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                 $this->mRevisionId = $revid;
 272                 $this->mOutputType = OT_HTML;
 273
 274                 //$text = $this->strip( $text, $this->mStripState );
 275                 // VOODOO MAGIC FIX! Sometimes the above segfaults in PHP5.
 276                 $x =& $this->mStripState;
 277
 278                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 279                 $text = $this->strip( $text, $x );
 280                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 281
 282                 # Hook to suspend the parser in this state
 283                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$x ) ) ) {
 284                         wfProfileOut( $fname );
 285                         return $text ;
 286                 }
 287
 288                 $text = $this->internalParse( $text );
 289
 290                 $text = $this->unstrip( $text, $this->mStripState );
 291
 292                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 293                 $fixtags = array(
 294                         # french spaces, last one Guillemet-left
 295                         # only if there is something before the space
 296                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 297                         # french spaces, Guillemet-right
 298                         '/(\\302\\253) /' => '\\1&nbsp;',
 299                 );
 300                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 301
 302                 # only once and last
 303                 $text = $this->doBlockLevels( $text, $linestart );
 304
 305                 $this->replaceLinkHolders( $text );
 306
 307                 # the position of the parserConvert() call should not be changed. it
 308                 # assumes that the links are all replaced and the only thing left
 309                 # is the <nowiki> mark.
 310                 # Side-effects: this calls $this->mOutput->setTitleText()
 311                 $text = $wgContLang->parserConvert( $text, $this );
 312
 313                 $text = $this->unstripNoWiki( $text, $this->mStripState );
 314
 315                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 316
 317                 $text = Sanitizer::normalizeCharReferences( $text );
 318
 319                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 320                         $text = Parser::tidy($text);
 321                 } else {
 322                         # attempt to sanitize at least some nesting problems
 323                         # (bug #2702 and quite a few others)
 324                         $tidyregs = array(
 325                                 # ''Something [http://www.cool.com cool''] -->
 326                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 327                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 328                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 329                                 # fix up an anchor inside another anchor, only
 330                                 # at least for a single single nested link (bug 3695)
 331                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 332                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 333                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 334                                 # contains a div, so fix it up here; replace
 335                                 # div with escaped text
 336                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 337                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 338                                 # remove empty italic or bold tag pairs, some
 339                                 # introduced by rules above
 340                                 '/<([bi])><\/\\1>/' => ''
 341                         );
 342
 343                         $text = preg_replace(
 344                                 array_keys( $tidyregs ),
 345                                 array_values( $tidyregs ),
 346                                 $text );
 347                 }
 348
 349                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 350
 351                 $this->mOutput->setText( $text );
 352                 wfProfileOut( $fname );
 353
 354                 return $this->mOutput;
 355         }
 356
 357         /**
 358          * Get a random string
 359          *
 360          * @private
 361          * @static
 362          */
 363         function getRandomString() {
 364                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 365         }
 366
 367         function &getTitle() { return $this->mTitle; }
 368         function getOptions() { return $this->mOptions; }
 369
 370         function getFunctionLang() {
 371                 global $wgLang, $wgContLang;
 372                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 373         }
 374
 375         /**
 376          * Replaces all occurrences of HTML-style comments and the given tags
 377          * in the text with a random marker and returns teh next text. The output
 378          * parameter $matches will be an associative array filled with data in
 379          * the form:
 380          *   'UNIQ-xxxxx' => array(
 381          *     'element',
 382          *     'tag content',
 383          *     array( 'param' => 'x' ),
 384          *     '<element param="x">tag content</element>' ) )
 385          *
 386          * @param $elements list of element names. Comments are always extracted.
 387          * @param $text Source text string.
 388          * @param $uniq_prefix
 389          *
 390          * @private
 391          * @static
 392          */
 393         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 394                 $rand = Parser::getRandomString();
 395                 $n = 1;
 396                 $stripped = '';
 397                 $matches = array();
 398
 399                 $taglist = implode( '|', $elements );
 400                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 401
 402                 while ( '' != $text ) {
 403                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 404                         $stripped .= $p[0];
 405                         if( count( $p ) < 5 ) {
 406                                 break;
 407                         }
 408                         if( count( $p ) > 5 ) {
 409                                 // comment
 410                                 $element    = $p[4];
 411                                 $attributes = '';
 412                                 $close      = '';
 413                                 $inside     = $p[5];
 414                         } else {
 415                                 // tag
 416                                 $element    = $p[1];
 417                                 $attributes = $p[2];
 418                                 $close      = $p[3];
 419                                 $inside     = $p[4];
 420                         }
 421
 422                         $marker = "$uniq_prefix-$element-$rand" . sprintf('%08X', $n++) . '-QINU';
 423                         $stripped .= $marker;
 424
 425                         if ( $close === '/>' ) {
 426                                 // Empty element tag, <tag />
 427                                 $content = null;
 428                                 $text = $inside;
 429                                 $tail = null;
 430                         } else {
 431                                 if( $element == '!--' ) {
 432                                         $end = '/(-->)/';
 433                                 } else {
 434                                         $end = "/(<\\/$element\\s*>)/i";
 435                                 }
 436                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 437                                 $content = $q[0];
 438                                 if( count( $q ) < 3 ) {
 439                                         # No end tag -- let it run out to the end of the text.
 440                                         $tail = '';
 441                                         $text = '';
 442                                 } else {
 443                                         $tail = $q[1];
 444                                         $text = $q[2];
 445                                 }
 446                         }
 447
 448                         $matches[$marker] = array( $element,
 449                                 $content,
 450                                 Sanitizer::decodeTagAttributes( $attributes ),
 451                                 "<$element$attributes$close$content$tail" );
 452                 }
 453                 return $stripped;
 454         }
 455
 456         /**
 457          * Strips and renders nowiki, pre, math, hiero
 458          * If $render is set, performs necessary rendering operations on plugins
 459          * Returns the text, and fills an array with data needed in unstrip()
 460          * If the $state is already a valid strip state, it adds to the state
 461          *
 462          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 463          *  will be stripped in addition to other tags. This is important
 464          *  for section editing, where these comments cause confusion when
 465          *  counting the sections in the wikisource
 466          *
 467          * @param array dontstrip contains tags which should not be stripped;
 468          *  used to prevent stipping of <gallery> when saving (fixes bug 2700)
 469          *
 470          * @private
 471          */
 472         function strip( $text, &$state, $stripcomments = false , $dontstrip = array () ) {
 473                 $render = ($this->mOutputType == OT_HTML);
 474
 475                 # Replace any instances of the placeholders
 476                 $uniq_prefix = $this->mUniqPrefix;
 477                 #$text = str_replace( $uniq_prefix, wfHtmlEscapeFirst( $uniq_prefix ), $text );
 478                 $commentState = array();
 479
 480                 $elements = array_merge(
 481                         array( 'nowiki', 'gallery' ),
 482                         array_keys( $this->mTagHooks ) );
 483                 global $wgRawHtml;
 484                 if( $wgRawHtml ) {
 485                         $elements[] = 'html';
 486                 }
 487                 if( $this->mOptions->getUseTeX() ) {
 488                         $elements[] = 'math';
 489                 }
 490
 491                 # Removing $dontstrip tags from $elements list (currently only 'gallery', fixing bug 2700)
 492                 foreach ( $elements AS $k => $v ) {
 493                         if ( !in_array ( $v , $dontstrip ) ) continue;
 494                         unset ( $elements[$k] );
 495                 }
 496
 497                 $matches = array();
 498                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 499
 500                 foreach( $matches as $marker => $data ) {
 501                         list( $element, $content, $params, $tag ) = $data;
 502                         if( $render ) {
 503                                 $tagName = strtolower( $element );
 504                                 switch( $tagName ) {
 505                                 case '!--':
 506                                         // Comment
 507                                         if( substr( $tag, -3 ) == '-->' ) {
 508                                                 $output = $tag;
 509                                         } else {
 510                                                 // Unclosed comment in input.
 511                                                 // Close it so later stripping can remove it
 512                                                 $output = "$tag-->";
 513                                         }
 514                                         break;
 515                                 case 'html':
 516                                         if( $wgRawHtml ) {
 517                                                 $output = $content;
 518                                                 break;
 519                                         }
 520                                         // Shouldn't happen otherwise. :)
 521                                 case 'nowiki':
 522                                         $output = wfEscapeHTMLTagsOnly( $content );
 523                                         break;
 524                                 case 'math':
 525                                         $output = MathRenderer::renderMath( $content );
 526                                         break;
 527                                 case 'gallery':
 528                                         $output = $this->renderImageGallery( $content, $params );
 529                                         break;
 530                                 default:
 531                                         if( isset( $this->mTagHooks[$tagName] ) ) {
 532                                                 $output = call_user_func_array( $this->mTagHooks[$tagName],
 533                                                         array( $content, $params, $this ) );
 534                                         } else {
 535                                                 throw new MWException( "Invalid call hook $element" );
 536                                         }
 537                                 }
 538                         } else {
 539                                 // Just stripping tags; keep the source
 540                                 $output = $tag;
 541                         }
 542                         if( !$stripcomments && $element == '!--' ) {
 543                                 $commentState[$marker] = $output;
 544                         } else {
 545                                 $state[$element][$marker] = $output;
 546                         }
 547                 }
 548
 549                 # Unstrip comments unless explicitly told otherwise.
 550                 # (The comments are always stripped prior to this point, so as to
 551                 # not invoke any extension tags / parser hooks contained within
 552                 # a comment.)
 553                 if ( !$stripcomments ) {
 554                         // Put them all back and forget them
 555                         $text = strtr( $text, $commentState );
 556                 }
 557
 558                 return $text;
 559         }
 560
 561         /**
 562          * Restores pre, math, and other extensions removed by strip()
 563          *
 564          * always call unstripNoWiki() after this one
 565          * @private
 566          */
 567         function unstrip( $text, &$state ) {
 568                 if ( !is_array( $state ) ) {
 569                         return $text;
 570                 }
 571
 572                 $replacements = array();
 573                 foreach( $state as $tag => $contentDict ) {
 574                         if( $tag != 'nowiki' && $tag != 'html' ) {
 575                                 foreach( $contentDict as $uniq => $content ) {
 576                                         $replacements[$uniq] = $content;
 577                                 }
 578                         }
 579                 }
 580                 $text = strtr( $text, $replacements );
 581
 582                 return $text;
 583         }
 584
 585         /**
 586          * Always call this after unstrip() to preserve the order
 587          *
 588          * @private
 589          */
 590         function unstripNoWiki( $text, &$state ) {
 591                 if ( !is_array( $state ) ) {
 592                         return $text;
 593                 }
 594
 595                 $replacements = array();
 596                 foreach( $state as $tag => $contentDict ) {
 597                         if( $tag == 'nowiki' || $tag == 'html' ) {
 598                                 foreach( $contentDict as $uniq => $content ) {
 599                                         $replacements[$uniq] = $content;
 600                                 }
 601                         }
 602                 }
 603                 $text = strtr( $text, $replacements );
 604
 605                 return $text;
 606         }
 607
 608         /**
 609          * Add an item to the strip state
 610          * Returns the unique tag which must be inserted into the stripped text
 611          * The tag will be replaced with the original text in unstrip()
 612          *
 613          * @private
 614          */
 615         function insertStripItem( $text, &$state ) {
 616                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 617                 if ( !$state ) {
 618                         $state = array();
 619                 }
 620                 $state['item'][$rnd] = $text;
 621                 return $rnd;
 622         }
 623
 624         /**
 625          * Interface with html tidy, used if $wgUseTidy = true.
 626          * If tidy isn't able to correct the markup, the original will be
 627          * returned in all its glory with a warning comment appended.
 628          *
 629          * Either the external tidy program or the in-process tidy extension
 630          * will be used depending on availability. Override the default
 631          * $wgTidyInternal setting to disable the internal if it's not working.
 632          *
 633          * @param string $text Hideous HTML input
 634          * @return string Corrected HTML output
 635          * @public
 636          * @static
 637          */
 638         function tidy( $text ) {
 639                 global $wgTidyInternal;
 640                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 641 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 642 '<head><title>test</title></head><body>'.$text.'</body></html>';
 643                 if( $wgTidyInternal ) {
 644                         $correctedtext = Parser::internalTidy( $wrappedtext );
 645                 } else {
 646                         $correctedtext = Parser::externalTidy( $wrappedtext );
 647                 }
 648                 if( is_null( $correctedtext ) ) {
 649                         wfDebug( "Tidy error detected!\n" );
 650                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 651                 }
 652                 return $correctedtext;
 653         }
 654
 655         /**
 656          * Spawn an external HTML tidy process and get corrected markup back from it.
 657          *
 658          * @private
 659          * @static
 660          */
 661         function externalTidy( $text ) {
 662                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 663                 $fname = 'Parser::externalTidy';
 664                 wfProfileIn( $fname );
 665
 666                 $cleansource = '';
 667                 $opts = ' -utf8';
 668
 669                 $descriptorspec = array(
 670                         0 => array('pipe', 'r'),
 671                         1 => array('pipe', 'w'),
 672                         2 => array('file', '/dev/null', 'a')
 673                 );
 674                 $pipes = array();
 675                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 676                 if (is_resource($process)) {
 677                         // Theoretically, this style of communication could cause a deadlock
 678                         // here. If the stdout buffer fills up, then writes to stdin could
 679                         // block. This doesn't appear to happen with tidy, because tidy only
 680                         // writes to stdout after it's finished reading from stdin. Search
 681                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 682                         fwrite($pipes[0], $text);
 683                         fclose($pipes[0]);
 684                         while (!feof($pipes[1])) {
 685                                 $cleansource .= fgets($pipes[1], 1024);
 686                         }
 687                         fclose($pipes[1]);
 688                         proc_close($process);
 689                 }
 690
 691                 wfProfileOut( $fname );
 692
 693                 if( $cleansource == '' && $text != '') {
 694                         // Some kind of error happened, so we couldn't get the corrected text.
 695                         // Just give up; we'll use the source text and append a warning.
 696                         return null;
 697                 } else {
 698                         return $cleansource;
 699                 }
 700         }
 701
 702         /**
 703          * Use the HTML tidy PECL extension to use the tidy library in-process,
 704          * saving the overhead of spawning a new process. Currently written to
 705          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 706          *
 707          * 'pear install tidy' should be able to compile the extension module.
 708          *
 709          * @private
 710          * @static
 711          */
 712         function internalTidy( $text ) {
 713                 global $wgTidyConf;
 714                 $fname = 'Parser::internalTidy';
 715                 wfProfileIn( $fname );
 716
 717                 tidy_load_config( $wgTidyConf );
 718                 tidy_set_encoding( 'utf8' );
 719                 tidy_parse_string( $text );
 720                 tidy_clean_repair();
 721                 if( tidy_get_status() == 2 ) {
 722                         // 2 is magic number for fatal error
 723                         // http://www.php.net/manual/en/function.tidy-get-status.php
 724                         $cleansource = null;
 725                 } else {
 726                         $cleansource = tidy_get_output();
 727                 }
 728                 wfProfileOut( $fname );
 729                 return $cleansource;
 730         }
 731
 732         /**
 733          * parse the wiki syntax used to render tables
 734          *
 735          * @private
 736          */
 737         function doTableStuff ( $t ) {
 738                 $fname = 'Parser::doTableStuff';
 739                 wfProfileIn( $fname );
 740
 741                 $t = explode ( "\n" , $t ) ;
 742                 $td = array () ; # Is currently a td tag open?
 743                 $ltd = array () ; # Was it TD or TH?
 744                 $tr = array () ; # Is currently a tr tag open?
 745                 $ltr = array () ; # tr attributes
 746                 $has_opened_tr = array(); # Did this table open a <tr> element?
 747                 $indent_level = 0; # indent level of the table
 748                 foreach ( $t AS $k => $x )
 749                 {
 750                         $x = trim ( $x ) ;
 751                         $fc = substr ( $x , 0 , 1 ) ;
 752                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 753                                 $indent_level = strlen( $matches[1] );
 754
 755                                 $attributes = $this->unstripForHTML( $matches[2] );
 756
 757                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 758                                         '<table' . Sanitizer::fixTagAttributes ( $attributes, 'table' ) . '>' ;
 759                                 array_push ( $td , false ) ;
 760                                 array_push ( $ltd , '' ) ;
 761                                 array_push ( $tr , false ) ;
 762                                 array_push ( $ltr , '' ) ;
 763                                 array_push ( $has_opened_tr, false );
 764                         }
 765                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 766                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 767                                 $z = "</table>" . substr ( $x , 2);
 768                                 $l = array_pop ( $ltd ) ;
 769                                 if ( !array_pop ( $has_opened_tr ) ) $z = "<tr><td></td></tr>" . $z ;
 770                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 771                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 772                                 array_pop ( $ltr ) ;
 773                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 774                         }
 775                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 776                                 $x = substr ( $x , 1 ) ;
 777                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 778                                 $z = '' ;
 779                                 $l = array_pop ( $ltd ) ;
 780                                 array_pop ( $has_opened_tr );
 781                                 array_push ( $has_opened_tr , true ) ;
 782                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 783                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 784                                 array_pop ( $ltr ) ;
 785                                 $t[$k] = $z ;
 786                                 array_push ( $tr , false ) ;
 787                                 array_push ( $td , false ) ;
 788                                 array_push ( $ltd , '' ) ;
 789                                 $attributes = $this->unstripForHTML( $x );
 790                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $attributes, 'tr' ) ) ;
 791                         }
 792                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 793                                 # $x is a table row
 794                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 795                                         $fc = '+' ;
 796                                         $x = substr ( $x , 1 ) ;
 797                                 }
 798                                 $after = substr ( $x , 1 ) ;
 799                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 800
 801                                 // Split up multiple cells on the same line.
 802                                 // FIXME: This can result in improper nesting of tags processed
 803                                 // by earlier parser steps, but should avoid splitting up eg
 804                                 // attribute values containing literal "||".
 805                                 $after = wfExplodeMarkup( '||', $after );
 806
 807                                 $t[$k] = '' ;
 808
 809                                 # Loop through each table cell
 810                                 foreach ( $after AS $theline )
 811                                 {
 812                                         $z = '' ;
 813                                         if ( $fc != '+' )
 814                                         {
 815                                                 $tra = array_pop ( $ltr ) ;
 816                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 817                                                 array_push ( $tr , true ) ;
 818                                                 array_push ( $ltr , '' ) ;
 819                                                 array_pop ( $has_opened_tr );
 820                                                 array_push ( $has_opened_tr , true ) ;
 821                                         }
 822
 823                                         $l = array_pop ( $ltd ) ;
 824                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 825                                         if ( $fc == '|' ) $l = 'td' ;
 826                                         else if ( $fc == '!' ) $l = 'th' ;
 827                                         else if ( $fc == '+' ) $l = 'caption' ;
 828                                         else $l = '' ;
 829                                         array_push ( $ltd , $l ) ;
 830
 831                                         # Cell parameters
 832                                         $y = explode ( '|' , $theline , 2 ) ;
 833                                         # Note that a '|' inside an invalid link should not
 834                                         # be mistaken as delimiting cell parameters
 835                                         if ( strpos( $y[0], '[[' ) !== false ) {
 836                                                 $y = array ($theline);
 837                                         }
 838                                         if ( count ( $y ) == 1 )
 839                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 840                                         else {
 841                                                 $attributes = $this->unstripForHTML( $y[0] );
 842                                                 $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($attributes, $l).">{$y[1]}" ;
 843                                         }
 844                                         $t[$k] .= $y ;
 845                                         array_push ( $td , true ) ;
 846                                 }
 847                         }
 848                 }
 849
 850                 # Closing open td, tr && table
 851                 while ( count ( $td ) > 0 )
 852                 {
 853                         $l = array_pop ( $ltd ) ;
 854                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 855                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 856                         if ( !array_pop ( $has_opened_tr ) ) $t[] = "<tr><td></td></tr>" ;
 857                         $t[] = '</table>' ;
 858                 }
 859
 860                 $t = implode ( "\n" , $t ) ;
 861                 # special case: don't return empty table
 862                 if($t == "<table>\n<tr><td></td></tr>\n</table>")
 863                         $t = '';
 864                 wfProfileOut( $fname );
 865                 return $t ;
 866         }
 867
 868         /**
 869          * Helper function for parse() that transforms wiki markup into
 870          * HTML. Only called for $mOutputType == OT_HTML.
 871          *
 872          * @private
 873          */
 874         function internalParse( $text ) {
 875                 $args = array();
 876                 $isMain = true;
 877                 $fname = 'Parser::internalParse';
 878                 wfProfileIn( $fname );
 879
 880                 # Remove <noinclude> tags and <includeonly> sections
 881                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 882                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 883                 $text = preg_replace( '/<includeonly>.*?<\/includeonly>/s', '', $text );
 884
 885                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
 886
 887                 $text = $this->replaceVariables( $text, $args );
 888
 889                 // Tables need to come after variable replacement for things to work
 890                 // properly; putting them before other transformations should keep
 891                 // exciting things like link expansions from showing up in surprising
 892                 // places.
 893                 $text = $this->doTableStuff( $text );
 894
 895                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 896
 897                 $text = $this->stripToc( $text );
 898                 $this->stripNoGallery( $text );
 899                 $text = $this->doHeadings( $text );
 900                 if($this->mOptions->getUseDynamicDates()) {
 901                         $df =& DateFormatter::getInstance();
 902                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 903                 }
 904                 $text = $this->doAllQuotes( $text );
 905                 $text = $this->replaceInternalLinks( $text );
 906                 $text = $this->replaceExternalLinks( $text );
 907
 908                 # replaceInternalLinks may sometimes leave behind
 909                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 910                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 911
 912                 $text = $this->doMagicLinks( $text );
 913                 $text = $this->formatHeadings( $text, $isMain );
 914
 915                 wfProfileOut( $fname );
 916                 return $text;
 917         }
 918
 919         /**
 920          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 921          * magic external links.
 922          *
 923          * @private
 924          */
 925         function &doMagicLinks( &$text ) {
 926                 $text = $this->magicISBN( $text );
 927                 $text = $this->magicRFC( $text, 'RFC ', 'rfcurl' );
 928                 $text = $this->magicRFC( $text, 'PMID ', 'pubmedurl' );
 929                 return $text;
 930         }
 931
 932         /**
 933          * Parse headers and return html
 934          *
 935          * @private
 936          */
 937         function doHeadings( $text ) {
 938                 $fname = 'Parser::doHeadings';
 939                 wfProfileIn( $fname );
 940                 for ( $i = 6; $i >= 1; --$i ) {
 941                         $h = str_repeat( '=', $i );
 942                         $text = preg_replace( "/^{$h}(.+){$h}\\s*$/m",
 943                           "<h{$i}>\\1</h{$i}>\\2", $text );
 944                 }
 945                 wfProfileOut( $fname );
 946                 return $text;
 947         }
 948
 949         /**
 950          * Replace single quotes with HTML markup
 951          * @private
 952          * @return string the altered text
 953          */
 954         function doAllQuotes( $text ) {
 955                 $fname = 'Parser::doAllQuotes';
 956                 wfProfileIn( $fname );
 957                 $outtext = '';
 958                 $lines = explode( "\n", $text );
 959                 foreach ( $lines as $line ) {
 960                         $outtext .= $this->doQuotes ( $line ) . "\n";
 961                 }
 962                 $outtext = substr($outtext, 0,-1);
 963                 wfProfileOut( $fname );
 964                 return $outtext;
 965         }
 966
 967         /**
 968          * Helper function for doAllQuotes()
 969          * @private
 970          */
 971         function doQuotes( $text ) {
 972                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
 973                 if ( count( $arr ) == 1 )
 974                         return $text;
 975                 else
 976                 {
 977                         # First, do some preliminary work. This may shift some apostrophes from
 978                         # being mark-up to being text. It also counts the number of occurrences
 979                         # of bold and italics mark-ups.
 980                         $i = 0;
 981                         $numbold = 0;
 982                         $numitalics = 0;
 983                         foreach ( $arr as $r )
 984                         {
 985                                 if ( ( $i % 2 ) == 1 )
 986                                 {
 987                                         # If there are ever four apostrophes, assume the first is supposed to
 988                                         # be text, and the remaining three constitute mark-up for bold text.
 989                                         if ( strlen( $arr[$i] ) == 4 )
 990                                         {
 991                                                 $arr[$i-1] .= "'";
 992                                                 $arr[$i] = "'''";
 993                                         }
 994                                         # If there are more than 5 apostrophes in a row, assume they're all
 995                                         # text except for the last 5.
 996                                         else if ( strlen( $arr[$i] ) > 5 )
 997                                         {
 998                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
 999                                                 $arr[$i] = "'''''";
1000                                         }
1001                                         # Count the number of occurrences of bold and italics mark-ups.
1002                                         # We are not counting sequences of five apostrophes.
1003                                         if ( strlen( $arr[$i] ) == 2 ) $numitalics++;  else
1004                                         if ( strlen( $arr[$i] ) == 3 ) $numbold++;     else
1005                                         if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1006                                 }
1007                                 $i++;
1008                         }
1009
1010                         # If there is an odd number of both bold and italics, it is likely
1011                         # that one of the bold ones was meant to be an apostrophe followed
1012                         # by italics. Which one we cannot know for certain, but it is more
1013                         # likely to be one that has a single-letter word before it.
1014                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1015                         {
1016                                 $i = 0;
1017                                 $firstsingleletterword = -1;
1018                                 $firstmultiletterword = -1;
1019                                 $firstspace = -1;
1020                                 foreach ( $arr as $r )
1021                                 {
1022                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1023                                         {
1024                                                 $x1 = substr ($arr[$i-1], -1);
1025                                                 $x2 = substr ($arr[$i-1], -2, 1);
1026                                                 if ($x1 == ' ') {
1027                                                         if ($firstspace == -1) $firstspace = $i;
1028                                                 } else if ($x2 == ' ') {
1029                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1030                                                 } else {
1031                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1032                                                 }
1033                                         }
1034                                         $i++;
1035                                 }
1036
1037                                 # If there is a single-letter word, use it!
1038                                 if ($firstsingleletterword > -1)
1039                                 {
1040                                         $arr [ $firstsingleletterword ] = "''";
1041                                         $arr [ $firstsingleletterword-1 ] .= "'";
1042                                 }
1043                                 # If not, but there's a multi-letter word, use that one.
1044                                 else if ($firstmultiletterword > -1)
1045                                 {
1046                                         $arr [ $firstmultiletterword ] = "''";
1047                                         $arr [ $firstmultiletterword-1 ] .= "'";
1048                                 }
1049                                 # ... otherwise use the first one that has neither.
1050                                 # (notice that it is possible for all three to be -1 if, for example,
1051                                 # there is only one pentuple-apostrophe in the line)
1052                                 else if ($firstspace > -1)
1053                                 {
1054                                         $arr [ $firstspace ] = "''";
1055                                         $arr [ $firstspace-1 ] .= "'";
1056                                 }
1057                         }
1058
1059                         # Now let's actually convert our apostrophic mush to HTML!
1060                         $output = '';
1061                         $buffer = '';
1062                         $state = '';
1063                         $i = 0;
1064                         foreach ($arr as $r)
1065                         {
1066                                 if (($i % 2) == 0)
1067                                 {
1068                                         if ($state == 'both')
1069                                                 $buffer .= $r;
1070                                         else
1071                                                 $output .= $r;
1072                                 }
1073                                 else
1074                                 {
1075                                         if (strlen ($r) == 2)
1076                                         {
1077                                                 if ($state == 'i')
1078                                                 { $output .= '</i>'; $state = ''; }
1079                                                 else if ($state == 'bi')
1080                                                 { $output .= '</i>'; $state = 'b'; }
1081                                                 else if ($state == 'ib')
1082                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1083                                                 else if ($state == 'both')
1084                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1085                                                 else # $state can be 'b' or ''
1086                                                 { $output .= '<i>'; $state .= 'i'; }
1087                                         }
1088                                         else if (strlen ($r) == 3)
1089                                         {
1090                                                 if ($state == 'b')
1091                                                 { $output .= '</b>'; $state = ''; }
1092                                                 else if ($state == 'bi')
1093                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1094                                                 else if ($state == 'ib')
1095                                                 { $output .= '</b>'; $state = 'i'; }
1096                                                 else if ($state == 'both')
1097                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1098                                                 else # $state can be 'i' or ''
1099                                                 { $output .= '<b>'; $state .= 'b'; }
1100                                         }
1101                                         else if (strlen ($r) == 5)
1102                                         {
1103                                                 if ($state == 'b')
1104                                                 { $output .= '</b><i>'; $state = 'i'; }
1105                                                 else if ($state == 'i')
1106                                                 { $output .= '</i><b>'; $state = 'b'; }
1107                                                 else if ($state == 'bi')
1108                                                 { $output .= '</i></b>'; $state = ''; }
1109                                                 else if ($state == 'ib')
1110                                                 { $output .= '</b></i>'; $state = ''; }
1111                                                 else if ($state == 'both')
1112                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1113                                                 else # ($state == '')
1114                                                 { $buffer = ''; $state = 'both'; }
1115                                         }
1116                                 }
1117                                 $i++;
1118                         }
1119                         # Now close all remaining tags.  Notice that the order is important.
1120                         if ($state == 'b' || $state == 'ib')
1121                                 $output .= '</b>';
1122                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1123                                 $output .= '</i>';
1124                         if ($state == 'bi')
1125                                 $output .= '</b>';
1126                         if ($state == 'both')
1127                                 $output .= '<b><i>'.$buffer.'</i></b>';
1128                         return $output;
1129                 }
1130         }
1131
1132         /**
1133          * Replace external links
1134          *
1135          * Note: this is all very hackish and the order of execution matters a lot.
1136          * Make sure to run maintenance/parserTests.php if you change this code.
1137          *
1138          * @private
1139          */
1140         function replaceExternalLinks( $text ) {
1141                 global $wgContLang;
1142                 $fname = 'Parser::replaceExternalLinks';
1143                 wfProfileIn( $fname );
1144
1145                 $sk =& $this->mOptions->getSkin();
1146
1147                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1148
1149                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1150
1151                 $i = 0;
1152                 while ( $i<count( $bits ) ) {
1153                         $url = $bits[$i++];
1154                         $protocol = $bits[$i++];
1155                         $text = $bits[$i++];
1156                         $trail = $bits[$i++];
1157
1158                         # The characters '<' and '>' (which were escaped by
1159                         # removeHTMLtags()) should not be included in
1160                         # URLs, per RFC 2396.
1161                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1162                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1163                                 $url = substr($url, 0, $m2[0][1]);
1164                         }
1165
1166                         # If the link text is an image URL, replace it with an <img> tag
1167                         # This happened by accident in the original parser, but some people used it extensively
1168                         $img = $this->maybeMakeExternalImage( $text );
1169                         if ( $img !== false ) {
1170                                 $text = $img;
1171                         }
1172
1173                         $dtrail = '';
1174
1175                         # Set linktype for CSS - if URL==text, link is essentially free
1176                         $linktype = ($text == $url) ? 'free' : 'text';
1177
1178                         # No link text, e.g. [http://domain.tld/some.link]
1179                         if ( $text == '' ) {
1180                                 # Autonumber if allowed. See bug #5918
1181                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1182                                         $text = '[' . ++$this->mAutonumber . ']';
1183                                         $linktype = 'autonumber';
1184                                 } else {
1185                                         # Otherwise just use the URL
1186                                         $text = htmlspecialchars( $url );
1187                                         $linktype = 'free';
1188                                 }
1189                         } else {
1190                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1191                                 # Check for trail
1192                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1193                         }
1194
1195                         $text = $wgContLang->markNoConversion($text);
1196
1197                         # Normalize any HTML entities in input. They will be
1198                         # re-escaped by makeExternalLink().
1199                         $url = Sanitizer::decodeCharReferences( $url );
1200
1201                         # Escape any control characters introduced by the above step
1202                         $url = preg_replace( '/[\][<>"\\x00-\\x20\\x7F]/e', "urlencode('\\0')", $url );
1203
1204                         # Process the trail (i.e. everything after this link up until start of the next link),
1205                         # replacing any non-bracketed links
1206                         $trail = $this->replaceFreeExternalLinks( $trail );
1207
1208                         # Use the encoded URL
1209                         # This means that users can paste URLs directly into the text
1210                         # Funny characters like &ouml; aren't valid in URLs anyway
1211                         # This was changed in August 2004
1212                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1213
1214                         # Register link in the output object.
1215                         # Replace unnecessary URL escape codes with the referenced character
1216                         # This prevents spammers from hiding links from the filters
1217                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1218                         $this->mOutput->addExternalLink( $pasteurized );
1219                 }
1220
1221                 wfProfileOut( $fname );
1222                 return $s;
1223         }
1224
1225         /**
1226          * Replace anything that looks like a URL with a link
1227          * @private
1228          */
1229         function replaceFreeExternalLinks( $text ) {
1230                 global $wgContLang;
1231                 $fname = 'Parser::replaceFreeExternalLinks';
1232                 wfProfileIn( $fname );
1233
1234                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1235                 $s = array_shift( $bits );
1236                 $i = 0;
1237
1238                 $sk =& $this->mOptions->getSkin();
1239
1240                 while ( $i < count( $bits ) ){
1241                         $protocol = $bits[$i++];
1242                         $remainder = $bits[$i++];
1243
1244                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1245                                 # Found some characters after the protocol that look promising
1246                                 $url = $protocol . $m[1];
1247                                 $trail = $m[2];
1248
1249                                 # special case: handle urls as url args:
1250                                 # http://www.example.com/foo?=http://www.example.com/bar
1251                                 if(strlen($trail) == 0 &&
1252                                         isset($bits[$i]) &&
1253                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1254                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1255                                 {
1256                                         # add protocol, arg
1257                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1258                                         $i += 2;
1259                                         $trail = $m[2];
1260                                 }
1261
1262                                 # The characters '<' and '>' (which were escaped by
1263                                 # removeHTMLtags()) should not be included in
1264                                 # URLs, per RFC 2396.
1265                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1266                                         $trail = substr($url, $m2[0][1]) . $trail;
1267                                         $url = substr($url, 0, $m2[0][1]);
1268                                 }
1269
1270                                 # Move trailing punctuation to $trail
1271                                 $sep = ',;\.:!?';
1272                                 # If there is no left bracket, then consider right brackets fair game too
1273                                 if ( strpos( $url, '(' ) === false ) {
1274                                         $sep .= ')';
1275                                 }
1276
1277                                 $numSepChars = strspn( strrev( $url ), $sep );
1278                                 if ( $numSepChars ) {
1279                                         $trail = substr( $url, -$numSepChars ) . $trail;
1280                                         $url = substr( $url, 0, -$numSepChars );
1281                                 }
1282
1283                                 # Normalize any HTML entities in input. They will be
1284                                 # re-escaped by makeExternalLink() or maybeMakeExternalImage()
1285                                 $url = Sanitizer::decodeCharReferences( $url );
1286
1287                                 # Escape any control characters introduced by the above step
1288                                 $url = preg_replace( '/[\][<>"\\x00-\\x20\\x7F]/e', "urlencode('\\0')", $url );
1289
1290                                 # Is this an external image?
1291                                 $text = $this->maybeMakeExternalImage( $url );
1292                                 if ( $text === false ) {
1293                                         # Not an image, make a link
1294                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1295                                         # Register it in the output object...
1296                                         # Replace unnecessary URL escape codes with their equivalent characters
1297                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1298                                         $this->mOutput->addExternalLink( $pasteurized );
1299                                 }
1300                                 $s .= $text . $trail;
1301                         } else {
1302                                 $s .= $protocol . $remainder;
1303                         }
1304                 }
1305                 wfProfileOut( $fname );
1306                 return $s;
1307         }
1308
1309         /**
1310          * Replace unusual URL escape codes with their equivalent characters
1311          * @param string
1312          * @return string
1313          * @static
1314          * @fixme This can merge genuinely required bits in the path or query string,
1315          *        breaking legit URLs. A proper fix would treat the various parts of
1316          *        the URL differently; as a workaround, just use the output for
1317          *        statistical records, not for actual linking/output.
1318          */
1319         function replaceUnusualEscapes( $url ) {
1320                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1321                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1322         }
1323
1324         /**
1325          * Callback function used in replaceUnusualEscapes().
1326          * Replaces unusual URL escape codes with their equivalent character
1327          * @static
1328          * @private
1329          */
1330         function replaceUnusualEscapesCallback( $matches ) {
1331                 $char = urldecode( $matches[0] );
1332                 $ord = ord( $char );
1333                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1334                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1335                         // No, shouldn't be escaped
1336                         return $char;
1337                 } else {
1338                         // Yes, leave it escaped
1339                         return $matches[0];
1340                 }
1341         }
1342
1343         /**
1344          * make an image if it's allowed, either through the global
1345          * option or through the exception
1346          * @private
1347          */
1348         function maybeMakeExternalImage( $url ) {
1349                 $sk =& $this->mOptions->getSkin();
1350                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1351                 $imagesexception = !empty($imagesfrom);
1352                 $text = false;
1353                 if ( $this->mOptions->getAllowExternalImages()
1354                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1355                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1356                                 # Image found
1357                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1358                         }
1359                 }
1360                 return $text;
1361         }
1362
1363         /**
1364          * Process [[ ]] wikilinks
1365          *
1366          * @private
1367          */
1368         function replaceInternalLinks( $s ) {
1369                 global $wgContLang;
1370                 static $fname = 'Parser::replaceInternalLinks' ;
1371
1372                 wfProfileIn( $fname );
1373
1374                 wfProfileIn( $fname.'-setup' );
1375                 static $tc = FALSE;
1376                 # the % is needed to support urlencoded titles as well
1377                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1378
1379                 $sk =& $this->mOptions->getSkin();
1380
1381                 #split the entire text string on occurences of [[
1382                 $a = explode( '[[', ' ' . $s );
1383                 #get the first element (all text up to first [[), and remove the space we added
1384                 $s = array_shift( $a );
1385                 $s = substr( $s, 1 );
1386
1387                 # Match a link having the form [[namespace:link|alternate]]trail
1388                 static $e1 = FALSE;
1389                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1390                 # Match cases where there is no "]]", which might still be images
1391                 static $e1_img = FALSE;
1392                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1393                 # Match the end of a line for a word that's not followed by whitespace,
1394                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1395                 $e2 = wfMsgForContent( 'linkprefix' );
1396
1397                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1398
1399                 if( is_null( $this->mTitle ) ) {
1400                         throw new MWException( 'nooo' );
1401                 }
1402                 $nottalk = !$this->mTitle->isTalkPage();
1403
1404                 if ( $useLinkPrefixExtension ) {
1405                         if ( preg_match( $e2, $s, $m ) ) {
1406                                 $first_prefix = $m[2];
1407                         } else {
1408                                 $first_prefix = false;
1409                         }
1410                 } else {
1411                         $prefix = '';
1412                 }
1413
1414                 $selflink = $this->mTitle->getPrefixedText();
1415                 wfProfileOut( $fname.'-setup' );
1416
1417                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1418                 $useSubpages = $this->areSubpagesAllowed();
1419
1420                 # Loop for each link
1421                 for ($k = 0; isset( $a[$k] ); $k++) {
1422                         $line = $a[$k];
1423                         if ( $useLinkPrefixExtension ) {
1424                                 wfProfileIn( $fname.'-prefixhandling' );
1425                                 if ( preg_match( $e2, $s, $m ) ) {
1426                                         $prefix = $m[2];
1427                                         $s = $m[1];
1428                                 } else {
1429                                         $prefix='';
1430                                 }
1431                                 # first link
1432                                 if($first_prefix) {
1433                                         $prefix = $first_prefix;
1434                                         $first_prefix = false;
1435                                 }
1436                                 wfProfileOut( $fname.'-prefixhandling' );
1437                         }
1438
1439                         $might_be_img = false;
1440
1441                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1442                                 $text = $m[2];
1443                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1444                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1445                                 # the real problem is with the $e1 regex
1446                                 # See bug 1300.
1447                                 #
1448                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1449                                 # and no image is in sight. See bug 2095.
1450                                 #
1451                                 if( $text !== '' &&
1452                                         preg_match( "/^\](.*)/s", $m[3], $n ) &&
1453                                         strpos($text, '[') !== false
1454                                 )
1455                                 {
1456                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1457                                         $m[3] = $n[1];
1458                                 }
1459                                 # fix up urlencoded title texts
1460                                 if(preg_match('/%/', $m[1] ))
1461                                         # Should anchors '#' also be rejected?
1462                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1463                                 $trail = $m[3];
1464                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1465                                 $might_be_img = true;
1466                                 $text = $m[2];
1467                                 if(preg_match('/%/', $m[1] )) $m[1] = urldecode($m[1]);
1468                                 $trail = "";
1469                         } else { # Invalid form; output directly
1470                                 $s .= $prefix . '[[' . $line ;
1471                                 continue;
1472                         }
1473
1474                         # Don't allow internal links to pages containing
1475                         # PROTO: where PROTO is a valid URL protocol; these
1476                         # should be external links.
1477                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1478                                 $s .= $prefix . '[[' . $line ;
1479                                 continue;
1480                         }
1481
1482                         # Make subpage if necessary
1483                         if( $useSubpages ) {
1484                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1485                         } else {
1486                                 $link = $m[1];
1487                         }
1488
1489                         $noforce = (substr($m[1], 0, 1) != ':');
1490                         if (!$noforce) {
1491                                 # Strip off leading ':'
1492                                 $link = substr($link, 1);
1493                         }
1494
1495                         $nt = Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1496                         if( !$nt ) {
1497                                 $s .= $prefix . '[[' . $line;
1498                                 continue;
1499                         }
1500
1501                         #check other language variants of the link
1502                         #if the article does not exist
1503                         if( $checkVariantLink
1504                             && $nt->getArticleID() == 0 ) {
1505                                 $wgContLang->findVariantLink($link, $nt);
1506                         }
1507
1508                         $ns = $nt->getNamespace();
1509                         $iw = $nt->getInterWiki();
1510
1511                         if ($might_be_img) { # if this is actually an invalid link
1512                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1513                                         $found = false;
1514                                         while (isset ($a[$k+1]) ) {
1515                                                 #look at the next 'line' to see if we can close it there
1516                                                 $spliced = array_splice( $a, $k + 1, 1 );
1517                                                 $next_line = array_shift( $spliced );
1518                                                 if( preg_match("/^(.*?]].*?)]](.*)$/sD", $next_line, $m) ) {
1519                                                 # the first ]] closes the inner link, the second the image
1520                                                         $found = true;
1521                                                         $text .= '[[' . $m[1];
1522                                                         $trail = $m[2];
1523                                                         break;
1524                                                 } elseif( preg_match("/^.*?]].*$/sD", $next_line, $m) ) {
1525                                                         #if there's exactly one ]] that's fine, we'll keep looking
1526                                                         $text .= '[[' . $m[0];
1527                                                 } else {
1528                                                         #if $next_line is invalid too, we need look no further
1529                                                         $text .= '[[' . $next_line;
1530                                                         break;
1531                                                 }
1532                                         }
1533                                         if ( !$found ) {
1534                                                 # we couldn't find the end of this imageLink, so output it raw
1535                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1536                                                 $text = $this->replaceInternalLinks($text);
1537                                                 $s .= $prefix . '[[' . $link . '|' . $text;
1538                                                 # note: no $trail, because without an end, there *is* no trail
1539                                                 continue;
1540                                         }
1541                                 } else { #it's not an image, so output it raw
1542                                         $s .= $prefix . '[[' . $link . '|' . $text;
1543                                         # note: no $trail, because without an end, there *is* no trail
1544                                         continue;
1545                                 }
1546                         }
1547
1548                         $wasblank = ( '' == $text );
1549                         if( $wasblank ) $text = $link;
1550
1551
1552                         # Link not escaped by : , create the various objects
1553                         if( $noforce ) {
1554
1555                                 # Interwikis
1556                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1557                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1558                                         $s = rtrim($s . "\n");
1559                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1560                                         continue;
1561                                 }
1562
1563                                 if ( $ns == NS_IMAGE ) {
1564                                         wfProfileIn( "$fname-image" );
1565                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1566                                                 # recursively parse links inside the image caption
1567                                                 # actually, this will parse them in any other parameters, too,
1568                                                 # but it might be hard to fix that, and it doesn't matter ATM
1569                                                 $text = $this->replaceExternalLinks($text);
1570                                                 $text = $this->replaceInternalLinks($text);
1571
1572                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1573                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1574                                                 $this->mOutput->addImage( $nt->getDBkey() );
1575
1576                                                 wfProfileOut( "$fname-image" );
1577                                                 continue;
1578                                         } else {
1579                                                 # We still need to record the image's presence on the page
1580                                                 $this->mOutput->addImage( $nt->getDBkey() );
1581                                         }
1582                                         wfProfileOut( "$fname-image" );
1583
1584                                 }
1585
1586                                 if ( $ns == NS_CATEGORY ) {
1587                                         wfProfileIn( "$fname-category" );
1588                                         $s = rtrim($s . "\n"); # bug 87
1589
1590                                         if ( $wasblank ) {
1591                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1592                                                         $sortkey = $this->mTitle->getText();
1593                                                 } else {
1594                                                         $sortkey = $this->mTitle->getPrefixedText();
1595                                                 }
1596                                         } else {
1597                                                 $sortkey = $text;
1598                                         }
1599                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1600                                         $sortkey = str_replace( "\n", '', $sortkey );
1601                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1602                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1603
1604                                         /**
1605                                          * Strip the whitespace Category links produce, see bug 87
1606                                          * @todo We might want to use trim($tmp, "\n") here.
1607                                          */
1608                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1609
1610                                         wfProfileOut( "$fname-category" );
1611                                         continue;
1612                                 }
1613                         }
1614
1615                         if( ( $nt->getPrefixedText() === $selflink ) &&
1616                             ( $nt->getFragment() === '' ) ) {
1617                                 # Self-links are handled specially; generally de-link and change to bold.
1618                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1619                                 continue;
1620                         }
1621
1622                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1623                         if( $ns == NS_MEDIA ) {
1624                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1625                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1626                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1627                                 $this->mOutput->addImage( $nt->getDBkey() );
1628                                 continue;
1629                         } elseif( $ns == NS_SPECIAL ) {
1630                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1631                                 continue;
1632                         } elseif( $ns == NS_IMAGE ) {
1633                                 $img = Image::newFromTitle( $nt );
1634                                 if( $img->exists() ) {
1635                                         // Force a blue link if the file exists; may be a remote
1636                                         // upload on the shared repository, and we want to see its
1637                                         // auto-generated page.
1638                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1639                                         continue;
1640                                 }
1641                         }
1642                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1643                 }
1644                 wfProfileOut( $fname );
1645                 return $s;
1646         }
1647
1648         /**
1649          * Make a link placeholder. The text returned can be later resolved to a real link with
1650          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1651          * parsing of interwiki links, and secondly to allow all extistence checks and
1652          * article length checks (for stub links) to be bundled into a single query.
1653          *
1654          */
1655         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1656                 if ( ! is_object($nt) ) {
1657                         # Fail gracefully
1658                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1659                 } else {
1660                         # Separate the link trail from the rest of the link
1661                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1662
1663                         if ( $nt->isExternal() ) {
1664                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1665                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1666                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1667                         } else {
1668                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1669                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1670                                 $this->mLinkHolders['queries'][] = $query;
1671                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1672                                 $this->mLinkHolders['titles'][] = $nt;
1673
1674                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1675                         }
1676                 }
1677                 return $retVal;
1678         }
1679
1680         /**
1681          * Render a forced-blue link inline; protect against double expansion of
1682          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1683          * Since this little disaster has to split off the trail text to avoid
1684          * breaking URLs in the following text without breaking trails on the
1685          * wiki links, it's been made into a horrible function.
1686          *
1687          * @param Title $nt
1688          * @param string $text
1689          * @param string $query
1690          * @param string $trail
1691          * @param string $prefix
1692          * @return string HTML-wikitext mix oh yuck
1693          */
1694         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1695                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1696                 $sk =& $this->mOptions->getSkin();
1697                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1698                 return $this->armorLinks( $link ) . $trail;
1699         }
1700
1701         /**
1702          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1703          * going to go through further parsing steps before inline URL expansion.
1704          *
1705          * In particular this is important when using action=render, which causes
1706          * full URLs to be included.
1707          *
1708          * Oh man I hate our multi-layer parser!
1709          *
1710          * @param string more-or-less HTML
1711          * @return string less-or-more HTML with NOPARSE bits
1712          */
1713         function armorLinks( $text ) {
1714                 return preg_replace( "/\b(" . wfUrlProtocols() . ')/',
1715                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1716         }
1717
1718         /**
1719          * Return true if subpage links should be expanded on this page.
1720          * @return bool
1721          */
1722         function areSubpagesAllowed() {
1723                 # Some namespaces don't allow subpages
1724                 global $wgNamespacesWithSubpages;
1725                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1726         }
1727
1728         /**
1729          * Handle link to subpage if necessary
1730          * @param string $target the source of the link
1731          * @param string &$text the link text, modified as necessary
1732          * @return string the full name of the link
1733          * @private
1734          */
1735         function maybeDoSubpageLink($target, &$text) {
1736                 # Valid link forms:
1737                 # Foobar -- normal
1738                 # :Foobar -- override special treatment of prefix (images, language links)
1739                 # /Foobar -- convert to CurrentPage/Foobar
1740                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1741                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1742                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1743
1744                 $fname = 'Parser::maybeDoSubpageLink';
1745                 wfProfileIn( $fname );
1746                 $ret = $target; # default return value is no change
1747
1748                 # Some namespaces don't allow subpages,
1749                 # so only perform processing if subpages are allowed
1750                 if( $this->areSubpagesAllowed() ) {
1751                         # Look at the first character
1752                         if( $target != '' && $target{0} == '/' ) {
1753                                 # / at end means we don't want the slash to be shown
1754                                 if( substr( $target, -1, 1 ) == '/' ) {
1755                                         $target = substr( $target, 1, -1 );
1756                                         $noslash = $target;
1757                                 } else {
1758                                         $noslash = substr( $target, 1 );
1759                                 }
1760
1761                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1762                                 if( '' === $text ) {
1763                                         $text = $target;
1764                                 } # this might be changed for ugliness reasons
1765                         } else {
1766                                 # check for .. subpage backlinks
1767                                 $dotdotcount = 0;
1768                                 $nodotdot = $target;
1769                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1770                                         ++$dotdotcount;
1771                                         $nodotdot = substr( $nodotdot, 3 );
1772                                 }
1773                                 if($dotdotcount > 0) {
1774                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1775                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1776                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1777                                                 # / at the end means don't show full path
1778                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1779                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1780                                                         if( '' === $text ) {
1781                                                                 $text = $nodotdot;
1782                                                         }
1783                                                 }
1784                                                 $nodotdot = trim( $nodotdot );
1785                                                 if( $nodotdot != '' ) {
1786                                                         $ret .= '/' . $nodotdot;
1787                                                 }
1788                                         }
1789                                 }
1790                         }
1791                 }
1792
1793                 wfProfileOut( $fname );
1794                 return $ret;
1795         }
1796
1797         /**#@+
1798          * Used by doBlockLevels()
1799          * @private
1800          */
1801         /* private */ function closeParagraph() {
1802                 $result = '';
1803                 if ( '' != $this->mLastSection ) {
1804                         $result = '</' . $this->mLastSection  . ">\n";
1805                 }
1806                 $this->mInPre = false;
1807                 $this->mLastSection = '';
1808                 return $result;
1809         }
1810         # getCommon() returns the length of the longest common substring
1811         # of both arguments, starting at the beginning of both.
1812         #
1813         /* private */ function getCommon( $st1, $st2 ) {
1814                 $fl = strlen( $st1 );
1815                 $shorter = strlen( $st2 );
1816                 if ( $fl < $shorter ) { $shorter = $fl; }
1817
1818                 for ( $i = 0; $i < $shorter; ++$i ) {
1819                         if ( $st1{$i} != $st2{$i} ) { break; }
1820                 }
1821                 return $i;
1822         }
1823         # These next three functions open, continue, and close the list
1824         # element appropriate to the prefix character passed into them.
1825         #
1826         /* private */ function openList( $char ) {
1827                 $result = $this->closeParagraph();
1828
1829                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1830                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1831                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1832                 else if ( ';' == $char ) {
1833                         $result .= '<dl><dt>';
1834                         $this->mDTopen = true;
1835                 }
1836                 else { $result = '<!-- ERR 1 -->'; }
1837
1838                 return $result;
1839         }
1840
1841         /* private */ function nextItem( $char ) {
1842                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1843                 else if ( ':' == $char || ';' == $char ) {
1844                         $close = '</dd>';
1845                         if ( $this->mDTopen ) { $close = '</dt>'; }
1846                         if ( ';' == $char ) {
1847                                 $this->mDTopen = true;
1848                                 return $close . '<dt>';
1849                         } else {
1850                                 $this->mDTopen = false;
1851                                 return $close . '<dd>';
1852                         }
1853                 }
1854                 return '<!-- ERR 2 -->';
1855         }
1856
1857         /* private */ function closeList( $char ) {
1858                 if ( '*' == $char ) { $text = '</li></ul>'; }
1859                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1860                 else if ( ':' == $char ) {
1861                         if ( $this->mDTopen ) {
1862                                 $this->mDTopen = false;
1863                                 $text = '</dt></dl>';
1864                         } else {
1865                                 $text = '</dd></dl>';
1866                         }
1867                 }
1868                 else {  return '<!-- ERR 3 -->'; }
1869                 return $text."\n";
1870         }
1871         /**#@-*/
1872
1873         /**
1874          * Make lists from lines starting with ':', '*', '#', etc.
1875          *
1876          * @private
1877          * @return string the lists rendered as HTML
1878          */
1879         function doBlockLevels( $text, $linestart ) {
1880                 $fname = 'Parser::doBlockLevels';
1881                 wfProfileIn( $fname );
1882
1883                 # Parsing through the text line by line.  The main thing
1884                 # happening here is handling of block-level elements p, pre,
1885                 # and making lists from lines starting with * # : etc.
1886                 #
1887                 $textLines = explode( "\n", $text );
1888
1889                 $lastPrefix = $output = '';
1890                 $this->mDTopen = $inBlockElem = false;
1891                 $prefixLength = 0;
1892                 $paragraphStack = false;
1893
1894                 if ( !$linestart ) {
1895                         $output .= array_shift( $textLines );
1896                 }
1897                 foreach ( $textLines as $oLine ) {
1898                         $lastPrefixLength = strlen( $lastPrefix );
1899                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1900                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1901                         if ( !$this->mInPre ) {
1902                                 # Multiple prefixes may abut each other for nested lists.
1903                                 $prefixLength = strspn( $oLine, '*#:;' );
1904                                 $pref = substr( $oLine, 0, $prefixLength );
1905
1906                                 # eh?
1907                                 $pref2 = str_replace( ';', ':', $pref );
1908                                 $t = substr( $oLine, $prefixLength );
1909                                 $this->mInPre = !empty($preOpenMatch);
1910                         } else {
1911                                 # Don't interpret any other prefixes in preformatted text
1912                                 $prefixLength = 0;
1913                                 $pref = $pref2 = '';
1914                                 $t = $oLine;
1915                         }
1916
1917                         # List generation
1918                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1919                                 # Same as the last item, so no need to deal with nesting or opening stuff
1920                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1921                                 $paragraphStack = false;
1922
1923                                 if ( substr( $pref, -1 ) == ';') {
1924                                         # The one nasty exception: definition lists work like this:
1925                                         # ; title : definition text
1926                                         # So we check for : in the remainder text to split up the
1927                                         # title and definition, without b0rking links.
1928                                         $term = $t2 = '';
1929                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1930                                                 $t = $t2;
1931                                                 $output .= $term . $this->nextItem( ':' );
1932                                         }
1933                                 }
1934                         } elseif( $prefixLength || $lastPrefixLength ) {
1935                                 # Either open or close a level...
1936                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
1937                                 $paragraphStack = false;
1938
1939                                 while( $commonPrefixLength < $lastPrefixLength ) {
1940                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
1941                                         --$lastPrefixLength;
1942                                 }
1943                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
1944                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
1945                                 }
1946                                 while ( $prefixLength > $commonPrefixLength ) {
1947                                         $char = substr( $pref, $commonPrefixLength, 1 );
1948                                         $output .= $this->openList( $char );
1949
1950                                         if ( ';' == $char ) {
1951                                                 # FIXME: This is dupe of code above
1952                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
1953                                                         $t = $t2;
1954                                                         $output .= $term . $this->nextItem( ':' );
1955                                                 }
1956                                         }
1957                                         ++$commonPrefixLength;
1958                                 }
1959                                 $lastPrefix = $pref2;
1960                         }
1961                         if( 0 == $prefixLength ) {
1962                                 wfProfileIn( "$fname-paragraph" );
1963                                 # No prefix (not in list)--go to paragraph mode
1964                                 // XXX: use a stack for nestable elements like span, table and div
1965                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/center|<\\/tr|<\\/td|<\\/th)/iS', $t );
1966                                 $closematch = preg_match(
1967                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
1968                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<center)/iS', $t );
1969                                 if ( $openmatch or $closematch ) {
1970                                         $paragraphStack = false;
1971                                         # TODO bug 5718: paragraph closed
1972                                         $output .= $this->closeParagraph();
1973                                         if ( $preOpenMatch and !$preCloseMatch ) {
1974                                                 $this->mInPre = true;
1975                                         }
1976                                         if ( $closematch ) {
1977                                                 $inBlockElem = false;
1978                                         } else {
1979                                                 $inBlockElem = true;
1980                                         }
1981                                 } else if ( !$inBlockElem && !$this->mInPre ) {
1982                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
1983                                                 // pre
1984                                                 if ($this->mLastSection != 'pre') {
1985                                                         $paragraphStack = false;
1986                                                         $output .= $this->closeParagraph().'<pre>';
1987                                                         $this->mLastSection = 'pre';
1988                                                 }
1989                                                 $t = substr( $t, 1 );
1990                                         } else {
1991                                                 // paragraph
1992                                                 if ( '' == trim($t) ) {
1993                                                         if ( $paragraphStack ) {
1994                                                                 $output .= $paragraphStack.'<br />';
1995                                                                 $paragraphStack = false;
1996                                                                 $this->mLastSection = 'p';
1997                                                         } else {
1998                                                                 if ($this->mLastSection != 'p' ) {
1999                                                                         $output .= $this->closeParagraph();
2000                                                                         $this->mLastSection = '';
2001                                                                         $paragraphStack = '<p>';
2002                                                                 } else {
2003                                                                         $paragraphStack = '</p><p>';
2004                                                                 }
2005                                                         }
2006                                                 } else {
2007                                                         if ( $paragraphStack ) {
2008                                                                 $output .= $paragraphStack;
2009                                                                 $paragraphStack = false;
2010                                                                 $this->mLastSection = 'p';
2011                                                         } else if ($this->mLastSection != 'p') {
2012                                                                 $output .= $this->closeParagraph().'<p>';
2013                                                                 $this->mLastSection = 'p';
2014                                                         }
2015                                                 }
2016                                         }
2017                                 }
2018                                 wfProfileOut( "$fname-paragraph" );
2019                         }
2020                         // somewhere above we forget to get out of pre block (bug 785)
2021                         if($preCloseMatch && $this->mInPre) {
2022                                 $this->mInPre = false;
2023                         }
2024                         if ($paragraphStack === false) {
2025                                 $output .= $t."\n";
2026                         }
2027                 }
2028                 while ( $prefixLength ) {
2029                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2030                         --$prefixLength;
2031                 }
2032                 if ( '' != $this->mLastSection ) {
2033                         $output .= '</' . $this->mLastSection . '>';
2034                         $this->mLastSection = '';
2035                 }
2036
2037                 wfProfileOut( $fname );
2038                 return $output;
2039         }
2040
2041         /**
2042          * Split up a string on ':', ignoring any occurences inside tags
2043          * to prevent illegal overlapping.
2044          * @param string $str the string to split
2045          * @param string &$before set to everything before the ':'
2046          * @param string &$after set to everything after the ':'
2047          * return string the position of the ':', or false if none found
2048          */
2049         function findColonNoLinks($str, &$before, &$after) {
2050                 $fname = 'Parser::findColonNoLinks';
2051                 wfProfileIn( $fname );
2052
2053                 $pos = strpos( $str, ':' );
2054                 if( $pos === false ) {
2055                         // Nothing to find!
2056                         wfProfileOut( $fname );
2057                         return false;
2058                 }
2059
2060                 $lt = strpos( $str, '<' );
2061                 if( $lt === false || $lt > $pos ) {
2062                         // Easy; no tag nesting to worry about
2063                         $before = substr( $str, 0, $pos );
2064                         $after = substr( $str, $pos+1 );
2065                         wfProfileOut( $fname );
2066                         return $pos;
2067                 }
2068
2069                 // Ugly state machine to walk through avoiding tags.
2070                 $state = MW_COLON_STATE_TEXT;
2071                 $stack = 0;
2072                 $len = strlen( $str );
2073                 for( $i = 0; $i < $len; $i++ ) {
2074                         $c = $str{$i};
2075
2076                         switch( $state ) {
2077                         // (Using the number is a performance hack for common cases)
2078                         case 0: // MW_COLON_STATE_TEXT:
2079                                 switch( $c ) {
2080                                 case "<":
2081                                         // Could be either a <start> tag or an </end> tag
2082                                         $state = MW_COLON_STATE_TAGSTART;
2083                                         break;
2084                                 case ":":
2085                                         if( $stack == 0 ) {
2086                                                 // We found it!
2087                                                 $before = substr( $str, 0, $i );
2088                                                 $after = substr( $str, $i + 1 );
2089                                                 wfProfileOut( $fname );
2090                                                 return $i;
2091                                         }
2092                                         // Embedded in a tag; don't break it.
2093                                         break;
2094                                 default:
2095                                         // Skip ahead looking for something interesting
2096                                         $colon = strpos( $str, ':', $i );
2097                                         if( $colon === false ) {
2098                                                 // Nothing else interesting
2099                                                 wfProfileOut( $fname );
2100                                                 return false;
2101                                         }
2102                                         $lt = strpos( $str, '<', $i );
2103                                         if( $stack === 0 ) {
2104                                                 if( $lt === false || $colon < $lt ) {
2105                                                         // We found it!
2106                                                         $before = substr( $str, 0, $colon );
2107                                                         $after = substr( $str, $colon + 1 );
2108                                                         wfProfileOut( $fname );
2109                                                         return $i;
2110                                                 }
2111                                         }
2112                                         if( $lt === false ) {
2113                                                 // Nothing else interesting to find; abort!
2114                                                 // We're nested, but there's no close tags left. Abort!
2115                                                 break 2;
2116                                         }
2117                                         // Skip ahead to next tag start
2118                                         $i = $lt;
2119                                         $state = MW_COLON_STATE_TAGSTART;
2120                                 }
2121                                 break;
2122                         case 1: // MW_COLON_STATE_TAG:
2123                                 // In a <tag>
2124                                 switch( $c ) {
2125                                 case ">":
2126                                         $stack++;
2127                                         $state = MW_COLON_STATE_TEXT;
2128                                         break;
2129                                 case "/":
2130                                         // Slash may be followed by >?
2131                                         $state = MW_COLON_STATE_TAGSLASH;
2132                                         break;
2133                                 default:
2134                                         // ignore
2135                                 }
2136                                 break;
2137                         case 2: // MW_COLON_STATE_TAGSTART:
2138                                 switch( $c ) {
2139                                 case "/":
2140                                         $state = MW_COLON_STATE_CLOSETAG;
2141                                         break;
2142                                 case "!":
2143                                         $state = MW_COLON_STATE_COMMENT;
2144                                         break;
2145                                 case ">":
2146                                         // Illegal early close? This shouldn't happen D:
2147                                         $state = MW_COLON_STATE_TEXT;
2148                                         break;
2149                                 default:
2150                                         $state = MW_COLON_STATE_TAG;
2151                                 }
2152                                 break;
2153                         case 3: // MW_COLON_STATE_CLOSETAG:
2154                                 // In a </tag>
2155                                 if( $c == ">" ) {
2156                                         $stack--;
2157                                         if( $stack < 0 ) {
2158                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2159                                                 wfProfileOut( $fname );
2160                                                 return false;
2161                                         }
2162                                         $state = MW_COLON_STATE_TEXT;
2163                                 }
2164                                 break;
2165                         case MW_COLON_STATE_TAGSLASH:
2166                                 if( $c == ">" ) {
2167                                         // Yes, a self-closed tag <blah/>
2168                                         $state = MW_COLON_STATE_TEXT;
2169                                 } else {
2170                                         // Probably we're jumping the gun, and this is an attribute
2171                                         $state = MW_COLON_STATE_TAG;
2172                                 }
2173                                 break;
2174                         case 5: // MW_COLON_STATE_COMMENT:
2175                                 if( $c == "-" ) {
2176                                         $state = MW_COLON_STATE_COMMENTDASH;
2177                                 }
2178                                 break;
2179                         case MW_COLON_STATE_COMMENTDASH:
2180                                 if( $c == "-" ) {
2181                                         $state = MW_COLON_STATE_COMMENTDASHDASH;
2182                                 } else {
2183                                         $state = MW_COLON_STATE_COMMENT;
2184                                 }
2185                                 break;
2186                         case MW_COLON_STATE_COMMENTDASHDASH:
2187                                 if( $c == ">" ) {
2188                                         $state = MW_COLON_STATE_TEXT;
2189                                 } else {
2190                                         $state = MW_COLON_STATE_COMMENT;
2191                                 }
2192                                 break;
2193                         default:
2194                                 throw new MWException( "State machine error in $fname" );
2195                         }
2196                 }
2197                 if( $stack > 0 ) {
2198                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2199                         return false;
2200                 }
2201                 wfProfileOut( $fname );
2202                 return false;
2203         }
2204
2205         /**
2206          * Return value of a magic variable (like PAGENAME)
2207          *
2208          * @private
2209          */
2210         function getVariableValue( $index ) {
2211                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2212
2213                 /**
2214                  * Some of these require message or data lookups and can be
2215                  * expensive to check many times.
2216                  */
2217                 static $varCache = array();
2218                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) )
2219                         if ( isset( $varCache[$index] ) )
2220                                 return $varCache[$index];
2221
2222                 $ts = time();
2223                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2224
2225                 switch ( $index ) {
2226                         case MAG_CURRENTMONTH:
2227                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
2228                         case MAG_CURRENTMONTHNAME:
2229                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2230                         case MAG_CURRENTMONTHNAMEGEN:
2231                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2232                         case MAG_CURRENTMONTHABBREV:
2233                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2234                         case MAG_CURRENTDAY:
2235                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2236                         case MAG_CURRENTDAY2:
2237                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2238                         case MAG_PAGENAME:
2239                                 return $this->mTitle->getText();
2240                         case MAG_PAGENAMEE:
2241                                 return $this->mTitle->getPartialURL();
2242                         case MAG_FULLPAGENAME:
2243                                 return $this->mTitle->getPrefixedText();
2244                         case MAG_FULLPAGENAMEE:
2245                                 return $this->mTitle->getPrefixedURL();
2246                         case MAG_SUBPAGENAME:
2247                                 return $this->mTitle->getSubpageText();
2248                         case MAG_SUBPAGENAMEE:
2249                                 return $this->mTitle->getSubpageUrlForm();
2250                         case MAG_BASEPAGENAME:
2251                                 return $this->mTitle->getBaseText();
2252                         case MAG_BASEPAGENAMEE:
2253                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2254                         case MAG_TALKPAGENAME:
2255                                 if( $this->mTitle->canTalk() ) {
2256                                         $talkPage = $this->mTitle->getTalkPage();
2257                                         return $talkPage->getPrefixedText();
2258                                 } else {
2259                                         return '';
2260                                 }
2261                         case MAG_TALKPAGENAMEE:
2262                                 if( $this->mTitle->canTalk() ) {
2263                                         $talkPage = $this->mTitle->getTalkPage();
2264                                         return $talkPage->getPrefixedUrl();
2265                                 } else {
2266                                         return '';
2267                                 }
2268                         case MAG_SUBJECTPAGENAME:
2269                                 $subjPage = $this->mTitle->getSubjectPage();
2270                                 return $subjPage->getPrefixedText();
2271                         case MAG_SUBJECTPAGENAMEE:
2272                                 $subjPage = $this->mTitle->getSubjectPage();
2273                                 return $subjPage->getPrefixedUrl();
2274                         case MAG_REVISIONID:
2275                                 return $this->mRevisionId;
2276                         case MAG_NAMESPACE:
2277                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2278                         case MAG_NAMESPACEE:
2279                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2280                         case MAG_TALKSPACE:
2281                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2282                         case MAG_TALKSPACEE:
2283                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2284                         case MAG_SUBJECTSPACE:
2285                                 return $this->mTitle->getSubjectNsText();
2286                         case MAG_SUBJECTSPACEE:
2287                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2288                         case MAG_CURRENTDAYNAME:
2289                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2290                         case MAG_CURRENTYEAR:
2291                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2292                         case MAG_CURRENTTIME:
2293                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2294                         case MAG_CURRENTWEEK:
2295                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2296                                 // int to remove the padding
2297                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2298                         case MAG_CURRENTDOW:
2299                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2300                         case MAG_NUMBEROFARTICLES:
2301                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
2302                         case MAG_NUMBEROFFILES:
2303                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
2304                         case MAG_NUMBEROFUSERS:
2305                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfUsers() );
2306                         case MAG_NUMBEROFPAGES:
2307                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfPages() );
2308                         case MAG_NUMBEROFADMINS:
2309                                 return $varCache[$index]  = $wgContLang->formatNum( wfNumberOfAdmins() );
2310                         case MAG_CURRENTTIMESTAMP:
2311                                 return $varCache[$index] = wfTimestampNow();
2312                         case MAG_CURRENTVERSION:
2313                                 global $wgVersion;
2314                                 return $wgVersion;
2315                         case MAG_SITENAME:
2316                                 return $wgSitename;
2317                         case MAG_SERVER:
2318                                 return $wgServer;
2319                         case MAG_SERVERNAME:
2320                                 return $wgServerName;
2321                         case MAG_SCRIPTPATH:
2322                                 return $wgScriptPath;
2323                         case MAG_DIRECTIONMARK:
2324                                 return $wgContLang->getDirMark();
2325                         case MAG_CONTENTLANGUAGE:
2326                                 global $wgContLanguageCode;
2327                                 return $wgContLanguageCode;
2328                         default:
2329                                 $ret = null;
2330                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2331                                         return $ret;
2332                                 else
2333                                         return null;
2334                 }
2335         }
2336
2337         /**
2338          * initialise the magic variables (like CURRENTMONTHNAME)
2339          *
2340          * @private
2341          */
2342         function initialiseVariables() {
2343                 $fname = 'Parser::initialiseVariables';
2344                 wfProfileIn( $fname );
2345                 global $wgVariableIDs;
2346
2347                 $this->mVariables = array();
2348                 foreach ( $wgVariableIDs as $id ) {
2349                         $mw =& MagicWord::get( $id );
2350                         $mw->addToArray( $this->mVariables, $id );
2351                 }
2352                 wfProfileOut( $fname );
2353         }
2354
2355         /**
2356          * parse any parentheses in format ((title|part|part))
2357          * and call callbacks to get a replacement text for any found piece
2358          *
2359          * @param string $text The text to parse
2360          * @param array $callbacks rules in form:
2361          *     '{' => array(                            # opening parentheses
2362          *                                      'end' => '}',   # closing parentheses
2363          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2364          *                                                                4 => callback         # replacement callback to call if {{{{..}}}} is found
2365          *                                                                )
2366          *                                      )
2367          * @private
2368          */
2369         function replace_callback ($text, $callbacks) {
2370                 wfProfileIn( __METHOD__ . '-self' );
2371                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2372                 $lastOpeningBrace = -1;                 # last not closed parentheses
2373
2374                 for ($i = 0; $i < strlen($text); $i++) {
2375                         # check for any opening brace
2376                         $rule = null;
2377                         $nextPos = -1;
2378                         foreach ($callbacks as $key => $value) {
2379                                 $pos = strpos ($text, $key, $i);
2380                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)) {
2381                                         $rule = $value;
2382                                         $nextPos = $pos;
2383                                 }
2384                         }
2385
2386                         if ($lastOpeningBrace >= 0) {
2387                                 $pos = strpos ($text, $openingBraceStack[$lastOpeningBrace]['braceEnd'], $i);
2388
2389                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2390                                         $rule = null;
2391                                         $nextPos = $pos;
2392                                 }
2393
2394                                 $pos = strpos ($text, '|', $i);
2395
2396                                 if (false !== $pos && (-1 == $nextPos || $pos < $nextPos)){
2397                                         $rule = null;
2398                                         $nextPos = $pos;
2399                                 }
2400                         }
2401
2402                         if ($nextPos == -1)
2403                                 break;
2404
2405                         $i = $nextPos;
2406
2407                         # found openning brace, lets add it to parentheses stack
2408                         if (null != $rule) {
2409                                 $piece = array('brace' => $text[$i],
2410                                                            'braceEnd' => $rule['end'],
2411                                                            'count' => 1,
2412                                                            'title' => '',
2413                                                            'parts' => null);
2414
2415                                 # count openning brace characters
2416                                 while ($i+1 < strlen($text) && $text[$i+1] == $piece['brace']) {
2417                                         $piece['count']++;
2418                                         $i++;
2419                                 }
2420
2421                                 $piece['startAt'] = $i+1;
2422                                 $piece['partStart'] = $i+1;
2423
2424                                 # we need to add to stack only if openning brace count is enough for any given rule
2425                                 foreach ($rule['cb'] as $cnt => $fn) {
2426                                         if ($piece['count'] >= $cnt) {
2427                                                 $lastOpeningBrace ++;
2428                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2429                                                 break;
2430                                         }
2431                                 }
2432
2433                                 continue;
2434                         }
2435                         else if ($lastOpeningBrace >= 0) {
2436                                 # first check if it is a closing brace
2437                                 if ($openingBraceStack[$lastOpeningBrace]['braceEnd'] == $text[$i]) {
2438                                         # lets check if it is enough characters for closing brace
2439                                         $count = 1;
2440                                         while ($i+$count < strlen($text) && $text[$i+$count] == $text[$i])
2441                                                 $count++;
2442
2443                                         # if there are more closing parentheses than opening ones, we parse less
2444                                         if ($openingBraceStack[$lastOpeningBrace]['count'] < $count)
2445                                                 $count = $openingBraceStack[$lastOpeningBrace]['count'];
2446
2447                                         # check for maximum matching characters (if there are 5 closing characters, we will probably need only 3 - depending on the rules)
2448                                         $matchingCount = 0;
2449                                         $matchingCallback = null;
2450                                         foreach ($callbacks[$openingBraceStack[$lastOpeningBrace]['brace']]['cb'] as $cnt => $fn) {
2451                                                 if ($count >= $cnt && $matchingCount < $cnt) {
2452                                                         $matchingCount = $cnt;
2453                                                         $matchingCallback = $fn;
2454                                                 }
2455                                         }
2456
2457                                         if ($matchingCount == 0) {
2458                                                 $i += $count - 1;
2459                                                 continue;
2460                                         }
2461
2462                                         # lets set a title or last part (if '|' was found)
2463                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts'])
2464                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2465                                         else
2466                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2467
2468                                         $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2469                                         $pieceEnd = $i + $matchingCount;
2470
2471                                         if( is_callable( $matchingCallback ) ) {
2472                                                 $cbArgs = array (
2473                                                                                  'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2474                                                                                  'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2475                                                                                  'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2476                                                                                  'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == "\n")),
2477                                                                                  );
2478                                                 # finally we can call a user callback and replace piece of text
2479                                                 wfProfileOut( __METHOD__ . '-self' );
2480                                                 $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2481                                                 wfProfileIn( __METHOD__ . '-self' );
2482                                                 $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2483                                                 $i = $pieceStart + strlen($replaceWith) - 1;
2484                                         }
2485                                         else {
2486                                                 # null value for callback means that parentheses should be parsed, but not replaced
2487                                                 $i += $matchingCount - 1;
2488                                         }
2489
2490                                         # reset last openning parentheses, but keep it in case there are unused characters
2491                                         $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2492                                                                    'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2493                                                                    'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2494                                                                    'title' => '',
2495                                                                    'parts' => null,
2496                                                                    'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2497                                         $openingBraceStack[$lastOpeningBrace--] = null;
2498
2499                                         if ($matchingCount < $piece['count']) {
2500                                                 $piece['count'] -= $matchingCount;
2501                                                 $piece['startAt'] -= $matchingCount;
2502                                                 $piece['partStart'] = $piece['startAt'];
2503                                                 # do we still qualify for any callback with remaining count?
2504                                                 foreach ($callbacks[$piece['brace']]['cb'] as $cnt => $fn) {
2505                                                         if ($piece['count'] >= $cnt) {
2506                                                                 $lastOpeningBrace ++;
2507                                                                 $openingBraceStack[$lastOpeningBrace] = $piece;
2508                                                                 break;
2509                                                         }
2510                                                 }
2511                                         }
2512                                         continue;
2513                                 }
2514
2515                                 # lets set a title if it is a first separator, or next part otherwise
2516                                 if ($text[$i] == '|') {
2517                                         if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2518                                                 $openingBraceStack[$lastOpeningBrace]['title'] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2519                                                 $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2520                                         }
2521                                         else
2522                                                 $openingBraceStack[$lastOpeningBrace]['parts'][] = substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'], $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2523
2524                                         $openingBraceStack[$lastOpeningBrace]['partStart'] = $i + 1;
2525                                 }
2526                         }
2527                 }
2528
2529                 wfProfileOut( __METHOD__ . '-self' );
2530                 return $text;
2531         }
2532
2533         /**
2534          * Replace magic variables, templates, and template arguments
2535          * with the appropriate text. Templates are substituted recursively,
2536          * taking care to avoid infinite loops.
2537          *
2538          * Note that the substitution depends on value of $mOutputType:
2539          *  OT_WIKI: only {{subst:}} templates
2540          *  OT_MSG: only magic variables
2541          *  OT_HTML: all templates and magic variables
2542          *
2543          * @param string $tex The text to transform
2544          * @param array $args Key-value pairs representing template parameters to substitute
2545          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2546          * @private
2547          */
2548         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2549                 # Prevent too big inclusions
2550                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
2551                         return $text;
2552                 }
2553
2554                 $fname = 'Parser::replaceVariables';
2555                 wfProfileIn( $fname );
2556
2557                 # This function is called recursively. To keep track of arguments we need a stack:
2558                 array_push( $this->mArgStack, $args );
2559
2560                 $braceCallbacks = array();
2561                 if ( !$argsOnly ) {
2562                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2563                 }
2564                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
2565                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2566                 }
2567                 $callbacks = array();
2568                 $callbacks['{'] = array('end' => '}', 'cb' => $braceCallbacks);
2569                 $callbacks['['] = array('end' => ']', 'cb' => array(2=>null));
2570                 $text = $this->replace_callback ($text, $callbacks);
2571
2572                 array_pop( $this->mArgStack );
2573
2574                 wfProfileOut( $fname );
2575                 return $text;
2576         }
2577
2578         /**
2579          * Replace magic variables
2580          * @private
2581          */
2582         function variableSubstitution( $matches ) {
2583                 $fname = 'Parser::variableSubstitution';
2584                 $varname = $matches[1];
2585                 wfProfileIn( $fname );
2586                 $skip = false;
2587                 if ( $this->mOutputType == OT_WIKI ) {
2588                         # Do only magic variables prefixed by SUBST
2589                         $mwSubst =& MagicWord::get( MAG_SUBST );
2590                         if (!$mwSubst->matchStartAndRemove( $varname ))
2591                                 $skip = true;
2592                         # Note that if we don't substitute the variable below,
2593                         # we don't remove the {{subst:}} magic word, in case
2594                         # it is a template rather than a magic variable.
2595                 }
2596                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2597                         $id = $this->mVariables[$varname];
2598                         $text = $this->getVariableValue( $id );
2599                         $this->mOutput->mContainsOldMagic = true;
2600                 } else {
2601                         $text = $matches[0];
2602                 }
2603                 wfProfileOut( $fname );
2604                 return $text;
2605         }
2606
2607         # Split template arguments
2608         function getTemplateArgs( $argsString ) {
2609                 if ( $argsString === '' ) {
2610                         return array();
2611                 }
2612
2613                 $args = explode( '|', substr( $argsString, 1 ) );
2614
2615                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2616                 # merged with the next arg because the '|' character between belongs
2617                 # to the link syntax and not the template parameter syntax.
2618                 $argc = count($args);
2619
2620                 for ( $i = 0; $i < $argc-1; $i++ ) {
2621                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2622                                 $args[$i] .= '|'.$args[$i+1];
2623                                 array_splice($args, $i+1, 1);
2624                                 $i--;
2625                                 $argc--;
2626                         }
2627                 }
2628
2629                 return $args;
2630         }
2631
2632         /**
2633          * Return the text of a template, after recursively
2634          * replacing any variables or templates within the template.
2635          *
2636          * @param array $piece The parts of the template
2637          *  $piece['text']: matched text
2638          *  $piece['title']: the title, i.e. the part before the |
2639          *  $piece['parts']: the parameter array
2640          * @return string the text of the template
2641          * @private
2642          */
2643         function braceSubstitution( $piece ) {
2644                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $action;
2645                 $fname = 'Parser::braceSubstitution';
2646                 wfProfileIn( $fname );
2647
2648                 # Flags
2649                 $found = false;             # $text has been filled
2650                 $nowiki = false;            # wiki markup in $text should be escaped
2651                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2652                 $noargs = false;            # Don't replace triple-brace arguments in $text
2653                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2654                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2655                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2656
2657                 # Title object, where $text came from
2658                 $title = NULL;
2659
2660                 $linestart = '';
2661
2662                 # $part1 is the bit before the first |, and must contain only title characters
2663                 # $args is a list of arguments, starting from index 0, not including $part1
2664
2665                 $part1 = $piece['title'];
2666                 # If the third subpattern matched anything, it will start with |
2667
2668                 if (null == $piece['parts']) {
2669                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2670                         if ($replaceWith != $piece['text']) {
2671                                 $text = $replaceWith;
2672                                 $found = true;
2673                                 $noparse = true;
2674                                 $noargs = true;
2675                         }
2676                 }
2677
2678                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2679                 $argc = count( $args );
2680
2681                 # SUBST
2682                 if ( !$found ) {
2683                         $mwSubst =& MagicWord::get( MAG_SUBST );
2684                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2685                                 # One of two possibilities is true:
2686                                 # 1) Found SUBST but not in the PST phase
2687                                 # 2) Didn't find SUBST and in the PST phase
2688                                 # In either case, return without further processing
2689                                 $text = $piece['text'];
2690                                 $found = true;
2691                                 $noparse = true;
2692                                 $noargs = true;
2693                         }
2694                 }
2695
2696                 # MSG, MSGNW, INT and RAW
2697                 if ( !$found ) {
2698                         # Check for MSGNW:
2699                         $mwMsgnw =& MagicWord::get( MAG_MSGNW );
2700                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2701                                 $nowiki = true;
2702                         } else {
2703                                 # Remove obsolete MSG:
2704                                 $mwMsg =& MagicWord::get( MAG_MSG );
2705                                 $mwMsg->matchStartAndRemove( $part1 );
2706                         }
2707
2708                         # Check for RAW:
2709                         $mwRaw =& MagicWord::get( MAG_RAW );
2710                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2711                                 $forceRawInterwiki = true;
2712                         }
2713
2714                         # Check if it is an internal message
2715                         $mwInt =& MagicWord::get( MAG_INT );
2716                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2717                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2718                                         $text = $linestart . wfMsgReal( $part1, $args, true );
2719                                         $found = true;
2720                                 }
2721                         }
2722                 }
2723
2724                 # Parser functions
2725                 if ( !$found ) {
2726                         wfProfileIn( __METHOD__ . '-pfunc' );
2727
2728                         $colonPos = strpos( $part1, ':' );
2729                         if ( $colonPos !== false ) {
2730                                 # Case sensitive functions
2731                                 $function = substr( $part1, 0, $colonPos );
2732                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2733                                         $function = $this->mFunctionSynonyms[1][$function];
2734                                 } else {
2735                                         # Case insensitive functions
2736                                         $function = strtolower( $function );
2737                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2738                                                 $function = $this->mFunctionSynonyms[0][$function];
2739                                         } else {
2740                                                 $function = false;
2741                                         }
2742                                 }
2743                                 if ( $function ) {
2744                                         $funcArgs = array_map( 'trim', $args );
2745                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
2746                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
2747                                         $found = true;
2748
2749                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
2750                                         //$noargs = true;
2751                                         //$noparse = true;
2752
2753                                         if ( is_array( $result ) ) {
2754                                                 if ( isset( $result[0] ) ) {
2755                                                         $text = $linestart . $result[0];
2756                                                         unset( $result[0] );
2757                                                 }
2758
2759                                                 // Extract flags into the local scope
2760                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
2761                                                 extract( $result );
2762                                         } else {
2763                                                 $text = $linestart . $result;
2764                                         }
2765                                 }
2766                         }
2767                         wfProfileOut( __METHOD__ . '-pfunc' );
2768                 }
2769
2770                 # Template table test
2771
2772                 # Did we encounter this template already? If yes, it is in the cache
2773                 # and we need to check for loops.
2774                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
2775                         $found = true;
2776
2777                         # Infinite loop test
2778                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2779                                 $noparse = true;
2780                                 $noargs = true;
2781                                 $found = true;
2782                                 $text = $linestart .
2783                                         '{{' . $part1 . '}}' .
2784                                         '<!-- WARNING: template loop detected -->';
2785                                 wfDebug( "$fname: template loop broken at '$part1'\n" );
2786                         } else {
2787                                 # set $text to cached message.
2788                                 $text = $linestart . $this->mTemplates[$piece['title']];
2789                         }
2790                 }
2791
2792                 # Load from database
2793                 $lastPathLevel = $this->mTemplatePath;
2794                 if ( !$found ) {
2795                         wfProfileIn( __METHOD__ . '-loadtpl' );
2796                         $ns = NS_TEMPLATE;
2797                         # declaring $subpage directly in the function call
2798                         # does not work correctly with references and breaks
2799                         # {{/subpage}}-style inclusions
2800                         $subpage = '';
2801                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2802                         if ($subpage !== '') {
2803                                 $ns = $this->mTitle->getNamespace();
2804                         }
2805                         $title = Title::newFromText( $part1, $ns );
2806
2807
2808                         if ( !is_null( $title ) ) {
2809                                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
2810                                 # Check for language variants if the template is not found
2811                                 if($checkVariantLink && $title->getArticleID() == 0){
2812                                         $wgContLang->findVariantLink($part1, $title);
2813                                 }
2814
2815                                 if ( !$title->isExternal() ) {
2816                                         # Check for excessive inclusion
2817                                         $dbk = $title->getPrefixedDBkey();
2818                                         if ( $this->incrementIncludeCount( $dbk ) ) {
2819                                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->mOutputType != OT_WIKI ) {
2820                                                         $text = SpecialPage::capturePath( $title );
2821                                                         if ( is_string( $text ) ) {
2822                                                                 $found = true;
2823                                                                 $noparse = true;
2824                                                                 $noargs = true;
2825                                                                 $isHTML = true;
2826                                                                 $this->disableCache();
2827                                                         }
2828                                                 } else {
2829                                                         $articleContent = $this->fetchTemplate( $title );
2830                                                         if ( $articleContent !== false ) {
2831                                                                 $found = true;
2832                                                                 $text = $articleContent;
2833                                                                 $replaceHeadings = true;
2834                                                         }
2835                                                 }
2836                                         }
2837
2838                                         # If the title is valid but undisplayable, make a link to it
2839                                         if ( $this->mOutputType == OT_HTML && !$found ) {
2840                                                 $text = '[['.$title->getPrefixedText().']]';
2841                                                 $found = true;
2842                                         }
2843                                 } elseif ( $title->isTrans() ) {
2844                                         // Interwiki transclusion
2845                                         if ( $this->mOutputType == OT_HTML && !$forceRawInterwiki ) {
2846                                                 $text = $this->interwikiTransclude( $title, 'render' );
2847                                                 $isHTML = true;
2848                                                 $noparse = true;
2849                                         } else {
2850                                                 $text = $this->interwikiTransclude( $title, 'raw' );
2851                                                 $replaceHeadings = true;
2852                                         }
2853                                         $found = true;
2854                                 }
2855
2856                                 # Template cache array insertion
2857                                 # Use the original $piece['title'] not the mangled $part1, so that
2858                                 # modifiers such as RAW: produce separate cache entries
2859                                 if( $found ) {
2860                                         if( $isHTML ) {
2861                                                 // A special page; don't store it in the template cache.
2862                                         } else {
2863                                                 $this->mTemplates[$piece['title']] = $text;
2864                                         }
2865                                         $text = $linestart . $text;
2866                                 }
2867                         }
2868                         wfProfileOut( __METHOD__ . '-loadtpl' );
2869                 }
2870
2871                 # Recursive parsing, escaping and link table handling
2872                 # Only for HTML output
2873                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2874                         $text = wfEscapeWikiText( $text );
2875                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found ) {
2876                         if ( $noargs ) {
2877                                 $assocArgs = array();
2878                         } else {
2879                                 # Clean up argument array
2880                                 $assocArgs = array();
2881                                 $index = 1;
2882                                 foreach( $args as $arg ) {
2883                                         $eqpos = strpos( $arg, '=' );
2884                                         if ( $eqpos === false ) {
2885                                                 $assocArgs[$index++] = $arg;
2886                                         } else {
2887                                                 $name = trim( substr( $arg, 0, $eqpos ) );
2888                                                 $value = trim( substr( $arg, $eqpos+1 ) );
2889                                                 if ( $value === false ) {
2890                                                         $value = '';
2891                                                 }
2892                                                 if ( $name !== false ) {
2893                                                         $assocArgs[$name] = $value;
2894                                                 }
2895                                         }
2896                                 }
2897
2898                                 # Add a new element to the templace recursion path
2899                                 $this->mTemplatePath[$part1] = 1;
2900                         }
2901
2902                         if ( !$noparse ) {
2903                                 # If there are any <onlyinclude> tags, only include them
2904                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
2905                                         preg_match_all( '/<onlyinclude>(.*?)\n?<\/onlyinclude>/s', $text, $m );
2906                                         $text = '';
2907                                         foreach ($m[1] as $piece)
2908                                                 $text .= $piece;
2909                                 }
2910                                 # Remove <noinclude> sections and <includeonly> tags
2911                                 $text = preg_replace( '/<noinclude>.*?<\/noinclude>/s', '', $text );
2912                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
2913
2914                                 if( $this->mOutputType == OT_HTML ) {
2915                                         # Strip <nowiki>, <pre>, etc.
2916                                         $text = $this->strip( $text, $this->mStripState );
2917                                         $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
2918                                 }
2919                                 $text = $this->replaceVariables( $text, $assocArgs );
2920
2921                                 # If the template begins with a table or block-level
2922                                 # element, it should be treated as beginning a new line.
2923                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) {
2924                                         $text = "\n" . $text;
2925                                 }
2926                         } elseif ( !$noargs ) {
2927                                 # $noparse and !$noargs
2928                                 # Just replace the arguments, not any double-brace items
2929                                 # This is used for rendered interwiki transclusion
2930                                 $text = $this->replaceVariables( $text, $assocArgs, true );
2931                         }
2932                 }
2933                 # Prune lower levels off the recursion check path
2934                 $this->mTemplatePath = $lastPathLevel;
2935
2936                 if ( !$found ) {
2937                         wfProfileOut( $fname );
2938                         return $piece['text'];
2939                 } else {
2940                         wfProfileIn( __METHOD__ . '-placeholders' );
2941                         if ( $isHTML ) {
2942                                 # Replace raw HTML by a placeholder
2943                                 # Add a blank line preceding, to prevent it from mucking up
2944                                 # immediately preceding headings
2945                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
2946                         } else {
2947                                 # replace ==section headers==
2948                                 # XXX this needs to go away once we have a better parser.
2949                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
2950                                         if( !is_null( $title ) )
2951                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
2952                                         else
2953                                                 $encodedname = base64_encode("");
2954                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
2955                                                 PREG_SPLIT_DELIM_CAPTURE);
2956                                         $text = '';
2957                                         $nsec = 0;
2958                                         for( $i = 0; $i < count($m); $i += 2 ) {
2959                                                 $text .= $m[$i];
2960                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
2961                                                 $hl = $m[$i + 1];
2962                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
2963                                                         $text .= $hl;
2964                                                         continue;
2965                                                 }
2966                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
2967                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
2968                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
2969
2970                                                 $nsec++;
2971                                         }
2972                                 }
2973                         }
2974                         wfProfileOut( __METHOD__ . '-placeholders' );
2975                 }
2976
2977                 # Prune lower levels off the recursion check path
2978                 $this->mTemplatePath = $lastPathLevel;
2979
2980                 if ( !$found ) {
2981                         wfProfileOut( $fname );
2982                         return $piece['text'];
2983                 } else {
2984                         wfProfileOut( $fname );
2985                         return $text;
2986                 }
2987         }
2988
2989         /**
2990          * Fetch the unparsed text of a template and register a reference to it.
2991          */
2992         function fetchTemplate( $title ) {
2993                 $text = false;
2994                 // Loop to fetch the article, with up to 1 redirect
2995                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
2996                         $rev = Revision::newFromTitle( $title );
2997                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
2998                         if ( !$rev ) {
2999                                 break;
3000                         }
3001                         $text = $rev->getText();
3002                         if ( $text === false ) {
3003                                 break;
3004                         }
3005                         // Redirect?
3006                         $title = Title::newFromRedirect( $text );
3007                 }
3008                 return $text;
3009         }
3010
3011         /**
3012          * Transclude an interwiki link.
3013          */
3014         function interwikiTransclude( $title, $action ) {
3015                 global $wgEnableScaryTranscluding, $wgCanonicalNamespaceNames;
3016
3017                 if (!$wgEnableScaryTranscluding)
3018                         return wfMsg('scarytranscludedisabled');
3019
3020                 // The namespace will actually only be 0 or 10, depending on whether there was a leading :
3021                 // But we'll handle it generally anyway
3022                 if ( $title->getNamespace() ) {
3023                         // Use the canonical namespace, which should work anywhere
3024                         $articleName = $wgCanonicalNamespaceNames[$title->getNamespace()] . ':' . $title->getDBkey();
3025                 } else {
3026                         $articleName = $title->getDBkey();
3027                 }
3028
3029                 $url = str_replace('$1', urlencode($articleName), Title::getInterwikiLink($title->getInterwiki()));
3030                 $url .= "?action=$action";
3031                 if (strlen($url) > 255)
3032                         return wfMsg('scarytranscludetoolong');
3033                 return $this->fetchScaryTemplateMaybeFromCache($url);
3034         }
3035
3036         function fetchScaryTemplateMaybeFromCache($url) {
3037                 global $wgTranscludeCacheExpiry;
3038                 $dbr =& wfGetDB(DB_SLAVE);
3039                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3040                                 array('tc_url' => $url));
3041                 if ($obj) {
3042                         $time = $obj->tc_time;
3043                         $text = $obj->tc_contents;
3044                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3045                                 return $text;
3046                         }
3047                 }
3048
3049                 $text = Http::get($url);
3050                 if (!$text)
3051                         return wfMsg('scarytranscludefailed', $url);
3052
3053                 $dbw =& wfGetDB(DB_MASTER);
3054                 $dbw->replace('transcache', array('tc_url'), array(
3055                         'tc_url' => $url,
3056                         'tc_time' => time(),
3057                         'tc_contents' => $text));
3058                 return $text;
3059         }
3060
3061
3062         /**
3063          * Triple brace replacement -- used for template arguments
3064          * @private
3065          */
3066         function argSubstitution( $matches ) {
3067                 $arg = trim( $matches['title'] );
3068                 $text = $matches['text'];
3069                 $inputArgs = end( $this->mArgStack );
3070
3071                 if ( array_key_exists( $arg, $inputArgs ) ) {
3072                         $text = $inputArgs[$arg];
3073                 } else if ($this->mOutputType == OT_HTML && null != $matches['parts'] && count($matches['parts']) > 0) {
3074                         $text = $matches['parts'][0];
3075                 }
3076
3077                 return $text;
3078         }
3079
3080         /**
3081          * Returns true if the function is allowed to include this entity
3082          * @private
3083          */
3084         function incrementIncludeCount( $dbk ) {
3085                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
3086                         $this->mIncludeCount[$dbk] = 0;
3087                 }
3088                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
3089                         return true;
3090                 } else {
3091                         return false;
3092                 }
3093         }
3094
3095         /**
3096          * Detect __NOGALLERY__ magic word and set a placeholder
3097          */
3098         function stripNoGallery( &$text ) {
3099                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3100                 # do not add TOC
3101                 $mw = MagicWord::get( MAG_NOGALLERY );
3102                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3103         }
3104
3105         /**
3106          * Detect __TOC__ magic word and set a placeholder
3107          */
3108         function stripToc( $text ) {
3109                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3110                 # do not add TOC
3111                 $mw = MagicWord::get( MAG_NOTOC );
3112                 if( $mw->matchAndRemove( $text ) ) {
3113                         $this->mShowToc = false;
3114                 }
3115
3116                 $mw = MagicWord::get( MAG_TOC );
3117                 if( $mw->match( $text ) ) {
3118                         $this->mShowToc = true;
3119                         $this->mForceTocPosition = true;
3120
3121                         // Set a placeholder. At the end we'll fill it in with the TOC.
3122                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3123
3124                         // Only keep the first one.
3125                         $text = $mw->replace( '', $text );
3126                 }
3127                 return $text;
3128         }
3129
3130         /**
3131          * This function accomplishes several tasks:
3132          * 1) Auto-number headings if that option is enabled
3133          * 2) Add an [edit] link to sections for logged in users who have enabled the option
3134          * 3) Add a Table of contents on the top for users who have enabled the option
3135          * 4) Auto-anchor headings
3136          *
3137          * It loops through all headlines, collects the necessary data, then splits up the
3138          * string and re-inserts the newly formatted headlines.
3139          *
3140          * @param string $text
3141          * @param boolean $isMain
3142          * @private
3143          */
3144         function formatHeadings( $text, $isMain=true ) {
3145                 global $wgMaxTocLevel, $wgContLang;
3146
3147                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3148                 if( !$this->mTitle->userCanEdit() ) {
3149                         $showEditLink = 0;
3150                 } else {
3151                         $showEditLink = $this->mOptions->getEditSection();
3152                 }
3153
3154                 # Inhibit editsection links if requested in the page
3155                 $esw =& MagicWord::get( MAG_NOEDITSECTION );
3156                 if( $esw->matchAndRemove( $text ) ) {
3157                         $showEditLink = 0;
3158                 }
3159
3160                 # Get all headlines for numbering them and adding funky stuff like [edit]
3161                 # links - this is for later, but we need the number of headlines right now
3162                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
3163
3164                 # if there are fewer than 4 headlines in the article, do not show TOC
3165                 # unless it's been explicitly enabled.
3166                 $enoughToc = $this->mShowToc &&
3167                         (($numMatches >= 4) || $this->mForceTocPosition);
3168
3169                 # Allow user to stipulate that a page should have a "new section"
3170                 # link added via __NEWSECTIONLINK__
3171                 $mw =& MagicWord::get( MAG_NEWSECTIONLINK );
3172                 if( $mw->matchAndRemove( $text ) )
3173                         $this->mOutput->setNewSection( true );
3174
3175                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3176                 # override above conditions and always show TOC above first header
3177                 $mw =& MagicWord::get( MAG_FORCETOC );
3178                 if ($mw->matchAndRemove( $text ) ) {
3179                         $this->mShowToc = true;
3180                         $enoughToc = true;
3181                 }
3182
3183                 # Never ever show TOC if no headers
3184                 if( $numMatches < 1 ) {
3185                         $enoughToc = false;
3186                 }
3187
3188                 # We need this to perform operations on the HTML
3189                 $sk =& $this->mOptions->getSkin();
3190
3191                 # headline counter
3192                 $headlineCount = 0;
3193                 $sectionCount = 0; # headlineCount excluding template sections
3194
3195                 # Ugh .. the TOC should have neat indentation levels which can be
3196                 # passed to the skin functions. These are determined here
3197                 $toc = '';
3198                 $full = '';
3199                 $head = array();
3200                 $sublevelCount = array();
3201                 $levelCount = array();
3202                 $toclevel = 0;
3203                 $level = 0;
3204                 $prevlevel = 0;
3205                 $toclevel = 0;
3206                 $prevtoclevel = 0;
3207
3208                 foreach( $matches[3] as $headline ) {
3209                         $istemplate = 0;
3210                         $templatetitle = '';
3211                         $templatesection = 0;
3212                         $numbering = '';
3213
3214                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3215                                 $istemplate = 1;
3216                                 $templatetitle = base64_decode($mat[1]);
3217                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3218                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3219                         }
3220
3221                         if( $toclevel ) {
3222                                 $prevlevel = $level;
3223                                 $prevtoclevel = $toclevel;
3224                         }
3225                         $level = $matches[1][$headlineCount];
3226
3227                         if( $doNumberHeadings || $enoughToc ) {
3228
3229                                 if ( $level > $prevlevel ) {
3230                                         # Increase TOC level
3231                                         $toclevel++;
3232                                         $sublevelCount[$toclevel] = 0;
3233                                         if( $toclevel<$wgMaxTocLevel ) {
3234                                                 $toc .= $sk->tocIndent();
3235                                         }
3236                                 }
3237                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3238                                         # Decrease TOC level, find level to jump to
3239
3240                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3241                                                 # Can only go down to level 1
3242                                                 $toclevel = 1;
3243                                         } else {
3244                                                 for ($i = $toclevel; $i > 0; $i--) {
3245                                                         if ( $levelCount[$i] == $level ) {
3246                                                                 # Found last matching level
3247                                                                 $toclevel = $i;
3248                                                                 break;
3249                                                         }
3250                                                         elseif ( $levelCount[$i] < $level ) {
3251                                                                 # Found first matching level below current level
3252                                                                 $toclevel = $i + 1;
3253                                                                 break;
3254                                                         }
3255                                                 }
3256                                         }
3257                                         if( $toclevel<$wgMaxTocLevel ) {
3258                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3259                                         }
3260                                 }
3261                                 else {
3262                                         # No change in level, end TOC line
3263                                         if( $toclevel<$wgMaxTocLevel ) {
3264                                                 $toc .= $sk->tocLineEnd();
3265                                         }
3266                                 }
3267
3268                                 $levelCount[$toclevel] = $level;
3269
3270                                 # count number of headlines for each level
3271                                 @$sublevelCount[$toclevel]++;
3272                                 $dot = 0;
3273                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3274                                         if( !empty( $sublevelCount[$i] ) ) {
3275                                                 if( $dot ) {
3276                                                         $numbering .= '.';
3277                                                 }
3278                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3279                                                 $dot = 1;
3280                                         }
3281                                 }
3282                         }
3283
3284                         # The canonized header is a version of the header text safe to use for links
3285                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3286                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
3287                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
3288
3289                         # Remove link placeholders by the link text.
3290                         #     <!--LINK number-->
3291                         # turns into
3292                         #     link text with suffix
3293                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3294                                                             "\$this->mLinkHolders['texts'][\$1]",
3295                                                             $canonized_headline );
3296                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3297                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3298                                                             $canonized_headline );
3299
3300                         # strip out HTML
3301                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3302                         $tocline = trim( $canonized_headline );
3303                         # Save headline for section edit hint before it's escaped
3304                         $headline_hint = trim( $canonized_headline );
3305                         $canonized_headline = Sanitizer::escapeId( $tocline );
3306                         $refers[$headlineCount] = $canonized_headline;
3307
3308                         # count how many in assoc. array so we can track dupes in anchors
3309                         @$refers[$canonized_headline]++;
3310                         $refcount[$headlineCount]=$refers[$canonized_headline];
3311
3312                         # Don't number the heading if it is the only one (looks silly)
3313                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3314                                 # the two are different if the line contains a link
3315                                 $headline=$numbering . ' ' . $headline;
3316                         }
3317
3318                         # Create the anchor for linking from the TOC to the section
3319                         $anchor = $canonized_headline;
3320                         if($refcount[$headlineCount] > 1 ) {
3321                                 $anchor .= '_' . $refcount[$headlineCount];
3322                         }
3323                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3324                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3325                         }
3326                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3327                                 if ( empty( $head[$headlineCount] ) ) {
3328                                         $head[$headlineCount] = '';
3329                                 }
3330                                 if( $istemplate )
3331                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
3332                                 else
3333                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3334                         }
3335
3336                         # give headline the correct <h#> tag
3337                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
3338
3339                         $headlineCount++;
3340                         if( !$istemplate )
3341                                 $sectionCount++;
3342                 }
3343
3344                 if( $enoughToc ) {
3345                         if( $toclevel<$wgMaxTocLevel ) {
3346                                 $toc .= $sk->tocUnindent( $toclevel - 1 );
3347                         }
3348                         $toc = $sk->tocList( $toc );
3349                 }
3350
3351                 # split up and insert constructed headlines
3352
3353                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3354                 $i = 0;
3355
3356                 foreach( $blocks as $block ) {
3357                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3358                                 # This is the [edit] link that appears for the top block of text when
3359                                 # section editing is enabled
3360
3361                                 # Disabled because it broke block formatting
3362                                 # For example, a bullet point in the top line
3363                                 # $full .= $sk->editSectionLink(0);
3364                         }
3365                         $full .= $block;
3366                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3367                                 # Top anchor now in skin
3368                                 $full = $full.$toc;
3369                         }
3370
3371                         if( !empty( $head[$i] ) ) {
3372                                 $full .= $head[$i];
3373                         }
3374                         $i++;
3375                 }
3376                 if( $this->mForceTocPosition ) {
3377                         return str_replace( '<!--MWTOC-->', $toc, $full );
3378                 } else {
3379                         return $full;
3380                 }
3381         }
3382
3383         /**
3384          * Return an HTML link for the "ISBN 123456" text
3385          * @private
3386          */
3387         function magicISBN( $text ) {
3388                 $fname = 'Parser::magicISBN';
3389                 wfProfileIn( $fname );
3390
3391                 $a = split( 'ISBN ', ' '.$text );
3392                 if ( count ( $a ) < 2 ) {
3393                         wfProfileOut( $fname );
3394                         return $text;
3395                 }
3396                 $text = substr( array_shift( $a ), 1);
3397                 $valid = '0123456789-Xx';
3398
3399                 foreach ( $a as $x ) {
3400                         # hack: don't replace inside thumbnail title/alt
3401                         # attributes
3402                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3403                                 $text .= "ISBN $x";
3404                                 continue;
3405                         }
3406
3407                         $isbn = $blank = '' ;
3408                         while ( $x !== '' && ' ' == $x{0} ) {
3409                                 $blank .= ' ';
3410                                 $x = substr( $x, 1 );
3411                         }
3412                         if ( $x == '' ) { # blank isbn
3413                                 $text .= "ISBN $blank";
3414                                 continue;
3415                         }
3416                         while ( strstr( $valid, $x{0} ) != false ) {
3417                                 $isbn .= $x{0};
3418                                 $x = substr( $x, 1 );
3419                         }
3420                         $num = str_replace( '-', '', $isbn );
3421                         $num = str_replace( ' ', '', $num );
3422                         $num = str_replace( 'x', 'X', $num );
3423
3424                         if ( '' == $num ) {
3425                                 $text .= "ISBN $blank$x";
3426                         } else {
3427                                 $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
3428                                 $text .= '<a href="' .
3429                                         $titleObj->escapeLocalUrl( 'isbn='.$num ) .
3430                                         "\" class=\"internal\">ISBN $isbn</a>";
3431                                 $text .= $x;
3432                         }
3433                 }
3434                 wfProfileOut( $fname );
3435                 return $text;
3436         }
3437
3438         /**
3439          * Return an HTML link for the "RFC 1234" text
3440          *
3441          * @private
3442          * @param string $text     Text to be processed
3443          * @param string $keyword  Magic keyword to use (default RFC)
3444          * @param string $urlmsg   Interface message to use (default rfcurl)
3445          * @return string
3446          */
3447         function magicRFC( $text, $keyword='RFC ', $urlmsg='rfcurl'  ) {
3448
3449                 $valid = '0123456789';
3450                 $internal = false;
3451
3452                 $a = split( $keyword, ' '.$text );
3453                 if ( count ( $a ) < 2 ) {
3454                         return $text;
3455                 }
3456                 $text = substr( array_shift( $a ), 1);
3457
3458                 /* Check if keyword is preceed by [[.
3459                  * This test is made here cause of the array_shift above
3460                  * that prevent the test to be done in the foreach.
3461                  */
3462                 if ( substr( $text, -2 ) == '[[' ) {
3463                         $internal = true;
3464                 }
3465
3466                 foreach ( $a as $x ) {
3467                         /* token might be empty if we have RFC RFC 1234 */
3468                         if ( $x=='' ) {
3469                                 $text.=$keyword;
3470                                 continue;
3471                                 }
3472
3473                         # hack: don't replace inside thumbnail title/alt
3474                         # attributes
3475                         if(preg_match('/<[^>]+(alt|title)="[^">]*$/', $text)) {
3476                                 $text .= $keyword . $x;
3477                                 continue;
3478                         }
3479
3480                         $id = $blank = '' ;
3481
3482                         /** remove and save whitespaces in $blank */
3483                         while ( $x{0} == ' ' ) {
3484                                 $blank .= ' ';
3485                                 $x = substr( $x, 1 );
3486                         }
3487
3488                         /** remove and save the rfc number in $id */
3489                         while ( strstr( $valid, $x{0} ) != false ) {
3490                                 $id .= $x{0};
3491                                 $x = substr( $x, 1 );
3492                         }
3493
3494                         if ( $id == '' ) {
3495                                 /* call back stripped spaces*/
3496                                 $text .= $keyword.$blank.$x;
3497                         } elseif( $internal ) {
3498                                 /* normal link */
3499                                 $text .= $keyword.$id.$x;
3500                         } else {
3501                                 /* build the external link*/
3502                                 $url = wfMsg( $urlmsg, $id);
3503                                 $sk =& $this->mOptions->getSkin();
3504                                 $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
3505                                 $text .= "<a href=\"{$url}\"{$la}>{$keyword}{$id}</a>{$x}";
3506                         }
3507
3508                         /* Check if the next RFC keyword is preceed by [[ */
3509                         $internal = ( substr($x,-2) == '[[' );
3510                 }
3511                 return $text;
3512         }
3513
3514         /**
3515          * Transform wiki markup when saving a page by doing \r\n -> \n
3516          * conversion, substitting signatures, {{subst:}} templates, etc.
3517          *
3518          * @param string $text the text to transform
3519          * @param Title &$title the Title object for the current article
3520          * @param User &$user the User object describing the current user
3521          * @param ParserOptions $options parsing options
3522          * @param bool $clearState whether to clear the parser state first
3523          * @return string the altered wiki markup
3524          * @public
3525          */
3526         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
3527                 $this->mOptions = $options;
3528                 $this->mTitle =& $title;
3529                 $this->mOutputType = OT_WIKI;
3530
3531                 if ( $clearState ) {
3532                         $this->clearState();
3533                 }
3534
3535                 $stripState = false;
3536                 $pairs = array(
3537                         "\r\n" => "\n",
3538                 );
3539                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3540                 $text = $this->strip( $text, $stripState, true, array( 'gallery' ) );
3541                 $text = $this->pstPass2( $text, $stripState, $user );
3542                 $text = $this->unstrip( $text, $stripState );
3543                 $text = $this->unstripNoWiki( $text, $stripState );
3544                 return $text;
3545         }
3546
3547         /**
3548          * Pre-save transform helper function
3549          * @private
3550          */
3551         function pstPass2( $text, &$stripState, &$user ) {
3552                 global $wgContLang, $wgLocaltimezone;
3553
3554                 /* Note: This is the timestamp saved as hardcoded wikitext to
3555                  * the database, we use $wgContLang here in order to give
3556                  * everyone the same signature and use the default one rather
3557                  * than the one selected in each user's preferences.
3558                  */
3559                 if ( isset( $wgLocaltimezone ) ) {
3560                         $oldtz = getenv( 'TZ' );
3561                         putenv( 'TZ='.$wgLocaltimezone );
3562                 }
3563                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3564                   ' (' . date( 'T' ) . ')';
3565                 if ( isset( $wgLocaltimezone ) ) {
3566                         putenv( 'TZ='.$oldtz );
3567                 }
3568
3569                 # Variable replacement
3570                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3571                 $text = $this->replaceVariables( $text );
3572
3573                 # Strip out <nowiki> etc. added via replaceVariables
3574                 $text = $this->strip( $text, $stripState, false, array( 'gallery' ) );
3575
3576                 # Signatures
3577                 $sigText = $this->getUserSig( $user );
3578                 $text = strtr( $text, array(
3579                         '~~~~~' => $d,
3580                         '~~~~' => "$sigText $d",
3581                         '~~~' => $sigText
3582                 ) );
3583
3584                 # Context links: [[|name]] and [[name (context)|]]
3585                 #
3586                 global $wgLegalTitleChars;
3587                 $tc = "[$wgLegalTitleChars]";
3588                 $np = str_replace( array( '(', ')' ), array( '', '' ), $tc ); # No parens
3589
3590                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3591                 $conpat = "/^({$np}+) \\(({$tc}+)\\)$/";
3592
3593                 $p1 = "/\[\[({$np}+) \\(({$np}+)\\)\\|]]/";             # [[page (context)|]]
3594                 $p2 = "/\[\[\\|({$tc}+)]]/";                                    # [[|page]]
3595                 $p3 = "/\[\[(:*$namespacechar+):({$np}+)\\|]]/";                # [[namespace:page|]] and [[:namespace:page|]]
3596                 $p4 = "/\[\[(:*$namespacechar+):({$np}+) \\(({$np}+)\\)\\|]]/"; # [[ns:page (cont)|]] and [[:ns:page (cont)|]]
3597                 $context = '';
3598                 $t = $this->mTitle->getText();
3599                 if ( preg_match( $conpat, $t, $m ) ) {
3600                         $context = $m[2];
3601                 }
3602                 $text = preg_replace( $p4, '[[\\1:\\2 (\\3)|\\2]]', $text );
3603                 $text = preg_replace( $p1, '[[\\1 (\\2)|\\1]]', $text );
3604                 $text = preg_replace( $p3, '[[\\1:\\2|\\2]]', $text );
3605
3606                 if ( '' == $context ) {
3607                         $text = preg_replace( $p2, '[[\\1]]', $text );
3608                 } else {
3609                         $text = preg_replace( $p2, "[[\\1 ({$context})|\\1]]", $text );
3610                 }
3611
3612                 # Trim trailing whitespace
3613                 # MAG_END (__END__) tag allows for trailing
3614                 # whitespace to be deliberately included
3615                 $text = rtrim( $text );
3616                 $mw =& MagicWord::get( MAG_END );
3617                 $mw->matchAndRemove( $text );
3618
3619                 return $text;
3620         }
3621
3622         /**
3623          * Fetch the user's signature text, if any, and normalize to
3624          * validated, ready-to-insert wikitext.
3625          *
3626          * @param User $user
3627          * @return string
3628          * @private
3629          */
3630         function getUserSig( &$user ) {
3631                 $username = $user->getName();
3632                 $nickname = $user->getOption( 'nickname' );
3633                 $nickname = $nickname === '' ? $username : $nickname;
3634
3635                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3636                         # Sig. might contain markup; validate this
3637                         if( $this->validateSig( $nickname ) !== false ) {
3638                                 # Validated; clean up (if needed) and return it
3639                                 return $this->cleanSig( $nickname, true );
3640                         } else {
3641                                 # Failed to validate; fall back to the default
3642                                 $nickname = $username;
3643                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3644                         }
3645                 }
3646
3647                 // Make sure nickname doesnt get a sig in a sig
3648                 $nickname = $this->cleanSigInSig( $nickname );
3649
3650                 # If we're still here, make it a link to the user page
3651                 $userpage = $user->getUserPage();
3652                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3653         }
3654
3655         /**
3656          * Check that the user's signature contains no bad XML
3657          *
3658          * @param string $text
3659          * @return mixed An expanded string, or false if invalid.
3660          */
3661         function validateSig( $text ) {
3662                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3663         }
3664
3665         /**
3666          * Clean up signature text
3667          *
3668          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3669          * 2) Substitute all transclusions
3670          *
3671          * @param string $text
3672          * @param $parsing Whether we're cleaning (preferences save) or parsing
3673          * @return string Signature text
3674          */
3675         function cleanSig( $text, $parsing = false ) {
3676                 global $wgTitle;
3677                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
3678
3679                 $substWord = MagicWord::get( MAG_SUBST );
3680                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3681                 $substText = '{{' . $substWord->getSynonym( 0 );
3682
3683                 $text = preg_replace( $substRegex, $substText, $text );
3684                 $text = $this->cleanSigInSig( $text );
3685                 $text = $this->replaceVariables( $text );
3686
3687                 $this->clearState();
3688                 return $text;
3689         }
3690
3691         /**
3692          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3693          * @param string $text
3694          * @return string Signature text with /~{3,5}/ removed
3695          */
3696         function cleanSigInSig( $text ) {
3697                 $text = preg_replace( '/~{3,5}/', '', $text );
3698                 return $text;
3699         }
3700
3701         /**
3702          * Set up some variables which are usually set up in parse()
3703          * so that an external function can call some class members with confidence
3704          * @public
3705          */
3706         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3707                 $this->mTitle =& $title;
3708                 $this->mOptions = $options;
3709                 $this->mOutputType = $outputType;
3710                 if ( $clearState ) {
3711                         $this->clearState();
3712                 }
3713         }
3714
3715         /**
3716          * Transform a MediaWiki message by replacing magic variables.
3717          *
3718          * @param string $text the text to transform
3719          * @param ParserOptions $options  options
3720          * @return string the text with variables substituted
3721          * @public
3722          */
3723         function transformMsg( $text, $options ) {
3724                 global $wgTitle;
3725                 static $executing = false;
3726
3727                 $fname = "Parser::transformMsg";
3728
3729                 # Guard against infinite recursion
3730                 if ( $executing ) {
3731                         return $text;
3732                 }
3733                 $executing = true;
3734
3735                 wfProfileIn($fname);
3736
3737                 $this->mTitle = $wgTitle;
3738                 $this->mOptions = $options;
3739                 $this->mOutputType = OT_MSG;
3740                 $this->clearState();
3741                 $text = $this->replaceVariables( $text );
3742
3743                 $executing = false;
3744                 wfProfileOut($fname);
3745                 return $text;
3746         }
3747
3748         /**
3749          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3750          * The callback should have the following form:
3751          *    function myParserHook( $text, $params, &$parser ) { ... }
3752          *
3753          * Transform and return $text. Use $parser for any required context, e.g. use
3754          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3755          *
3756          * @public
3757          *
3758          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3759          * @param mixed $callback The callback function (and object) to use for the tag
3760          *
3761          * @return The old value of the mTagHooks array associated with the hook
3762          */
3763         function setHook( $tag, $callback ) {
3764                 $tag = strtolower( $tag );
3765                 $oldVal = @$this->mTagHooks[$tag];
3766                 $this->mTagHooks[$tag] = $callback;
3767
3768                 return $oldVal;
3769         }
3770
3771         /**
3772          * Create a function, e.g. {{sum:1|2|3}}
3773          * The callback function should have the form:
3774          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3775          *
3776          * The callback may either return the text result of the function, or an array with the text
3777          * in element 0, and a number of flags in the other elements. The names of the flags are
3778          * specified in the keys. Valid flags are:
3779          *   found                     The text returned is valid, stop processing the template. This
3780          *                             is on by default.
3781          *   nowiki                    Wiki markup in the return value should be escaped
3782          *   noparse                   Unsafe HTML tags should not be stripped, etc.
3783          *   noargs                    Don't replace triple-brace arguments in the return value
3784          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3785          *
3786          * @public
3787          *
3788          * @param mixed $id The magic word ID, or (deprecated) the function name. Function names are case-insensitive.
3789          * @param mixed $callback The callback function (and object) to use
3790          * @param integer $flags a combination of the following flags:
3791          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
3792          *
3793          * @return The old callback function for this name, if any
3794          */
3795         function setFunctionHook( $id, $callback, $flags = 0 ) {
3796                 if( is_string( $id ) ) {
3797                         $id = strtolower( $id );
3798                 }
3799                 $oldVal = @$this->mFunctionHooks[$id];
3800                 $this->mFunctionHooks[$id] = $callback;
3801
3802                 # Add to function cache
3803                 if ( is_int( $id ) ) {
3804                         $mw = MagicWord::get( $id );
3805                         $synonyms = $mw->getSynonyms();
3806                         $sensitive = intval( $mw->isCaseSensitive() );
3807                 } else {
3808                         $synonyms = array( $id );
3809                         $sensitive = 0;
3810                 }
3811
3812                 foreach ( $synonyms as $syn ) {
3813                         # Case
3814                         if ( !$sensitive ) {
3815                                 $syn = strtolower( $syn );
3816                         }
3817                         # Add leading hash
3818                         if ( !( $flags & SFH_NO_HASH ) ) {
3819                                 $syn = '#' . $syn;
3820                         }
3821                         # Remove trailing colon
3822                         if ( substr( $syn, -1, 1 ) == ':' ) {
3823                                 $syn = substr( $syn, 0, -1 );
3824                         }
3825                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
3826                 }
3827                 return $oldVal;
3828         }
3829
3830         /**
3831          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3832          * Placeholders created in Skin::makeLinkObj()
3833          * Returns an array of links found, indexed by PDBK:
3834          *  0 - broken
3835          *  1 - normal link
3836          *  2 - stub
3837          * $options is a bit field, RLH_FOR_UPDATE to select for update
3838          */
3839         function replaceLinkHolders( &$text, $options = 0 ) {
3840                 global $wgUser;
3841                 global $wgOutputReplace;
3842
3843                 $fname = 'Parser::replaceLinkHolders';
3844                 wfProfileIn( $fname );
3845
3846                 $pdbks = array();
3847                 $colours = array();
3848                 $sk =& $this->mOptions->getSkin();
3849                 $linkCache =& LinkCache::singleton();
3850
3851                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3852                         wfProfileIn( $fname.'-check' );
3853                         $dbr =& wfGetDB( DB_SLAVE );
3854                         $page = $dbr->tableName( 'page' );
3855                         $threshold = $wgUser->getOption('stubthreshold');
3856
3857                         # Sort by namespace
3858                         asort( $this->mLinkHolders['namespaces'] );
3859
3860                         # Generate query
3861                         $query = false;
3862                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3863                                 # Make title object
3864                                 $title = $this->mLinkHolders['titles'][$key];
3865
3866                                 # Skip invalid entries.
3867                                 # Result will be ugly, but prevents crash.
3868                                 if ( is_null( $title ) ) {
3869                                         continue;
3870                                 }
3871                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3872
3873                                 # Check if it's a static known link, e.g. interwiki
3874                                 if ( $title->isAlwaysKnown() ) {
3875                                         $colours[$pdbk] = 1;
3876                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3877                                         $colours[$pdbk] = 1;
3878                                         $this->mOutput->addLink( $title, $id );
3879                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3880                                         $colours[$pdbk] = 0;
3881                                 } else {
3882                                         # Not in the link cache, add it to the query
3883                                         if ( !isset( $current ) ) {
3884                                                 $current = $ns;
3885                                                 $query =  "SELECT page_id, page_namespace, page_title";
3886                                                 if ( $threshold > 0 ) {
3887                                                         $query .= ', page_len, page_is_redirect';
3888                                                 }
3889                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
3890                                         } elseif ( $current != $ns ) {
3891                                                 $current = $ns;
3892                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
3893                                         } else {
3894                                                 $query .= ', ';
3895                                         }
3896
3897                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3898                                 }
3899                         }
3900                         if ( $query ) {
3901                                 $query .= '))';
3902                                 if ( $options & RLH_FOR_UPDATE ) {
3903                                         $query .= ' FOR UPDATE';
3904                                 }
3905
3906                                 $res = $dbr->query( $query, $fname );
3907
3908                                 # Fetch data and form into an associative array
3909                                 # non-existent = broken
3910                                 # 1 = known
3911                                 # 2 = stub
3912                                 while ( $s = $dbr->fetchObject($res) ) {
3913                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
3914                                         $pdbk = $title->getPrefixedDBkey();
3915                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
3916                                         $this->mOutput->addLink( $title, $s->page_id );
3917
3918                                         if ( $threshold >  0 ) {
3919                                                 $size = $s->page_len;
3920                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
3921                                                         $colours[$pdbk] = 1;
3922                                                 } else {
3923                                                         $colours[$pdbk] = 2;
3924                                                 }
3925                                         } else {
3926                                                 $colours[$pdbk] = 1;
3927                                         }
3928                                 }
3929                         }
3930                         wfProfileOut( $fname.'-check' );
3931
3932                         # Construct search and replace arrays
3933                         wfProfileIn( $fname.'-construct' );
3934                         $wgOutputReplace = array();
3935                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3936                                 $pdbk = $pdbks[$key];
3937                                 $searchkey = "<!--LINK $key-->";
3938                                 $title = $this->mLinkHolders['titles'][$key];
3939                                 if ( empty( $colours[$pdbk] ) ) {
3940                                         $linkCache->addBadLinkObj( $title );
3941                                         $colours[$pdbk] = 0;
3942                                         $this->mOutput->addLink( $title, 0 );
3943                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
3944                                                                         $this->mLinkHolders['texts'][$key],
3945                                                                         $this->mLinkHolders['queries'][$key] );
3946                                 } elseif ( $colours[$pdbk] == 1 ) {
3947                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
3948                                                                         $this->mLinkHolders['texts'][$key],
3949                                                                         $this->mLinkHolders['queries'][$key] );
3950                                 } elseif ( $colours[$pdbk] == 2 ) {
3951                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
3952                                                                         $this->mLinkHolders['texts'][$key],
3953                                                                         $this->mLinkHolders['queries'][$key] );
3954                                 }
3955                         }
3956                         wfProfileOut( $fname.'-construct' );
3957
3958                         # Do the thing
3959                         wfProfileIn( $fname.'-replace' );
3960
3961                         $text = preg_replace_callback(
3962                                 '/(<!--LINK .*?-->)/',
3963                                 "wfOutputReplaceMatches",
3964                                 $text);
3965
3966                         wfProfileOut( $fname.'-replace' );
3967                 }
3968
3969                 # Now process interwiki link holders
3970                 # This is quite a bit simpler than internal links
3971                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
3972                         wfProfileIn( $fname.'-interwiki' );
3973                         # Make interwiki link HTML
3974                         $wgOutputReplace = array();
3975                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
3976                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
3977                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
3978                         }
3979
3980                         $text = preg_replace_callback(
3981                                 '/<!--IWLINK (.*?)-->/',
3982                                 "wfOutputReplaceMatches",
3983                                 $text );
3984                         wfProfileOut( $fname.'-interwiki' );
3985                 }
3986
3987                 wfProfileOut( $fname );
3988                 return $colours;
3989         }
3990
3991         /**
3992          * Replace <!--LINK--> link placeholders with plain text of links
3993          * (not HTML-formatted).
3994          * @param string $text
3995          * @return string
3996          */
3997         function replaceLinkHoldersText( $text ) {
3998                 $fname = 'Parser::replaceLinkHoldersText';
3999                 wfProfileIn( $fname );
4000
4001                 $text = preg_replace_callback(
4002                         '/<!--(LINK|IWLINK) (.*?)-->/',
4003                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4004                         $text );
4005
4006                 wfProfileOut( $fname );
4007                 return $text;
4008         }
4009
4010         /**
4011          * @param array $matches
4012          * @return string
4013          * @private
4014          */
4015         function replaceLinkHoldersTextCallback( $matches ) {
4016                 $type = $matches[1];
4017                 $key  = $matches[2];
4018                 if( $type == 'LINK' ) {
4019                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4020                                 return $this->mLinkHolders['texts'][$key];
4021                         }
4022                 } elseif( $type == 'IWLINK' ) {
4023                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4024                                 return $this->mInterwikiLinkHolders['texts'][$key];
4025                         }
4026                 }
4027                 return $matches[0];
4028         }
4029
4030         /**
4031          * Tag hook handler for 'pre'.
4032          */
4033         function renderPreTag( $text, $attribs, $parser ) {
4034                 // Backwards-compatibility hack
4035                 $content = preg_replace( '!<nowiki>(.*?)</nowiki>!is', '\\1', $text );
4036
4037                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4038                 return wfOpenElement( 'pre', $attribs ) .
4039                         wfEscapeHTMLTagsOnly( $content ) .
4040                         '</pre>';
4041         }
4042
4043         /**
4044          * Renders an image gallery from a text with one line per image.
4045          * text labels may be given by using |-style alternative text. E.g.
4046          *   Image:one.jpg|The number "1"
4047          *   Image:tree.jpg|A tree
4048          * given as text will return the HTML of a gallery with two images,
4049          * labeled 'The number "1"' and
4050          * 'A tree'.
4051          */
4052         function renderImageGallery( $text, $params ) {
4053                 $ig = new ImageGallery();
4054                 $ig->setShowBytes( false );
4055                 $ig->setShowFilename( false );
4056                 $ig->setParsing();
4057                 $ig->useSkin( $this->mOptions->getSkin() );
4058
4059                 if( isset( $params['caption'] ) )
4060                         $ig->setCaption( $params['caption'] );
4061
4062                 $lines = explode( "\n", $text );
4063                 foreach ( $lines as $line ) {
4064                         # match lines like these:
4065                         # Image:someimage.jpg|This is some image
4066                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4067                         # Skip empty lines
4068                         if ( count( $matches ) == 0 ) {
4069                                 continue;
4070                         }
4071                         $nt =& Title::newFromText( $matches[1] );
4072                         if( is_null( $nt ) ) {
4073                                 # Bogus title. Ignore these so we don't bomb out later.
4074                                 continue;
4075                         }
4076                         if ( isset( $matches[3] ) ) {
4077                                 $label = $matches[3];
4078                         } else {
4079                                 $label = '';
4080                         }
4081
4082                         $pout = $this->parse( $label,
4083                                 $this->mTitle,
4084                                 $this->mOptions,
4085                                 false, // Strip whitespace...?
4086                                 false  // Don't clear state!
4087                         );
4088                         $html = $pout->getText();
4089
4090                         $ig->add( new Image( $nt ), $html );
4091
4092                         # Only add real images (bug #5586)
4093                         if ( $nt->getNamespace() == NS_IMAGE ) {
4094                                 $this->mOutput->addImage( $nt->getDBkey() );
4095                         }
4096                 }
4097                 return $ig->toHTML();
4098         }
4099
4100         /**
4101          * Parse image options text and use it to make an image
4102          */
4103         function makeImage( &$nt, $options ) {
4104                 global $wgUseImageResize;
4105
4106                 $align = '';
4107
4108                 # Check if the options text is of the form "options|alt text"
4109                 # Options are:
4110                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4111                 #  * left               no resizing, just left align. label is used for alt= only
4112                 #  * right              same, but right aligned
4113                 #  * none               same, but not aligned
4114                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4115                 #  * center             center the image
4116                 #  * framed             Keep original image size, no magnify-button.
4117
4118                 $part = explode( '|', $options);
4119
4120                 $mwThumb  =& MagicWord::get( MAG_IMG_THUMBNAIL );
4121                 $mwManualThumb =& MagicWord::get( MAG_IMG_MANUALTHUMB );
4122                 $mwLeft   =& MagicWord::get( MAG_IMG_LEFT );
4123                 $mwRight  =& MagicWord::get( MAG_IMG_RIGHT );
4124                 $mwNone   =& MagicWord::get( MAG_IMG_NONE );
4125                 $mwWidth  =& MagicWord::get( MAG_IMG_WIDTH );
4126                 $mwCenter =& MagicWord::get( MAG_IMG_CENTER );
4127                 $mwFramed =& MagicWord::get( MAG_IMG_FRAMED );
4128                 $caption = '';
4129
4130                 $width = $height = $framed = $thumb = false;
4131                 $manual_thumb = '' ;
4132
4133                 foreach( $part as $key => $val ) {
4134                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
4135                                 $thumb=true;
4136                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
4137                                 # use manually specified thumbnail
4138                                 $thumb=true;
4139                                 $manual_thumb = $match;
4140                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
4141                                 # remember to set an alignment, don't render immediately
4142                                 $align = 'right';
4143                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
4144                                 # remember to set an alignment, don't render immediately
4145                                 $align = 'left';
4146                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
4147                                 # remember to set an alignment, don't render immediately
4148                                 $align = 'center';
4149                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
4150                                 # remember to set an alignment, don't render immediately
4151                                 $align = 'none';
4152                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
4153                                 wfDebug( "MAG_IMG_WIDTH match: $match\n" );
4154                                 # $match is the image width in pixels
4155                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
4156                                         $width = intval( $m[1] );
4157                                         $height = intval( $m[2] );
4158                                 } else {
4159                                         $width = intval($match);
4160                                 }
4161                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
4162                                 $framed=true;
4163                         } else {
4164                                 $caption = $val;
4165                         }
4166                 }
4167                 # Strip bad stuff out of the alt text
4168                 $alt = $this->replaceLinkHoldersText( $caption );
4169
4170                 # make sure there are no placeholders in thumbnail attributes
4171                 # that are later expanded to html- so expand them now and
4172                 # remove the tags
4173                 $alt = $this->unstrip($alt, $this->mStripState);
4174                 $alt = Sanitizer::stripAllTags( $alt );
4175
4176                 # Linker does the rest
4177                 $sk =& $this->mOptions->getSkin();
4178                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
4179         }
4180
4181         /**
4182          * Set a flag in the output object indicating that the content is dynamic and
4183          * shouldn't be cached.
4184          */
4185         function disableCache() {
4186                 wfDebug( "Parser output marked as uncacheable.\n" );
4187                 $this->mOutput->mCacheTime = -1;
4188         }
4189
4190         /**#@+
4191          * Callback from the Sanitizer for expanding items found in HTML attribute
4192          * values, so they can be safely tested and escaped.
4193          * @param string $text
4194          * @param array $args
4195          * @return string
4196          * @private
4197          */
4198         function attributeStripCallback( &$text, $args ) {
4199                 $text = $this->replaceVariables( $text, $args );
4200                 $text = $this->unstripForHTML( $text );
4201                 return $text;
4202         }
4203
4204         function unstripForHTML( $text ) {
4205                 $text = $this->unstrip( $text, $this->mStripState );
4206                 $text = $this->unstripNoWiki( $text, $this->mStripState );
4207                 return $text;
4208         }
4209         /**#@-*/
4210
4211         /**#@+
4212          * Accessor/mutator
4213          */
4214         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4215         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4216         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4217         /**#@-*/
4218
4219         /**#@+
4220          * Accessor
4221          */
4222         function getTags() { return array_keys( $this->mTagHooks ); }
4223         /**#@-*/
4224
4225
4226         /**
4227          * Break wikitext input into sections, and either pull or replace
4228          * some particular section's text.
4229          *
4230          * External callers should use the getSection and replaceSection methods.
4231          *
4232          * @param $text Page wikitext
4233          * @param $section Numbered section. 0 pulls the text before the first
4234          *                 heading; other numbers will pull the given section
4235          *                 along with its lower-level subsections.
4236          * @param $mode One of "get" or "replace"
4237          * @param $newtext Replacement text for section data.
4238          * @return string for "get", the extracted section text.
4239          *                for "replace", the whole page with the section replaced.
4240          */
4241         private function extractSections( $text, $section, $mode, $newtext='' ) {
4242                 # strip NOWIKI etc. to avoid confusion (true-parameter causes HTML
4243                 # comments to be stripped as well)
4244                 $striparray = array();
4245
4246                 $oldOutputType = $this->mOutputType;
4247                 $oldOptions = $this->mOptions;
4248                 $this->mOptions = new ParserOptions();
4249                 $this->mOutputType = OT_WIKI;
4250
4251                 $striptext = $this->strip( $text, $striparray, true );
4252
4253                 $this->mOutputType = $oldOutputType;
4254                 $this->mOptions = $oldOptions;
4255
4256                 # now that we can be sure that no pseudo-sections are in the source,
4257                 # split it up by section
4258                 $uniq = preg_quote( $this->uniqPrefix(), '/' );
4259                 $comment = "(?:$uniq-!--.*?QINU)";
4260                 $secs = preg_split(
4261                 /*
4262                         "/
4263                         ^(
4264                         (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4265                         (?:
4266                                 (=+) # Should this be limited to 6?
4267                                 .+?  # Section title...
4268                                 \\2  # Ending = count must match start
4269                         |
4270                                 ^
4271                                 <h([1-6])\b.*?>
4272                                 .*?
4273                                 <\/h\\3\s*>
4274                         )
4275                         (?:$comment|<\/?noinclude>|\s+)* # Trailing whitespace ok
4276                         )$
4277                         /mix",
4278                 */
4279                         "/
4280                         (
4281                                 ^
4282                                 (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4283                                 (=+) # Should this be limited to 6?
4284                                 .+?  # Section title...
4285                                 \\2  # Ending = count must match start
4286                                 (?:$comment|<\/?noinclude>|[ \\t]+)* # Trailing whitespace ok
4287                                 $
4288                         |
4289                                 <h([1-6])\b.*?>
4290                                 .*?
4291                                 <\/h\\3\s*>
4292                         )
4293                         /mix",
4294                         $striptext, -1,
4295                         PREG_SPLIT_DELIM_CAPTURE);
4296
4297                 if( $mode == "get" ) {
4298                         if( $section == 0 ) {
4299                                 // "Section 0" returns the content before any other section.
4300                                 $rv = $secs[0];
4301                         } else {
4302                                 $rv = "";
4303                         }
4304                 } elseif( $mode == "replace" ) {
4305                         if( $section == 0 ) {
4306                                 $rv = $newtext . "\n\n";
4307                                 $remainder = true;
4308                         } else {
4309                                 $rv = $secs[0];
4310                                 $remainder = false;
4311                         }
4312                 }
4313                 $count = 0;
4314                 $sectionLevel = 0;
4315                 for( $index = 1; $index < count( $secs ); ) {
4316                         $headerLine = $secs[$index++];
4317                         if( $secs[$index] ) {
4318                                 // A wiki header
4319                                 $headerLevel = strlen( $secs[$index++] );
4320                         } else {
4321                                 // An HTML header
4322                                 $index++;
4323                                 $headerLevel = intval( $secs[$index++] );
4324                         }
4325                         $content = $secs[$index++];
4326
4327                         $count++;
4328                         if( $mode == "get" ) {
4329                                 if( $count == $section ) {
4330                                         $rv = $headerLine . $content;
4331                                         $sectionLevel = $headerLevel;
4332                                 } elseif( $count > $section ) {
4333                                         if( $sectionLevel && $headerLevel > $sectionLevel ) {
4334                                                 $rv .= $headerLine . $content;
4335                                         } else {
4336                                                 // Broke out to a higher-level section
4337                                                 break;
4338                                         }
4339                                 }
4340                         } elseif( $mode == "replace" ) {
4341                                 if( $count < $section ) {
4342                                         $rv .= $headerLine . $content;
4343                                 } elseif( $count == $section ) {
4344                                         $rv .= $newtext . "\n\n";
4345                                         $sectionLevel = $headerLevel;
4346                                 } elseif( $count > $section ) {
4347                                         if( $headerLevel <= $sectionLevel ) {
4348                                                 // Passed the section's sub-parts.
4349                                                 $remainder = true;
4350                                         }
4351                                         if( $remainder ) {
4352                                                 $rv .= $headerLine . $content;
4353                                         }
4354                                 }
4355                         }
4356                 }
4357                 # reinsert stripped tags
4358                 $rv = $this->unstrip( $rv, $striparray );
4359                 $rv = $this->unstripNoWiki( $rv, $striparray );
4360                 $rv = trim( $rv );
4361                 return $rv;
4362         }
4363
4364         /**
4365          * This function returns the text of a section, specified by a number ($section).
4366          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4367          * the first section before any such heading (section 0).
4368          *
4369          * If a section contains subsections, these are also returned.
4370          *
4371          * @param $text String: text to look in
4372          * @param $section Integer: section number
4373          * @return string text of the requested section
4374          */
4375         function getSection( $text, $section ) {
4376                 return $this->extractSections( $text, $section, "get" );
4377         }
4378
4379         function replaceSection( $oldtext, $section, $text ) {
4380                 return $this->extractSections( $oldtext, $section, "replace", $text );
4381         }
4382
4383 }
4384
4385 /**
4386  * @todo document
4387  * @package MediaWiki
4388  */
4389 class ParserOutput
4390 {
4391         var $mText,             # The output text
4392                 $mLanguageLinks,    # List of the full text of language links, in the order they appear
4393                 $mCategories,       # Map of category names to sort keys
4394                 $mContainsOldMagic, # Boolean variable indicating if the input contained variables like {{CURRENTDAY}}
4395                 $mCacheTime,        # Time when this object was generated, or -1 for uncacheable. Used in ParserCache.
4396                 $mVersion,          # Compatibility check
4397                 $mTitleText,        # title text of the chosen language variant
4398                 $mLinks,            # 2-D map of NS/DBK to ID for the links in the document. ID=zero for broken.
4399                 $mTemplates,        # 2-D map of NS/DBK to ID for the template references. ID=zero for broken.
4400                 $mImages,           # DB keys of the images used, in the array key only
4401                 $mExternalLinks,    # External link URLs, in the key only
4402                 $mHTMLtitle,            # Display HTML title
4403                 $mSubtitle,                     # Additional subtitle
4404                 $mNewSection,           # Show a new section link?
4405                 $mNoGallery;            # No gallery on category page? (__NOGALLERY__)
4406
4407         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
4408                 $containsOldMagic = false, $titletext = '' )
4409         {
4410                 $this->mText = $text;
4411                 $this->mLanguageLinks = $languageLinks;
4412                 $this->mCategories = $categoryLinks;
4413                 $this->mContainsOldMagic = $containsOldMagic;
4414                 $this->mCacheTime = '';
4415                 $this->mVersion = MW_PARSER_VERSION;
4416                 $this->mTitleText = $titletext;
4417                 $this->mLinks = array();
4418                 $this->mTemplates = array();
4419                 $this->mImages = array();
4420                 $this->mExternalLinks = array();
4421                 $this->mHTMLtitle = "" ;
4422                 $this->mSubtitle = "" ;
4423                 $this->mNewSection = false;
4424                 $this->mNoGallery = false;
4425         }
4426
4427         function getText()                   { return $this->mText; }
4428         function &getLanguageLinks()          { return $this->mLanguageLinks; }
4429         function getCategoryLinks()          { return array_keys( $this->mCategories ); }
4430         function &getCategories()            { return $this->mCategories; }
4431         function getCacheTime()              { return $this->mCacheTime; }
4432         function getTitleText()              { return $this->mTitleText; }
4433         function &getLinks()                 { return $this->mLinks; }
4434         function &getTemplates()             { return $this->mTemplates; }
4435         function &getImages()                { return $this->mImages; }
4436         function &getExternalLinks()         { return $this->mExternalLinks; }
4437         function getNoGallery()              { return $this->mNoGallery; }
4438
4439         function containsOldMagic()          { return $this->mContainsOldMagic; }
4440         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
4441         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
4442         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategories, $cl ); }
4443         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
4444         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
4445         function setTitleText( $t )          { return wfSetVar ($this->mTitleText, $t); }
4446
4447         function addCategory( $c, $sort )    { $this->mCategories[$c] = $sort; }
4448         function addImage( $name )           { $this->mImages[$name] = 1; }
4449         function addLanguageLink( $t )       { $this->mLanguageLinks[] = $t; }
4450         function addExternalLink( $url )     { $this->mExternalLinks[$url] = 1; }
4451
4452         function setNewSection( $value ) {
4453                 $this->mNewSection = (bool)$value;
4454         }
4455         function getNewSection() {
4456                 return (bool)$this->mNewSection;
4457         }
4458
4459         function addLink( $title, $id ) {
4460                 $ns = $title->getNamespace();
4461                 $dbk = $title->getDBkey();
4462                 if ( !isset( $this->mLinks[$ns] ) ) {
4463                         $this->mLinks[$ns] = array();
4464                 }
4465                 $this->mLinks[$ns][$dbk] = $id;
4466         }
4467
4468         function addTemplate( $title, $id ) {
4469                 $ns = $title->getNamespace();
4470                 $dbk = $title->getDBkey();
4471                 if ( !isset( $this->mTemplates[$ns] ) ) {
4472                         $this->mTemplates[$ns] = array();
4473                 }
4474                 $this->mTemplates[$ns][$dbk] = $id;
4475         }
4476
4477         /**
4478          * Return true if this cached output object predates the global or
4479          * per-article cache invalidation timestamps, or if it comes from
4480          * an incompatible older version.
4481          *
4482          * @param string $touched the affected article's last touched timestamp
4483          * @return bool
4484          * @public
4485          */
4486         function expired( $touched ) {
4487                 global $wgCacheEpoch;
4488                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
4489                        $this->getCacheTime() < $touched ||
4490                        $this->getCacheTime() <= $wgCacheEpoch ||
4491                        !isset( $this->mVersion ) ||
4492                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
4493         }
4494 }
4495
4496 /**
4497  * Set options of the Parser
4498  * @todo document
4499  * @package MediaWiki
4500  */
4501 class ParserOptions
4502 {
4503         # All variables are private
4504         var $mUseTeX;                    # Use texvc to expand <math> tags
4505         var $mUseDynamicDates;           # Use DateFormatter to format dates
4506         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
4507         var $mAllowExternalImages;       # Allow external images inline
4508         var $mAllowExternalImagesFrom;   # If not, any exception?
4509         var $mSkin;                      # Reference to the preferred skin
4510         var $mDateFormat;                # Date format index
4511         var $mEditSection;               # Create "edit section" links
4512         var $mNumberHeadings;            # Automatically number headings
4513         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
4514         var $mTidy;                      # Ask for tidy cleanup
4515         var $mInterfaceMessage;          # Which lang to call for PLURAL and GRAMMAR
4516
4517         var $mUser;                      # Stored user object, just used to initialise the skin
4518
4519         function getUseTeX()                        { return $this->mUseTeX; }
4520         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
4521         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
4522         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
4523         function getAllowExternalImagesFrom()       { return $this->mAllowExternalImagesFrom; }
4524         function getDateFormat()                    { return $this->mDateFormat; }
4525         function getEditSection()                   { return $this->mEditSection; }
4526         function getNumberHeadings()                { return $this->mNumberHeadings; }
4527         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
4528         function getTidy()                          { return $this->mTidy; }
4529         function getInterfaceMessage()              { return $this->mInterfaceMessage; }
4530
4531         function &getSkin() {
4532                 if ( !isset( $this->mSkin ) ) {
4533                         $this->mSkin = $this->mUser->getSkin();
4534                 }
4535                 return $this->mSkin;
4536         }
4537
4538         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
4539         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
4540         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
4541         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
4542         function setAllowExternalImagesFrom( $x )   { return wfSetVar( $this->mAllowExternalImagesFrom, $x ); }
4543         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
4544         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
4545         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
4546         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
4547         function setTidy( $x )                      { return wfSetVar( $this->mTidy, $x); }
4548         function setSkin( &$x ) { $this->mSkin =& $x; }
4549         function setInterfaceMessage( $x )          { return wfSetVar( $this->mInterfaceMessage, $x); }
4550
4551         function ParserOptions( $user = null ) {
4552                 $this->initialiseFromUser( $user );
4553         }
4554
4555         /**
4556          * Get parser options
4557          * @static
4558          */
4559         function newFromUser( &$user ) {
4560                 return new ParserOptions( $user );
4561         }
4562
4563         /** Get user options */
4564         function initialiseFromUser( &$userInput ) {
4565                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages;
4566                 global $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion;
4567                 $fname = 'ParserOptions::initialiseFromUser';
4568                 wfProfileIn( $fname );
4569                 if ( !$userInput ) {
4570                         global $wgUser;
4571                         if ( isset( $wgUser ) ) {
4572                                 $user = $wgUser;
4573                         } else {
4574                                 $user = new User;
4575                                 $user->setLoaded( true );
4576                         }
4577                 } else {
4578                         $user =& $userInput;
4579                 }
4580
4581                 $this->mUser = $user;
4582
4583                 $this->mUseTeX = $wgUseTeX;
4584                 $this->mUseDynamicDates = $wgUseDynamicDates;
4585                 $this->mInterwikiMagic = $wgInterwikiMagic;
4586                 $this->mAllowExternalImages = $wgAllowExternalImages;
4587                 $this->mAllowExternalImagesFrom = $wgAllowExternalImagesFrom;
4588                 $this->mSkin = null; # Deferred
4589                 $this->mDateFormat = $user->getOption( 'date' );
4590                 $this->mEditSection = true;
4591                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
4592                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
4593                 $this->mTidy = false;
4594                 $this->mInterfaceMessage = false;
4595                 wfProfileOut( $fname );
4596         }
4597 }
4598
4599 /**
4600  * Callback function used by Parser::replaceLinkHolders()
4601  * to substitute link placeholders.
4602  */
4603 function &wfOutputReplaceMatches( $matches ) {
4604         global $wgOutputReplace;
4605         return $wgOutputReplace[$matches[1]];
4606 }
4607
4608 /**
4609  * Return the total number of articles
4610  */
4611 function wfNumberOfArticles() {
4612         global $wgNumberOfArticles;
4613
4614         wfLoadSiteStats();
4615         return $wgNumberOfArticles;
4616 }
4617
4618 /**
4619  * Return the number of files
4620  */
4621 function wfNumberOfFiles() {
4622         $fname = 'wfNumberOfFiles';
4623
4624         wfProfileIn( $fname );
4625         $dbr =& wfGetDB( DB_SLAVE );
4626         $numImages = $dbr->selectField('site_stats', 'ss_images', array(), $fname );
4627         wfProfileOut( $fname );
4628
4629         return $numImages;
4630 }
4631
4632 /**
4633  * Return the number of user accounts
4634  * @return integer
4635  */
4636 function wfNumberOfUsers() {
4637         wfProfileIn( 'wfNumberOfUsers' );
4638         $dbr =& wfGetDB( DB_SLAVE );
4639         $count = $dbr->selectField( 'site_stats', 'ss_users', array(), 'wfNumberOfUsers' );
4640         wfProfileOut( 'wfNumberOfUsers' );
4641         return (int)$count;
4642 }
4643
4644 /**
4645  * Return the total number of pages
4646  * @return integer
4647  */
4648 function wfNumberOfPages() {
4649         wfProfileIn( 'wfNumberOfPages' );
4650         $dbr =& wfGetDB( DB_SLAVE );
4651         $count = $dbr->selectField( 'site_stats', 'ss_total_pages', array(), 'wfNumberOfPages' );
4652         wfProfileOut( 'wfNumberOfPages' );
4653         return (int)$count;
4654 }
4655
4656 /**
4657  * Return the total number of admins
4658  *
4659  * @return integer
4660  */
4661 function wfNumberOfAdmins() {
4662         static $admins = -1;
4663         wfProfileIn( 'wfNumberOfAdmins' );
4664         if( $admins == -1 ) {
4665                 $dbr =& wfGetDB( DB_SLAVE );
4666                 $admins = $dbr->selectField( 'user_groups', 'COUNT(*)', array( 'ug_group' => 'sysop' ), 'wfNumberOfAdmins' );
4667         }
4668         wfProfileOut( 'wfNumberOfAdmins' );
4669         return (int)$admins;
4670 }
4671
4672 /**
4673  * Count the number of pages in a particular namespace
4674  *
4675  * @param $ns Namespace
4676  * @return integer
4677  */
4678 function wfPagesInNs( $ns ) {
4679         static $pageCount = array();
4680         wfProfileIn( 'wfPagesInNs' );
4681         if( !isset( $pageCount[$ns] ) ) {
4682                 $dbr =& wfGetDB( DB_SLAVE );
4683                 $pageCount[$ns] = $dbr->selectField( 'page', 'COUNT(*)', array( 'page_namespace' => $ns ), 'wfPagesInNs' );
4684         }
4685         wfProfileOut( 'wfPagesInNs' );
4686         return (int)$pageCount[$ns];
4687 }
4688
4689 /**
4690  * Get various statistics from the database
4691  * @private
4692  */
4693 function wfLoadSiteStats() {
4694         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
4695         $fname = 'wfLoadSiteStats';
4696
4697         if ( -1 != $wgNumberOfArticles ) return;
4698         $dbr =& wfGetDB( DB_SLAVE );
4699         $s = $dbr->selectRow( 'site_stats',
4700                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
4701                 array( 'ss_row_id' => 1 ), $fname
4702         );
4703
4704         if ( $s === false ) {
4705                 return;
4706         } else {
4707                 $wgTotalViews = $s->ss_total_views;
4708                 $wgTotalEdits = $s->ss_total_edits;
4709                 $wgNumberOfArticles = $s->ss_good_articles;
4710         }
4711 }
4712
4713 /**
4714  * Escape html tags
4715  * Basically replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
4716  *
4717  * @param $in String: text that might contain HTML tags.
4718  * @return string Escaped string
4719  */
4720 function wfEscapeHTMLTagsOnly( $in ) {
4721         return str_replace(
4722                 array( '"', '>', '<' ),
4723                 array( '&quot;', '&gt;', '&lt;' ),
4724                 $in );
4725 }
4726
4727 ?>