         /* Check if the next RFC keyword is preceed by [[ */
3513                         $internal = ( substr($x,-2) == '[[' );
3514                 }
3515                 return $text;
3516         }
3517
3518         /**
3519          * Transform wiki markup when saving a page by doing \r\n -> \n
3520          * conversion, substitting signatures, {{subst:}} templates, etc.
3521          *
3522          * @param string $text the text to transform
3523          * @param Title &$title the Title object for the current article
3524          * @param User &$user the User object describing the current user
3525          * @param ParserOptions $options parsing options
3526          * @param bool $clearState whether to clear the parser state first
3527          * @return string the altered wiki markup
3528          * @public
3529          */
3530         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
3531                 $this->mOptions = $options;
3532                 $this->mTitle =& $title;
3533                 $this->mOutputType = OT_WIKI;
3534
3535                 if ( $clearState ) {
3536                         $this->clearState();
3537                 }
3538
3539                 $stripState = false;
3540                 $pairs = array(
3541                         "\r\n" => "\n",
3542                 );
3543                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3544                 $text = $this->strip( $text, $stripState, true, array( 'gallery' ) );
3545                 $text = $this->pstPass2( $text, $stripState, $user );
3546                 $text = $this->unstrip( $text, $stripState );
3547                 $text = $this->unstripNoWiki( $text, $stripState );
3548                 return $text;
3549         }
3550
3551         /**
3552          * Pre-save transform helper function
3553          * @private
3554          */
3555         function pstPass2( $text, &$stripState, &$user ) {
3556                 global $wgContLang, $wgLocaltimezone;
3557
3558                 /* Note: This is the timestamp saved as hardcoded wikitext to
3559                  * the database, we use $wgContLang here in order to give
3560                  * everyone the same signature and use the default one rather
3561                  * than the one selected in each user's preferences.
3562                  */
3563                 if ( isset( $wgLocaltimezone ) ) {
3564                         $oldtz = getenv( 'TZ' );
3565                         putenv( 'TZ='.$wgLocaltimezone );
3566                 }
3567                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3568                   ' (' . date( 'T' ) . ')';
3569                 if ( isset( $wgLocaltimezone ) ) {
3570                         putenv( 'TZ='.$oldtz );
3571                 }
3572
3573                 # Variable replacement
3574                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3575                 $text = $this->replaceVariables( $text );
3576
3577                 # Strip out <nowiki> etc. added via replaceVariables
3578                 $text = $this->strip( $text, $stripState, false, array( 'gallery' ) );
3579
3580                 # Signatures
3581                 $sigText = $this->getUserSig( $user );
3582                 $text = strtr( $text, array(
3583                         '~~~~~' => $d,
3584                         '~~~~' => "$sigText $d",
3585                         '~~~' => $sigText
3586                 ) );
3587
3588                 # Context links: [[|name]] and [[name (context)|]]
3589                 #
3590                 global $wgLegalTitleChars;
3591                 $tc = "[$wgLegalTitleChars]";
3592                 $np = str_replace( array( '(', ')' ), array( '', '' ), $tc ); # No parens
3593
3594                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3595                 $conpat = "/^({$np}+) \\(({$tc}+)\\)$/";
3596
3597                 $p1 = "/\[\[({$np}+) \\(({$np}+)\\)\\|]]/";             # [[page (context)|]]
3598                 $p2 = "/\[\[\\|({$tc}+)]]/";                                    # [[|page]]
3599                 $p3 = "/\[\[(:*$namespacechar+):({$np}+)\\|]]/";                # [[namespace:page|]] and [[:namespace:page|]]
3600                 $p4 = "/\[\[(:*$namespacechar+):({$np}+) \\(({$np}+)\\)\\|]]/"; # [[ns:page (cont)|]] and [[:ns:page (cont)|]]
3601                 $context = '';
3602                 $t = $this->mTitle->getText();
3603                 if ( preg_match( $conpat, $t, $m ) ) {
3604                         $context = $m[2];
3605                 }
3606                 $text = preg_replace( $p4, '[[\\1:\\2 (\\3)|\\2]]', $text );
3607                 $text = preg_replace( $p1, '[[\\1 (\\2)|\\1]]', $text );
3608                 $text = preg_replace( $p3, '[[\\1:\\2|\\2]]', $text );
3609
3610                 if ( '' == $context ) {
3611                         $text = preg_replace( $p2, '[[\\1]]', $text );
3612                 } else {
3613                         $text = preg_replace( $p2, "[[\\1 ({$context})|\\1]]", $text );
3614                 }
3615
3616                 # Trim trailing whitespace
3617                 # MAG_END (__END__) tag allows for trailing
3618                 # whitespace to be deliberately included
3619                 $text = rtrim( $text );
3620                 $mw =& MagicWord::get( MAG_END );
3621                 $mw->matchAndRemove( $text );
3622
3623                 return $text;
3624         }
3625
3626         /**
3627          * Fetch the user's signature text, if any, and normalize to
3628          * validated, ready-to-insert wikitext.
3629          *
3630          * @param User $user
3631          * @return string
3632          * @private
3633          */
3634         function getUserSig( &$user ) {
3635                 $username = $user->getName();
3636                 $nickname = $user->getOption( 'nickname' );
3637                 $nickname = $nickname === '' ? $username : $nickname;
3638
3639                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3640                         # Sig. might contain markup; validate this
3641                         if( $this->validateSig( $nickname ) !== false ) {
3642                                 # Validated; clean up (if needed) and return it
3643                                 return $this->cleanSig( $nickname, true );
3644                         } else {
3645                                 # Failed to validate; fall back to the default
3646                                 $nickname = $username;
3647                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3648                         }
3649                 }
3650
3651                 // Make sure nickname doesnt get a sig in a sig
3652                 $nickname = $this->cleanSigInSig( $nickname );
3653
3654                 # If we're still here, make it a link to the user page
3655                 $userpage = $user->getUserPage();
3656                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3657         }
3658
3659         /**
3660          * Check that the user's signature contains no bad XML
3661          *
3662          * @param string $text
3663          * @return mixed An expanded string, or false if invalid.
3664          */
3665         function validateSig( $text ) {
3666                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3667         }
3668
3669         /**
3670          * Clean up signature text
3671          *
3672          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3673          * 2) Substitute all transclusions
3674          *
3675          * @param string $text
3676          * @param $parsing Whether we're cleaning (preferences save) or parsing
3677          * @return string Signature text
3678          */
3679         function cleanSig( $text, $parsing = false ) {
3680                 global $wgTitle;
3681                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
3682
3683                 $substWord = MagicWord::get( MAG_SUBST );
3684                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3685                 $substText = '{{' . $substWord->getSynonym( 0 );
3686
3687                 $text = preg_replace( $substRegex, $substText, $text );
3688                 $text = $this->cleanSigInSig( $text );
3689                 $text = $this->replaceVariables( $text );
3690
3691                 $this->clearState();
3692                 return $text;
3693         }
3694
3695         /**
3696          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3697          * @param string $text
3698          * @return string Signature text with /~{3,5}/ removed
3699          */
3700         function cleanSigInSig( $text ) {
3701                 $text = preg_replace( '/~{3,5}/', '', $text );
3702                 return $text;
3703         }
3704
3705         /**
3706          * Set up some variables which are usually set up in parse()
3707          * so that an external function can call some class members with confidence
3708          * @public
3709          */
3710         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3711                 $this->mTitle =& $title;
3712                 $this->mOptions = $options;
3713                 $this->mOutputType = $outputType;
3714                 if ( $clearState ) {
3715                         $this->clearState();
3716                 }
3717         }
3718
3719         /**
3720          * Transform a MediaWiki message by replacing magic variables.
3721          *
3722          * @param string $text the text to transform
3723          * @param ParserOptions $options  options
3724          * @return string the text with variables substituted
3725          * @public
3726          */
3727         function transformMsg( $text, $options ) {
3728                 global $wgTitle;
3729                 static $executing = false;
3730
3731                 $fname = "Parser::transformMsg";
3732
3733                 # Guard against infinite recursion
3734                 if ( $executing ) {
3735                         return $text;
3736                 }
3737                 $executing = true;
3738
3739                 wfProfileIn($fname);
3740
3741                 $this->mTitle = $wgTitle;
3742                 $this->mOptions = $options;
3743                 $this->mOutputType = OT_MSG;
3744                 $this->clearState();
3745                 $text = $this->replaceVariables( $text );
3746
3747                 $executing = false;
3748                 wfProfileOut($fname);
3749                 return $text;
3750         }
3751
3752         /**
3753          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3754          * The callback should have the following form:
3755          *    function myParserHook( $text, $params, &$parser ) { ... }
3756          *
3757          * Transform and return $text. Use $parser for any required context, e.g. use
3758          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3759          *
3760          * @public
3761          *
3762          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3763          * @param mixed $callback The callback function (and object) to use for the tag
3764          *
3765          * @return The old value of the mTagHooks array associated with the hook
3766          */
3767         function setHook( $tag, $callback ) {
3768                 $tag = strtolower( $tag );
3769                 $oldVal = @$this->mTagHooks[$tag];
3770                 $this->mTagHooks[$tag] = $callback;
3771
3772                 return $oldVal;
3773         }
3774
3775         /**
3776          * Create a function, e.g. {{sum:1|2|3}}
3777          * The callback function should have the form:
3778          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3779          *
3780          * The callback may either return the text result of the function, or an array with the text
3781          * in element 0, and a number of flags in the other elements. The names of the flags are
3782          * specified in the keys. Valid flags are:
3783          *   found                     The text returned is valid, stop processing the template. This
3784          *                             is on by default.
3785          *   nowiki                    Wiki markup in the return value should be escaped
3786          *   noparse                   Unsafe HTML tags should not be stripped, etc.
3787          *   noargs                    Don't replace triple-brace arguments in the return value
3788          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3789          *
3790          * @public
3791          *
3792          * @param mixed $id The magic word ID, or (deprecated) the function name. Function names are case-insensitive.
3793          * @param mixed $callback The callback function (and object) to use
3794          * @param integer $flags a combination of the following flags:
3795          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
3796          *
3797          * @return The old callback function for this name, if any
3798          */
3799         function setFunctionHook( $id, $callback, $flags = 0 ) {
3800                 if( is_string( $id ) ) {
3801                         $id = strtolower( $id );
3802                 }
3803                 $oldVal = @$this->mFunctionHooks[$id];
3804                 $this->mFunctionHooks[$id] = $callback;
3805
3806                 # Add to function cache
3807                 if ( is_int( $id ) ) {
3808                         $mw = MagicWord::get( $id );
3809                         $synonyms = $mw->getSynonyms();
3810                         $sensitive = intval( $mw->isCaseSensitive() );
3811                 } else {
3812                         $synonyms = array( $id );
3813                         $sensitive = 0;
3814                 }
3815
3816                 foreach ( $synonyms as $syn ) {
3817                         # Case
3818                         if ( !$sensitive ) {
3819                                 $syn = strtolower( $syn );
3820                         }
3821                         # Add leading hash
3822                         if ( !( $flags & SFH_NO_HASH ) ) {
3823                                 $syn = '#' . $syn;
3824                         }
3825                         # Remove trailing colon
3826                         if ( substr( $syn, -1, 1 ) == ':' ) {
3827                                 $syn = substr( $syn, 0, -1 );
3828                         }
3829                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
3830                 }
3831                 return $oldVal;
3832         }
3833
3834         /**
3835          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3836          * Placeholders created in Skin::makeLinkObj()
3837          * Returns an array of links found, indexed by PDBK:
3838          *  0 - broken
3839          *  1 - normal link
3840          *  2 - stub
3841          * $options is a bit field, RLH_FOR_UPDATE to select for update
3842          */
3843         function replaceLinkHolders( &$text, $options = 0 ) {
3844                 global $wgUser;
3845                 global $wgOutputReplace;
3846
3847                 $fname = 'Parser::replaceLinkHolders';
3848                 wfProfileIn( $fname );
3849
3850                 $pdbks = array();
3851                 $colours = array();
3852                 $sk =& $this->mOptions->getSkin();
3853                 $linkCache =& LinkCache::singleton();
3854
3855                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3856                         wfProfileIn( $fname.'-check' );
3857                         $dbr =& wfGetDB( DB_SLAVE );
3858                         $page = $dbr->tableName( 'page' );
3859                         $threshold = $wgUser->getOption('stubthreshold');
3860
3861                         # Sort by namespace
3862                         asort( $this->mLinkHolders['namespaces'] );
3863
3864                         # Generate query
3865                         $query = false;
3866                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3867                                 # Make title object
3868                                 $title = $this->mLinkHolders['titles'][$key];
3869
3870                                 # Skip invalid entries.
3871                                 # Result will be ugly, but prevents crash.
3872                                 if ( is_null( $title ) ) {
3873                                         continue;
3874                                 }
3875                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3876
3877                                 # Check if it's a static known link, e.g. interwiki
3878                                 if ( $title->isAlwaysKnown() ) {
3879                                         $colours[$pdbk] = 1;
3880                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3881                                         $colours[$pdbk] = 1;
3882                                         $this->mOutput->addLink( $title, $id );
3883                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3884                                         $colours[$pdbk] = 0;
3885                                 } else {
3886                                         # Not in the link cache, add it to the query
3887                                         if ( !isset( $current ) ) {
3888                                                 $current = $ns;
3889                                                 $query =  "SELECT page_id, page_namespace, page_title";
3890                                                 if ( $threshold > 0 ) {
3891                                                         $query .= ', page_len, page_is_redirect';
3892                                                 }
3893                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
3894                                         } elseif ( $current != $ns ) {
3895                                                 $current = $ns;
3896                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
3897                                         } else {
3898                                                 $query .= ', ';
3899                                         }
3900
3901                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3902                                 }
3903                         }
3904                         if ( $query ) {
3905                                 $query .= '))';
3906                                 if ( $options & RLH_FOR_UPDATE ) {
3907                                         $query .= ' FOR UPDATE';
3908                                 }
3909
3910                                 $res = $dbr->query( $query, $fname );
3911
3912                                 # Fetch data and form into an associative array
3913                                 # non-existent = broken
3914                                 # 1 = known
3915                                 # 2 = stub
3916                                 while ( $s = $dbr->fetchObject($res) ) {
3917                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
3918                                         $pdbk = $title->getPrefixedDBkey();
3919                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
3920                                         $this->mOutput->addLink( $title, $s->page_id );
3921
3922                                         if ( $threshold >  0 ) {
3923                                                 $size = $s->page_len;
3924                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
3925                                                         $colours[$pdbk] = 1;
3926                                                 } else {
3927                                                         $colours[$pdbk] = 2;
3928                                                 }
3929                                         } else {
3930                                                 $colours[$pdbk] = 1;
3931                                         }
3932                                 }
3933                         }
3934                         wfProfileOut( $fname.'-check' );
3935
3936                         # Construct search and replace arrays
3937                         wfProfileIn( $fname.'-construct' );
3938                         $wgOutputReplace = array();
3939                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3940                                 $pdbk = $pdbks[$key];
3941                                 $searchkey = "<!--LINK $key-->";
3942                                 $title = $this->mLinkHolders['titles'][$key];
3943                                 if ( empty( $colours[$pdbk] ) ) {
3944                                         $linkCache->addBadLinkObj( $title );
3945                                         $colours[$pdbk] = 0;
3946                                         $this->mOutput->addLink( $title, 0 );
3947                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
3948                                                                         $this->mLinkHolders['texts'][$key],
3949                                                                         $this->mLinkHolders['queries'][$key] );
3950                                 } elseif ( $colours[$pdbk] == 1 ) {
3951                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
3952                                                                         $this->mLinkHolders['texts'][$key],
3953                                                                         $this->mLinkHolders['queries'][$key] );
3954                                 } elseif ( $colours[$pdbk] == 2 ) {
3955                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
3956                                                                         $this->mLinkHolders['texts'][$key],
3957                                                                         $this->mLinkHolders['queries'][$key] );
3958                                 }
3959                         }
3960                         wfProfileOut( $fname.'-construct' );
3961
3962                         # Do the thing
3963                         wfProfileIn( $fname.'-replace' );
3964
3965                         $text = preg_replace_callback(
3966                                 '/(<!--LINK .*?-->)/',
3967                                 "wfOutputReplaceMatches",
3968                                 $text);
3969
3970                         wfProfileOut( $fname.'-replace' );
3971                 }
3972
3973                 # Now process interwiki link holders
3974                 # This is quite a bit simpler than internal links
3975                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
3976                         wfProfileIn( $fname.'-interwiki' );
3977                         # Make interwiki link HTML
3978                         $wgOutputReplace = array();
3979                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
3980                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
3981                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
3982                         }
3983
3984                         $text = preg_replace_callback(
3985                                 '/<!--IWLINK (.*?)-->/',
3986                                 "wfOutputReplaceMatches",
3987                                 $text );
3988                         wfProfileOut( $fname.'-interwiki' );
3989                 }
3990
3991                 wfProfileOut( $fname );
3992                 return $colours;
3993         }
3994
3995         /**
3996          * Replace <!--LINK--> link placeholders with plain text of links
3997          * (not HTML-formatted).
3998          * @param string $text
3999          * @return string
4000          */
4001         function replaceLinkHoldersText( $text ) {
4002                 $fname = 'Parser::replaceLinkHoldersText';
4003                 wfProfileIn( $fname );
4004
4005                 $text = preg_replace_callback(
4006                         '/<!--(LINK|IWLINK) (.*?)-->/',
4007                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4008                         $text );
4009
4010                 wfProfileOut( $fname );
4011                 return $text;
4012         }
4013
4014         /**
4015          * @param array $matches
4016          * @return string
4017          * @private
4018          */
4019         function replaceLinkHoldersTextCallback( $matches ) {
4020                 $type = $matches[1];
4021                 $key  = $matches[2];
4022                 if( $type == 'LINK' ) {
4023                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4024                                 return $this->mLinkHolders['texts'][$key];
4025                         }
4026                 } elseif( $type == 'IWLINK' ) {
4027                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4028                                 return $this->mInterwikiLinkHolders['texts'][$key];
4029                         }
4030                 }
4031                 return $matches[0];
4032         }
4033
4034         /**
4035          * Tag hook handler for 'pre'.
4036          */
4037         function renderPreTag( $text, $attribs, $parser ) {
4038                 // Backwards-compatibility hack
4039                 $content = preg_replace( '!<nowiki>(.*?)</nowiki>!is', '\\1', $text );
4040
4041                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4042                 return wfOpenElement( 'pre', $attribs ) .
4043                         wfEscapeHTMLTagsOnly( $content ) .
4044                         '</pre>';
4045         }
4046
4047         /**
4048          * Renders an image gallery from a text with one line per image.
4049          * text labels may be given by using |-style alternative text. E.g.
4050          *   Image:one.jpg|The number "1"
4051          *   Image:tree.jpg|A tree
4052          * given as text will return the HTML of a gallery with two images,
4053          * labeled 'The number "1"' and
4054          * 'A tree'.
4055          */
4056         function renderImageGallery( $text, $params ) {
4057                 $ig = new ImageGallery();
4058                 $ig->setShowBytes( false );
4059                 $ig->setShowFilename( false );
4060                 $ig->setParsing();
4061                 $ig->useSkin( $this->mOptions->getSkin() );
4062
4063                 if( isset( $params['caption'] ) )
4064                         $ig->setCaption( $params['caption'] );
4065
4066                 $lines = explode( "\n", $text );
4067                 foreach ( $lines as $line ) {
4068                         # match lines like these:
4069                         # Image:someimage.jpg|This is some image
4070                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4071                         # Skip empty lines
4072                         if ( count( $matches ) == 0 ) {
4073                                 continue;
4074                         }
4075                         $nt =& Title::newFromText( $matches[1] );
4076                         if( is_null( $nt ) ) {
4077                                 # Bogus title. Ignore these so we don't bomb out later.
4078                                 continue;
4079                         }
4080                         if ( isset( $matches[3] ) ) {
4081                                 $label = $matches[3];
4082                         } else {
4083                                 $label = '';
4084                         }
4085
4086                         $pout = $this->parse( $label,
4087                                 $this->mTitle,
4088                                 $this->mOptions,
4089                                 false, // Strip whitespace...?
4090                                 false  // Don't clear state!
4091                         );
4092                         $html = $pout->getText();
4093
4094                         $ig->add( new Image( $nt ), $html );
4095
4096                         # Only add real images (bug #5586)
4097                         if ( $nt->getNamespace() == NS_IMAGE ) {
4098                                 $this->mOutput->addImage( $nt->getDBkey() );
4099                         }
4100                 }
4101                 return $ig->toHTML();
4102         }
4103
4104         /**
4105          * Parse image options text and use it to make an image
4106          */
4107         function makeImage( &$nt, $options ) {
4108                 global $wgUseImageResize;
4109
4110                 $align = '';
4111
4112                 # Check if the options text is of the form "options|alt text"
4113                 # Options are:
4114                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4115                 #  * left               no resizing, just left align. label is used for alt= only
4116                 #  * right              same, but right aligned
4117                 #  * none               same, but not aligned
4118                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4119                 #  * center             center the image
4120                 #  * framed             Keep original image size, no magnify-button.
4121
4122                 $part = explode( '|', $options);
4123
4124                 $mwThumb  =& MagicWord::get( MAG_IMG_THUMBNAIL );
4125                 $mwManualThumb =& MagicWord::get( MAG_IMG_MANUALTHUMB );
4126                 $mwLeft   =& MagicWord::get( MAG_IMG_LEFT );
4127                 $mwRight  =& MagicWord::get( MAG_IMG_RIGHT );
4128                 $mwNone   =& MagicWord::get( MAG_IMG_NONE );
4129                 $mwWidth  =& MagicWord::get( MAG_IMG_WIDTH );
4130                 $mwCenter =& MagicWord::get( MAG_IMG_CENTER );
4131                 $mwFramed =& MagicWord::get( MAG_IMG_FRAMED );
4132                 $caption = '';
4133
4134                 $width = $height = $framed = $thumb = false;
4135                 $manual_thumb = '' ;
4136
4137                 foreach( $part as $key => $val ) {
4138                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
4139                                 $thumb=true;
4140                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
4141                                 # use manually specified thumbnail
4142                                 $thumb=true;
4143                                 $manual_thumb = $match;
4144                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
4145                                 # remember to set an alignment, don't render immediately
4146                                 $align = 'right';
4147                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
4148                                 # remember to set an alignment, don't render immediately
4149                                 $align = 'left';
4150                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
4151                                 # remember to set an alignment, don't render immediately
4152                                 $align = 'center';
4153                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
4154                                 # remember to set an alignment, don't render immediately
4155                                 $align = 'none';
4156                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
4157                                 wfDebug( "MAG_IMG_WIDTH match: $match\n" );
4158                                 # $match is the image width in pixels
4159                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
4160                                         $width = intval( $m[1] );
4161                                         $height = intval( $m[2] );
4162                                 } else {
4163                                         $width = intval($match);
4164                                 }
4165                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
4166                                 $framed=true;
4167                         } else {
4168                                 $caption = $val;
4169                         }
4170                 }
4171                 # Strip bad stuff out of the alt text
4172                 $alt = $this->replaceLinkHoldersText( $caption );
4173
4174                 # make sure there are no placeholders in thumbnail attributes
4175                 # that are later expanded to html- so expand them now and
4176                 # remove the tags
4177                 $alt = $this->unstrip($alt, $this->mStripState);
4178                 $alt = Sanitizer::stripAllTags( $alt );
4179
4180                 # Linker does the rest
4181                 $sk =& $this->mOptions->getSkin();
4182                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
4183         }
4184
4185         /**
4186          * Set a flag in the output object indicating that the content is dynamic and
4187          * shouldn't be cached.
4188          */
4189         function disableCache() {
4190                 wfDebug( "Parser output marked as uncacheable.\n" );
4191                 $this->mOutput->mCacheTime = -1;
4192         }
4193
4194         /**#@+
4195          * Callback from the Sanitizer for expanding items found in HTML attribute
4196          * values, so they can be safely tested and escaped.
4197          * @param string $text
4198          * @param array $args
4199          * @return string
4200          * @private
4201          */
4202         function attributeStripCallback( &$text, $args ) {
4203                 $text = $this->replaceVariables( $text, $args );
4204                 $text = $this->unstripForHTML( $text );
4205                 return $text;
4206         }
4207
4208         function unstripForHTML( $text ) {
4209                 $text = $this->unstrip( $text, $this->mStripState );
4210                 $text = $this->unstripNoWiki( $text, $this->mStripState );
4211                 return $text;
4212         }
4213         /**#@-*/
4214
4215         /**#@+
4216          * Accessor/mutator
4217          */
4218         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4219         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4220         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4221         /**#@-*/
4222
4223         /**#@+
4224          * Accessor
4225          */
4226         function getTags() { return array_keys( $this->mTagHooks ); }
4227         /**#@-*/
4228
4229
4230         /**
4231          * Break wikitext input into sections, and either pull or replace
4232          * some particular section's text.
4233          *
4234          * External callers should use the getSection and replaceSection methods.
4235          *
4236          * @param $text Page wikitext
4237          * @param $section Numbered section. 0 pulls the text before the first
4238          *                 heading; other numbers will pull the given section
4239          *                 along with its lower-level subsections.
4240          * @param $mode One of "get" or "replace"
4241          * @param $newtext Replacement text for section data.
4242          * @return string for "get", the extracted section text.
4243          *                for "replace", the whole page with the section replaced.
4244          */
4245         private function extractSections( $text, $section, $mode, $newtext='' ) {
4246                 # strip NOWIKI etc. to avoid confusion (true-parameter causes HTML
4247                 # comments to be stripped as well)
4248                 $striparray = array();
4249
4250                 $oldOutputType = $this->mOutputType;
4251                 $oldOptions = $this->mOptions;
4252                 $this->mOptions = new ParserOptions();
4253                 $this->mOutputType = OT_WIKI;
4254
4255                 $striptext = $this->strip( $text, $striparray, true );
4256
4257                 $this->mOutputType = $oldOutputType;
4258                 $this->mOptions = $oldOptions;
4259
4260                 # now that we can be sure that no pseudo-sections are in the source,
4261                 # split it up by section
4262                 $uniq = preg_quote( $this->uniqPrefix(), '/' );
4263                 $comment = "(?:$uniq-!--.*?QINU)";
4264                 $secs = preg_split(
4265                 /*
4266                         "/
4267                         ^(
4268                         (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4269                         (?:
4270                                 (=+) # Should this be limited to 6?
4271                                 .+?  # Section title...
4272                                 \\2  # Ending = count must match start
4273                         |
4274                                 ^
4275                                 <h([1-6])\b.*?>
4276                                 .*?
4277                                 <\/h\\3\s*>
4278                         )
4279                         (?:$comment|<\/?noinclude>|\s+)* # Trailing whitespace ok
4280                         )$
4281                         /mix",
4282                 */
4283                         "/
4284                         (
4285                                 ^
4286                                 (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4287                                 (=+) # Should this be limited to 6?
4288                                 .+?  # Section title...
4289                                 \\2  # Ending = count must match start
4290                                 (?:$comment|<\/?noinclude>|[ \\t]+)* # Trailing whitespace ok
4291                                 $
4292                         |
4293                                 <h([1-6])\b.*?>
4294                                 .*?
4295                                 <\/h\\3\s*>
4296                         )
4297                         /mix",
4298                         $striptext, -1,
4299                         PREG_SPLIT_DELIM_CAPTURE);
4300
4301                 if( $mode == "get" ) {
4302                         if( $section == 0 ) {
4303                                 // "Section 0" returns the content before any other section.
4304                                 $rv = $secs[0];
4305                         } else {
4306                                 $rv = "";
4307                         }
4308                 } elseif( $mode == "replace" ) {
4309                         if( $section == 0 ) {
4310                                 $rv = $newtext . "\n\n";
4311                                 $remainder = true;
4312                         } else {
4313                                 $rv = $secs[0];
4314                                 $remainder = false;
4315                         }
4316                 }
4317                 $count = 0;
4318                 $sectionLevel = 0;
4319                 for( $index = 1; $index < count( $secs ); ) {
4320                         $headerLine = $secs[$index++];
4321                         if( $secs[$index] ) {
4322                                 // A wiki header
4323                                 $headerLevel = strlen( $secs[$index++] );
4324                         } else {
4325                                 // An HTML header
4326                                 $index++;
4327                                 $headerLevel = intval( $secs[$index++] );
4328                         }
4329                         $content = $secs[$index++];
4330
4331                         $count++;
4332                         if( $mode == "get" ) {
4333                                 if( $count == $section ) {
4334                                         $rv = $headerLine . $content;
4335                                         $sectionLevel = $headerLevel;
4336                                 } elseif( $count > $section ) {
4337                                         if( $sectionLevel && $headerLevel > $sectionLevel ) {
4338                                                 $rv .= $headerLine . $content;
4339                                         } else {
4340                                                 // Broke out to a higher-level section
4341                                                 break;
4342                                         }
4343                                 }
4344                         } elseif( $mode == "replace" ) {
4345                                 if( $count < $section ) {
4346                                         $rv .= $headerLine . $content;
4347                                 } elseif( $count == $section ) {
4348                                         $rv .= $newtext . "\n\n";
4349                                         $sectionLevel = $headerLevel;
4350                                 } elseif( $count > $section ) {
4351                                         if( $headerLevel <= $sectionLevel ) {
4352                                                 // Passed the section's sub-parts.
4353                                                 $remainder = true;
4354                                         }
4355                                         if( $remainder ) {
4356                                                 $rv .= $headerLine . $content;
4357                                         }
4358                                 }
4359                         }
4360                 }
4361                 # reinsert stripped tags
4362                 $rv = $this->unstrip( $rv, $striparray );
4363                 $rv = $this->unstripNoWiki( $rv, $striparray );
4364                 $rv = trim( $rv );
4365                 return $rv;
4366         }
4367
4368         /**
4369          * This function returns the text of a section, specified by a number ($section).
4370          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4371          * the first section before any such heading (section 0).
4372          *
4373          * If a section contains subsections, these are also returned.
4374          *
4375          * @param $text String: text to look in
4376          * @param $section Integer: section number
4377          * @return string text of the requested section
4378          */
4379         function getSection( $text, $section ) {
4380                 return $this->extractSections( $text, $section, "get" );
4381         }
4382
4383         function replaceSection( $oldtext, $section, $text ) {
4384                 return $this->extractSections( $oldtext, $section, "replace", $text );
4385         }
4386
4387 }
4388
4389 /**
4390  * @todo document
4391  * @package MediaWiki
4392  */
4393 class ParserOutput
4394 {
4395         var $mText,             # The output text
4396                 $mLanguageLinks,    # List of the full text of language links, in the order they appear
4397                 $mCategories,       # Map of category names to sort keys
4398                 $mContainsOldMagic, # Boolean variable indicating if the input contained variables like {{CURRENTDAY}}
4399                 $mCacheTime,        # Time when this object was generated, or -1 for uncacheable. Used in ParserCache.
4400                 $mVersion,          # Compatibility check
4401                 $mTitleText,        # title text of the chosen language variant
4402                 $mLinks,            # 2-D map of NS/DBK to ID for the links in the document. ID=zero for broken.
4403                 $mTemplates,        # 2-D map of NS/DBK to ID for the template references. ID=zero for broken.
4404                 $mImages,           # DB keys of the images used, in the array key only
4405                 $mExternalLinks,    # External link URLs, in the key only
4406                 $mHTMLtitle,            # Display HTML title
4407                 $mSubtitle,                     # Additional subtitle
4408                 $mNewSection,           # Show a new section link?
4409                 $mNoGallery;            # No gallery on category page? (__NOGALLERY__)
4410
4411         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
4412                 $containsOldMagic = false, $titletext = '' )
4413         {
4414                 $this->mText = $text;
4415                 $this->mLanguageLinks = $languageLinks;
4416                 $this->mCategories = $categoryLinks;
4417                 $this->mContainsOldMagic = $containsOldMagic;
4418                 $this->mCacheTime = '';
4419                 $this->mVersion = MW_PARSER_VERSION;
4420                 $this->mTitleText = $titletext;
4421                 $this->mLinks = array();
4422                 $this->mTemplates = array();
4423                 $this->mImages = array();
4424                 $this->mExternalLinks = array();
4425                 $this->mHTMLtitle = "" ;
4426                 $this->mSubtitle = "" ;
4427                 $this->mNewSection = false;
4428                 $this->mNoGallery = false;
4429         }
4430
4431         function getText()                   { return $this->mText; }
4432         function &getLanguageLinks()          { return $this->mLanguageLinks; }
4433         function getCategoryLinks()          { return array_keys( $this->mCategories ); }
4434         function &getCategories()            { return $this->mCategories; }
4435         function getCacheTime()              { return $this->mCacheTime; }
4436         function getTitleText()              { return $this->mTitleText; }
4437         function &getLinks()                 { return $this->mLinks; }
4438         function &getTemplates()             { return $this->mTemplates; }
4439         function &getImages()                { return $this->mImages; }
4440         function &getExternalLinks()         { return $this->mExternalLinks; }
4441         function getNoGallery()              { return $this->mNoGallery; }
4442
4443         function containsOldMagic()          { return $this->mContainsOldMagic; }
4444         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
4445         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
4446         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategories, $cl ); }
4447         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
4448         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
4449         function setTitleText( $t )          { return wfSetVar ($this->mTitleText, $t); }
4450
4451         function addCategory( $c, $sort )    { $this->mCategories[$c] = $sort; }
4452         function addImage( $name )           { $this->mImages[$name] = 1; }
4453         function addLanguageLink( $t )       { $this->mLanguageLinks[] = $t; }
4454         function addExternalLink( $url )     { $this->mExternalLinks[$url] = 1; }
4455
4456         function setNewSection( $value ) {
4457                 $this->mNewSection = (bool)$value;
4458         }
4459         function getNewSection() {
4460                 return (bool)$this->mNewSection;
4461         }
4462
4463         function addLink( $title, $id ) {
4464                 $ns = $title->getNamespace();
4465                 $dbk = $title->getDBkey();
4466                 if ( !isset( $this->mLinks[$ns] ) ) {
4467                         $this->mLinks[$ns] = array();
4468                 }
4469                 $this->mLinks[$ns][$dbk] = $id;
4470         }
4471
4472         function addTemplate( $title, $id ) {
4473                 $ns = $title->getNamespace();
4474                 $dbk = $title->getDBkey();
4475                 if ( !isset( $this->mTemplates[$ns] ) ) {
4476                         $this->mTemplates[$ns] = array();
4477                 }
4478                 $this->mTemplates[$ns][$dbk] = $id;
4479         }
4480
4481         /**
4482          * Return true if this cached output object predates the global or
4483          * per-article cache invalidation timestamps, or if it comes from
4484          * an incompatible older version.
4485          *
4486          * @param string $touched the affected article's last touched timestamp
4487          * @return bool
4488          * @public
4489          */
4490         function expired( $touched ) {
4491                 global $wgCacheEpoch;
4492                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
4493                        $this->getCacheTime() < $touched ||
4494                        $this->getCacheTime() <= $wgCacheEpoch ||
4495                        !isset( $this->mVersion ) ||
4496                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
4497         }
4498 }
4499
4500 /**
4501  * Set options of the Parser
4502  * @todo document
4503  * @package MediaWiki
4504  */
4505 class ParserOptions
4506 {
4507         # All variables are private
4508         var $mUseTeX;                    # Use texvc to expand <math> tags
4509         var $mUseDynamicDates;           # Use DateFormatter to format dates
4510         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
4511         var $mAllowExternalImages;       # Allow external images inline
4512         var $mAllowExternalImagesFrom;   # If not, any exception?
4513         var $mSkin;                      # Reference to the preferred skin
4514         var $mDateFormat;                # Date format index
4515         var $mEditSection;               # Create "edit section" links
4516         var $mNumberHeadings;            # Automatically number headings
4517         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
4518         var $mTidy;                      # Ask for tidy cleanup
4519         var $mInterfaceMessage;          # Which lang to call for PLURAL and GRAMMAR
4520
4521         var $mUser;                      # Stored user object, just used to initialise the skin
4522
4523         function getUseTeX()                        { return $this->mUseTeX; }
4524         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
4525         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
4526         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
4527         function getAllowExternalImagesFrom()       { return $this->mAllowExternalImagesFrom; }
4528         function getDateFormat()                    { return $this->mDateFormat; }
4529         function getEditSection()                   { return $this->mEditSection; }
4530         function getNumberHeadings()                { return $this->mNumberHeadings; }
4531         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
4532         function getTidy()                          { return $this->mTidy; }
4533         function getInterfaceMessage()              { return $this->mInterfaceMessage; }
4534
4535         function &getSkin() {
4536                 if ( !isset( $this->mSkin ) ) {
4537                         $this->mSkin = $this->mUser->getSkin();
4538                 }
4539                 return $this->mSkin;
4540         }
4541
4542         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
4543         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
4544         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
4545         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
4546         function setAllowExternalImagesFrom( $x )   { return wfSetVar( $this->mAllowExternalImagesFrom, $x ); }
4547         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
4548         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
4549         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
4550         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
4551         function setTidy( $x )                      { return wfSetVar( $this->mTidy, $x); }
4552         function setSkin( &$x ) { $this->mSkin =& $x; }
4553         function setInterfaceMessage( $x )          { return wfSetVar( $this->mInterfaceMessage, $x); }
4554
4555         function ParserOptions( $user = null ) {
4556                 $this->initialiseFromUser( $user );
4557         }
4558
4559         /**
4560          * Get parser options
4561          * @static
4562          */
4563         static function newFromUser( &$user ) {
4564                 return new ParserOptions( $user );
4565         }
4566
4567         /** Get user options */
4568         function initialiseFromUser( &$userInput ) {
4569                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages;
4570                 global $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion;
4571                 $fname = 'ParserOptions::initialiseFromUser';
4572                 wfProfileIn( $fname );
4573                 if ( !$userInput ) {
4574                         global $wgUser;
4575                         if ( isset( $wgUser ) ) {
4576                                 $user = $wgUser;
4577                         } else {
4578                                 $user = new User;
4579                                 $user->setLoaded( true );
4580                         }
4581                 } else {
4582                         $user =& $userInput;
4583                 }
4584
4585                 $this->mUser = $user;
4586
4587                 $this->mUseTeX = $wgUseTeX;
4588                 $this->mUseDynamicDates = $wgUseDynamicDates;
4589                 $this->mInterwikiMagic = $wgInterwikiMagic;
4590                 $this->mAllowExternalImages = $wgAllowExternalImages;
4591                 $this->mAllowExternalImagesFrom = $wgAllowExternalImagesFrom;
4592                 $this->mSkin = null; # Deferred
4593                 $this->mDateFormat = $user->getOption( 'date' );
4594                 $this->mEditSection = true;
4595                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
4596                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
4597                 $this->mTidy = false;
4598                 $this->mInterfaceMessage = false;
4599                 wfProfileOut( $fname );
4600         }
4601 }
4602
4603 /**
4604  * Callback function used by Parser::replaceLinkHolders()
4605  * to substitute link placeholders.
4606  */
4607 function &wfOutputReplaceMatches( $matches ) {
4608         global $wgOutputReplace;
4609         return $wgOutputReplace[$matches[1]];
4610 }
4611
4612 /**
4613  * Return the total number of articles
4614  */
4615 function wfNumberOfArticles() {
4616         global $wgNumberOfArticles;
4617
4618         wfLoadSiteStats();
4619         return $wgNumberOfArticles;
4620 }
4621
4622 /**
4623  * Return the number of files
4624  */
4625 function wfNumberOfFiles() {
4626         $fname = 'wfNumberOfFiles';
4627
4628         wfProfileIn( $fname );
4629         $dbr =& wfGetDB( DB_SLAVE );
4630         $numImages = $dbr->selectField('site_stats', 'ss_images', array(), $fname );
4631         wfProfileOut( $fname );
4632
4633         return $numImages;
4634 }
4635
4636 /**
4637  * Return the number of user accounts
4638  * @return integer
4639  */
4640 function wfNumberOfUsers() {
4641         wfProfileIn( 'wfNumberOfUsers' );
4642         $dbr =& wfGetDB( DB_SLAVE );
4643         $count = $dbr->selectField( 'site_stats', 'ss_users', array(), 'wfNumberOfUsers' );
4644         wfProfileOut( 'wfNumberOfUsers' );
4645         return (int)$count;
4646 }
4647
4648 /**
4649  * Return the total number of pages
4650  * @return integer
4651  */
4652 function wfNumberOfPages() {
4653         wfProfileIn( 'wfNumberOfPages' );
4654         $dbr =& wfGetDB( DB_SLAVE );
4655         $count = $dbr->selectField( 'site_stats', 'ss_total_pages', array(), 'wfNumberOfPages' );
4656         wfProfileOut( 'wfNumberOfPages' );
4657         return (int)$count;
4658 }
4659
4660 /**
4661  * Return the total number of admins
4662  *
4663  * @return integer
4664  */
4665 function wfNumberOfAdmins() {
4666         static $admins = -1;
4667         wfProfileIn( 'wfNumberOfAdmins' );
4668         if( $admins == -1 ) {
4669                 $dbr =& wfGetDB( DB_SLAVE );
4670                 $admins = $dbr->selectField( 'user_groups', 'COUNT(*)', array( 'ug_group' => 'sysop' ), 'wfNumberOfAdmins' );
4671         }
4672         wfProfileOut( 'wfNumberOfAdmins' );
4673         return (int)$admins;
4674 }
4675
4676 /**
4677  * Count the number of pages in a particular namespace
4678  *
4679  * @param $ns Namespace
4680  * @return integer
4681  */
4682 function wfPagesInNs( $ns ) {
4683         static $pageCount = array();
4684         wfProfileIn( 'wfPagesInNs' );
4685         if( !isset( $pageCount[$ns] ) ) {
4686                 $dbr =& wfGetDB( DB_SLAVE );
4687                 $pageCount[$ns] = $dbr->selectField( 'page', 'COUNT(*)', array( 'page_namespace' => $ns ), 'wfPagesInNs' );
4688         }
4689         wfProfileOut( 'wfPagesInNs' );
4690         return (int)$pageCount[$ns];
4691 }
4692
4693 /**
4694  * Get various statistics from the database
4695  * @private
4696  */
4697 function wfLoadSiteStats() {
4698         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
4699         $fname = 'wfLoadSiteStats';
4700
4701         if ( -1 != $wgNumberOfArticles ) return;
4702         $dbr =& wfGetDB( DB_SLAVE );
4703         $s = $dbr->selectRow( 'site_stats',
4704                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
4705                 array( 'ss_row_id' => 1 ), $fname
4706         );
4707
4708         if ( $s === false ) {
4709                 return;
4710         } else {
4711                 $wgTotalViews = $s->ss_total_views;
4712                 $wgTotalEdits = $s->ss_total_edits;
4713                 $wgNumberOfArticles = $s->ss_good_articles;
4714         }
4715 }
4716
4717 /**
4718  * Escape html tags
4719  * Basically replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
4720  *
4721  * @param $in String: text that might contain HTML tags.
4722  * @return string Escaped string
4723  */
4724 function wfEscapeHTMLTagsOnly( $in ) {
4725         return str_replace(
4726                 array( '"', '>', '<' ),
4727                 array( '&quot;', '&gt;', '&lt;' ),
4728                 $in );
4729 }
4730
4731 ?>