includes/Parser.php

   1 <?php
   2 /**
   3  * File for Parser and related classes
   4  *
   5  * @package MediaWiki
   6  * @subpackage Parser
   7  */
   8
   9 /**
  10  * Update this version number when the ParserOutput format
  11  * changes in an incompatible way, so the parser cache
  12  * can automatically discard old data.
  13  */
  14 define( 'MW_PARSER_VERSION', '1.6.1' );
  15
  16 /**
  17  * Variable substitution O(N^2) attack
  18  *
  19  * Without countermeasures, it would be possible to attack the parser by saving
  20  * a page filled with a large number of inclusions of large pages. The size of
  21  * the generated page would be proportional to the square of the input size.
  22  * Hence, we limit the number of inclusions of any given page, thus bringing any
  23  * attack back to O(N).
  24  */
  25
  26 define( 'MAX_INCLUDE_REPEAT', 100 );
  27 define( 'MAX_INCLUDE_SIZE', 1000000 ); // 1 Million
  28
  29 define( 'RLH_FOR_UPDATE', 1 );
  30
  31 # Allowed values for $mOutputType
  32 define( 'OT_HTML', 1 );
  33 define( 'OT_WIKI', 2 );
  34 define( 'OT_MSG' , 3 );
  35
  36 # Flags for setFunctionHook
  37 define( 'SFH_NO_HASH', 1 );
  38
  39 # string parameter for extractTags which will cause it
  40 # to strip HTML comments in addition to regular
  41 # <XML>-style tags. This should not be anything we
  42 # may want to use in wikisyntax
  43 define( 'STRIP_COMMENTS', 'HTMLCommentStrip' );
  44
  45 # Constants needed for external link processing
  46 define( 'HTTP_PROTOCOLS', 'http:\/\/|https:\/\/' );
  47 # Everything except bracket, space, or control characters
  48 define( 'EXT_LINK_URL_CLASS', '[^][<>"\\x00-\\x20\\x7F]' );
  49 # Including space, but excluding newlines
  50 define( 'EXT_LINK_TEXT_CLASS', '[^\]\\x0a\\x0d]' );
  51 define( 'EXT_IMAGE_FNAME_CLASS', '[A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]' );
  52 define( 'EXT_IMAGE_EXTENSIONS', 'gif|png|jpg|jpeg' );
  53 define( 'EXT_LINK_BRACKETED',  '/\[(\b(' . wfUrlProtocols() . ')'.
  54         EXT_LINK_URL_CLASS.'+) *('.EXT_LINK_TEXT_CLASS.'*?)\]/S' );
  55 define( 'EXT_IMAGE_REGEX',
  56         '/^('.HTTP_PROTOCOLS.')'.  # Protocol
  57         '('.EXT_LINK_URL_CLASS.'+)\\/'.  # Hostname and path
  58         '('.EXT_IMAGE_FNAME_CLASS.'+)\\.((?i)'.EXT_IMAGE_EXTENSIONS.')$/S' # Filename
  59 );
  60
  61 // State constants for the definition list colon extraction
  62 define( 'MW_COLON_STATE_TEXT', 0 );
  63 define( 'MW_COLON_STATE_TAG', 1 );
  64 define( 'MW_COLON_STATE_TAGSTART', 2 );
  65 define( 'MW_COLON_STATE_CLOSETAG', 3 );
  66 define( 'MW_COLON_STATE_TAGSLASH', 4 );
  67 define( 'MW_COLON_STATE_COMMENT', 5 );
  68 define( 'MW_COLON_STATE_COMMENTDASH', 6 );
  69 define( 'MW_COLON_STATE_COMMENTDASHDASH', 7 );
  70
  71 /**
  72  * PHP Parser
  73  *
  74  * Processes wiki markup
  75  *
  76  * <pre>
  77  * There are three main entry points into the Parser class:
  78  * parse()
  79  *   produces HTML output
  80  * preSaveTransform().
  81  *   produces altered wiki markup.
  82  * transformMsg()
  83  *   performs brace substitution on MediaWiki messages
  84  *
  85  * Globals used:
  86  *    objects:   $wgLang, $wgContLang
  87  *
  88  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  89  *
  90  * settings:
  91  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  92  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  93  *  $wgLocaltimezone, $wgAllowSpecialInclusion*
  94  *
  95  *  * only within ParserOptions
  96  * </pre>
  97  *
  98  * @package MediaWiki
  99  */
 100 class Parser
 101 {
 102         /**#@+
 103          * @private
 104          */
 105         # Persistent:
 106         var $mTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables;
 107
 108         # Cleared with clearState():
 109         var $mOutput, $mAutonumber, $mDTopen, $mStripState = array();
 110         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 111         var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
 112         var $mTemplates,        // cache of already loaded templates, avoids
 113                                 // multiple SQL queries for the same string
 114             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 115                                 // in this path. Used for loop detection.
 116
 117         # Temporary
 118         # These are variables reset at least once per parse regardless of $clearState
 119         var $mOptions,      // ParserOptions object
 120                 $mTitle,        // Title context, used for self-link rendering and similar things
 121                 $mOutputType,   // Output type, one of the OT_xxx constants
 122                 $mRevisionId;   // ID to display in {{REVISIONID}} tags
 123
 124         /**#@-*/
 125
 126         /**
 127          * Constructor
 128          *
 129          * @public
 130          */
 131         function Parser() {
 132                 $this->mTagHooks = array();
 133                 $this->mFunctionHooks = array();
 134                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 135                 $this->mFirstCall = true;
 136         }
 137
 138         /**
 139          * Do various kinds of initialisation on the first call of the parser
 140          */
 141         function firstCallInit() {
 142                 if ( !$this->mFirstCall ) {
 143                         return;
 144                 }
 145
 146                 wfProfileIn( __METHOD__ );
 147                 global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
 148
 149                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 150
 151                 $this->setFunctionHook( 'ns', array( 'CoreParserFunctions', 'ns' ), SFH_NO_HASH );
 152                 $this->setFunctionHook( 'urlencode', array( 'CoreParserFunctions', 'urlencode' ), SFH_NO_HASH );
 153                 $this->setFunctionHook( 'lcfirst', array( 'CoreParserFunctions', 'lcfirst' ), SFH_NO_HASH );
 154                 $this->setFunctionHook( 'ucfirst', array( 'CoreParserFunctions', 'ucfirst' ), SFH_NO_HASH );
 155                 $this->setFunctionHook( 'lc', array( 'CoreParserFunctions', 'lc' ), SFH_NO_HASH );
 156                 $this->setFunctionHook( 'uc', array( 'CoreParserFunctions', 'uc' ), SFH_NO_HASH );
 157                 $this->setFunctionHook( 'localurl', array( 'CoreParserFunctions', 'localurl' ), SFH_NO_HASH );
 158                 $this->setFunctionHook( 'localurle', array( 'CoreParserFunctions', 'localurle' ), SFH_NO_HASH );
 159                 $this->setFunctionHook( 'fullurl', array( 'CoreParserFunctions', 'fullurl' ), SFH_NO_HASH );
 160                 $this->setFunctionHook( 'fullurle', array( 'CoreParserFunctions', 'fullurle' ), SFH_NO_HASH );
 161                 $this->setFunctionHook( 'formatnum', array( 'CoreParserFunctions', 'formatnum' ), SFH_NO_HASH );
 162                 $this->setFunctionHook( 'grammar', array( 'CoreParserFunctions', 'grammar' ), SFH_NO_HASH );
 163                 $this->setFunctionHook( 'plural', array( 'CoreParserFunctions', 'plural' ), SFH_NO_HASH );
 164                 $this->setFunctionHook( 'numberofpages', array( 'CoreParserFunctions', 'numberofpages' ), SFH_NO_HASH );
 165                 $this->setFunctionHook( 'numberofusers', array( 'CoreParserFunctions', 'numberofusers' ), SFH_NO_HASH );
 166                 $this->setFunctionHook( 'numberofarticles', array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 167                 $this->setFunctionHook( 'numberoffiles', array( 'CoreParserFunctions', 'numberoffiles' ), SFH_NO_HASH );
 168                 $this->setFunctionHook( 'numberofadmins', array( 'CoreParserFunctions', 'numberofadmins' ), SFH_NO_HASH );
 169                 $this->setFunctionHook( 'language', array( 'CoreParserFunctions', 'language' ), SFH_NO_HASH );
 170
 171                 if ( $wgAllowDisplayTitle ) {
 172                         $this->setFunctionHook( 'displaytitle', array( 'CoreParserFunctions', 'displaytitle' ), SFH_NO_HASH );
 173                 }
 174                 if ( $wgAllowSlowParserFunctions ) {
 175                         $this->setFunctionHook( 'pagesinnamespace', array( 'CoreParserFunctions', 'pagesinnamespace' ), SFH_NO_HASH );
 176                 }
 177
 178                 $this->initialiseVariables();
 179
 180                 $this->mFirstCall = false;
 181                 wfProfileOut( __METHOD__ );
 182         }
 183
 184         /**
 185          * Clear Parser state
 186          *
 187          * @private
 188          */
 189         function clearState() {
 190                 wfProfileIn( __METHOD__ );
 191                 if ( $this->mFirstCall ) {
 192                         $this->firstCallInit();
 193                 }
 194                 $this->mOutput = new ParserOutput;
 195                 $this->mAutonumber = 0;
 196                 $this->mLastSection = '';
 197                 $this->mDTopen = false;
 198                 $this->mIncludeCount = array();
 199                 $this->mStripState = array();
 200                 $this->mArgStack = array();
 201                 $this->mInPre = false;
 202                 $this->mInterwikiLinkHolders = array(
 203                         'texts' => array(),
 204                         'titles' => array()
 205                 );
 206                 $this->mLinkHolders = array(
 207                         'namespaces' => array(),
 208                         'dbkeys' => array(),
 209                         'queries' => array(),
 210                         'texts' => array(),
 211                         'titles' => array()
 212                 );
 213                 $this->mRevisionId = null;
 214
 215                 /**
 216                  * Prefix for temporary replacement strings for the multipass parser.
 217                  * \x07 should never appear in input as it's disallowed in XML.
 218                  * Using it at the front also gives us a little extra robustness
 219                  * since it shouldn't match when butted up against identifier-like
 220                  * string constructs.
 221                  */
 222                 $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 223
 224                 # Clear these on every parse, bug 4549
 225                 $this->mTemplates = array();
 226                 $this->mTemplatePath = array();
 227
 228                 $this->mShowToc = true;
 229                 $this->mForceTocPosition = false;
 230
 231                 wfRunHooks( 'ParserClearState', array( &$this ) );
 232                 wfProfileOut( __METHOD__ );
 233         }
 234
 235         /**
 236          * Accessor for mUniqPrefix.
 237          *
 238          * @public
 239          */
 240         function uniqPrefix() {
 241                 return $this->mUniqPrefix;
 242         }
 243
 244         /**
 245          * Convert wikitext to HTML
 246          * Do not call this function recursively.
 247          *
 248          * @private
 249          * @param string $text Text we want to parse
 250          * @param Title &$title A title object
 251          * @param array $options
 252          * @param boolean $linestart
 253          * @param boolean $clearState
 254          * @param int $revid number to pass in {{REVISIONID}}
 255          * @return ParserOutput a ParserOutput
 256          */
 257         function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 258                 /**
 259                  * First pass--just handle <nowiki> sections, pass the rest off
 260                  * to internalParse() which does all the real work.
 261                  */
 262
 263                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 264                 $fname = 'Parser::parse';
 265                 wfProfileIn( $fname );
 266
 267                 if ( $clearState ) {
 268                         $this->clearState();
 269                 }
 270
 271                 $this->mOptions = $options;
 272                 $this->mTitle =& $title;
 273                 $oldRevisionId = $this->mRevisionId;
 274                 if( $revid !== null ) {
 275                         $this->mRevisionId = $revid;
 276                 }
 277                 $this->mOutputType = OT_HTML;
 278
 279                 //$text = $this->strip( $text, $this->mStripState );
 280                 // VOODOO MAGIC FIX! Sometimes the above segfaults in PHP5.
 281                 $x =& $this->mStripState;
 282
 283                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 284                 $text = $this->strip( $text, $x );
 285                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 286
 287                 $text = $this->internalParse( $text );
 288
 289                 $text = $this->unstrip( $text, $this->mStripState );
 290
 291                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 292                 $fixtags = array(
 293                         # french spaces, last one Guillemet-left
 294                         # only if there is something before the space
 295                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 296                         # french spaces, Guillemet-right
 297                         '/(\\302\\253) /' => '\\1&nbsp;',
 298                 );
 299                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 300
 301                 # only once and last
 302                 $text = $this->doBlockLevels( $text, $linestart );
 303
 304                 $this->replaceLinkHolders( $text );
 305
 306                 # the position of the parserConvert() call should not be changed. it
 307                 # assumes that the links are all replaced and the only thing left
 308                 # is the <nowiki> mark.
 309                 # Side-effects: this calls $this->mOutput->setTitleText()
 310                 $text = $wgContLang->parserConvert( $text, $this );
 311
 312                 $text = $this->unstripNoWiki( $text, $this->mStripState );
 313
 314                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 315
 316                 $text = Sanitizer::normalizeCharReferences( $text );
 317
 318                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 319                         $text = Parser::tidy($text);
 320                 } else {
 321                         # attempt to sanitize at least some nesting problems
 322                         # (bug #2702 and quite a few others)
 323                         $tidyregs = array(
 324                                 # ''Something [http://www.cool.com cool''] -->
 325                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 326                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 327                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 328                                 # fix up an anchor inside another anchor, only
 329                                 # at least for a single single nested link (bug 3695)
 330                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 331                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 332                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 333                                 # contains a div, so fix it up here; replace
 334                                 # div with escaped text
 335                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 336                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 337                                 # remove empty italic or bold tag pairs, some
 338                                 # introduced by rules above
 339                                 '/<([bi])><\/\\1>/' => '',
 340                         );
 341
 342                         $text = preg_replace(
 343                                 array_keys( $tidyregs ),
 344                                 array_values( $tidyregs ),
 345                                 $text );
 346                 }
 347
 348                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 349
 350                 $this->mOutput->setText( $text );
 351                 $this->mRevisionId = $oldRevisionId;
 352                 wfProfileOut( $fname );
 353
 354                 return $this->mOutput;
 355         }
 356
 357         /**
 358          * Recursive parser entry point that can be called from an extension tag
 359          * hook.
 360          */
 361         function recursiveTagParse( $text ) {
 362                 wfProfileIn( __METHOD__ );
 363                 $x =& $this->mStripState;
 364                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 365                 $text = $this->strip( $text, $x );
 366                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 367                 $text = $this->internalParse( $text );
 368                 wfProfileOut( __METHOD__ );
 369                 return $text;
 370         }
 371
 372         /**
 373          * Get a random string
 374          *
 375          * @private
 376          * @static
 377          */
 378         function getRandomString() {
 379                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 380         }
 381
 382         function &getTitle() { return $this->mTitle; }
 383         function getOptions() { return $this->mOptions; }
 384
 385         function getFunctionLang() {
 386                 global $wgLang, $wgContLang;
 387                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 388         }
 389
 390         /**
 391          * Replaces all occurrences of HTML-style comments and the given tags
 392          * in the text with a random marker and returns teh next text. The output
 393          * parameter $matches will be an associative array filled with data in
 394          * the form:
 395          *   'UNIQ-xxxxx' => array(
 396          *     'element',
 397          *     'tag content',
 398          *     array( 'param' => 'x' ),
 399          *     '<element param="x">tag content</element>' ) )
 400          *
 401          * @param $elements list of element names. Comments are always extracted.
 402          * @param $text Source text string.
 403          * @param $uniq_prefix
 404          *
 405          * @private
 406          * @static
 407          */
 408         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 409                 static $n = 1;
 410                 $stripped = '';
 411                 $matches = array();
 412
 413                 $taglist = implode( '|', $elements );
 414                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 415
 416                 while ( '' != $text ) {
 417                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 418                         $stripped .= $p[0];
 419                         if( count( $p ) < 5 ) {
 420                                 break;
 421                         }
 422                         if( count( $p ) > 5 ) {
 423                                 // comment
 424                                 $element    = $p[4];
 425                                 $attributes = '';
 426                                 $close      = '';
 427                                 $inside     = $p[5];
 428                         } else {
 429                                 // tag
 430                                 $element    = $p[1];
 431                                 $attributes = $p[2];
 432                                 $close      = $p[3];
 433                                 $inside     = $p[4];
 434                         }
 435
 436                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . '-QINU';
 437                         $stripped .= $marker;
 438
 439                         if ( $close === '/>' ) {
 440                                 // Empty element tag, <tag />
 441                                 $content = null;
 442                                 $text = $inside;
 443                                 $tail = null;
 444                         } else {
 445                                 if( $element == '!--' ) {
 446                                         $end = '/(-->)/';
 447                                 } else {
 448                                         $end = "/(<\\/$element\\s*>)/i";
 449                                 }
 450                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 451                                 $content = $q[0];
 452                                 if( count( $q ) < 3 ) {
 453                                         # No end tag -- let it run out to the end of the text.
 454                                         $tail = '';
 455                                         $text = '';
 456                                 } else {
 457                                         $tail = $q[1];
 458                                         $text = $q[2];
 459                                 }
 460                         }
 461
 462                         $matches[$marker] = array( $element,
 463                                 $content,
 464                                 Sanitizer::decodeTagAttributes( $attributes ),
 465                                 "<$element$attributes$close$content$tail" );
 466                 }
 467                 return $stripped;
 468         }
 469
 470         /**
 471          * Strips and renders nowiki, pre, math, hiero
 472          * If $render is set, performs necessary rendering operations on plugins
 473          * Returns the text, and fills an array with data needed in unstrip()
 474          * If the $state is already a valid strip state, it adds to the state
 475          *
 476          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 477          *  will be stripped in addition to other tags. This is important
 478          *  for section editing, where these comments cause confusion when
 479          *  counting the sections in the wikisource
 480          *
 481          * @param array dontstrip contains tags which should not be stripped;
 482          *  used to prevent stipping of <gallery> when saving (fixes bug 2700)
 483          *
 484          * @private
 485          */
 486         function strip( $text, &$state, $stripcomments = false , $dontstrip = array () ) {
 487                 wfProfileIn( __METHOD__ );
 488                 $render = ($this->mOutputType == OT_HTML);
 489
 490                 # Replace any instances of the placeholders
 491                 $uniq_prefix = $this->mUniqPrefix;
 492                 #$text = str_replace( $uniq_prefix, wfHtmlEscapeFirst( $uniq_prefix ), $text );
 493                 $commentState = array();
 494
 495                 $elements = array_merge(
 496                         array( 'nowiki', 'gallery' ),
 497                         array_keys( $this->mTagHooks ) );
 498                 global $wgRawHtml;
 499                 if( $wgRawHtml ) {
 500                         $elements[] = 'html';
 501                 }
 502                 if( $this->mOptions->getUseTeX() ) {
 503                         $elements[] = 'math';
 504                 }
 505
 506                 # Removing $dontstrip tags from $elements list (currently only 'gallery', fixing bug 2700)
 507                 foreach ( $elements AS $k => $v ) {
 508                         if ( !in_array ( $v , $dontstrip ) ) continue;
 509                         unset ( $elements[$k] );
 510                 }
 511
 512                 $matches = array();
 513                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 514
 515                 foreach( $matches as $marker => $data ) {
 516                         list( $element, $content, $params, $tag ) = $data;
 517                         if( $render ) {
 518                                 $tagName = strtolower( $element );
 519                                 wfProfileIn( __METHOD__."-render-$tagName" );
 520                                 switch( $tagName ) {
 521                                 case '!--':
 522                                         // Comment
 523                                         if( substr( $tag, -3 ) == '-->' ) {
 524                                                 $output = $tag;
 525                                         } else {
 526                                                 // Unclosed comment in input.
 527                                                 // Close it so later stripping can remove it
 528                                                 $output = "$tag-->";
 529                                         }
 530                                         break;
 531                                 case 'html':
 532                                         if( $wgRawHtml ) {
 533                                                 $output = $content;
 534                                                 break;
 535                                         }
 536                                         // Shouldn't happen otherwise. :)
 537                                 case 'nowiki':
 538                                         $output = wfEscapeHTMLTagsOnly( $content );
 539                                         break;
 540                                 case 'math':
 541                                         $output = MathRenderer::renderMath( $content );
 542                                         break;
 543                                 case 'gallery':
 544                                         $output = $this->renderImageGallery( $content, $params );
 545                                         break;
 546                                 default:
 547                                         if( isset( $this->mTagHooks[$tagName] ) ) {
 548                                                 $output = call_user_func_array( $this->mTagHooks[$tagName],
 549                                                         array( $content, $params, $this ) );
 550                                         } else {
 551                                                 throw new MWException( "Invalid call hook $element" );
 552                                         }
 553                                 }
 554                                 wfProfileOut( __METHOD__."-render-$tagName" );
 555                         } else {
 556                                 // Just stripping tags; keep the source
 557                                 $output = $tag;
 558                         }
 559
 560                         // Unstrip the output, because unstrip() is no longer recursive so
 561                         // it won't do it itself
 562                         $output = $this->unstrip( $output, $state );
 563
 564                         if( !$stripcomments && $element == '!--' ) {
 565                                 $commentState[$marker] = $output;
 566                         } elseif ( $element == 'html' || $element == 'nowiki' ) {
 567                                 $state['nowiki'][$marker] = $output;
 568                         } else {
 569                                 $state['general'][$marker] = $output;
 570                         }
 571                 }
 572
 573                 # Unstrip comments unless explicitly told otherwise.
 574                 # (The comments are always stripped prior to this point, so as to
 575                 # not invoke any extension tags / parser hooks contained within
 576                 # a comment.)
 577                 if ( !$stripcomments ) {
 578                         // Put them all back and forget them
 579                         $text = strtr( $text, $commentState );
 580                 }
 581
 582                 wfProfileOut( __METHOD__ );
 583                 return $text;
 584         }
 585
 586         /**
 587          * Restores pre, math, and other extensions removed by strip()
 588          *
 589          * always call unstripNoWiki() after this one
 590          * @private
 591          */
 592         function unstrip( $text, &$state ) {
 593                 if ( !isset( $state['general'] ) ) {
 594                         return $text;
 595                 }
 596
 597                 wfProfileIn( __METHOD__ );
 598                 # TODO: good candidate for FSS
 599                 $text = strtr( $text, $state['general'] );
 600                 wfProfileOut( __METHOD__ );
 601                 return $text;
 602         }
 603
 604         /**
 605          * Always call this after unstrip() to preserve the order
 606          *
 607          * @private
 608          */
 609         function unstripNoWiki( $text, &$state ) {
 610                 if ( !isset( $state['nowiki'] ) ) {
 611                         return $text;
 612                 }
 613
 614                 wfProfileIn( __METHOD__ );
 615                 # TODO: good candidate for FSS
 616                 $text = strtr( $text, $state['nowiki'] );
 617                 wfProfileOut( __METHOD__ );
 618
 619                 return $text;
 620         }
 621
 622         /**
 623          * Add an item to the strip state
 624          * Returns the unique tag which must be inserted into the stripped text
 625          * The tag will be replaced with the original text in unstrip()
 626          *
 627          * @private
 628          */
 629         function insertStripItem( $text, &$state ) {
 630                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 631                 if ( !$state ) {
 632                         $state = array();
 633                 }
 634                 $state['general'][$rnd] = $text;
 635                 return $rnd;
 636         }
 637
 638         /**
 639          * Interface with html tidy, used if $wgUseTidy = true.
 640          * If tidy isn't able to correct the markup, the original will be
 641          * returned in all its glory with a warning comment appended.
 642          *
 643          * Either the external tidy program or the in-process tidy extension
 644          * will be used depending on availability. Override the default
 645          * $wgTidyInternal setting to disable the internal if it's not working.
 646          *
 647          * @param string $text Hideous HTML input
 648          * @return string Corrected HTML output
 649          * @public
 650          * @static
 651          */
 652         function tidy( $text ) {
 653                 global $wgTidyInternal;
 654                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 655 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 656 '<head><title>test</title></head><body>'.$text.'</body></html>';
 657                 if( $wgTidyInternal ) {
 658                         $correctedtext = Parser::internalTidy( $wrappedtext );
 659                 } else {
 660                         $correctedtext = Parser::externalTidy( $wrappedtext );
 661                 }
 662                 if( is_null( $correctedtext ) ) {
 663                         wfDebug( "Tidy error detected!\n" );
 664                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 665                 }
 666                 return $correctedtext;
 667         }
 668
 669         /**
 670          * Spawn an external HTML tidy process and get corrected markup back from it.
 671          *
 672          * @private
 673          * @static
 674          */
 675         function externalTidy( $text ) {
 676                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 677                 $fname = 'Parser::externalTidy';
 678                 wfProfileIn( $fname );
 679
 680                 $cleansource = '';
 681                 $opts = ' -utf8';
 682
 683                 $descriptorspec = array(
 684                         0 => array('pipe', 'r'),
 685                         1 => array('pipe', 'w'),
 686                         2 => array('file', '/dev/null', 'a')
 687                 );
 688                 $pipes = array();
 689                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 690                 if (is_resource($process)) {
 691                         // Theoretically, this style of communication could cause a deadlock
 692                         // here. If the stdout buffer fills up, then writes to stdin could
 693                         // block. This doesn't appear to happen with tidy, because tidy only
 694                         // writes to stdout after it's finished reading from stdin. Search
 695                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 696                         fwrite($pipes[0], $text);
 697                         fclose($pipes[0]);
 698                         while (!feof($pipes[1])) {
 699                                 $cleansource .= fgets($pipes[1], 1024);
 700                         }
 701                         fclose($pipes[1]);
 702                         proc_close($process);
 703                 }
 704
 705                 wfProfileOut( $fname );
 706
 707                 if( $cleansource == '' && $text != '') {
 708                         // Some kind of error happened, so we couldn't get the corrected text.
 709                         // Just give up; we'll use the source text and append a warning.
 710                         return null;
 711                 } else {
 712                         return $cleansource;
 713                 }
 714         }
 715
 716         /**
 717          * Use the HTML tidy PECL extension to use the tidy library in-process,
 718          * saving the overhead of spawning a new process. Currently written to
 719          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 720          *
 721          * 'pear install tidy' should be able to compile the extension module.
 722          *
 723          * @private
 724          * @static
 725          */
 726         function internalTidy( $text ) {
 727                 global $wgTidyConf;
 728                 $fname = 'Parser::internalTidy';
 729                 wfProfileIn( $fname );
 730
 731                 tidy_load_config( $wgTidyConf );
 732                 tidy_set_encoding( 'utf8' );
 733                 tidy_parse_string( $text );
 734                 tidy_clean_repair();
 735                 if( tidy_get_status() == 2 ) {
 736                         // 2 is magic number for fatal error
 737                         // http://www.php.net/manual/en/function.tidy-get-status.php
 738                         $cleansource = null;
 739                 } else {
 740                         $cleansource = tidy_get_output();
 741                 }
 742                 wfProfileOut( $fname );
 743                 return $cleansource;
 744         }
 745
 746         /**
 747          * parse the wiki syntax used to render tables
 748          *
 749          * @private
 750          */
 751         function doTableStuff ( $t ) {
 752                 $fname = 'Parser::doTableStuff';
 753                 wfProfileIn( $fname );
 754
 755                 $t = explode ( "\n" , $t ) ;
 756                 $td = array () ; # Is currently a td tag open?
 757                 $ltd = array () ; # Was it TD or TH?
 758                 $tr = array () ; # Is currently a tr tag open?
 759                 $ltr = array () ; # tr attributes
 760                 $has_opened_tr = array(); # Did this table open a <tr> element?
 761                 $indent_level = 0; # indent level of the table
 762                 foreach ( $t AS $k => $x )
 763                 {
 764                         $x = trim ( $x ) ;
 765                         $fc = substr ( $x , 0 , 1 ) ;
 766                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 767                                 $indent_level = strlen( $matches[1] );
 768
 769                                 $attributes = $this->unstripForHTML( $matches[2] );
 770
 771                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 772                                         '<table' . Sanitizer::fixTagAttributes ( $attributes, 'table' ) . '>' ;
 773                                 array_push ( $td , false ) ;
 774                                 array_push ( $ltd , '' ) ;
 775                                 array_push ( $tr , false ) ;
 776                                 array_push ( $ltr , '' ) ;
 777                                 array_push ( $has_opened_tr, false );
 778                         }
 779                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 780                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 781                                 $z = "</table>" . substr ( $x , 2);
 782                                 $l = array_pop ( $ltd ) ;
 783                                 if ( !array_pop ( $has_opened_tr ) ) $z = "<tr><td></td></tr>" . $z ;
 784                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 785                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 786                                 array_pop ( $ltr ) ;
 787                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 788                         }
 789                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 790                                 $x = substr ( $x , 1 ) ;
 791                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 792                                 $z = '' ;
 793                                 $l = array_pop ( $ltd ) ;
 794                                 array_pop ( $has_opened_tr );
 795                                 array_push ( $has_opened_tr , true ) ;
 796                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 797                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 798                                 array_pop ( $ltr ) ;
 799                                 $t[$k] = $z ;
 800                                 array_push ( $tr , false ) ;
 801                                 array_push ( $td , false ) ;
 802                                 array_push ( $ltd , '' ) ;
 803                                 $attributes = $this->unstripForHTML( $x );
 804                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $attributes, 'tr' ) ) ;
 805                         }
 806                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 807                                 # $x is a table row
 808                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 809                                         $fc = '+' ;
 810                                         $x = substr ( $x , 1 ) ;
 811                                 }
 812                                 $after = substr ( $x , 1 ) ;
 813                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 814
 815                                 // Split up multiple cells on the same line.
 816                                 // FIXME: This can result in improper nesting of tags processed
 817                                 // by earlier parser steps, but should avoid splitting up eg
 818                                 // attribute values containing literal "||".
 819                                 $after = wfExplodeMarkup( '||', $after );
 820
 821                                 $t[$k] = '' ;
 822
 823                                 # Loop through each table cell
 824                                 foreach ( $after AS $theline )
 825                                 {
 826                                         $z = '' ;
 827                                         if ( $fc != '+' )
 828                                         {
 829                                                 $tra = array_pop ( $ltr ) ;
 830                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 831                                                 array_push ( $tr , true ) ;
 832                                                 array_push ( $ltr , '' ) ;
 833                                                 array_pop ( $has_opened_tr );
 834                                                 array_push ( $has_opened_tr , true ) ;
 835                                         }
 836
 837                                         $l = array_pop ( $ltd ) ;
 838                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 839                                         if ( $fc == '|' ) $l = 'td' ;
 840                                         else if ( $fc == '!' ) $l = 'th' ;
 841                                         else if ( $fc == '+' ) $l = 'caption' ;
 842                                         else $l = '' ;
 843                                         array_push ( $ltd , $l ) ;
 844
 845                                         # Cell parameters
 846                                         $y = explode ( '|' , $theline , 2 ) ;
 847                                         # Note that a '|' inside an invalid link should not
 848                                         # be mistaken as delimiting cell parameters
 849                                         if ( strpos( $y[0], '[[' ) !== false ) {
 850                                                 $y = array ($theline);
 851                                         }
 852                                         if ( count ( $y ) == 1 )
 853                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 854                                         else {
 855                                                 $attributes = $this->unstripForHTML( $y[0] );
 856                                                 $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($attributes, $l).">{$y[1]}" ;
 857                                         }
 858                                         $t[$k] .= $y ;
 859                                         array_push ( $td , true ) ;
 860                                 }
 861                         }
 862                 }
 863
 864                 # Closing open td, tr && table
 865                 while ( count ( $td ) > 0 )
 866                 {
 867                         $l = array_pop ( $ltd ) ;
 868                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 869                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 870                         if ( !array_pop ( $has_opened_tr ) ) $t[] = "<tr><td></td></tr>" ;
 871                         $t[] = '</table>' ;
 872                 }
 873
 874                 $t = implode ( "\n" , $t ) ;
 875                 # special case: don't return empty table
 876                 if($t == "<table>\n<tr><td></td></tr>\n</table>")
 877                         $t = '';
 878                 wfProfileOut( $fname );
 879                 return $t ;
 880         }
 881
 882         /**
 883          * Helper function for parse() that transforms wiki markup into
 884          * HTML. Only called for $mOutputType == OT_HTML.
 885          *
 886          * @private
 887          */
 888         function internalParse( $text ) {
 889                 $args = array();
 890                 $isMain = true;
 891                 $fname = 'Parser::internalParse';
 892                 wfProfileIn( $fname );
 893
 894                 # Hook to suspend the parser in this state
 895                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$x ) ) ) {
 896                         wfProfileOut( $fname );
 897                         return $text ;
 898                 }
 899
 900                 # Remove <noinclude> tags and <includeonly> sections
 901                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 902                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 903                 $text = preg_replace( '/<includeonly>.*?<\/includeonly>/s', '', $text );
 904
 905                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
 906
 907                 $text = $this->replaceVariables( $text, $args );
 908
 909                 // Tables need to come after variable replacement for things to work
 910                 // properly; putting them before other transformations should keep
 911                 // exciting things like link expansions from showing up in surprising
 912                 // places.
 913                 $text = $this->doTableStuff( $text );
 914
 915                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 916
 917                 $text = $this->stripToc( $text );
 918                 $this->stripNoGallery( $text );
 919                 $text = $this->doHeadings( $text );
 920                 if($this->mOptions->getUseDynamicDates()) {
 921                         $df =& DateFormatter::getInstance();
 922                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 923                 }
 924                 $text = $this->doAllQuotes( $text );
 925                 $text = $this->replaceInternalLinks( $text );
 926                 $text = $this->replaceExternalLinks( $text );
 927
 928                 # replaceInternalLinks may sometimes leave behind
 929                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 930                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 931
 932                 $text = $this->doMagicLinks( $text );
 933                 $text = $this->formatHeadings( $text, $isMain );
 934
 935                 wfProfileOut( $fname );
 936                 return $text;
 937         }
 938
 939         /**
 940          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 941          * magic external links.
 942          *
 943          * @private
 944          */
 945         function &doMagicLinks( &$text ) {
 946                 wfProfileIn( __METHOD__ );
 947                 $text = preg_replace_callback(
 948                         '!(?:                           # Start cases
 949                             <a.*?</a>                   # Skip link text
 950                             <.*?> |                     # Skip stuff inside HTML elements
 951                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
 952                             ISBN\s+([0-9Xx-]+)          # ISBN, capture number as m[2]
 953                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 954                 wfProfileOut( __METHOD__ );
 955                 return $text;
 956         }
 957
 958         function magicLinkCallback( $m ) {
 959                 if ( substr( $m[0], 0, 1 ) == '<' ) {
 960                         # Skip HTML element
 961                         return $m[0];
 962                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
 963                         $isbn = $m[2];
 964                         $num = strtr( $isbn, array(
 965                                 '-' => '',
 966                                 ' ' => '',
 967                                 'x' => 'X',
 968                         ));
 969                         $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
 970                         $text = '<a href="' .
 971                                 $titleObj->escapeLocalUrl( "isbn=$num" ) .
 972                                 "\" class=\"internal\">ISBN $isbn</a>";
 973                 } else {
 974                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
 975                                 $keyword = 'RFC';
 976                                 $urlmsg = 'rfcurl';
 977                                 $id = $m[1];
 978                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
 979                                 $keyword = 'PMID';
 980                                 $urlmsg = 'pubmedurl';
 981                                 $id = $m[1];
 982                         } else {
 983                                 throw new MWException( __METHOD__.': unrecognised match type "' .
 984                                         substr($m[0], 0, 20 ) . '"' );
 985                         }
 986
 987                         $url = wfMsg( $urlmsg, $id);
 988                         $sk =& $this->mOptions->getSkin();
 989                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
 990                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
 991                 }
 992                 return $text;
 993         }
 994
 995         /**
 996          * Parse headers and return html
 997          *
 998          * @private
 999          */
1000         function doHeadings( $text ) {
1001                 $fname = 'Parser::doHeadings';
1002                 wfProfileIn( $fname );
1003                 for ( $i = 6; $i >= 1; --$i ) {
1004                         $h = str_repeat( '=', $i );
1005                         $text = preg_replace( "/^{$h}(.+){$h}\\s*$/m",
1006                           "<h{$i}>\\1</h{$i}>\\2", $text );
1007                 }
1008                 wfProfileOut( $fname );
1009                 return $text;
1010         }
1011
1012         /**
1013          * Replace single quotes with HTML markup
1014          * @private
1015          * @return string the altered text
1016          */
1017         function doAllQuotes( $text ) {
1018                 $fname = 'Parser::doAllQuotes';
1019                 wfProfileIn( $fname );
1020                 $outtext = '';
1021                 $lines = explode( "\n", $text );
1022                 foreach ( $lines as $line ) {
1023                         $outtext .= $this->doQuotes ( $line ) . "\n";
1024                 }
1025                 $outtext = substr($outtext, 0,-1);
1026                 wfProfileOut( $fname );
1027                 return $outtext;
1028         }
1029
1030         /**
1031          * Helper function for doAllQuotes()
1032          * @private
1033          */
1034         function doQuotes( $text ) {
1035                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1036                 if ( count( $arr ) == 1 )
1037                         return $text;
1038                 else
1039                 {
1040                         # First, do some preliminary work. This may shift some apostrophes from
1041                         # being mark-up to being text. It also counts the number of occurrences
1042                         # of bold and italics mark-ups.
1043                         $i = 0;
1044                         $numbold = 0;
1045                         $numitalics = 0;
1046                         foreach ( $arr as $r )
1047                         {
1048                                 if ( ( $i % 2 ) == 1 )
1049                                 {
1050                                         # If there are ever four apostrophes, assume the first is supposed to
1051                                         # be text, and the remaining three constitute mark-up for bold text.
1052                                         if ( strlen( $arr[$i] ) == 4 )
1053                                         {
1054                                                 $arr[$i-1] .= "'";
1055                                                 $arr[$i] = "'''";
1056                                         }
1057                                         # If there are more than 5 apostrophes in a row, assume they're all
1058                                         # text except for the last 5.
1059                                         else if ( strlen( $arr[$i] ) > 5 )
1060                                         {
1061                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1062                                                 $arr[$i] = "'''''";
1063                                         }
1064                                         # Count the number of occurrences of bold and italics mark-ups.
1065                                         # We are not counting sequences of five apostrophes.
1066                                         if ( strlen( $arr[$i] ) == 2 ) $numitalics++;  else
1067                                         if ( strlen( $arr[$i] ) == 3 ) $numbold++;     else
1068                                         if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1069                                 }
1070                                 $i++;
1071                         }
1072
1073                         # If there is an odd number of both bold and italics, it is likely
1074                         # that one of the bold ones was meant to be an apostrophe followed
1075                         # by italics. Which one we cannot know for certain, but it is more
1076                         # likely to be one that has a single-letter word before it.
1077                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1078                         {
1079                                 $i = 0;
1080                                 $firstsingleletterword = -1;
1081                                 $firstmultiletterword = -1;
1082                                 $firstspace = -1;
1083                                 foreach ( $arr as $r )
1084                                 {
1085                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1086                                         {
1087                                                 $x1 = substr ($arr[$i-1], -1);
1088                                                 $x2 = substr ($arr[$i-1], -2, 1);
1089                                                 if ($x1 == ' ') {
1090                                                         if ($firstspace == -1) $firstspace = $i;
1091                                                 } else if ($x2 == ' ') {
1092                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1093                                                 } else {
1094                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1095                                                 }
1096                                         }
1097                                         $i++;
1098                                 }
1099
1100                                 # If there is a single-letter word, use it!
1101                                 if ($firstsingleletterword > -1)
1102                                 {
1103                                         $arr [ $firstsingleletterword ] = "''";
1104                                         $arr [ $firstsingleletterword-1 ] .= "'";
1105                                 }
1106                                 # If not, but there's a multi-letter word, use that one.
1107                                 else if ($firstmultiletterword > -1)
1108                                 {
1109                                         $arr [ $firstmultiletterword ] = "''";
1110                                         $arr [ $firstmultiletterword-1 ] .= "'";
1111                                 }
1112                                 # ... otherwise use the first one that has neither.
1113                                 # (notice that it is possible for all three to be -1 if, for example,
1114                                 # there is only one pentuple-apostrophe in the line)
1115                                 else if ($firstspace > -1)
1116                                 {
1117                                         $arr [ $firstspace ] = "''";
1118                                         $arr [ $firstspace-1 ] .= "'";
1119                                 }
1120                         }
1121
1122                         # Now let's actually convert our apostrophic mush to HTML!
1123                         $output = '';
1124                         $buffer = '';
1125                         $state = '';
1126                         $i = 0;
1127                         foreach ($arr as $r)
1128                         {
1129                                 if (($i % 2) == 0)
1130                                 {
1131                                         if ($state == 'both')
1132                                                 $buffer .= $r;
1133                                         else
1134                                                 $output .= $r;
1135                                 }
1136                                 else
1137                                 {
1138                                         if (strlen ($r) == 2)
1139                                         {
1140                                                 if ($state == 'i')
1141                                                 { $output .= '</i>'; $state = ''; }
1142                                                 else if ($state == 'bi')
1143                                                 { $output .= '</i>'; $state = 'b'; }
1144                                                 else if ($state == 'ib')
1145                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1146                                                 else if ($state == 'both')
1147                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1148                                                 else # $state can be 'b' or ''
1149                                                 { $output .= '<i>'; $state .= 'i'; }
1150                                         }
1151                                         else if (strlen ($r) == 3)
1152                                         {
1153                                                 if ($state == 'b')
1154                                                 { $output .= '</b>'; $state = ''; }
1155                                                 else if ($state == 'bi')
1156                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1157                                                 else if ($state == 'ib')
1158                                                 { $output .= '</b>'; $state = 'i'; }
1159                                                 else if ($state == 'both')
1160                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1161                                                 else # $state can be 'i' or ''
1162                                                 { $output .= '<b>'; $state .= 'b'; }
1163                                         }
1164                                         else if (strlen ($r) == 5)
1165                                         {
1166                                                 if ($state == 'b')
1167                                                 { $output .= '</b><i>'; $state = 'i'; }
1168                                                 else if ($state == 'i')
1169                                                 { $output .= '</i><b>'; $state = 'b'; }
1170                                                 else if ($state == 'bi')
1171                                                 { $output .= '</i></b>'; $state = ''; }
1172                                                 else if ($state == 'ib')
1173                                                 { $output .= '</b></i>'; $state = ''; }
1174                                                 else if ($state == 'both')
1175                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1176                                                 else # ($state == '')
1177                                                 { $buffer = ''; $state = 'both'; }
1178                                         }
1179                                 }
1180                                 $i++;
1181                         }
1182                         # Now close all remaining tags.  Notice that the order is important.
1183                         if ($state == 'b' || $state == 'ib')
1184                                 $output .= '</b>';
1185                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1186                                 $output .= '</i>';
1187                         if ($state == 'bi')
1188                                 $output .= '</b>';
1189                         if ($state == 'both')
1190                                 $output .= '<b><i>'.$buffer.'</i></b>';
1191                         return $output;
1192                 }
1193         }
1194
1195         /**
1196          * Replace external links
1197          *
1198          * Note: this is all very hackish and the order of execution matters a lot.
1199          * Make sure to run maintenance/parserTests.php if you change this code.
1200          *
1201          * @private
1202          */
1203         function replaceExternalLinks( $text ) {
1204                 global $wgContLang;
1205                 $fname = 'Parser::replaceExternalLinks';
1206                 wfProfileIn( $fname );
1207
1208                 $sk =& $this->mOptions->getSkin();
1209
1210                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1211
1212                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1213
1214                 $i = 0;
1215                 while ( $i<count( $bits ) ) {
1216                         $url = $bits[$i++];
1217                         $protocol = $bits[$i++];
1218                         $text = $bits[$i++];
1219                         $trail = $bits[$i++];
1220
1221                         # The characters '<' and '>' (which were escaped by
1222                         # removeHTMLtags()) should not be included in
1223                         # URLs, per RFC 2396.
1224                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1225                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1226                                 $url = substr($url, 0, $m2[0][1]);
1227                         }
1228
1229                         # If the link text is an image URL, replace it with an <img> tag
1230                         # This happened by accident in the original parser, but some people used it extensively
1231                         $img = $this->maybeMakeExternalImage( $text );
1232                         if ( $img !== false ) {
1233                                 $text = $img;
1234                         }
1235
1236                         $dtrail = '';
1237
1238                         # Set linktype for CSS - if URL==text, link is essentially free
1239                         $linktype = ($text == $url) ? 'free' : 'text';
1240
1241                         # No link text, e.g. [http://domain.tld/some.link]
1242                         if ( $text == '' ) {
1243                                 # Autonumber if allowed. See bug #5918
1244                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1245                                         $text = '[' . ++$this->mAutonumber . ']';
1246                                         $linktype = 'autonumber';
1247                                 } else {
1248                                         # Otherwise just use the URL
1249                                         $text = htmlspecialchars( $url );
1250                                         $linktype = 'free';
1251                                 }
1252                         } else {
1253                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1254                                 # Check for trail
1255                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1256                         }
1257
1258                         $text = $wgContLang->markNoConversion($text);
1259
1260                         $url = Sanitizer::cleanUrl( $url );
1261
1262                         # Process the trail (i.e. everything after this link up until start of the next link),
1263                         # replacing any non-bracketed links
1264                         $trail = $this->replaceFreeExternalLinks( $trail );
1265
1266                         # Use the encoded URL
1267                         # This means that users can paste URLs directly into the text
1268                         # Funny characters like &ouml; aren't valid in URLs anyway
1269                         # This was changed in August 2004
1270                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1271
1272                         # Register link in the output object.
1273                         # Replace unnecessary URL escape codes with the referenced character
1274                         # This prevents spammers from hiding links from the filters
1275                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1276                         $this->mOutput->addExternalLink( $pasteurized );
1277                 }
1278
1279                 wfProfileOut( $fname );
1280                 return $s;
1281         }
1282
1283         /**
1284          * Replace anything that looks like a URL with a link
1285          * @private
1286          */
1287         function replaceFreeExternalLinks( $text ) {
1288                 global $wgContLang;
1289                 $fname = 'Parser::replaceFreeExternalLinks';
1290                 wfProfileIn( $fname );
1291
1292                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1293                 $s = array_shift( $bits );
1294                 $i = 0;
1295
1296                 $sk =& $this->mOptions->getSkin();
1297
1298                 while ( $i < count( $bits ) ){
1299                         $protocol = $bits[$i++];
1300                         $remainder = $bits[$i++];
1301
1302                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1303                                 # Found some characters after the protocol that look promising
1304                                 $url = $protocol . $m[1];
1305                                 $trail = $m[2];
1306
1307                                 # special case: handle urls as url args:
1308                                 # http://www.example.com/foo?=http://www.example.com/bar
1309                                 if(strlen($trail) == 0 &&
1310                                         isset($bits[$i]) &&
1311                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1312                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1313                                 {
1314                                         # add protocol, arg
1315                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1316                                         $i += 2;
1317                                         $trail = $m[2];
1318                                 }
1319
1320                                 # The characters '<' and '>' (which were escaped by
1321                                 # removeHTMLtags()) should not be included in
1322                                 # URLs, per RFC 2396.
1323                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1324                                         $trail = substr($url, $m2[0][1]) . $trail;
1325                                         $url = substr($url, 0, $m2[0][1]);
1326                                 }
1327
1328                                 # Move trailing punctuation to $trail
1329                                 $sep = ',;\.:!?';
1330                                 # If there is no left bracket, then consider right brackets fair game too
1331                                 if ( strpos( $url, '(' ) === false ) {
1332                                         $sep .= ')';
1333                                 }
1334
1335                                 $numSepChars = strspn( strrev( $url ), $sep );
1336                                 if ( $numSepChars ) {
1337                                         $trail = substr( $url, -$numSepChars ) . $trail;
1338                                         $url = substr( $url, 0, -$numSepChars );
1339                                 }
1340
1341                                 $url = Sanitizer::cleanUrl( $url );
1342
1343                                 # Is this an external image?
1344                                 $text = $this->maybeMakeExternalImage( $url );
1345                                 if ( $text === false ) {
1346                                         # Not an image, make a link
1347                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1348                                         # Register it in the output object...
1349                                         # Replace unnecessary URL escape codes with their equivalent characters
1350                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1351                                         $this->mOutput->addExternalLink( $pasteurized );
1352                                 }
1353                                 $s .= $text . $trail;
1354                         } else {
1355                                 $s .= $protocol . $remainder;
1356                         }
1357                 }
1358                 wfProfileOut( $fname );
1359                 return $s;
1360         }
1361
1362         /**
1363          * Replace unusual URL escape codes with their equivalent characters
1364          * @param string
1365          * @return string
1366          * @static
1367          * @fixme This can merge genuinely required bits in the path or query string,
1368          *        breaking legit URLs. A proper fix would treat the various parts of
1369          *        the URL differently; as a workaround, just use the output for
1370          *        statistical records, not for actual linking/output.
1371          */
1372         static function replaceUnusualEscapes( $url ) {
1373                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1374                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1375         }
1376
1377         /**
1378          * Callback function used in replaceUnusualEscapes().
1379          * Replaces unusual URL escape codes with their equivalent character
1380          * @static
1381          * @private
1382          */
1383         private static function replaceUnusualEscapesCallback( $matches ) {
1384                 $char = urldecode( $matches[0] );
1385                 $ord = ord( $char );
1386                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1387                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1388                         // No, shouldn't be escaped
1389                         return $char;
1390                 } else {
1391                         // Yes, leave it escaped
1392                         return $matches[0];
1393                 }
1394         }
1395
1396         /**
1397          * make an image if it's allowed, either through the global
1398          * option or through the exception
1399          * @private
1400          */
1401         function maybeMakeExternalImage( $url ) {
1402                 $sk =& $this->mOptions->getSkin();
1403                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1404                 $imagesexception = !empty($imagesfrom);
1405                 $text = false;
1406                 if ( $this->mOptions->getAllowExternalImages()
1407                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1408                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1409                                 # Image found
1410                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1411                         }
1412                 }
1413                 return $text;
1414         }
1415
1416         /**
1417          * Process [[ ]] wikilinks
1418          *
1419          * @private
1420          */
1421         function replaceInternalLinks( $s ) {
1422                 global $wgContLang;
1423                 static $fname = 'Parser::replaceInternalLinks' ;
1424
1425                 wfProfileIn( $fname );
1426
1427                 wfProfileIn( $fname.'-setup' );
1428                 static $tc = FALSE;
1429                 # the % is needed to support urlencoded titles as well
1430                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1431
1432                 $sk =& $this->mOptions->getSkin();
1433
1434                 #split the entire text string on occurences of [[
1435                 $a = explode( '[[', ' ' . $s );
1436                 #get the first element (all text up to first [[), and remove the space we added
1437                 $s = array_shift( $a );
1438                 $s = substr( $s, 1 );
1439
1440                 # Match a link having the form [[namespace:link|alternate]]trail
1441                 static $e1 = FALSE;
1442                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1443                 # Match cases where there is no "]]", which might still be images
1444                 static $e1_img = FALSE;
1445                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1446                 # Match the end of a line for a word that's not followed by whitespace,
1447                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1448                 $e2 = wfMsgForContent( 'linkprefix' );
1449
1450                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1451
1452                 if( is_null( $this->mTitle ) ) {
1453                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1454                 }
1455                 $nottalk = !$this->mTitle->isTalkPage();
1456
1457                 if ( $useLinkPrefixExtension ) {
1458                         if ( preg_match( $e2, $s, $m ) ) {
1459                                 $first_prefix = $m[2];
1460                         } else {
1461                                 $first_prefix = false;
1462                         }
1463                 } else {
1464                         $prefix = '';
1465                 }
1466
1467                 $selflink = $this->mTitle->getPrefixedText();
1468                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1469                 $useSubpages = $this->areSubpagesAllowed();
1470                 wfProfileOut( $fname.'-setup' );
1471
1472                 # Loop for each link
1473                 for ($k = 0; isset( $a[$k] ); $k++) {
1474                         $line = $a[$k];
1475                         if ( $useLinkPrefixExtension ) {
1476                                 wfProfileIn( $fname.'-prefixhandling' );
1477                                 if ( preg_match( $e2, $s, $m ) ) {
1478                                         $prefix = $m[2];
1479                                         $s = $m[1];
1480                                 } else {
1481                                         $prefix='';
1482                                 }
1483                                 # first link
1484                                 if($first_prefix) {
1485                                         $prefix = $first_prefix;
1486                                         $first_prefix = false;
1487                                 }
1488                                 wfProfileOut( $fname.'-prefixhandling' );
1489                         }
1490
1491                         $might_be_img = false;
1492
1493                         wfProfileIn( "$fname-e1" );
1494                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1495                                 $text = $m[2];
1496                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1497                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1498                                 # the real problem is with the $e1 regex
1499                                 # See bug 1300.
1500                                 #
1501                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1502                                 # and no image is in sight. See bug 2095.
1503                                 #
1504                                 if( $text !== '' &&
1505                                         substr( $m[3], 0, 1 ) === ']' &&
1506                                         strpos($text, '[') !== false
1507                                 )
1508                                 {
1509                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1510                                         $m[3] = substr( $m[3], 1 );
1511                                 }
1512                                 # fix up urlencoded title texts
1513                                 if( strpos( $m[1], '%' ) !== false ) {
1514                                         # Should anchors '#' also be rejected?
1515                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1516                                 }
1517                                 $trail = $m[3];
1518                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1519                                 $might_be_img = true;
1520                                 $text = $m[2];
1521                                 if ( strpos( $m[1], '%' ) !== false ) {
1522                                        $m[1] = urldecode($m[1]);
1523                                 }
1524                                 $trail = "";
1525                         } else { # Invalid form; output directly
1526                                 $s .= $prefix . '[[' . $line ;
1527                                 wfProfileOut( "$fname-e1" );
1528                                 continue;
1529                         }
1530                         wfProfileOut( "$fname-e1" );
1531                         wfProfileIn( "$fname-misc" );
1532
1533                         # Don't allow internal links to pages containing
1534                         # PROTO: where PROTO is a valid URL protocol; these
1535                         # should be external links.
1536                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1537                                 $s .= $prefix . '[[' . $line ;
1538                                 continue;
1539                         }
1540
1541                         # Make subpage if necessary
1542                         if( $useSubpages ) {
1543                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1544                         } else {
1545                                 $link = $m[1];
1546                         }
1547
1548                         $noforce = (substr($m[1], 0, 1) != ':');
1549                         if (!$noforce) {
1550                                 # Strip off leading ':'
1551                                 $link = substr($link, 1);
1552                         }
1553
1554                         wfProfileOut( "$fname-misc" );
1555                         wfProfileIn( "$fname-title" );
1556                         $nt = Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1557                         if( !$nt ) {
1558                                 $s .= $prefix . '[[' . $line;
1559                                 wfProfileOut( "$fname-title" );
1560                                 continue;
1561                         }
1562
1563                         #check other language variants of the link
1564                         #if the article does not exist
1565                         if( $checkVariantLink
1566                             && $nt->getArticleID() == 0 ) {
1567                                 $wgContLang->findVariantLink($link, $nt);
1568                         }
1569
1570                         $ns = $nt->getNamespace();
1571                         $iw = $nt->getInterWiki();
1572                         wfProfileOut( "$fname-title" );
1573
1574                         if ($might_be_img) { # if this is actually an invalid link
1575                                 wfProfileIn( "$fname-might_be_img" );
1576                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1577                                         $found = false;
1578                                         while (isset ($a[$k+1]) ) {
1579                                                 #look at the next 'line' to see if we can close it there
1580                                                 $spliced = array_splice( $a, $k + 1, 1 );
1581                                                 $next_line = array_shift( $spliced );
1582                                                 $m = explode( ']]', $next_line, 3 );
1583                                                 if ( count( $m ) == 3 ) {
1584                                                         # the first ]] closes the inner link, the second the image
1585                                                         $found = true;
1586                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1587                                                         $trail = $m[2];
1588                                                         break;
1589                                                 } elseif ( count( $m ) == 2 ) {
1590                                                         #if there's exactly one ]] that's fine, we'll keep looking
1591                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1592                                                 } else {
1593                                                         #if $next_line is invalid too, we need look no further
1594                                                         $text .= '[[' . $next_line;
1595                                                         break;
1596                                                 }
1597                                         }
1598                                         if ( !$found ) {
1599                                                 # we couldn't find the end of this imageLink, so output it raw
1600                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1601                                                 $text = $this->replaceInternalLinks($text);
1602                                                 $s .= "{$prefix}[[$link|$text";
1603                                                 # note: no $trail, because without an end, there *is* no trail
1604                                                 wfProfileOut( "$fname-might_be_img" );
1605                                                 continue;
1606                                         }
1607                                 } else { #it's not an image, so output it raw
1608                                         $s .= "{$prefix}[[$link|$text";
1609                                         # note: no $trail, because without an end, there *is* no trail
1610                                         wfProfileOut( "$fname-might_be_img" );
1611                                         continue;
1612                                 }
1613                                 wfProfileOut( "$fname-might_be_img" );
1614                         }
1615
1616                         $wasblank = ( '' == $text );
1617                         if( $wasblank ) $text = $link;
1618
1619                         # Link not escaped by : , create the various objects
1620                         if( $noforce ) {
1621
1622                                 # Interwikis
1623                                 wfProfileIn( "$fname-interwiki" );
1624                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1625                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1626                                         $s = rtrim($s . "\n");
1627                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1628                                         wfProfileOut( "$fname-interwiki" );
1629                                         continue;
1630                                 }
1631                                 wfProfileOut( "$fname-interwiki" );
1632
1633                                 if ( $ns == NS_IMAGE ) {
1634                                         wfProfileIn( "$fname-image" );
1635                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1636                                                 # recursively parse links inside the image caption
1637                                                 # actually, this will parse them in any other parameters, too,
1638                                                 # but it might be hard to fix that, and it doesn't matter ATM
1639                                                 $text = $this->replaceExternalLinks($text);
1640                                                 $text = $this->replaceInternalLinks($text);
1641
1642                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1643                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1644                                                 $this->mOutput->addImage( $nt->getDBkey() );
1645
1646                                                 wfProfileOut( "$fname-image" );
1647                                                 continue;
1648                                         } else {
1649                                                 # We still need to record the image's presence on the page
1650                                                 $this->mOutput->addImage( $nt->getDBkey() );
1651                                         }
1652                                         wfProfileOut( "$fname-image" );
1653
1654                                 }
1655
1656                                 if ( $ns == NS_CATEGORY ) {
1657                                         wfProfileIn( "$fname-category" );
1658                                         $s = rtrim($s . "\n"); # bug 87
1659
1660                                         if ( $wasblank ) {
1661                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1662                                                         $sortkey = $this->mTitle->getText();
1663                                                 } else {
1664                                                         $sortkey = $this->mTitle->getPrefixedText();
1665                                                 }
1666                                         } else {
1667                                                 $sortkey = $text;
1668                                         }
1669                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1670                                         $sortkey = str_replace( "\n", '', $sortkey );
1671                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1672                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1673
1674                                         /**
1675                                          * Strip the whitespace Category links produce, see bug 87
1676                                          * @todo We might want to use trim($tmp, "\n") here.
1677                                          */
1678                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1679
1680                                         wfProfileOut( "$fname-category" );
1681                                         continue;
1682                                 }
1683                         }
1684
1685                         if( ( $nt->getPrefixedText() === $selflink ) &&
1686                             ( $nt->getFragment() === '' ) ) {
1687                                 # Self-links are handled specially; generally de-link and change to bold.
1688                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1689                                 continue;
1690                         }
1691
1692                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1693                         if( $ns == NS_MEDIA ) {
1694                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1695                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1696                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1697                                 $this->mOutput->addImage( $nt->getDBkey() );
1698                                 continue;
1699                         } elseif( $ns == NS_SPECIAL ) {
1700                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1701                                 continue;
1702                         } elseif( $ns == NS_IMAGE ) {
1703                                 $img = new Image( $nt );
1704                                 if( $img->exists() ) {
1705                                         // Force a blue link if the file exists; may be a remote
1706                                         // upload on the shared repository, and we want to see its
1707                                         // auto-generated page.
1708                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1709                                         continue;
1710                                 }
1711                         }
1712                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1713                 }
1714                 wfProfileOut( $fname );
1715                 return $s;
1716         }
1717
1718         /**
1719          * Make a link placeholder. The text returned can be later resolved to a real link with
1720          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1721          * parsing of interwiki links, and secondly to allow all existence checks and
1722          * article length checks (for stub links) to be bundled into a single query.
1723          *
1724          */
1725         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1726                 wfProfileIn( __METHOD__ );
1727                 if ( ! is_object($nt) ) {
1728                         # Fail gracefully
1729                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1730                 } else {
1731                         # Separate the link trail from the rest of the link
1732                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1733
1734                         if ( $nt->isExternal() ) {
1735                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1736                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1737                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1738                         } else {
1739                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1740                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1741                                 $this->mLinkHolders['queries'][] = $query;
1742                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1743                                 $this->mLinkHolders['titles'][] = $nt;
1744
1745                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1746                         }
1747                 }
1748                 wfProfileOut( __METHOD__ );
1749                 return $retVal;
1750         }
1751
1752         /**
1753          * Render a forced-blue link inline; protect against double expansion of
1754          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1755          * Since this little disaster has to split off the trail text to avoid
1756          * breaking URLs in the following text without breaking trails on the
1757          * wiki links, it's been made into a horrible function.
1758          *
1759          * @param Title $nt
1760          * @param string $text
1761          * @param string $query
1762          * @param string $trail
1763          * @param string $prefix
1764          * @return string HTML-wikitext mix oh yuck
1765          */
1766         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1767                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1768                 $sk =& $this->mOptions->getSkin();
1769                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1770                 return $this->armorLinks( $link ) . $trail;
1771         }
1772
1773         /**
1774          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1775          * going to go through further parsing steps before inline URL expansion.
1776          *
1777          * In particular this is important when using action=render, which causes
1778          * full URLs to be included.
1779          *
1780          * Oh man I hate our multi-layer parser!
1781          *
1782          * @param string more-or-less HTML
1783          * @return string less-or-more HTML with NOPARSE bits
1784          */
1785         function armorLinks( $text ) {
1786                 return preg_replace( "/\b(" . wfUrlProtocols() . ')/',
1787                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1788         }
1789
1790         /**
1791          * Return true if subpage links should be expanded on this page.
1792          * @return bool
1793          */
1794         function areSubpagesAllowed() {
1795                 # Some namespaces don't allow subpages
1796                 global $wgNamespacesWithSubpages;
1797                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1798         }
1799
1800         /**
1801          * Handle link to subpage if necessary
1802          * @param string $target the source of the link
1803          * @param string &$text the link text, modified as necessary
1804          * @return string the full name of the link
1805          * @private
1806          */
1807         function maybeDoSubpageLink($target, &$text) {
1808                 # Valid link forms:
1809                 # Foobar -- normal
1810                 # :Foobar -- override special treatment of prefix (images, language links)
1811                 # /Foobar -- convert to CurrentPage/Foobar
1812                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1813                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1814                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1815
1816                 $fname = 'Parser::maybeDoSubpageLink';
1817                 wfProfileIn( $fname );
1818                 $ret = $target; # default return value is no change
1819
1820                 # Some namespaces don't allow subpages,
1821                 # so only perform processing if subpages are allowed
1822                 if( $this->areSubpagesAllowed() ) {
1823                         # Look at the first character
1824                         if( $target != '' && $target{0} == '/' ) {
1825                                 # / at end means we don't want the slash to be shown
1826                                 if( substr( $target, -1, 1 ) == '/' ) {
1827                                         $target = substr( $target, 1, -1 );
1828                                         $noslash = $target;
1829                                 } else {
1830                                         $noslash = substr( $target, 1 );
1831                                 }
1832
1833                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1834                                 if( '' === $text ) {
1835                                         $text = $target;
1836                                 } # this might be changed for ugliness reasons
1837                         } else {
1838                                 # check for .. subpage backlinks
1839                                 $dotdotcount = 0;
1840                                 $nodotdot = $target;
1841                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1842                                         ++$dotdotcount;
1843                                         $nodotdot = substr( $nodotdot, 3 );
1844                                 }
1845                                 if($dotdotcount > 0) {
1846                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1847                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1848                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1849                                                 # / at the end means don't show full path
1850                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1851                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1852                                                         if( '' === $text ) {
1853                                                                 $text = $nodotdot;
1854                                                         }
1855                                                 }
1856                                                 $nodotdot = trim( $nodotdot );
1857                                                 if( $nodotdot != '' ) {
1858                                                         $ret .= '/' . $nodotdot;
1859                                                 }
1860                                         }
1861                                 }
1862                         }
1863                 }
1864
1865                 wfProfileOut( $fname );
1866                 return $ret;
1867         }
1868
1869         /**#@+
1870          * Used by doBlockLevels()
1871          * @private
1872          */
1873         /* private */ function closeParagraph() {
1874                 $result = '';
1875                 if ( '' != $this->mLastSection ) {
1876                         $result = '</' . $this->mLastSection  . ">\n";
1877                 }
1878                 $this->mInPre = false;
1879                 $this->mLastSection = '';
1880                 return $result;
1881         }
1882         # getCommon() returns the length of the longest common substring
1883         # of both arguments, starting at the beginning of both.
1884         #
1885         /* private */ function getCommon( $st1, $st2 ) {
1886                 $fl = strlen( $st1 );
1887                 $shorter = strlen( $st2 );
1888                 if ( $fl < $shorter ) { $shorter = $fl; }
1889
1890                 for ( $i = 0; $i < $shorter; ++$i ) {
1891                         if ( $st1{$i} != $st2{$i} ) { break; }
1892                 }
1893                 return $i;
1894         }
1895         # These next three functions open, continue, and close the list
1896         # element appropriate to the prefix character passed into them.
1897         #
1898         /* private */ function openList( $char ) {
1899                 $result = $this->closeParagraph();
1900
1901                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1902                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1903                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1904                 else if ( ';' == $char ) {
1905                         $result .= '<dl><dt>';
1906                         $this->mDTopen = true;
1907                 }
1908                 else { $result = '<!-- ERR 1 -->'; }
1909
1910                 return $result;
1911         }
1912
1913         /* private */ function nextItem( $char ) {
1914                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1915                 else if ( ':' == $char || ';' == $char ) {
1916                         $close = '</dd>';
1917                         if ( $this->mDTopen ) { $close = '</dt>'; }
1918                         if ( ';' == $char ) {
1919                                 $this->mDTopen = true;
1920                                 return $close . '<dt>';
1921                         } else {
1922                                 $this->mDTopen = false;
1923                                 return $close . '<dd>';
1924                         }
1925                 }
1926                 return '<!-- ERR 2 -->';
1927         }
1928
1929         /* private */ function closeList( $char ) {
1930                 if ( '*' == $char ) { $text = '</li></ul>'; }
1931                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1932                 else if ( ':' == $char ) {
1933                         if ( $this->mDTopen ) {
1934                                 $this->mDTopen = false;
1935                                 $text = '</dt></dl>';
1936                         } else {
1937                                 $text = '</dd></dl>';
1938                         }
1939                 }
1940                 else {  return '<!-- ERR 3 -->'; }
1941                 return $text."\n";
1942         }
1943         /**#@-*/
1944
1945         /**
1946          * Make lists from lines starting with ':', '*', '#', etc.
1947          *
1948          * @private
1949          * @return string the lists rendered as HTML
1950          */
1951         function doBlockLevels( $text, $linestart ) {
1952                 $fname = 'Parser::doBlockLevels';
1953                 wfProfileIn( $fname );
1954
1955                 # Parsing through the text line by line.  The main thing
1956                 # happening here is handling of block-level elements p, pre,
1957                 # and making lists from lines starting with * # : etc.
1958                 #
1959                 $textLines = explode( "\n", $text );
1960
1961                 $lastPrefix = $output = '';
1962                 $this->mDTopen = $inBlockElem = false;
1963                 $prefixLength = 0;
1964                 $paragraphStack = false;
1965
1966                 if ( !$linestart ) {
1967                         $output .= array_shift( $textLines );
1968                 }
1969                 foreach ( $textLines as $oLine ) {
1970                         $lastPrefixLength = strlen( $lastPrefix );
1971                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1972                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1973                         if ( !$this->mInPre ) {
1974                                 # Multiple prefixes may abut each other for nested lists.
1975                                 $prefixLength = strspn( $oLine, '*#:;' );
1976                                 $pref = substr( $oLine, 0, $prefixLength );
1977
1978                                 # eh?
1979                                 $pref2 = str_replace( ';', ':', $pref );
1980                                 $t = substr( $oLine, $prefixLength );
1981                                 $this->mInPre = !empty($preOpenMatch);
1982                         } else {
1983                                 # Don't interpret any other prefixes in preformatted text
1984                                 $prefixLength = 0;
1985                                 $pref = $pref2 = '';
1986                                 $t = $oLine;
1987                         }
1988
1989                         # List generation
1990                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1991                                 # Same as the last item, so no need to deal with nesting or opening stuff
1992                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1993                                 $paragraphStack = false;
1994
1995                                 if ( substr( $pref, -1 ) == ';') {
1996                                         # The one nasty exception: definition lists work like this:
1997                                         # ; title : definition text
1998                                         # So we check for : in the remainder text to split up the
1999                                         # title and definition, without b0rking links.
2000                                         $term = $t2 = '';
2001                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2002                                                 $t = $t2;
2003                                                 $output .= $term . $this->nextItem( ':' );
2004                                         }
2005                                 }
2006                         } elseif( $prefixLength || $lastPrefixLength ) {
2007                                 # Either open or close a level...
2008                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2009                                 $paragraphStack = false;
2010
2011                                 while( $commonPrefixLength < $lastPrefixLength ) {
2012                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2013                                         --$lastPrefixLength;
2014                                 }
2015                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2016                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2017                                 }
2018                                 while ( $prefixLength > $commonPrefixLength ) {
2019                                         $char = substr( $pref, $commonPrefixLength, 1 );
2020                                         $output .= $this->openList( $char );
2021
2022                                         if ( ';' == $char ) {
2023                                                 # FIXME: This is dupe of code above
2024                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2025                                                         $t = $t2;
2026                                                         $output .= $term . $this->nextItem( ':' );
2027                                                 }
2028                                         }
2029                                         ++$commonPrefixLength;
2030                                 }
2031                                 $lastPrefix = $pref2;
2032                         }
2033                         if( 0 == $prefixLength ) {
2034                                 wfProfileIn( "$fname-paragraph" );
2035                                 # No prefix (not in list)--go to paragraph mode
2036                                 // XXX: use a stack for nestable elements like span, table and div
2037                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/center|<\\/tr|<\\/td|<\\/th)/iS', $t );
2038                                 $closematch = preg_match(
2039                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2040                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<center)/iS', $t );
2041                                 if ( $openmatch or $closematch ) {
2042                                         $paragraphStack = false;
2043                                         # TODO bug 5718: paragraph closed
2044                                         $output .= $this->closeParagraph();
2045                                         if ( $preOpenMatch and !$preCloseMatch ) {
2046                                                 $this->mInPre = true;
2047                                         }
2048                                         if ( $closematch ) {
2049                                                 $inBlockElem = false;
2050                                         } else {
2051                                                 $inBlockElem = true;
2052                                         }
2053                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2054                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2055                                                 // pre
2056                                                 if ($this->mLastSection != 'pre') {
2057                                                         $paragraphStack = false;
2058                                                         $output .= $this->closeParagraph().'<pre>';
2059                                                         $this->mLastSection = 'pre';
2060                                                 }
2061                                                 $t = substr( $t, 1 );
2062                                         } else {
2063                                                 // paragraph
2064                                                 if ( '' == trim($t) ) {
2065                                                         if ( $paragraphStack ) {
2066                                                                 $output .= $paragraphStack.'<br />';
2067                                                                 $paragraphStack = false;
2068                                                                 $this->mLastSection = 'p';
2069                                                         } else {
2070                                                                 if ($this->mLastSection != 'p' ) {
2071                                                                         $output .= $this->closeParagraph();
2072                                                                         $this->mLastSection = '';
2073                                                                         $paragraphStack = '<p>';
2074                                                                 } else {
2075                                                                         $paragraphStack = '</p><p>';
2076                                                                 }
2077                                                         }
2078                                                 } else {
2079                                                         if ( $paragraphStack ) {
2080                                                                 $output .= $paragraphStack;
2081                                                                 $paragraphStack = false;
2082                                                                 $this->mLastSection = 'p';
2083                                                         } else if ($this->mLastSection != 'p') {
2084                                                                 $output .= $this->closeParagraph().'<p>';
2085                                                                 $this->mLastSection = 'p';
2086                                                         }
2087                                                 }
2088                                         }
2089                                 }
2090                                 wfProfileOut( "$fname-paragraph" );
2091                         }
2092                         // somewhere above we forget to get out of pre block (bug 785)
2093                         if($preCloseMatch && $this->mInPre) {
2094                                 $this->mInPre = false;
2095                         }
2096                         if ($paragraphStack === false) {
2097                                 $output .= $t."\n";
2098                         }
2099                 }
2100                 while ( $prefixLength ) {
2101                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2102                         --$prefixLength;
2103                 }
2104                 if ( '' != $this->mLastSection ) {
2105                         $output .= '</' . $this->mLastSection . '>';
2106                         $this->mLastSection = '';
2107                 }
2108
2109                 wfProfileOut( $fname );
2110                 return $output;
2111         }
2112
2113         /**
2114          * Split up a string on ':', ignoring any occurences inside tags
2115          * to prevent illegal overlapping.
2116          * @param string $str the string to split
2117          * @param string &$before set to everything before the ':'
2118          * @param string &$after set to everything after the ':'
2119          * return string the position of the ':', or false if none found
2120          */
2121         function findColonNoLinks($str, &$before, &$after) {
2122                 $fname = 'Parser::findColonNoLinks';
2123                 wfProfileIn( $fname );
2124
2125                 $pos = strpos( $str, ':' );
2126                 if( $pos === false ) {
2127                         // Nothing to find!
2128                         wfProfileOut( $fname );
2129                         return false;
2130                 }
2131
2132                 $lt = strpos( $str, '<' );
2133                 if( $lt === false || $lt > $pos ) {
2134                         // Easy; no tag nesting to worry about
2135                         $before = substr( $str, 0, $pos );
2136                         $after = substr( $str, $pos+1 );
2137                         wfProfileOut( $fname );
2138                         return $pos;
2139                 }
2140
2141                 // Ugly state machine to walk through avoiding tags.
2142                 $state = MW_COLON_STATE_TEXT;
2143                 $stack = 0;
2144                 $len = strlen( $str );
2145                 for( $i = 0; $i < $len; $i++ ) {
2146                         $c = $str{$i};
2147
2148                         switch( $state ) {
2149                         // (Using the number is a performance hack for common cases)
2150                         case 0: // MW_COLON_STATE_TEXT:
2151                                 switch( $c ) {
2152                                 case "<":
2153                                         // Could be either a <start> tag or an </end> tag
2154                                         $state = MW_COLON_STATE_TAGSTART;
2155                                         break;
2156                                 case ":":
2157                                         if( $stack == 0 ) {
2158                                                 // We found it!
2159                                                 $before = substr( $str, 0, $i );
2160                                                 $after = substr( $str, $i + 1 );
2161                                                 wfProfileOut( $fname );
2162                                                 return $i;
2163                                         }
2164                                         // Embedded in a tag; don't break it.
2165                                         break;
2166                                 default:
2167                                         // Skip ahead looking for something interesting
2168                                         $colon = strpos( $str, ':', $i );
2169                                         if( $colon === false ) {
2170                                                 // Nothing else interesting
2171                                                 wfProfileOut( $fname );
2172                                                 return false;
2173                                         }
2174                                         $lt = strpos( $str, '<', $i );
2175                                         if( $stack === 0 ) {
2176                                                 if( $lt === false || $colon < $lt ) {
2177                                                         // We found it!
2178                                                         $before = substr( $str, 0, $colon );
2179                                                         $after = substr( $str, $colon + 1 );
2180                                                         wfProfileOut( $fname );
2181                                                         return $i;
2182                                                 }
2183                                         }
2184                                         if( $lt === false ) {
2185                                                 // Nothing else interesting to find; abort!
2186                                                 // We're nested, but there's no close tags left. Abort!
2187                                                 break 2;
2188                                         }
2189                                         // Skip ahead to next tag start
2190                                         $i = $lt;
2191                                         $state = MW_COLON_STATE_TAGSTART;
2192                                 }
2193                                 break;
2194                         case 1: // MW_COLON_STATE_TAG:
2195                                 // In a <tag>
2196                                 switch( $c ) {
2197                                 case ">":
2198                                         $stack++;
2199                                         $state = MW_COLON_STATE_TEXT;
2200                                         break;
2201                                 case "/":
2202                                         // Slash may be followed by >?
2203                                         $state = MW_COLON_STATE_TAGSLASH;
2204                                         break;
2205                                 default:
2206                                         // ignore
2207                                 }
2208                                 break;
2209                         case 2: // MW_COLON_STATE_TAGSTART:
2210                                 switch( $c ) {
2211                                 case "/":
2212                                         $state = MW_COLON_STATE_CLOSETAG;
2213                                         break;
2214                                 case "!":
2215                                         $state = MW_COLON_STATE_COMMENT;
2216                                         break;
2217                                 case ">":
2218                                         // Illegal early close? This shouldn't happen D:
2219                                         $state = MW_COLON_STATE_TEXT;
2220                                         break;
2221                                 default:
2222                                         $state = MW_COLON_STATE_TAG;
2223                                 }
2224                                 break;
2225                         case 3: // MW_COLON_STATE_CLOSETAG:
2226                                 // In a </tag>
2227                                 if( $c == ">" ) {
2228                                         $stack--;
2229                                         if( $stack < 0 ) {
2230                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2231                                                 wfProfileOut( $fname );
2232                                                 return false;
2233                                         }
2234                                         $state = MW_COLON_STATE_TEXT;
2235                                 }
2236                                 break;
2237                         case MW_COLON_STATE_TAGSLASH:
2238                                 if( $c == ">" ) {
2239                                         // Yes, a self-closed tag <blah/>
2240                                         $state = MW_COLON_STATE_TEXT;
2241                                 } else {
2242                                         // Probably we're jumping the gun, and this is an attribute
2243                                         $state = MW_COLON_STATE_TAG;
2244                                 }
2245                                 break;
2246                         case 5: // MW_COLON_STATE_COMMENT:
2247                                 if( $c == "-" ) {
2248                                         $state = MW_COLON_STATE_COMMENTDASH;
2249                                 }
2250                                 break;
2251                         case MW_COLON_STATE_COMMENTDASH:
2252                                 if( $c == "-" ) {
2253                                         $state = MW_COLON_STATE_COMMENTDASHDASH;
2254                                 } else {
2255                                         $state = MW_COLON_STATE_COMMENT;
2256                                 }
2257                                 break;
2258                         case MW_COLON_STATE_COMMENTDASHDASH:
2259                                 if( $c == ">" ) {
2260                                         $state = MW_COLON_STATE_TEXT;
2261                                 } else {
2262                                         $state = MW_COLON_STATE_COMMENT;
2263                                 }
2264                                 break;
2265                         default:
2266                                 throw new MWException( "State machine error in $fname" );
2267                         }
2268                 }
2269                 if( $stack > 0 ) {
2270                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2271                         return false;
2272                 }
2273                 wfProfileOut( $fname );
2274                 return false;
2275         }
2276
2277         /**
2278          * Return value of a magic variable (like PAGENAME)
2279          *
2280          * @private
2281          */
2282         function getVariableValue( $index ) {
2283                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2284
2285                 /**
2286                  * Some of these require message or data lookups and can be
2287                  * expensive to check many times.
2288                  */
2289                 static $varCache = array();
2290                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) )
2291                         if ( isset( $varCache[$index] ) )
2292                                 return $varCache[$index];
2293
2294                 $ts = time();
2295                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2296
2297                 switch ( $index ) {
2298                         case 'currentmonth':
2299                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
2300                         case 'currentmonthname':
2301                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2302                         case 'currentmonthnamegen':
2303                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2304                         case 'currentmonthabbrev':
2305                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2306                         case 'currentday':
2307                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2308                         case 'currentday2':
2309                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2310                         case 'pagename':
2311                                 return $this->mTitle->getText();
2312                         case 'pagenamee':
2313                                 return $this->mTitle->getPartialURL();
2314                         case 'fullpagename':
2315                                 return $this->mTitle->getPrefixedText();
2316                         case 'fullpagenamee':
2317                                 return $this->mTitle->getPrefixedURL();
2318                         case 'subpagename':
2319                                 return $this->mTitle->getSubpageText();
2320                         case 'subpagenamee':
2321                                 return $this->mTitle->getSubpageUrlForm();
2322                         case 'basepagename':
2323                                 return $this->mTitle->getBaseText();
2324                         case 'basepagenamee':
2325                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2326                         case 'talkpagename':
2327                                 if( $this->mTitle->canTalk() ) {
2328                                         $talkPage = $this->mTitle->getTalkPage();
2329                                         return $talkPage->getPrefixedText();
2330                                 } else {
2331                                         return '';
2332                                 }
2333                         case 'talkpagenamee':
2334                                 if( $this->mTitle->canTalk() ) {
2335                                         $talkPage = $this->mTitle->getTalkPage();
2336                                         return $talkPage->getPrefixedUrl();
2337                                 } else {
2338                                         return '';
2339                                 }
2340                         case 'subjectpagename':
2341                                 $subjPage = $this->mTitle->getSubjectPage();
2342                                 return $subjPage->getPrefixedText();
2343                         case 'subjectpagenamee':
2344                                 $subjPage = $this->mTitle->getSubjectPage();
2345                                 return $subjPage->getPrefixedUrl();
2346                         case 'revisionid':
2347                                 return $this->mRevisionId;
2348                         case 'namespace':
2349                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2350                         case 'namespacee':
2351                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2352                         case 'talkspace':
2353                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2354                         case 'talkspacee':
2355                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2356                         case 'subjectspace':
2357                                 return $this->mTitle->getSubjectNsText();
2358                         case 'subjectspacee':
2359                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2360                         case 'currentdayname':
2361                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2362                         case 'currentyear':
2363                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2364                         case 'currenttime':
2365                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2366                         case 'currentweek':
2367                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2368                                 // int to remove the padding
2369                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2370                         case 'currentdow':
2371                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2372                         case 'numberofarticles':
2373                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
2374                         case 'numberoffiles':
2375                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
2376                         case 'numberofusers':
2377                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfUsers() );
2378                         case 'numberofpages':
2379                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfPages() );
2380                         case 'numberofadmins':
2381                                 return $varCache[$index]  = $wgContLang->formatNum( wfNumberOfAdmins() );
2382                         case 'currenttimestamp':
2383                                 return $varCache[$index] = wfTimestampNow();
2384                         case 'currentversion':
2385                                 global $wgVersion;
2386                                 return $wgVersion;
2387                         case 'sitename':
2388                                 return $wgSitename;
2389                         case 'server':
2390                                 return $wgServer;
2391                         case 'servername':
2392                                 return $wgServerName;
2393                         case 'scriptpath':
2394                                 return $wgScriptPath;
2395                         case 'directionmark':
2396                                 return $wgContLang->getDirMark();
2397                         case 'contentlanguage':
2398                                 global $wgContLanguageCode;
2399                                 return $wgContLanguageCode;
2400                         default:
2401                                 $ret = null;
2402                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2403                                         return $ret;
2404                                 else
2405                                         return null;
2406                 }
2407         }
2408
2409         /**
2410          * initialise the magic variables (like CURRENTMONTHNAME)
2411          *
2412          * @private
2413          */
2414         function initialiseVariables() {
2415                 $fname = 'Parser::initialiseVariables';
2416                 wfProfileIn( $fname );
2417                 $variableIDs = MagicWord::getVariableIDs();
2418
2419                 $this->mVariables = array();
2420                 foreach ( $variableIDs as $id ) {
2421                         $mw =& MagicWord::get( $id );
2422                         $mw->addToArray( $this->mVariables, $id );
2423                 }
2424                 wfProfileOut( $fname );
2425         }
2426
2427         /**
2428          * parse any parentheses in format ((title|part|part))
2429          * and call callbacks to get a replacement text for any found piece
2430          *
2431          * @param string $text The text to parse
2432          * @param array $callbacks rules in form:
2433          *     '{' => array(                            # opening parentheses
2434          *                                      'end' => '}',   # closing parentheses
2435          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2436          *                                                                3 => callback         # replacement callback to call if {{{..}}} is found
2437          *                                                                )
2438          *                                      )
2439          *                                      'min' => 2,     # Minimum parenthesis count in cb
2440          *                                      'max' => 3,     # Maximum parenthesis count in cb
2441          * @private
2442          */
2443         function replace_callback ($text, $callbacks) {
2444                 wfProfileIn( __METHOD__ );
2445                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2446                 $lastOpeningBrace = -1;                 # last not closed parentheses
2447
2448                 $validOpeningBraces = implode( '', array_keys( $callbacks ) );
2449
2450                 $i = 0;
2451                 while ( $i < strlen( $text ) ) {
2452                         # Find next opening brace, closing brace or pipe
2453                         if ( $lastOpeningBrace == -1 ) {
2454                                 $currentClosing = '';
2455                                 $search = $validOpeningBraces;
2456                         } else {
2457                                 $currentClosing = $openingBraceStack[$lastOpeningBrace]['braceEnd'];
2458                                 $search = $validOpeningBraces . '|' . $currentClosing;
2459                         }
2460                         $rule = null;
2461                         $i += strcspn( $text, $search, $i );
2462                         if ( $i < strlen( $text ) ) {
2463                                 if ( $text[$i] == '|' ) {
2464                                         $found = 'pipe';
2465                                 } elseif ( $text[$i] == $currentClosing ) {
2466                                         $found = 'close';
2467                                 } else {
2468                                         $found = 'open';
2469                                         $rule = $callbacks[$text[$i]];
2470                                 }
2471                         } else {
2472                                 # All done
2473                                 break;
2474                         }
2475
2476                         if ( $found == 'open' ) {
2477                                 # found opening brace, let's add it to parentheses stack
2478                                 $piece = array('brace' => $text[$i],
2479                                                            'braceEnd' => $rule['end'],
2480                                                            'title' => '',
2481                                                            'parts' => null);
2482
2483                                 # count opening brace characters
2484                                 $piece['count'] = strspn( $text, $piece['brace'], $i );
2485                                 $piece['startAt'] = $piece['partStart'] = $i + $piece['count'];
2486                                 $i += $piece['count'];
2487
2488                                 # we need to add to stack only if opening brace count is enough for one of the rules
2489                                 if ( $piece['count'] >= $rule['min'] ) {
2490                                         $lastOpeningBrace ++;
2491                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2492                                 }
2493                         } elseif ( $found == 'close' ) {
2494                                 # lets check if it is enough characters for closing brace
2495                                 $maxCount = $openingBraceStack[$lastOpeningBrace]['count'];
2496                                 $count = strspn( $text, $text[$i], $i, $maxCount );
2497
2498                                 # check for maximum matching characters (if there are 5 closing
2499                                 # characters, we will probably need only 3 - depending on the rules)
2500                                 $matchingCount = 0;
2501                                 $matchingCallback = null;
2502                                 $cbType = $callbacks[$openingBraceStack[$lastOpeningBrace]['brace']];
2503                                 if ( $count > $cbType['max'] ) {
2504                                         # The specified maximum exists in the callback array, unless the caller
2505                                         # has made an error
2506                                         $matchingCount = $cbType['max'];
2507                                 } else {
2508                                         # Count is less than the maximum
2509                                         # Skip any gaps in the callback array to find the true largest match
2510                                         # Need to use array_key_exists not isset because the callback can be null
2511                                         $matchingCount = $count;
2512                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $cbType['cb'] ) ) {
2513                                                 --$matchingCount;
2514                                         }
2515                                 }
2516
2517                                 if ($matchingCount <= 0) {
2518                                         $i += $count;
2519                                         continue;
2520                                 }
2521                                 $matchingCallback = $cbType['cb'][$matchingCount];
2522
2523                                 # let's set a title or last part (if '|' was found)
2524                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2525                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2526                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2527                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2528                                 } else {
2529                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2530                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2531                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2532                                 }
2533
2534                                 $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2535                                 $pieceEnd = $i + $matchingCount;
2536
2537                                 if( is_callable( $matchingCallback ) ) {
2538                                         $cbArgs = array (
2539                                                                          'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2540                                                                          'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2541                                                                          'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2542                                                                          'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == "\n")),
2543                                                                          );
2544                                         # finally we can call a user callback and replace piece of text
2545                                         $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2546                                         $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2547                                         $i = $pieceStart + strlen($replaceWith);
2548                                 } else {
2549                                         # null value for callback means that parentheses should be parsed, but not replaced
2550                                         $i += $matchingCount;
2551                                 }
2552
2553                                 # reset last opening parentheses, but keep it in case there are unused characters
2554                                 $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2555                                                            'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2556                                                            'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2557                                                            'title' => '',
2558                                                            'parts' => null,
2559                                                            'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2560                                 $openingBraceStack[$lastOpeningBrace--] = null;
2561
2562                                 if ($matchingCount < $piece['count']) {
2563                                         $piece['count'] -= $matchingCount;
2564                                         $piece['startAt'] -= $matchingCount;
2565                                         $piece['partStart'] = $piece['startAt'];
2566                                         # do we still qualify for any callback with remaining count?
2567                                         $currentCbList = $callbacks[$piece['brace']]['cb'];
2568                                         while ( $piece['count'] ) {
2569                                                 if ( array_key_exists( $piece['count'], $currentCbList ) ) {
2570                                                         $lastOpeningBrace++;
2571                                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2572                                                         break;
2573                                                 }
2574                                                 --$piece['count'];
2575                                         }
2576                                 }
2577                         } elseif ( $found == 'pipe' ) {
2578                                 # lets set a title if it is a first separator, or next part otherwise
2579                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2580                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2581                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2582                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2583                                         $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2584                                 } else {
2585                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2586                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2587                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2588                                 }
2589                                 $openingBraceStack[$lastOpeningBrace]['partStart'] = ++$i;
2590                         }
2591                 }
2592
2593                 wfProfileOut( __METHOD__ );
2594                 return $text;
2595         }
2596
2597         /**
2598          * Replace magic variables, templates, and template arguments
2599          * with the appropriate text. Templates are substituted recursively,
2600          * taking care to avoid infinite loops.
2601          *
2602          * Note that the substitution depends on value of $mOutputType:
2603          *  OT_WIKI: only {{subst:}} templates
2604          *  OT_MSG: only magic variables
2605          *  OT_HTML: all templates and magic variables
2606          *
2607          * @param string $tex The text to transform
2608          * @param array $args Key-value pairs representing template parameters to substitute
2609          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2610          * @private
2611          */
2612         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2613                 # Prevent too big inclusions
2614                 if( strlen( $text ) > MAX_INCLUDE_SIZE ) {
2615                         return $text;
2616                 }
2617
2618                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2619                 wfProfileIn( $fname );
2620
2621                 # This function is called recursively. To keep track of arguments we need a stack:
2622                 array_push( $this->mArgStack, $args );
2623
2624                 $braceCallbacks = array();
2625                 if ( !$argsOnly ) {
2626                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2627                 }
2628                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
2629                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2630                 }
2631                 if ( $braceCallbacks ) {
2632                         $callbacks = array(
2633                                 '{' => array(
2634                                         'end' => '}',
2635                                         'cb' => $braceCallbacks,
2636                                         'min' => $argsOnly ? 3 : 2,
2637                                         'max' => isset( $braceCallbacks[3] ) ? 3 : 2,
2638                                 ),
2639                                 '[' => array(
2640                                         'end' => ']',
2641                                         'cb' => array(2=>null),
2642                                         'min' => 2,
2643                                         'max' => 2,
2644                                 )
2645                         );
2646                         $text = $this->replace_callback ($text, $callbacks);
2647
2648                         array_pop( $this->mArgStack );
2649                 }
2650                 wfProfileOut( $fname );
2651                 return $text;
2652         }
2653
2654         /**
2655          * Replace magic variables
2656          * @private
2657          */
2658         function variableSubstitution( $matches ) {
2659                 $fname = 'Parser::variableSubstitution';
2660                 $varname = $matches[1];
2661                 wfProfileIn( $fname );
2662                 $skip = false;
2663                 if ( $this->mOutputType == OT_WIKI ) {
2664                         # Do only magic variables prefixed by SUBST
2665                         $mwSubst =& MagicWord::get( 'subst' );
2666                         if (!$mwSubst->matchStartAndRemove( $varname ))
2667                                 $skip = true;
2668                         # Note that if we don't substitute the variable below,
2669                         # we don't remove the {{subst:}} magic word, in case
2670                         # it is a template rather than a magic variable.
2671                 }
2672                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2673                         $id = $this->mVariables[$varname];
2674                         $text = $this->getVariableValue( $id );
2675                         $this->mOutput->mContainsOldMagic = true;
2676                 } else {
2677                         $text = $matches[0];
2678                 }
2679                 wfProfileOut( $fname );
2680                 return $text;
2681         }
2682
2683         # Split template arguments
2684         function getTemplateArgs( $argsString ) {
2685                 if ( $argsString === '' ) {
2686                         return array();
2687                 }
2688
2689                 $args = explode( '|', substr( $argsString, 1 ) );
2690
2691                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2692                 # merged with the next arg because the '|' character between belongs
2693                 # to the link syntax and not the template parameter syntax.
2694                 $argc = count($args);
2695
2696                 for ( $i = 0; $i < $argc-1; $i++ ) {
2697                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2698                                 $args[$i] .= '|'.$args[$i+1];
2699                                 array_splice($args, $i+1, 1);
2700                                 $i--;
2701                                 $argc--;
2702                         }
2703                 }
2704
2705                 return $args;
2706         }
2707
2708         /**
2709          * Return the text of a template, after recursively
2710          * replacing any variables or templates within the template.
2711          *
2712          * @param array $piece The parts of the template
2713          *  $piece['text']: matched text
2714          *  $piece['title']: the title, i.e. the part before the |
2715          *  $piece['parts']: the parameter array
2716          * @return string the text of the template
2717          * @private
2718          */
2719         function braceSubstitution( $piece ) {
2720                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $action;
2721                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2722                 wfProfileIn( $fname );
2723                 wfProfileIn( __METHOD__.'-setup' );
2724
2725                 # Flags
2726                 $found = false;             # $text has been filled
2727                 $nowiki = false;            # wiki markup in $text should be escaped
2728                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2729                 $noargs = false;            # Don't replace triple-brace arguments in $text
2730                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2731                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2732                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2733
2734                 # Title object, where $text came from
2735                 $title = NULL;
2736
2737                 $linestart = '';
2738
2739                 # $part1 is the bit before the first |, and must contain only title characters
2740                 # $args is a list of arguments, starting from index 0, not including $part1
2741
2742                 $part1 = $piece['title'];
2743                 # If the third subpattern matched anything, it will start with |
2744
2745                 if (null == $piece['parts']) {
2746                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2747                         if ($replaceWith != $piece['text']) {
2748                                 $text = $replaceWith;
2749                                 $found = true;
2750                                 $noparse = true;
2751                                 $noargs = true;
2752                         }
2753                 }
2754
2755                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2756                 $argc = count( $args );
2757                 wfProfileOut( __METHOD__.'-setup' );
2758
2759                 # SUBST
2760                 wfProfileIn( __METHOD__.'-modifiers' );
2761                 if ( !$found ) {
2762                         $mwSubst =& MagicWord::get( 'subst' );
2763                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2764                                 # One of two possibilities is true:
2765                                 # 1) Found SUBST but not in the PST phase
2766                                 # 2) Didn't find SUBST and in the PST phase
2767                                 # In either case, return without further processing
2768                                 $text = $piece['text'];
2769                                 $found = true;
2770                                 $noparse = true;
2771                                 $noargs = true;
2772                         }
2773                 }
2774
2775                 # MSG, MSGNW, INT and RAW
2776                 if ( !$found ) {
2777                         # Check for MSGNW:
2778                         $mwMsgnw =& MagicWord::get( 'msgnw' );
2779                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2780                                 $nowiki = true;
2781                         } else {
2782                                 # Remove obsolete MSG:
2783                                 $mwMsg =& MagicWord::get( 'msg' );
2784                                 $mwMsg->matchStartAndRemove( $part1 );
2785                         }
2786
2787                         # Check for RAW:
2788                         $mwRaw =& MagicWord::get( 'raw' );
2789                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2790                                 $forceRawInterwiki = true;
2791                         }
2792
2793                         # Check if it is an internal message
2794                         $mwInt =& MagicWord::get( 'int' );
2795                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2796                                 if ( $this->incrementIncludeCount( 'int:'.$part1 ) ) {
2797                                         $text = $linestart . wfMsgReal( $part1, $args, true );
2798                                         $found = true;
2799                                 }
2800                         }
2801                 }
2802                 wfProfileOut( __METHOD__.'-modifiers' );
2803
2804                 # Parser functions
2805                 if ( !$found ) {
2806                         wfProfileIn( __METHOD__ . '-pfunc' );
2807
2808                         $colonPos = strpos( $part1, ':' );
2809                         if ( $colonPos !== false ) {
2810                                 # Case sensitive functions
2811                                 $function = substr( $part1, 0, $colonPos );
2812                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2813                                         $function = $this->mFunctionSynonyms[1][$function];
2814                                 } else {
2815                                         # Case insensitive functions
2816                                         $function = strtolower( $function );
2817                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2818                                                 $function = $this->mFunctionSynonyms[0][$function];
2819                                         } else {
2820                                                 $function = false;
2821                                         }
2822                                 }
2823                                 if ( $function ) {
2824                                         $funcArgs = array_map( 'trim', $args );
2825                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
2826                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
2827                                         $found = true;
2828
2829                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
2830                                         //$noargs = true;
2831                                         //$noparse = true;
2832
2833                                         if ( is_array( $result ) ) {
2834                                                 if ( isset( $result[0] ) ) {
2835                                                         $text = $linestart . $result[0];
2836                                                         unset( $result[0] );
2837                                                 }
2838
2839                                                 // Extract flags into the local scope
2840                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
2841                                                 extract( $result );
2842                                         } else {
2843                                                 $text = $linestart . $result;
2844                                         }
2845                                 }
2846                         }
2847                         wfProfileOut( __METHOD__ . '-pfunc' );
2848                 }
2849
2850                 # Template table test
2851
2852                 # Did we encounter this template already? If yes, it is in the cache
2853                 # and we need to check for loops.
2854                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
2855                         $found = true;
2856
2857                         # Infinite loop test
2858                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2859                                 $noparse = true;
2860                                 $noargs = true;
2861                                 $found = true;
2862                                 $text = $linestart .
2863                                         '{{' . $part1 . '}}' .
2864                                         '<!-- WARNING: template loop detected -->';
2865                                 wfDebug( __METHOD__.": template loop broken at '$part1'\n" );
2866                         } else {
2867                                 # set $text to cached message.
2868                                 $text = $linestart . $this->mTemplates[$piece['title']];
2869                         }
2870                 }
2871
2872                 # Load from database
2873                 $lastPathLevel = $this->mTemplatePath;
2874                 if ( !$found ) {
2875                         wfProfileIn( __METHOD__ . '-loadtpl' );
2876                         $ns = NS_TEMPLATE;
2877                         # declaring $subpage directly in the function call
2878                         # does not work correctly with references and breaks
2879                         # {{/subpage}}-style inclusions
2880                         $subpage = '';
2881                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2882                         if ($subpage !== '') {
2883                                 $ns = $this->mTitle->getNamespace();
2884                         }
2885                         $title = Title::newFromText( $part1, $ns );
2886
2887
2888                         if ( !is_null( $title ) ) {
2889                                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
2890                                 # Check for language variants if the template is not found
2891                                 if($checkVariantLink && $title->getArticleID() == 0){
2892                                         $wgContLang->findVariantLink($part1, $title);
2893                                 }
2894
2895                                 if ( !$title->isExternal() ) {
2896                                         # Check for excessive inclusion
2897                                         $dbk = $title->getPrefixedDBkey();
2898                                         if ( $this->incrementIncludeCount( $dbk ) ) {
2899                                                 if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->mOutputType != OT_WIKI ) {
2900                                                         $text = SpecialPage::capturePath( $title );
2901                                                         if ( is_string( $text ) ) {
2902                                                                 $found = true;
2903                                                                 $noparse = true;
2904                                                                 $noargs = true;
2905                                                                 $isHTML = true;
2906                                                                 $this->disableCache();
2907                                                         }
2908                                                 } else {
2909                                                         $articleContent = $this->fetchTemplate( $title );
2910                                                         if ( $articleContent !== false ) {
2911                                                                 $found = true;
2912                                                                 $text = $articleContent;
2913                                                                 $replaceHeadings = true;
2914                                                         }
2915                                                 }
2916                                         }
2917
2918                                         # If the title is valid but undisplayable, make a link to it
2919                                         if ( $this->mOutputType == OT_HTML && !$found ) {
2920                                                 $text = '[['.$title->getPrefixedText().']]';
2921                                                 $found = true;
2922                                         }
2923                                 } elseif ( $title->isTrans() ) {
2924                                         // Interwiki transclusion
2925                                         if ( $this->mOutputType == OT_HTML && !$forceRawInterwiki ) {
2926                                                 $text = $this->interwikiTransclude( $title, 'render' );
2927                                                 $isHTML = true;
2928                                                 $noparse = true;
2929                                         } else {
2930                                                 $text = $this->interwikiTransclude( $title, 'raw' );
2931                                                 $replaceHeadings = true;
2932                                         }
2933                                         $found = true;
2934                                 }
2935
2936                                 # Template cache array insertion
2937                                 # Use the original $piece['title'] not the mangled $part1, so that
2938                                 # modifiers such as RAW: produce separate cache entries
2939                                 if( $found ) {
2940                                         if( $isHTML ) {
2941                                                 // A special page; don't store it in the template cache.
2942                                         } else {
2943                                                 $this->mTemplates[$piece['title']] = $text;
2944                                         }
2945                                         $text = $linestart . $text;
2946                                 }
2947                         }
2948                         wfProfileOut( __METHOD__ . '-loadtpl' );
2949                 }
2950
2951                 # Recursive parsing, escaping and link table handling
2952                 # Only for HTML output
2953                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2954                         $text = wfEscapeWikiText( $text );
2955                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found ) {
2956                         if ( $noargs ) {
2957                                 $assocArgs = array();
2958                         } else {
2959                                 # Clean up argument array
2960                                 $assocArgs = array();
2961                                 $index = 1;
2962                                 foreach( $args as $arg ) {
2963                                         $eqpos = strpos( $arg, '=' );
2964                                         if ( $eqpos === false ) {
2965                                                 $assocArgs[$index++] = $arg;
2966                                         } else {
2967                                                 $name = trim( substr( $arg, 0, $eqpos ) );
2968                                                 $value = trim( substr( $arg, $eqpos+1 ) );
2969                                                 if ( $value === false ) {
2970                                                         $value = '';
2971                                                 }
2972                                                 if ( $name !== false ) {
2973                                                         $assocArgs[$name] = $value;
2974                                                 }
2975                                         }
2976                                 }
2977
2978                                 # Add a new element to the templace recursion path
2979                                 $this->mTemplatePath[$part1] = 1;
2980                         }
2981
2982                         if ( !$noparse ) {
2983                                 # If there are any <onlyinclude> tags, only include them
2984                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
2985                                         preg_match_all( '/<onlyinclude>(.*?)\n?<\/onlyinclude>/s', $text, $m );
2986                                         $text = '';
2987                                         foreach ($m[1] as $piece)
2988                                                 $text .= $piece;
2989                                 }
2990                                 # Remove <noinclude> sections and <includeonly> tags
2991                                 $text = preg_replace( '/<noinclude>.*?<\/noinclude>/s', '', $text );
2992                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
2993
2994                                 if( $this->mOutputType == OT_HTML ) {
2995                                         # Strip <nowiki>, <pre>, etc.
2996                                         $text = $this->strip( $text, $this->mStripState );
2997                                         $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
2998                                 }
2999                                 $text = $this->replaceVariables( $text, $assocArgs );
3000
3001                                 # If the template begins with a table or block-level
3002                                 # element, it should be treated as beginning a new line.
3003                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) {
3004                                         $text = "\n" . $text;
3005                                 }
3006                         } elseif ( !$noargs ) {
3007                                 # $noparse and !$noargs
3008                                 # Just replace the arguments, not any double-brace items
3009                                 # This is used for rendered interwiki transclusion
3010                                 $text = $this->replaceVariables( $text, $assocArgs, true );
3011                         }
3012                 }
3013                 # Prune lower levels off the recursion check path
3014                 $this->mTemplatePath = $lastPathLevel;
3015
3016                 if ( !$found ) {
3017                         wfProfileOut( $fname );
3018                         return $piece['text'];
3019                 } else {
3020                         wfProfileIn( __METHOD__ . '-placeholders' );
3021                         if ( $isHTML ) {
3022                                 # Replace raw HTML by a placeholder
3023                                 # Add a blank line preceding, to prevent it from mucking up
3024                                 # immediately preceding headings
3025                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
3026                         } else {
3027                                 # replace ==section headers==
3028                                 # XXX this needs to go away once we have a better parser.
3029                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
3030                                         if( !is_null( $title ) )
3031                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
3032                                         else
3033                                                 $encodedname = base64_encode("");
3034                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
3035                                                 PREG_SPLIT_DELIM_CAPTURE);
3036                                         $text = '';
3037                                         $nsec = 0;
3038                                         for( $i = 0; $i < count($m); $i += 2 ) {
3039                                                 $text .= $m[$i];
3040                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
3041                                                 $hl = $m[$i + 1];
3042                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
3043                                                         $text .= $hl;
3044                                                         continue;
3045                                                 }
3046                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
3047                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
3048                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
3049
3050                                                 $nsec++;
3051                                         }
3052                                 }
3053                         }
3054                         wfProfileOut( __METHOD__ . '-placeholders' );
3055                 }
3056
3057                 # Prune lower levels off the recursion check path
3058                 $this->mTemplatePath = $lastPathLevel;
3059
3060                 if ( !$found ) {
3061                         wfProfileOut( $fname );
3062                         return $piece['text'];
3063                 } else {
3064                         wfProfileOut( $fname );
3065                         return $text;
3066                 }
3067         }
3068
3069         /**
3070          * Fetch the unparsed text of a template and register a reference to it.
3071          */
3072         function fetchTemplate( $title ) {
3073                 $text = false;
3074                 // Loop to fetch the article, with up to 1 redirect
3075                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3076                         $rev = Revision::newFromTitle( $title );
3077                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
3078                         if ( !$rev ) {
3079                                 break;
3080                         }
3081                         $text = $rev->getText();
3082                         if ( $text === false ) {
3083                                 break;
3084                         }
3085                         // Redirect?
3086                         $title = Title::newFromRedirect( $text );
3087                 }
3088                 return $text;
3089         }
3090
3091         /**
3092          * Transclude an interwiki link.
3093          */
3094         function interwikiTransclude( $title, $action ) {
3095                 global $wgEnableScaryTranscluding, $wgCanonicalNamespaceNames;
3096
3097                 if (!$wgEnableScaryTranscluding)
3098                         return wfMsg('scarytranscludedisabled');
3099
3100                 // The namespace will actually only be 0 or 10, depending on whether there was a leading :
3101                 // But we'll handle it generally anyway
3102                 if ( $title->getNamespace() ) {
3103                         // Use the canonical namespace, which should work anywhere
3104                         $articleName = $wgCanonicalNamespaceNames[$title->getNamespace()] . ':' . $title->getDBkey();
3105                 } else {
3106                         $articleName = $title->getDBkey();
3107                 }
3108
3109                 $url = str_replace('$1', urlencode($articleName), Title::getInterwikiLink($title->getInterwiki()));
3110                 $url .= "?action=$action";
3111                 if (strlen($url) > 255)
3112                         return wfMsg('scarytranscludetoolong');
3113                 return $this->fetchScaryTemplateMaybeFromCache($url);
3114         }
3115
3116         function fetchScaryTemplateMaybeFromCache($url) {
3117                 global $wgTranscludeCacheExpiry;
3118                 $dbr =& wfGetDB(DB_SLAVE);
3119                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3120                                 array('tc_url' => $url));
3121                 if ($obj) {
3122                         $time = $obj->tc_time;
3123                         $text = $obj->tc_contents;
3124                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3125                                 return $text;
3126                         }
3127                 }
3128
3129                 $text = Http::get($url);
3130                 if (!$text)
3131                         return wfMsg('scarytranscludefailed', $url);
3132
3133                 $dbw =& wfGetDB(DB_MASTER);
3134                 $dbw->replace('transcache', array('tc_url'), array(
3135                         'tc_url' => $url,
3136                         'tc_time' => time(),
3137                         'tc_contents' => $text));
3138                 return $text;
3139         }
3140
3141
3142         /**
3143          * Triple brace replacement -- used for template arguments
3144          * @private
3145          */
3146         function argSubstitution( $matches ) {
3147                 $arg = trim( $matches['title'] );
3148                 $text = $matches['text'];
3149                 $inputArgs = end( $this->mArgStack );
3150
3151                 if ( array_key_exists( $arg, $inputArgs ) ) {
3152                         $text = $inputArgs[$arg];
3153                 } else if ($this->mOutputType == OT_HTML && null != $matches['parts'] && count($matches['parts']) > 0) {
3154                         $text = $matches['parts'][0];
3155                 }
3156
3157                 return $text;
3158         }
3159
3160         /**
3161          * Returns true if the function is allowed to include this entity
3162          * @private
3163          */
3164         function incrementIncludeCount( $dbk ) {
3165                 if ( !array_key_exists( $dbk, $this->mIncludeCount ) ) {
3166                         $this->mIncludeCount[$dbk] = 0;
3167                 }
3168                 if ( ++$this->mIncludeCount[$dbk] <= MAX_INCLUDE_REPEAT ) {
3169                         return true;
3170                 } else {
3171                         return false;
3172                 }
3173         }
3174
3175         /**
3176          * Detect __NOGALLERY__ magic word and set a placeholder
3177          */
3178         function stripNoGallery( &$text ) {
3179                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3180                 # do not add TOC
3181                 $mw = MagicWord::get( 'nogallery' );
3182                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3183         }
3184
3185         /**
3186          * Detect __TOC__ magic word and set a placeholder
3187          */
3188         function stripToc( $text ) {
3189                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3190                 # do not add TOC
3191                 $mw = MagicWord::get( 'notoc' );
3192                 if( $mw->matchAndRemove( $text ) ) {
3193                         $this->mShowToc = false;
3194                 }
3195
3196                 $mw = MagicWord::get( 'toc' );
3197                 if( $mw->match( $text ) ) {
3198                         $this->mShowToc = true;
3199                         $this->mForceTocPosition = true;
3200
3201                         // Set a placeholder. At the end we'll fill it in with the TOC.
3202                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3203
3204                         // Only keep the first one.
3205                         $text = $mw->replace( '', $text );
3206                 }
3207                 return $text;
3208         }
3209
3210         /**
3211          * This function accomplishes several tasks:
3212          * 1) Auto-number headings if that option is enabled
3213          * 2) Add an [edit] link to sections for logged in users who have enabled the option
3214          * 3) Add a Table of contents on the top for users who have enabled the option
3215          * 4) Auto-anchor headings
3216          *
3217          * It loops through all headlines, collects the necessary data, then splits up the
3218          * string and re-inserts the newly formatted headlines.
3219          *
3220          * @param string $text
3221          * @param boolean $isMain
3222          * @private
3223          */
3224         function formatHeadings( $text, $isMain=true ) {
3225                 global $wgMaxTocLevel, $wgContLang;
3226
3227                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3228                 if( !$this->mTitle->userCanEdit() ) {
3229                         $showEditLink = 0;
3230                 } else {
3231                         $showEditLink = $this->mOptions->getEditSection();
3232                 }
3233
3234                 # Inhibit editsection links if requested in the page
3235                 $esw =& MagicWord::get( 'noeditsection' );
3236                 if( $esw->matchAndRemove( $text ) ) {
3237                         $showEditLink = 0;
3238                 }
3239
3240                 # Get all headlines for numbering them and adding funky stuff like [edit]
3241                 # links - this is for later, but we need the number of headlines right now
3242                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
3243
3244                 # if there are fewer than 4 headlines in the article, do not show TOC
3245                 # unless it's been explicitly enabled.
3246                 $enoughToc = $this->mShowToc &&
3247                         (($numMatches >= 4) || $this->mForceTocPosition);
3248
3249                 # Allow user to stipulate that a page should have a "new section"
3250                 # link added via __NEWSECTIONLINK__
3251                 $mw =& MagicWord::get( 'newsectionlink' );
3252                 if( $mw->matchAndRemove( $text ) )
3253                         $this->mOutput->setNewSection( true );
3254
3255                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3256                 # override above conditions and always show TOC above first header
3257                 $mw =& MagicWord::get( 'forcetoc' );
3258                 if ($mw->matchAndRemove( $text ) ) {
3259                         $this->mShowToc = true;
3260                         $enoughToc = true;
3261                 }
3262
3263                 # Never ever show TOC if no headers
3264                 if( $numMatches < 1 ) {
3265                         $enoughToc = false;
3266                 }
3267
3268                 # We need this to perform operations on the HTML
3269                 $sk =& $this->mOptions->getSkin();
3270
3271                 # headline counter
3272                 $headlineCount = 0;
3273                 $sectionCount = 0; # headlineCount excluding template sections
3274
3275                 # Ugh .. the TOC should have neat indentation levels which can be
3276                 # passed to the skin functions. These are determined here
3277                 $toc = '';
3278                 $full = '';
3279                 $head = array();
3280                 $sublevelCount = array();
3281                 $levelCount = array();
3282                 $toclevel = 0;
3283                 $level = 0;
3284                 $prevlevel = 0;
3285                 $toclevel = 0;
3286                 $prevtoclevel = 0;
3287
3288                 foreach( $matches[3] as $headline ) {
3289                         $istemplate = 0;
3290                         $templatetitle = '';
3291                         $templatesection = 0;
3292                         $numbering = '';
3293
3294                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3295                                 $istemplate = 1;
3296                                 $templatetitle = base64_decode($mat[1]);
3297                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3298                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3299                         }
3300
3301                         if( $toclevel ) {
3302                                 $prevlevel = $level;
3303                                 $prevtoclevel = $toclevel;
3304                         }
3305                         $level = $matches[1][$headlineCount];
3306
3307                         if( $doNumberHeadings || $enoughToc ) {
3308
3309                                 if ( $level > $prevlevel ) {
3310                                         # Increase TOC level
3311                                         $toclevel++;
3312                                         $sublevelCount[$toclevel] = 0;
3313                                         if( $toclevel<$wgMaxTocLevel ) {
3314                                                 $toc .= $sk->tocIndent();
3315                                         }
3316                                 }
3317                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3318                                         # Decrease TOC level, find level to jump to
3319
3320                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3321                                                 # Can only go down to level 1
3322                                                 $toclevel = 1;
3323                                         } else {
3324                                                 for ($i = $toclevel; $i > 0; $i--) {
3325                                                         if ( $levelCount[$i] == $level ) {
3326                                                                 # Found last matching level
3327                                                                 $toclevel = $i;
3328                                                                 break;
3329                                                         }
3330                                                         elseif ( $levelCount[$i] < $level ) {
3331                                                                 # Found first matching level below current level
3332                                                                 $toclevel = $i + 1;
3333                                                                 break;
3334                                                         }
3335                                                 }
3336                                         }
3337                                         if( $toclevel<$wgMaxTocLevel ) {
3338                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3339                                         }
3340                                 }
3341                                 else {
3342                                         # No change in level, end TOC line
3343                                         if( $toclevel<$wgMaxTocLevel ) {
3344                                                 $toc .= $sk->tocLineEnd();
3345                                         }
3346                                 }
3347
3348                                 $levelCount[$toclevel] = $level;
3349
3350                                 # count number of headlines for each level
3351                                 @$sublevelCount[$toclevel]++;
3352                                 $dot = 0;
3353                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3354                                         if( !empty( $sublevelCount[$i] ) ) {
3355                                                 if( $dot ) {
3356                                                         $numbering .= '.';
3357                                                 }
3358                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3359                                                 $dot = 1;
3360                                         }
3361                                 }
3362                         }
3363
3364                         # The canonized header is a version of the header text safe to use for links
3365                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3366                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
3367                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
3368
3369                         # Remove link placeholders by the link text.
3370                         #     <!--LINK number-->
3371                         # turns into
3372                         #     link text with suffix
3373                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3374                                                             "\$this->mLinkHolders['texts'][\$1]",
3375                                                             $canonized_headline );
3376                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3377                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3378                                                             $canonized_headline );
3379
3380                         # strip out HTML
3381                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3382                         $tocline = trim( $canonized_headline );
3383                         # Save headline for section edit hint before it's escaped
3384                         $headline_hint = trim( $canonized_headline );
3385                         $canonized_headline = Sanitizer::escapeId( $tocline );
3386                         $refers[$headlineCount] = $canonized_headline;
3387
3388                         # count how many in assoc. array so we can track dupes in anchors
3389                         @$refers[$canonized_headline]++;
3390                         $refcount[$headlineCount]=$refers[$canonized_headline];
3391
3392                         # Don't number the heading if it is the only one (looks silly)
3393                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3394                                 # the two are different if the line contains a link
3395                                 $headline=$numbering . ' ' . $headline;
3396                         }
3397
3398                         # Create the anchor for linking from the TOC to the section
3399                         $anchor = $canonized_headline;
3400                         if($refcount[$headlineCount] > 1 ) {
3401                                 $anchor .= '_' . $refcount[$headlineCount];
3402                         }
3403                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3404                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3405                         }
3406                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3407                                 if ( empty( $head[$headlineCount] ) ) {
3408                                         $head[$headlineCount] = '';
3409                                 }
3410                                 if( $istemplate )
3411                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
3412                                 else
3413                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3414                         }
3415
3416                         # give headline the correct <h#> tag
3417                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
3418
3419                         $headlineCount++;
3420                         if( !$istemplate )
3421                                 $sectionCount++;
3422                 }
3423
3424                 if( $enoughToc ) {
3425                         if( $toclevel<$wgMaxTocLevel ) {
3426                                 $toc .= $sk->tocUnindent( $toclevel - 1 );
3427                         }
3428                         $toc = $sk->tocList( $toc );
3429                 }
3430
3431                 # split up and insert constructed headlines
3432
3433                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3434                 $i = 0;
3435
3436                 foreach( $blocks as $block ) {
3437                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3438                                 # This is the [edit] link that appears for the top block of text when
3439                                 # section editing is enabled
3440
3441                                 # Disabled because it broke block formatting
3442                                 # For example, a bullet point in the top line
3443                                 # $full .= $sk->editSectionLink(0);
3444                         }
3445                         $full .= $block;
3446                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3447                                 # Top anchor now in skin
3448                                 $full = $full.$toc;
3449                         }
3450
3451                         if( !empty( $head[$i] ) ) {
3452                                 $full .= $head[$i];
3453                         }
3454                         $i++;
3455                 }
3456                 if( $this->mForceTocPosition ) {
3457                         return str_replace( '<!--MWTOC-->', $toc, $full );
3458                 } else {
3459                         return $full;
3460                 }
3461         }
3462
3463         /**
3464          * Transform wiki markup when saving a page by doing \r\n -> \n
3465          * conversion, substitting signatures, {{subst:}} templates, etc.
3466          *
3467          * @param string $text the text to transform
3468          * @param Title &$title the Title object for the current article
3469          * @param User &$user the User object describing the current user
3470          * @param ParserOptions $options parsing options
3471          * @param bool $clearState whether to clear the parser state first
3472          * @return string the altered wiki markup
3473          * @public
3474          */
3475         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
3476                 $this->mOptions = $options;
3477                 $this->mTitle =& $title;
3478                 $this->mOutputType = OT_WIKI;
3479
3480                 if ( $clearState ) {
3481                         $this->clearState();
3482                 }
3483
3484                 $stripState = false;
3485                 $pairs = array(
3486                         "\r\n" => "\n",
3487                 );
3488                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3489                 $text = $this->strip( $text, $stripState, true, array( 'gallery' ) );
3490                 $text = $this->pstPass2( $text, $stripState, $user );
3491                 $text = $this->unstrip( $text, $stripState );
3492                 $text = $this->unstripNoWiki( $text, $stripState );
3493                 return $text;
3494         }
3495
3496         /**
3497          * Pre-save transform helper function
3498          * @private
3499          */
3500         function pstPass2( $text, &$stripState, &$user ) {
3501                 global $wgContLang, $wgLocaltimezone;
3502
3503                 /* Note: This is the timestamp saved as hardcoded wikitext to
3504                  * the database, we use $wgContLang here in order to give
3505                  * everyone the same signature and use the default one rather
3506                  * than the one selected in each user's preferences.
3507                  */
3508                 if ( isset( $wgLocaltimezone ) ) {
3509                         $oldtz = getenv( 'TZ' );
3510                         putenv( 'TZ='.$wgLocaltimezone );
3511                 }
3512                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3513                   ' (' . date( 'T' ) . ')';
3514                 if ( isset( $wgLocaltimezone ) ) {
3515                         putenv( 'TZ='.$oldtz );
3516                 }
3517
3518                 # Variable replacement
3519                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3520                 $text = $this->replaceVariables( $text );
3521
3522                 # Strip out <nowiki> etc. added via replaceVariables
3523                 $text = $this->strip( $text, $stripState, false, array( 'gallery' ) );
3524
3525                 # Signatures
3526                 $sigText = $this->getUserSig( $user );
3527                 $text = strtr( $text, array(
3528                         '~~~~~' => $d,
3529                         '~~~~' => "$sigText $d",
3530                         '~~~' => $sigText
3531                 ) );
3532
3533                 # Context links: [[|name]] and [[name (context)|]]
3534                 #
3535                 global $wgLegalTitleChars;
3536                 $tc = "[$wgLegalTitleChars]";
3537
3538                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3539                 $conpat = "/^{$tc}+?( \\({$tc}+\\)|)$/";
3540
3541                 $p1 = "/\[\[(:?$namespacechar+:|:|)({$tc}+?)( \\({$tc}+\\)|)\\|]]/";    # [[ns:page (context)|]]
3542                 $p2 = "/\[\[\\|({$tc}+)]]/";                                            # [[|page]]
3543
3544                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3545
3546                 $t = $this->mTitle->getText();
3547                 if ( preg_match( $conpat, $t, $m ) && '' != $m[1] ) {
3548                         $text = preg_replace( $p2, "[[\\1{$m[1]}|\\1]]", $text );
3549                 } else {
3550                         # if $m[1] is empty, don't bother duplicating the title
3551                         $text = preg_replace( $p2, '[[\\1]]', $text );
3552                 }
3553
3554                 # Trim trailing whitespace
3555                 # __END__ tag allows for trailing
3556                 # whitespace to be deliberately included
3557                 $text = rtrim( $text );
3558                 $mw =& MagicWord::get( 'end' );
3559                 $mw->matchAndRemove( $text );
3560
3561                 return $text;
3562         }
3563
3564         /**
3565          * Fetch the user's signature text, if any, and normalize to
3566          * validated, ready-to-insert wikitext.
3567          *
3568          * @param User $user
3569          * @return string
3570          * @private
3571          */
3572         function getUserSig( &$user ) {
3573                 $username = $user->getName();
3574                 $nickname = $user->getOption( 'nickname' );
3575                 $nickname = $nickname === '' ? $username : $nickname;
3576
3577                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3578                         # Sig. might contain markup; validate this
3579                         if( $this->validateSig( $nickname ) !== false ) {
3580                                 # Validated; clean up (if needed) and return it
3581                                 return $this->cleanSig( $nickname, true );
3582                         } else {
3583                                 # Failed to validate; fall back to the default
3584                                 $nickname = $username;
3585                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3586                         }
3587                 }
3588
3589                 // Make sure nickname doesnt get a sig in a sig
3590                 $nickname = $this->cleanSigInSig( $nickname );
3591
3592                 # If we're still here, make it a link to the user page
3593                 $userpage = $user->getUserPage();
3594                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3595         }
3596
3597         /**
3598          * Check that the user's signature contains no bad XML
3599          *
3600          * @param string $text
3601          * @return mixed An expanded string, or false if invalid.
3602          */
3603         function validateSig( $text ) {
3604                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3605         }
3606
3607         /**
3608          * Clean up signature text
3609          *
3610          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3611          * 2) Substitute all transclusions
3612          *
3613          * @param string $text
3614          * @param $parsing Whether we're cleaning (preferences save) or parsing
3615          * @return string Signature text
3616          */
3617         function cleanSig( $text, $parsing = false ) {
3618                 global $wgTitle;
3619                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
3620
3621                 $substWord = MagicWord::get( 'subst' );
3622                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3623                 $substText = '{{' . $substWord->getSynonym( 0 );
3624
3625                 $text = preg_replace( $substRegex, $substText, $text );
3626                 $text = $this->cleanSigInSig( $text );
3627                 $text = $this->replaceVariables( $text );
3628
3629                 $this->clearState();
3630                 return $text;
3631         }
3632
3633         /**
3634          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3635          * @param string $text
3636          * @return string Signature text with /~{3,5}/ removed
3637          */
3638         function cleanSigInSig( $text ) {
3639                 $text = preg_replace( '/~{3,5}/', '', $text );
3640                 return $text;
3641         }
3642
3643         /**
3644          * Set up some variables which are usually set up in parse()
3645          * so that an external function can call some class members with confidence
3646          * @public
3647          */
3648         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3649                 $this->mTitle =& $title;
3650                 $this->mOptions = $options;
3651                 $this->mOutputType = $outputType;
3652                 if ( $clearState ) {
3653                         $this->clearState();
3654                 }
3655         }
3656
3657         /**
3658          * Transform a MediaWiki message by replacing magic variables.
3659          *
3660          * @param string $text the text to transform
3661          * @param ParserOptions $options  options
3662          * @return string the text with variables substituted
3663          * @public
3664          */
3665         function transformMsg( $text, $options ) {
3666                 global $wgTitle;
3667                 static $executing = false;
3668
3669                 $fname = "Parser::transformMsg";
3670
3671                 # Guard against infinite recursion
3672                 if ( $executing ) {
3673                         return $text;
3674                 }
3675                 $executing = true;
3676
3677                 wfProfileIn($fname);
3678
3679                 $this->mTitle = $wgTitle;
3680                 $this->mOptions = $options;
3681                 $this->mOutputType = OT_MSG;
3682                 $this->clearState();
3683                 $text = $this->replaceVariables( $text );
3684
3685                 $executing = false;
3686                 wfProfileOut($fname);
3687                 return $text;
3688         }
3689
3690         /**
3691          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3692          * The callback should have the following form:
3693          *    function myParserHook( $text, $params, &$parser ) { ... }
3694          *
3695          * Transform and return $text. Use $parser for any required context, e.g. use
3696          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3697          *
3698          * @public
3699          *
3700          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3701          * @param mixed $callback The callback function (and object) to use for the tag
3702          *
3703          * @return The old value of the mTagHooks array associated with the hook
3704          */
3705         function setHook( $tag, $callback ) {
3706                 $tag = strtolower( $tag );
3707                 $oldVal = @$this->mTagHooks[$tag];
3708                 $this->mTagHooks[$tag] = $callback;
3709
3710                 return $oldVal;
3711         }
3712
3713         /**
3714          * Create a function, e.g. {{sum:1|2|3}}
3715          * The callback function should have the form:
3716          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3717          *
3718          * The callback may either return the text result of the function, or an array with the text
3719          * in element 0, and a number of flags in the other elements. The names of the flags are
3720          * specified in the keys. Valid flags are:
3721          *   found                     The text returned is valid, stop processing the template. This
3722          *                             is on by default.
3723          *   nowiki                    Wiki markup in the return value should be escaped
3724          *   noparse                   Unsafe HTML tags should not be stripped, etc.
3725          *   noargs                    Don't replace triple-brace arguments in the return value
3726          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3727          *
3728          * @public
3729          *
3730          * @param string $id The magic word ID
3731          * @param mixed $callback The callback function (and object) to use
3732          * @param integer $flags a combination of the following flags:
3733          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
3734          *
3735          * @return The old callback function for this name, if any
3736          */
3737         function setFunctionHook( $id, $callback, $flags = 0 ) {
3738                 $oldVal = @$this->mFunctionHooks[$id];
3739                 $this->mFunctionHooks[$id] = $callback;
3740
3741                 # Add to function cache
3742                 $mw = MagicWord::get( $id );
3743                 if ( !$mw ) {
3744                         throw new MWException( 'The calling convention to Parser::setFunctionHook() has changed, ' .
3745                                 'it is now required to pass a MagicWord ID as the first parameter.' );
3746                 }
3747
3748                 $synonyms = $mw->getSynonyms();
3749                 $sensitive = intval( $mw->isCaseSensitive() );
3750
3751                 foreach ( $synonyms as $syn ) {
3752                         # Case
3753                         if ( !$sensitive ) {
3754                                 $syn = strtolower( $syn );
3755                         }
3756                         # Add leading hash
3757                         if ( !( $flags & SFH_NO_HASH ) ) {
3758                                 $syn = '#' . $syn;
3759                         }
3760                         # Remove trailing colon
3761                         if ( substr( $syn, -1, 1 ) == ':' ) {
3762                                 $syn = substr( $syn, 0, -1 );
3763                         }
3764                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
3765                 }
3766                 return $oldVal;
3767         }
3768
3769         /**
3770          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3771          * Placeholders created in Skin::makeLinkObj()
3772          * Returns an array of links found, indexed by PDBK:
3773          *  0 - broken
3774          *  1 - normal link
3775          *  2 - stub
3776          * $options is a bit field, RLH_FOR_UPDATE to select for update
3777          */
3778         function replaceLinkHolders( &$text, $options = 0 ) {
3779                 global $wgUser;
3780                 global $wgOutputReplace;
3781
3782                 $fname = 'Parser::replaceLinkHolders';
3783                 wfProfileIn( $fname );
3784
3785                 $pdbks = array();
3786                 $colours = array();
3787                 $sk =& $this->mOptions->getSkin();
3788                 $linkCache =& LinkCache::singleton();
3789
3790                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3791                         wfProfileIn( $fname.'-check' );
3792                         $dbr =& wfGetDB( DB_SLAVE );
3793                         $page = $dbr->tableName( 'page' );
3794                         $threshold = $wgUser->getOption('stubthreshold');
3795
3796                         # Sort by namespace
3797                         asort( $this->mLinkHolders['namespaces'] );
3798
3799                         # Generate query
3800                         $query = false;
3801                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3802                                 # Make title object
3803                                 $title = $this->mLinkHolders['titles'][$key];
3804
3805                                 # Skip invalid entries.
3806                                 # Result will be ugly, but prevents crash.
3807                                 if ( is_null( $title ) ) {
3808                                         continue;
3809                                 }
3810                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3811
3812                                 # Check if it's a static known link, e.g. interwiki
3813                                 if ( $title->isAlwaysKnown() ) {
3814                                         $colours[$pdbk] = 1;
3815                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3816                                         $colours[$pdbk] = 1;
3817                                         $this->mOutput->addLink( $title, $id );
3818                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3819                                         $colours[$pdbk] = 0;
3820                                 } else {
3821                                         # Not in the link cache, add it to the query
3822                                         if ( !isset( $current ) ) {
3823                                                 $current = $ns;
3824                                                 $query =  "SELECT page_id, page_namespace, page_title";
3825                                                 if ( $threshold > 0 ) {
3826                                                         $query .= ', page_len, page_is_redirect';
3827                                                 }
3828                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
3829                                         } elseif ( $current != $ns ) {
3830                                                 $current = $ns;
3831                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
3832                                         } else {
3833                                                 $query .= ', ';
3834                                         }
3835
3836                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3837                                 }
3838                         }
3839                         if ( $query ) {
3840                                 $query .= '))';
3841                                 if ( $options & RLH_FOR_UPDATE ) {
3842                                         $query .= ' FOR UPDATE';
3843                                 }
3844
3845                                 $res = $dbr->query( $query, $fname );
3846
3847                                 # Fetch data and form into an associative array
3848                                 # non-existent = broken
3849                                 # 1 = known
3850                                 # 2 = stub
3851                                 while ( $s = $dbr->fetchObject($res) ) {
3852                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
3853                                         $pdbk = $title->getPrefixedDBkey();
3854                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
3855                                         $this->mOutput->addLink( $title, $s->page_id );
3856
3857                                         if ( $threshold >  0 ) {
3858                                                 $size = $s->page_len;
3859                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
3860                                                         $colours[$pdbk] = 1;
3861                                                 } else {
3862                                                         $colours[$pdbk] = 2;
3863                                                 }
3864                                         } else {
3865                                                 $colours[$pdbk] = 1;
3866                                         }
3867                                 }
3868                         }
3869                         wfProfileOut( $fname.'-check' );
3870
3871                         # Construct search and replace arrays
3872                         wfProfileIn( $fname.'-construct' );
3873                         $wgOutputReplace = array();
3874                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3875                                 $pdbk = $pdbks[$key];
3876                                 $searchkey = "<!--LINK $key-->";
3877                                 $title = $this->mLinkHolders['titles'][$key];
3878                                 if ( empty( $colours[$pdbk] ) ) {
3879                                         $linkCache->addBadLinkObj( $title );
3880                                         $colours[$pdbk] = 0;
3881                                         $this->mOutput->addLink( $title, 0 );
3882                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
3883                                                                         $this->mLinkHolders['texts'][$key],
3884                                                                         $this->mLinkHolders['queries'][$key] );
3885                                 } elseif ( $colours[$pdbk] == 1 ) {
3886                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
3887                                                                         $this->mLinkHolders['texts'][$key],
3888                                                                         $this->mLinkHolders['queries'][$key] );
3889                                 } elseif ( $colours[$pdbk] == 2 ) {
3890                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
3891                                                                         $this->mLinkHolders['texts'][$key],
3892                                                                         $this->mLinkHolders['queries'][$key] );
3893                                 }
3894                         }
3895                         wfProfileOut( $fname.'-construct' );
3896
3897                         # Do the thing
3898                         wfProfileIn( $fname.'-replace' );
3899
3900                         $text = preg_replace_callback(
3901                                 '/(<!--LINK .*?-->)/',
3902                                 "wfOutputReplaceMatches",
3903                                 $text);
3904
3905                         wfProfileOut( $fname.'-replace' );
3906                 }
3907
3908                 # Now process interwiki link holders
3909                 # This is quite a bit simpler than internal links
3910                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
3911                         wfProfileIn( $fname.'-interwiki' );
3912                         # Make interwiki link HTML
3913                         $wgOutputReplace = array();
3914                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
3915                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
3916                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
3917                         }
3918
3919                         $text = preg_replace_callback(
3920                                 '/<!--IWLINK (.*?)-->/',
3921                                 "wfOutputReplaceMatches",
3922                                 $text );
3923                         wfProfileOut( $fname.'-interwiki' );
3924                 }
3925
3926                 wfProfileOut( $fname );
3927                 return $colours;
3928         }
3929
3930         /**
3931          * Replace <!--LINK--> link placeholders with plain text of links
3932          * (not HTML-formatted).
3933          * @param string $text
3934          * @return string
3935          */
3936         function replaceLinkHoldersText( $text ) {
3937                 $fname = 'Parser::replaceLinkHoldersText';
3938                 wfProfileIn( $fname );
3939
3940                 $text = preg_replace_callback(
3941                         '/<!--(LINK|IWLINK) (.*?)-->/',
3942                         array( &$this, 'replaceLinkHoldersTextCallback' ),
3943                         $text );
3944
3945                 wfProfileOut( $fname );
3946                 return $text;
3947         }
3948
3949         /**
3950          * @param array $matches
3951          * @return string
3952          * @private
3953          */
3954         function replaceLinkHoldersTextCallback( $matches ) {
3955                 $type = $matches[1];
3956                 $key  = $matches[2];
3957                 if( $type == 'LINK' ) {
3958                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
3959                                 return $this->mLinkHolders['texts'][$key];
3960                         }
3961                 } elseif( $type == 'IWLINK' ) {
3962                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
3963                                 return $this->mInterwikiLinkHolders['texts'][$key];
3964                         }
3965                 }
3966                 return $matches[0];
3967         }
3968
3969         /**
3970          * Tag hook handler for 'pre'.
3971          */
3972         function renderPreTag( $text, $attribs, $parser ) {
3973                 // Backwards-compatibility hack
3974                 $content = preg_replace( '!<nowiki>(.*?)</nowiki>!is', '\\1', $text );
3975
3976                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
3977                 return wfOpenElement( 'pre', $attribs ) .
3978                         wfEscapeHTMLTagsOnly( $content ) .
3979                         '</pre>';
3980         }
3981
3982         /**
3983          * Renders an image gallery from a text with one line per image.
3984          * text labels may be given by using |-style alternative text. E.g.
3985          *   Image:one.jpg|The number "1"
3986          *   Image:tree.jpg|A tree
3987          * given as text will return the HTML of a gallery with two images,
3988          * labeled 'The number "1"' and
3989          * 'A tree'.
3990          */
3991         function renderImageGallery( $text, $params ) {
3992                 $ig = new ImageGallery();
3993                 $ig->setShowBytes( false );
3994                 $ig->setShowFilename( false );
3995                 $ig->setParsing();
3996                 $ig->useSkin( $this->mOptions->getSkin() );
3997
3998                 if( isset( $params['caption'] ) )
3999                         $ig->setCaption( $params['caption'] );
4000
4001                 $lines = explode( "\n", $text );
4002                 foreach ( $lines as $line ) {
4003                         # match lines like these:
4004                         # Image:someimage.jpg|This is some image
4005                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4006                         # Skip empty lines
4007                         if ( count( $matches ) == 0 ) {
4008                                 continue;
4009                         }
4010                         $tp = Title::newFromText( $matches[1] );
4011                         $nt =& $tp;
4012                         if( is_null( $nt ) ) {
4013                                 # Bogus title. Ignore these so we don't bomb out later.
4014                                 continue;
4015                         }
4016                         if ( isset( $matches[3] ) ) {
4017                                 $label = $matches[3];
4018                         } else {
4019                                 $label = '';
4020                         }
4021
4022                         $pout = $this->parse( $label,
4023                                 $this->mTitle,
4024                                 $this->mOptions,
4025                                 false, // Strip whitespace...?
4026                                 false  // Don't clear state!
4027                         );
4028                         $html = $pout->getText();
4029
4030                         $ig->add( new Image( $nt ), $html );
4031
4032                         # Only add real images (bug #5586)
4033                         if ( $nt->getNamespace() == NS_IMAGE ) {
4034                                 $this->mOutput->addImage( $nt->getDBkey() );
4035                         }
4036                 }
4037                 return $ig->toHTML();
4038         }
4039
4040         /**
4041          * Parse image options text and use it to make an image
4042          */
4043         function makeImage( &$nt, $options ) {
4044                 global $wgUseImageResize;
4045
4046                 $align = '';
4047
4048                 # Check if the options text is of the form "options|alt text"
4049                 # Options are:
4050                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4051                 #  * left               no resizing, just left align. label is used for alt= only
4052                 #  * right              same, but right aligned
4053                 #  * none               same, but not aligned
4054                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4055                 #  * center             center the image
4056                 #  * framed             Keep original image size, no magnify-button.
4057
4058                 $part = explode( '|', $options);
4059
4060                 $mwThumb  =& MagicWord::get( 'img_thumbnail' );
4061                 $mwManualThumb =& MagicWord::get( 'img_manualthumb' );
4062                 $mwLeft   =& MagicWord::get( 'img_left' );
4063                 $mwRight  =& MagicWord::get( 'img_right' );
4064                 $mwNone   =& MagicWord::get( 'img_none' );
4065                 $mwWidth  =& MagicWord::get( 'img_width' );
4066                 $mwCenter =& MagicWord::get( 'img_center' );
4067                 $mwFramed =& MagicWord::get( 'img_framed' );
4068                 $caption = '';
4069
4070                 $width = $height = $framed = $thumb = false;
4071                 $manual_thumb = '' ;
4072
4073                 foreach( $part as $key => $val ) {
4074                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
4075                                 $thumb=true;
4076                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
4077                                 # use manually specified thumbnail
4078                                 $thumb=true;
4079                                 $manual_thumb = $match;
4080                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
4081                                 # remember to set an alignment, don't render immediately
4082                                 $align = 'right';
4083                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
4084                                 # remember to set an alignment, don't render immediately
4085                                 $align = 'left';
4086                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
4087                                 # remember to set an alignment, don't render immediately
4088                                 $align = 'center';
4089                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
4090                                 # remember to set an alignment, don't render immediately
4091                                 $align = 'none';
4092                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
4093                                 wfDebug( "img_width match: $match\n" );
4094                                 # $match is the image width in pixels
4095                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
4096                                         $width = intval( $m[1] );
4097                                         $height = intval( $m[2] );
4098                                 } else {
4099                                         $width = intval($match);
4100                                 }
4101                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
4102                                 $framed=true;
4103                         } else {
4104                                 $caption = $val;
4105                         }
4106                 }
4107                 # Strip bad stuff out of the alt text
4108                 $alt = $this->replaceLinkHoldersText( $caption );
4109
4110                 # make sure there are no placeholders in thumbnail attributes
4111                 # that are later expanded to html- so expand them now and
4112                 # remove the tags
4113                 $alt = $this->unstrip($alt, $this->mStripState);
4114                 $alt = Sanitizer::stripAllTags( $alt );
4115
4116                 # Linker does the rest
4117                 $sk =& $this->mOptions->getSkin();
4118                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
4119         }
4120
4121         /**
4122          * Set a flag in the output object indicating that the content is dynamic and
4123          * shouldn't be cached.
4124          */
4125         function disableCache() {
4126                 wfDebug( "Parser output marked as uncacheable.\n" );
4127                 $this->mOutput->mCacheTime = -1;
4128         }
4129
4130         /**#@+
4131          * Callback from the Sanitizer for expanding items found in HTML attribute
4132          * values, so they can be safely tested and escaped.
4133          * @param string $text
4134          * @param array $args
4135          * @return string
4136          * @private
4137          */
4138         function attributeStripCallback( &$text, $args ) {
4139                 $text = $this->replaceVariables( $text, $args );
4140                 $text = $this->unstripForHTML( $text );
4141                 return $text;
4142         }
4143
4144         function unstripForHTML( $text ) {
4145                 $text = $this->unstrip( $text, $this->mStripState );
4146                 $text = $this->unstripNoWiki( $text, $this->mStripState );
4147                 return $text;
4148         }
4149         /**#@-*/
4150
4151         /**#@+
4152          * Accessor/mutator
4153          */
4154         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4155         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4156         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4157         /**#@-*/
4158
4159         /**#@+
4160          * Accessor
4161          */
4162         function getTags() { return array_keys( $this->mTagHooks ); }
4163         /**#@-*/
4164
4165
4166         /**
4167          * Break wikitext input into sections, and either pull or replace
4168          * some particular section's text.
4169          *
4170          * External callers should use the getSection and replaceSection methods.
4171          *
4172          * @param $text Page wikitext
4173          * @param $section Numbered section. 0 pulls the text before the first
4174          *                 heading; other numbers will pull the given section
4175          *                 along with its lower-level subsections.
4176          * @param $mode One of "get" or "replace"
4177          * @param $newtext Replacement text for section data.
4178          * @return string for "get", the extracted section text.
4179          *                for "replace", the whole page with the section replaced.
4180          */
4181         private function extractSections( $text, $section, $mode, $newtext='' ) {
4182                 # strip NOWIKI etc. to avoid confusion (true-parameter causes HTML
4183                 # comments to be stripped as well)
4184                 $striparray = array();
4185
4186                 $oldOutputType = $this->mOutputType;
4187                 $oldOptions = $this->mOptions;
4188                 $this->mOptions = new ParserOptions();
4189                 $this->mOutputType = OT_WIKI;
4190
4191                 $striptext = $this->strip( $text, $striparray, true );
4192
4193                 $this->mOutputType = $oldOutputType;
4194                 $this->mOptions = $oldOptions;
4195
4196                 # now that we can be sure that no pseudo-sections are in the source,
4197                 # split it up by section
4198                 $uniq = preg_quote( $this->uniqPrefix(), '/' );
4199                 $comment = "(?:$uniq-!--.*?QINU)";
4200                 $secs = preg_split(
4201                 /*
4202                         "/
4203                         ^(
4204                         (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4205                         (?:
4206                                 (=+) # Should this be limited to 6?
4207                                 .+?  # Section title...
4208                                 \\2  # Ending = count must match start
4209                         |
4210                                 ^
4211                                 <h([1-6])\b.*?>
4212                                 .*?
4213                                 <\/h\\3\s*>
4214                         )
4215                         (?:$comment|<\/?noinclude>|\s+)* # Trailing whitespace ok
4216                         )$
4217                         /mix",
4218                 */
4219                         "/
4220                         (
4221                                 ^
4222                                 (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4223                                 (=+) # Should this be limited to 6?
4224                                 .+?  # Section title...
4225                                 \\2  # Ending = count must match start
4226                                 (?:$comment|<\/?noinclude>|[ \\t]+)* # Trailing whitespace ok
4227                                 $
4228                         |
4229                                 <h([1-6])\b.*?>
4230                                 .*?
4231                                 <\/h\\3\s*>
4232                         )
4233                         /mix",
4234                         $striptext, -1,
4235                         PREG_SPLIT_DELIM_CAPTURE);
4236
4237                 if( $mode == "get" ) {
4238                         if( $section == 0 ) {
4239                                 // "Section 0" returns the content before any other section.
4240                                 $rv = $secs[0];
4241                         } else {
4242                                 $rv = "";
4243                         }
4244                 } elseif( $mode == "replace" ) {
4245                         if( $section == 0 ) {
4246                                 $rv = $newtext . "\n\n";
4247                                 $remainder = true;
4248                         } else {
4249                                 $rv = $secs[0];
4250                                 $remainder = false;
4251                         }
4252                 }
4253                 $count = 0;
4254                 $sectionLevel = 0;
4255                 for( $index = 1; $index < count( $secs ); ) {
4256                         $headerLine = $secs[$index++];
4257                         if( $secs[$index] ) {
4258                                 // A wiki header
4259                                 $headerLevel = strlen( $secs[$index++] );
4260                         } else {
4261                                 // An HTML header
4262                                 $index++;
4263                                 $headerLevel = intval( $secs[$index++] );
4264                         }
4265                         $content = $secs[$index++];
4266
4267                         $count++;
4268                         if( $mode == "get" ) {
4269                                 if( $count == $section ) {
4270                                         $rv = $headerLine . $content;
4271                                         $sectionLevel = $headerLevel;
4272                                 } elseif( $count > $section ) {
4273                                         if( $sectionLevel && $headerLevel > $sectionLevel ) {
4274                                                 $rv .= $headerLine . $content;
4275                                         } else {
4276                                                 // Broke out to a higher-level section
4277                                                 break;
4278                                         }
4279                                 }
4280                         } elseif( $mode == "replace" ) {
4281                                 if( $count < $section ) {
4282                                         $rv .= $headerLine . $content;
4283                                 } elseif( $count == $section ) {
4284                                         $rv .= $newtext . "\n\n";
4285                                         $sectionLevel = $headerLevel;
4286                                 } elseif( $count > $section ) {
4287                                         if( $headerLevel <= $sectionLevel ) {
4288                                                 // Passed the section's sub-parts.
4289                                                 $remainder = true;
4290                                         }
4291                                         if( $remainder ) {
4292                                                 $rv .= $headerLine . $content;
4293                                         }
4294                                 }
4295                         }
4296                 }
4297                 # reinsert stripped tags
4298                 $rv = $this->unstrip( $rv, $striparray );
4299                 $rv = $this->unstripNoWiki( $rv, $striparray );
4300                 $rv = trim( $rv );
4301                 return $rv;
4302         }
4303
4304         /**
4305          * This function returns the text of a section, specified by a number ($section).
4306          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4307          * the first section before any such heading (section 0).
4308          *
4309          * If a section contains subsections, these are also returned.
4310          *
4311          * @param $text String: text to look in
4312          * @param $section Integer: section number
4313          * @return string text of the requested section
4314          */
4315         function getSection( $text, $section ) {
4316                 return $this->extractSections( $text, $section, "get" );
4317         }
4318
4319         function replaceSection( $oldtext, $section, $text ) {
4320                 return $this->extractSections( $oldtext, $section, "replace", $text );
4321         }
4322
4323 }
4324
4325 /**
4326  * @todo document
4327  * @package MediaWiki
4328  */
4329 class ParserOutput
4330 {
4331         var $mText,             # The output text
4332                 $mLanguageLinks,    # List of the full text of language links, in the order they appear
4333                 $mCategories,       # Map of category names to sort keys
4334                 $mContainsOldMagic, # Boolean variable indicating if the input contained variables like {{CURRENTDAY}}
4335                 $mCacheTime,        # Time when this object was generated, or -1 for uncacheable. Used in ParserCache.
4336                 $mVersion,          # Compatibility check
4337                 $mTitleText,        # title text of the chosen language variant
4338                 $mLinks,            # 2-D map of NS/DBK to ID for the links in the document. ID=zero for broken.
4339                 $mTemplates,        # 2-D map of NS/DBK to ID for the template references. ID=zero for broken.
4340                 $mImages,           # DB keys of the images used, in the array key only
4341                 $mExternalLinks,    # External link URLs, in the key only
4342                 $mHTMLtitle,            # Display HTML title
4343                 $mSubtitle,                     # Additional subtitle
4344                 $mNewSection,           # Show a new section link?
4345                 $mNoGallery;            # No gallery on category page? (__NOGALLERY__)
4346
4347         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
4348                 $containsOldMagic = false, $titletext = '' )
4349         {
4350                 $this->mText = $text;
4351                 $this->mLanguageLinks = $languageLinks;
4352                 $this->mCategories = $categoryLinks;
4353                 $this->mContainsOldMagic = $containsOldMagic;
4354                 $this->mCacheTime = '';
4355                 $this->mVersion = MW_PARSER_VERSION;
4356                 $this->mTitleText = $titletext;
4357                 $this->mLinks = array();
4358                 $this->mTemplates = array();
4359                 $this->mImages = array();
4360                 $this->mExternalLinks = array();
4361                 $this->mHTMLtitle = "" ;
4362                 $this->mSubtitle = "" ;
4363                 $this->mNewSection = false;
4364                 $this->mNoGallery = false;
4365         }
4366
4367         function getText()                   { return $this->mText; }
4368         function &getLanguageLinks()          { return $this->mLanguageLinks; }
4369         function getCategoryLinks()          { return array_keys( $this->mCategories ); }
4370         function &getCategories()            { return $this->mCategories; }
4371         function getCacheTime()              { return $this->mCacheTime; }
4372         function getTitleText()              { return $this->mTitleText; }
4373         function &getLinks()                 { return $this->mLinks; }
4374         function &getTemplates()             { return $this->mTemplates; }
4375         function &getImages()                { return $this->mImages; }
4376         function &getExternalLinks()         { return $this->mExternalLinks; }
4377         function getNoGallery()              { return $this->mNoGallery; }
4378         function getSubtitle()               { return $this->mSubtitle; }
4379
4380         function containsOldMagic()          { return $this->mContainsOldMagic; }
4381         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
4382         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
4383         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategories, $cl ); }
4384         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
4385         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
4386         function setTitleText( $t )          { return wfSetVar($this->mTitleText, $t); }
4387         function setSubtitle( $st )          { return wfSetVar( $this->mSubtitle, $st ); }
4388
4389         function addCategory( $c, $sort )    { $this->mCategories[$c] = $sort; }
4390         function addImage( $name )           { $this->mImages[$name] = 1; }
4391         function addLanguageLink( $t )       { $this->mLanguageLinks[] = $t; }
4392         function addExternalLink( $url )     { $this->mExternalLinks[$url] = 1; }
4393
4394         function setNewSection( $value ) {
4395                 $this->mNewSection = (bool)$value;
4396         }
4397         function getNewSection() {
4398                 return (bool)$this->mNewSection;
4399         }
4400
4401         function addLink( $title, $id ) {
4402                 $ns = $title->getNamespace();
4403                 $dbk = $title->getDBkey();
4404                 if ( !isset( $this->mLinks[$ns] ) ) {
4405                         $this->mLinks[$ns] = array();
4406                 }
4407                 $this->mLinks[$ns][$dbk] = $id;
4408         }
4409
4410         function addTemplate( $title, $id ) {
4411                 $ns = $title->getNamespace();
4412                 $dbk = $title->getDBkey();
4413                 if ( !isset( $this->mTemplates[$ns] ) ) {
4414                         $this->mTemplates[$ns] = array();
4415                 }
4416                 $this->mTemplates[$ns][$dbk] = $id;
4417         }
4418
4419         /**
4420          * Return true if this cached output object predates the global or
4421          * per-article cache invalidation timestamps, or if it comes from
4422          * an incompatible older version.
4423          *
4424          * @param string $touched the affected article's last touched timestamp
4425          * @return bool
4426          * @public
4427          */
4428         function expired( $touched ) {
4429                 global $wgCacheEpoch;
4430                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
4431                        $this->getCacheTime() < $touched ||
4432                        $this->getCacheTime() <= $wgCacheEpoch ||
4433                        !isset( $this->mVersion ) ||
4434                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
4435         }
4436 }
4437
4438 /**
4439  * Set options of the Parser
4440  * @todo document
4441  * @package MediaWiki
4442  */
4443 class ParserOptions
4444 {
4445         # All variables are private
4446         var $mUseTeX;                    # Use texvc to expand <math> tags
4447         var $mUseDynamicDates;           # Use DateFormatter to format dates
4448         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
4449         var $mAllowExternalImages;       # Allow external images inline
4450         var $mAllowExternalImagesFrom;   # If not, any exception?
4451         var $mSkin;                      # Reference to the preferred skin
4452         var $mDateFormat;                # Date format index
4453         var $mEditSection;               # Create "edit section" links
4454         var $mNumberHeadings;            # Automatically number headings
4455         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
4456         var $mTidy;                      # Ask for tidy cleanup
4457         var $mInterfaceMessage;          # Which lang to call for PLURAL and GRAMMAR
4458
4459         var $mUser;                      # Stored user object, just used to initialise the skin
4460
4461         function getUseTeX()                        { return $this->mUseTeX; }
4462         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
4463         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
4464         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
4465         function getAllowExternalImagesFrom()       { return $this->mAllowExternalImagesFrom; }
4466         function getEditSection()                   { return $this->mEditSection; }
4467         function getNumberHeadings()                { return $this->mNumberHeadings; }
4468         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
4469         function getTidy()                          { return $this->mTidy; }
4470         function getInterfaceMessage()              { return $this->mInterfaceMessage; }
4471
4472         function &getSkin() {
4473                 if ( !isset( $this->mSkin ) ) {
4474                         $this->mSkin = $this->mUser->getSkin();
4475                 }
4476                 return $this->mSkin;
4477         }
4478
4479         function getDateFormat() {
4480                 if ( !isset( $this->mDateFormat ) ) {
4481                         $this->mDateFormat = $this->mUser->getDatePreference();
4482                 }
4483                 return $this->mDateFormat;
4484         }
4485
4486         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
4487         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
4488         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
4489         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
4490         function setAllowExternalImagesFrom( $x )   { return wfSetVar( $this->mAllowExternalImagesFrom, $x ); }
4491         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
4492         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
4493         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
4494         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
4495         function setTidy( $x )                      { return wfSetVar( $this->mTidy, $x); }
4496         function setSkin( &$x ) { $this->mSkin =& $x; }
4497         function setInterfaceMessage( $x )          { return wfSetVar( $this->mInterfaceMessage, $x); }
4498
4499         function ParserOptions( $user = null ) {
4500                 $this->initialiseFromUser( $user );
4501         }
4502
4503         /**
4504          * Get parser options
4505          * @static
4506          */
4507         static function newFromUser( $user ) {
4508                 return new ParserOptions( $user );
4509         }
4510
4511         /** Get user options */
4512         function initialiseFromUser( $userInput ) {
4513                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages;
4514                 global $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion;
4515                 $fname = 'ParserOptions::initialiseFromUser';
4516                 wfProfileIn( $fname );
4517                 if ( !$userInput ) {
4518                         global $wgUser;
4519                         if ( isset( $wgUser ) ) {
4520                                 $user = $wgUser;
4521                         } else {
4522                                 $user = new User;
4523                                 $user->setLoaded( true );
4524                         }
4525                 } else {
4526                         $user =& $userInput;
4527                 }
4528
4529                 $this->mUser = $user;
4530
4531                 $this->mUseTeX = $wgUseTeX;
4532                 $this->mUseDynamicDates = $wgUseDynamicDates;
4533                 $this->mInterwikiMagic = $wgInterwikiMagic;
4534                 $this->mAllowExternalImages = $wgAllowExternalImages;
4535                 $this->mAllowExternalImagesFrom = $wgAllowExternalImagesFrom;
4536                 $this->mSkin = null; # Deferred
4537                 $this->mDateFormat = null; # Deferred
4538                 $this->mEditSection = true;
4539                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
4540                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
4541                 $this->mTidy = false;
4542                 $this->mInterfaceMessage = false;
4543                 wfProfileOut( $fname );
4544         }
4545 }
4546
4547 /**
4548  * Callback function used by Parser::replaceLinkHolders()
4549  * to substitute link placeholders.
4550  */
4551 function &wfOutputReplaceMatches( $matches ) {
4552         global $wgOutputReplace;
4553         return $wgOutputReplace[$matches[1]];
4554 }
4555
4556 /**
4557  * Return the total number of articles
4558  */
4559 function wfNumberOfArticles() {
4560         global $wgNumberOfArticles;
4561
4562         wfLoadSiteStats();
4563         return $wgNumberOfArticles;
4564 }
4565
4566 /**
4567  * Return the number of files
4568  */
4569 function wfNumberOfFiles() {
4570         $fname = 'wfNumberOfFiles';
4571
4572         wfProfileIn( $fname );
4573         $dbr =& wfGetDB( DB_SLAVE );
4574         $numImages = $dbr->selectField('site_stats', 'ss_images', array(), $fname );
4575         wfProfileOut( $fname );
4576
4577         return $numImages;
4578 }
4579
4580 /**
4581  * Return the number of user accounts
4582  * @return integer
4583  */
4584 function wfNumberOfUsers() {
4585         wfProfileIn( 'wfNumberOfUsers' );
4586         $dbr =& wfGetDB( DB_SLAVE );
4587         $count = $dbr->selectField( 'site_stats', 'ss_users', array(), 'wfNumberOfUsers' );
4588         wfProfileOut( 'wfNumberOfUsers' );
4589         return (int)$count;
4590 }
4591
4592 /**
4593  * Return the total number of pages
4594  * @return integer
4595  */
4596 function wfNumberOfPages() {
4597         wfProfileIn( 'wfNumberOfPages' );
4598         $dbr =& wfGetDB( DB_SLAVE );
4599         $count = $dbr->selectField( 'site_stats', 'ss_total_pages', array(), 'wfNumberOfPages' );
4600         wfProfileOut( 'wfNumberOfPages' );
4601         return (int)$count;
4602 }
4603
4604 /**
4605  * Return the total number of admins
4606  *
4607  * @return integer
4608  */
4609 function wfNumberOfAdmins() {
4610         static $admins = -1;
4611         wfProfileIn( 'wfNumberOfAdmins' );
4612         if( $admins == -1 ) {
4613                 $dbr =& wfGetDB( DB_SLAVE );
4614                 $admins = $dbr->selectField( 'user_groups', 'COUNT(*)', array( 'ug_group' => 'sysop' ), 'wfNumberOfAdmins' );
4615         }
4616         wfProfileOut( 'wfNumberOfAdmins' );
4617         return (int)$admins;
4618 }
4619
4620 /**
4621  * Count the number of pages in a particular namespace
4622  *
4623  * @param $ns Namespace
4624  * @return integer
4625  */
4626 function wfPagesInNs( $ns ) {
4627         static $pageCount = array();
4628         wfProfileIn( 'wfPagesInNs' );
4629         if( !isset( $pageCount[$ns] ) ) {
4630                 $dbr =& wfGetDB( DB_SLAVE );
4631                 $pageCount[$ns] = $dbr->selectField( 'page', 'COUNT(*)', array( 'page_namespace' => $ns ), 'wfPagesInNs' );
4632         }
4633         wfProfileOut( 'wfPagesInNs' );
4634         return (int)$pageCount[$ns];
4635 }
4636
4637 /**
4638  * Get various statistics from the database
4639  * @private
4640  */
4641 function wfLoadSiteStats() {
4642         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
4643         $fname = 'wfLoadSiteStats';
4644
4645         if ( -1 != $wgNumberOfArticles ) return;
4646         $dbr =& wfGetDB( DB_SLAVE );
4647         $s = $dbr->selectRow( 'site_stats',
4648                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
4649                 array( 'ss_row_id' => 1 ), $fname
4650         );
4651
4652         if ( $s === false ) {
4653                 return;
4654         } else {
4655                 $wgTotalViews = $s->ss_total_views;
4656                 $wgTotalEdits = $s->ss_total_edits;
4657                 $wgNumberOfArticles = $s->ss_good_articles;
4658         }
4659 }
4660
4661 /**
4662  * Escape html tags
4663  * Basically replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
4664  *
4665  * @param $in String: text that might contain HTML tags.
4666  * @return string Escaped string
4667  */
4668 function wfEscapeHTMLTagsOnly( $in ) {
4669         return str_replace(
4670                 array( '"', '>', '<' ),
4671                 array( '&quot;', '&gt;', '&lt;' ),
4672                 $in );
4673 }
4674
4675 ?>