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 define( 'RLH_FOR_UPDATE', 1 );
  17
  18 # Allowed values for $mOutputType
  19 define( 'OT_HTML', 1 );
  20 define( 'OT_WIKI', 2 );
  21 define( 'OT_MSG' , 3 );
  22
  23 # Flags for setFunctionHook
  24 define( 'SFH_NO_HASH', 1 );
  25
  26 # string parameter for extractTags which will cause it
  27 # to strip HTML comments in addition to regular
  28 # <XML>-style tags. This should not be anything we
  29 # may want to use in wikisyntax
  30 define( 'STRIP_COMMENTS', 'HTMLCommentStrip' );
  31
  32 # Constants needed for external link processing
  33 define( 'HTTP_PROTOCOLS', 'http:\/\/|https:\/\/' );
  34 # Everything except bracket, space, or control characters
  35 define( 'EXT_LINK_URL_CLASS', '[^][<>"\\x00-\\x20\\x7F]' );
  36 # Including space, but excluding newlines
  37 define( 'EXT_LINK_TEXT_CLASS', '[^\]\\x0a\\x0d]' );
  38 define( 'EXT_IMAGE_FNAME_CLASS', '[A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]' );
  39 define( 'EXT_IMAGE_EXTENSIONS', 'gif|png|jpg|jpeg' );
  40 define( 'EXT_LINK_BRACKETED',  '/\[(\b(' . wfUrlProtocols() . ')'.
  41         EXT_LINK_URL_CLASS.'+) *('.EXT_LINK_TEXT_CLASS.'*?)\]/S' );
  42 define( 'EXT_IMAGE_REGEX',
  43         '/^('.HTTP_PROTOCOLS.')'.  # Protocol
  44         '('.EXT_LINK_URL_CLASS.'+)\\/'.  # Hostname and path
  45         '('.EXT_IMAGE_FNAME_CLASS.'+)\\.((?i)'.EXT_IMAGE_EXTENSIONS.')$/S' # Filename
  46 );
  47
  48 // State constants for the definition list colon extraction
  49 define( 'MW_COLON_STATE_TEXT', 0 );
  50 define( 'MW_COLON_STATE_TAG', 1 );
  51 define( 'MW_COLON_STATE_TAGSTART', 2 );
  52 define( 'MW_COLON_STATE_CLOSETAG', 3 );
  53 define( 'MW_COLON_STATE_TAGSLASH', 4 );
  54 define( 'MW_COLON_STATE_COMMENT', 5 );
  55 define( 'MW_COLON_STATE_COMMENTDASH', 6 );
  56 define( 'MW_COLON_STATE_COMMENTDASHDASH', 7 );
  57
  58 /**
  59  * PHP Parser
  60  *
  61  * Processes wiki markup
  62  *
  63  * <pre>
  64  * There are three main entry points into the Parser class:
  65  * parse()
  66  *   produces HTML output
  67  * preSaveTransform().
  68  *   produces altered wiki markup.
  69  * transformMsg()
  70  *   performs brace substitution on MediaWiki messages
  71  *
  72  * Globals used:
  73  *    objects:   $wgLang, $wgContLang
  74  *
  75  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  76  *
  77  * settings:
  78  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  79  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  80  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  81  *  $wgMaxArticleSize*
  82  *
  83  *  * only within ParserOptions
  84  * </pre>
  85  *
  86  * @package MediaWiki
  87  */
  88 class Parser
  89 {
  90         /**#@+
  91          * @private
  92          */
  93         # Persistent:
  94         var $mTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables;
  95
  96         # Cleared with clearState():
  97         var $mOutput, $mAutonumber, $mDTopen, $mStripState = array();
  98         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
  99         var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
 100         var $mIncludeSizes;
 101         var $mTemplates,        // cache of already loaded templates, avoids
 102                                 // multiple SQL queries for the same string
 103             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 104                                 // in this path. Used for loop detection.
 105
 106         # Temporary
 107         # These are variables reset at least once per parse regardless of $clearState
 108         var $mOptions,      // ParserOptions object
 109                 $mTitle,        // Title context, used for self-link rendering and similar things
 110                 $mOutputType,   // Output type, one of the OT_xxx constants
 111                 $mRevisionId;   // ID to display in {{REVISIONID}} tags
 112
 113         /**#@-*/
 114
 115         /**
 116          * Constructor
 117          *
 118          * @public
 119          */
 120         function Parser() {
 121                 $this->mTagHooks = array();
 122                 $this->mFunctionHooks = array();
 123                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 124                 $this->mFirstCall = true;
 125         }
 126
 127         /**
 128          * Do various kinds of initialisation on the first call of the parser
 129          */
 130         function firstCallInit() {
 131                 if ( !$this->mFirstCall ) {
 132                         return;
 133                 }
 134
 135                 wfProfileIn( __METHOD__ );
 136                 global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
 137
 138                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 139
 140                 $this->setFunctionHook( 'ns', array( 'CoreParserFunctions', 'ns' ), SFH_NO_HASH );
 141                 $this->setFunctionHook( 'urlencode', array( 'CoreParserFunctions', 'urlencode' ), SFH_NO_HASH );
 142                 $this->setFunctionHook( 'lcfirst', array( 'CoreParserFunctions', 'lcfirst' ), SFH_NO_HASH );
 143                 $this->setFunctionHook( 'ucfirst', array( 'CoreParserFunctions', 'ucfirst' ), SFH_NO_HASH );
 144                 $this->setFunctionHook( 'lc', array( 'CoreParserFunctions', 'lc' ), SFH_NO_HASH );
 145                 $this->setFunctionHook( 'uc', array( 'CoreParserFunctions', 'uc' ), SFH_NO_HASH );
 146                 $this->setFunctionHook( 'localurl', array( 'CoreParserFunctions', 'localurl' ), SFH_NO_HASH );
 147                 $this->setFunctionHook( 'localurle', array( 'CoreParserFunctions', 'localurle' ), SFH_NO_HASH );
 148                 $this->setFunctionHook( 'fullurl', array( 'CoreParserFunctions', 'fullurl' ), SFH_NO_HASH );
 149                 $this->setFunctionHook( 'fullurle', array( 'CoreParserFunctions', 'fullurle' ), SFH_NO_HASH );
 150                 $this->setFunctionHook( 'formatnum', array( 'CoreParserFunctions', 'formatnum' ), SFH_NO_HASH );
 151                 $this->setFunctionHook( 'grammar', array( 'CoreParserFunctions', 'grammar' ), SFH_NO_HASH );
 152                 $this->setFunctionHook( 'plural', array( 'CoreParserFunctions', 'plural' ), SFH_NO_HASH );
 153                 $this->setFunctionHook( 'numberofpages', array( 'CoreParserFunctions', 'numberofpages' ), SFH_NO_HASH );
 154                 $this->setFunctionHook( 'numberofusers', array( 'CoreParserFunctions', 'numberofusers' ), SFH_NO_HASH );
 155                 $this->setFunctionHook( 'numberofarticles', array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 156                 $this->setFunctionHook( 'numberoffiles', array( 'CoreParserFunctions', 'numberoffiles' ), SFH_NO_HASH );
 157                 $this->setFunctionHook( 'numberofadmins', array( 'CoreParserFunctions', 'numberofadmins' ), SFH_NO_HASH );
 158                 $this->setFunctionHook( 'language', array( 'CoreParserFunctions', 'language' ), SFH_NO_HASH );
 159
 160                 if ( $wgAllowDisplayTitle ) {
 161                         $this->setFunctionHook( 'displaytitle', array( 'CoreParserFunctions', 'displaytitle' ), SFH_NO_HASH );
 162                 }
 163                 if ( $wgAllowSlowParserFunctions ) {
 164                         $this->setFunctionHook( 'pagesinnamespace', array( 'CoreParserFunctions', 'pagesinnamespace' ), SFH_NO_HASH );
 165                 }
 166
 167                 $this->initialiseVariables();
 168
 169                 $this->mFirstCall = false;
 170                 wfProfileOut( __METHOD__ );
 171         }
 172
 173         /**
 174          * Clear Parser state
 175          *
 176          * @private
 177          */
 178         function clearState() {
 179                 wfProfileIn( __METHOD__ );
 180                 if ( $this->mFirstCall ) {
 181                         $this->firstCallInit();
 182                 }
 183                 $this->mOutput = new ParserOutput;
 184                 $this->mAutonumber = 0;
 185                 $this->mLastSection = '';
 186                 $this->mDTopen = false;
 187                 $this->mIncludeCount = array();
 188                 $this->mStripState = array();
 189                 $this->mArgStack = array();
 190                 $this->mInPre = false;
 191                 $this->mInterwikiLinkHolders = array(
 192                         'texts' => array(),
 193                         'titles' => array()
 194                 );
 195                 $this->mLinkHolders = array(
 196                         'namespaces' => array(),
 197                         'dbkeys' => array(),
 198                         'queries' => array(),
 199                         'texts' => array(),
 200                         'titles' => array()
 201                 );
 202                 $this->mRevisionId = null;
 203
 204                 /**
 205                  * Prefix for temporary replacement strings for the multipass parser.
 206                  * \x07 should never appear in input as it's disallowed in XML.
 207                  * Using it at the front also gives us a little extra robustness
 208                  * since it shouldn't match when butted up against identifier-like
 209                  * string constructs.
 210                  */
 211                 $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 212
 213                 # Clear these on every parse, bug 4549
 214                 $this->mTemplates = array();
 215                 $this->mTemplatePath = array();
 216
 217                 $this->mShowToc = true;
 218                 $this->mForceTocPosition = false;
 219                 $this->mIncludeSizes = array(
 220                         'pre-expand' => 0,
 221                         'post-expand' => 0,
 222                         'arg' => 0
 223                 );
 224
 225                 wfRunHooks( 'ParserClearState', array( &$this ) );
 226                 wfProfileOut( __METHOD__ );
 227         }
 228
 229         /**
 230          * Accessor for mUniqPrefix.
 231          *
 232          * @public
 233          */
 234         function uniqPrefix() {
 235                 return $this->mUniqPrefix;
 236         }
 237
 238         /**
 239          * Convert wikitext to HTML
 240          * Do not call this function recursively.
 241          *
 242          * @private
 243          * @param string $text Text we want to parse
 244          * @param Title &$title A title object
 245          * @param array $options
 246          * @param boolean $linestart
 247          * @param boolean $clearState
 248          * @param int $revid number to pass in {{REVISIONID}}
 249          * @return ParserOutput a ParserOutput
 250          */
 251         function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 252                 /**
 253                  * First pass--just handle <nowiki> sections, pass the rest off
 254                  * to internalParse() which does all the real work.
 255                  */
 256
 257                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 258                 $fname = 'Parser::parse';
 259                 wfProfileIn( $fname );
 260
 261                 if ( $clearState ) {
 262                         $this->clearState();
 263                 }
 264
 265                 $this->mOptions = $options;
 266                 $this->mTitle =& $title;
 267                 $oldRevisionId = $this->mRevisionId;
 268                 if( $revid !== null ) {
 269                         $this->mRevisionId = $revid;
 270                 }
 271                 $this->mOutputType = OT_HTML;
 272
 273                 //$text = $this->strip( $text, $this->mStripState );
 274                 // VOODOO MAGIC FIX! Sometimes the above segfaults in PHP5.
 275                 $x =& $this->mStripState;
 276
 277                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 278                 $text = $this->strip( $text, $x );
 279                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 280
 281                 $text = $this->internalParse( $text );
 282
 283                 $text = $this->unstrip( $text, $this->mStripState );
 284
 285                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 286                 $fixtags = array(
 287                         # french spaces, last one Guillemet-left
 288                         # only if there is something before the space
 289                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 290                         # french spaces, Guillemet-right
 291                         '/(\\302\\253) /' => '\\1&nbsp;',
 292                 );
 293                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 294
 295                 # only once and last
 296                 $text = $this->doBlockLevels( $text, $linestart );
 297
 298                 $this->replaceLinkHolders( $text );
 299
 300                 # the position of the parserConvert() call should not be changed. it
 301                 # assumes that the links are all replaced and the only thing left
 302                 # is the <nowiki> mark.
 303                 # Side-effects: this calls $this->mOutput->setTitleText()
 304                 $text = $wgContLang->parserConvert( $text, $this );
 305
 306                 $text = $this->unstripNoWiki( $text, $this->mStripState );
 307
 308                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 309
 310                 $text = Sanitizer::normalizeCharReferences( $text );
 311
 312                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 313                         $text = Parser::tidy($text);
 314                 } else {
 315                         # attempt to sanitize at least some nesting problems
 316                         # (bug #2702 and quite a few others)
 317                         $tidyregs = array(
 318                                 # ''Something [http://www.cool.com cool''] -->
 319                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 320                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 321                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 322                                 # fix up an anchor inside another anchor, only
 323                                 # at least for a single single nested link (bug 3695)
 324                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 325                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 326                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 327                                 # contains a div, so fix it up here; replace
 328                                 # div with escaped text
 329                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 330                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 331                                 # remove empty italic or bold tag pairs, some
 332                                 # introduced by rules above
 333                                 '/<([bi])><\/\\1>/' => '',
 334                         );
 335
 336                         $text = preg_replace(
 337                                 array_keys( $tidyregs ),
 338                                 array_values( $tidyregs ),
 339                                 $text );
 340                 }
 341
 342                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 343
 344                 # Information on include size limits, for the benefit of users who try to skirt them
 345                 if ( max( $this->mIncludeSizes ) > 1000 ) {
 346                         $max = $this->mOptions->getMaxIncludeSize();
 347                         $text .= "<!-- \n" .
 348                                 "Pre-expand include size: {$this->mIncludeSizes['pre-expand']} bytes\n" .
 349                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']} bytes\n" .
 350                                 "Template argument size: {$this->mIncludeSizes['arg']} bytes\n" .
 351                                 "Maximum: $max bytes\n" .
 352                                 "-->\n";
 353                 }
 354                 $this->mOutput->setText( $text );
 355                 $this->mRevisionId = $oldRevisionId;
 356                 wfProfileOut( $fname );
 357
 358                 return $this->mOutput;
 359         }
 360
 361         /**
 362          * Recursive parser entry point that can be called from an extension tag
 363          * hook.
 364          */
 365         function recursiveTagParse( $text ) {
 366                 wfProfileIn( __METHOD__ );
 367                 $x =& $this->mStripState;
 368                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 369                 $text = $this->strip( $text, $x );
 370                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 371                 $text = $this->internalParse( $text );
 372                 wfProfileOut( __METHOD__ );
 373                 return $text;
 374         }
 375
 376         /**
 377          * Get a random string
 378          *
 379          * @private
 380          * @static
 381          */
 382         function getRandomString() {
 383                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 384         }
 385
 386         function &getTitle() { return $this->mTitle; }
 387         function getOptions() { return $this->mOptions; }
 388
 389         function getFunctionLang() {
 390                 global $wgLang, $wgContLang;
 391                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 392         }
 393
 394         /**
 395          * Replaces all occurrences of HTML-style comments and the given tags
 396          * in the text with a random marker and returns teh next text. The output
 397          * parameter $matches will be an associative array filled with data in
 398          * the form:
 399          *   'UNIQ-xxxxx' => array(
 400          *     'element',
 401          *     'tag content',
 402          *     array( 'param' => 'x' ),
 403          *     '<element param="x">tag content</element>' ) )
 404          *
 405          * @param $elements list of element names. Comments are always extracted.
 406          * @param $text Source text string.
 407          * @param $uniq_prefix
 408          *
 409          * @private
 410          * @static
 411          */
 412         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 413                 static $n = 1;
 414                 $stripped = '';
 415                 $matches = array();
 416
 417                 $taglist = implode( '|', $elements );
 418                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 419
 420                 while ( '' != $text ) {
 421                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 422                         $stripped .= $p[0];
 423                         if( count( $p ) < 5 ) {
 424                                 break;
 425                         }
 426                         if( count( $p ) > 5 ) {
 427                                 // comment
 428                                 $element    = $p[4];
 429                                 $attributes = '';
 430                                 $close      = '';
 431                                 $inside     = $p[5];
 432                         } else {
 433                                 // tag
 434                                 $element    = $p[1];
 435                                 $attributes = $p[2];
 436                                 $close      = $p[3];
 437                                 $inside     = $p[4];
 438                         }
 439
 440                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . '-QINU';
 441                         $stripped .= $marker;
 442
 443                         if ( $close === '/>' ) {
 444                                 // Empty element tag, <tag />
 445                                 $content = null;
 446                                 $text = $inside;
 447                                 $tail = null;
 448                         } else {
 449                                 if( $element == '!--' ) {
 450                                         $end = '/(-->)/';
 451                                 } else {
 452                                         $end = "/(<\\/$element\\s*>)/i";
 453                                 }
 454                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 455                                 $content = $q[0];
 456                                 if( count( $q ) < 3 ) {
 457                                         # No end tag -- let it run out to the end of the text.
 458                                         $tail = '';
 459                                         $text = '';
 460                                 } else {
 461                                         $tail = $q[1];
 462                                         $text = $q[2];
 463                                 }
 464                         }
 465
 466                         $matches[$marker] = array( $element,
 467                                 $content,
 468                                 Sanitizer::decodeTagAttributes( $attributes ),
 469                                 "<$element$attributes$close$content$tail" );
 470                 }
 471                 return $stripped;
 472         }
 473
 474         /**
 475          * Strips and renders nowiki, pre, math, hiero
 476          * If $render is set, performs necessary rendering operations on plugins
 477          * Returns the text, and fills an array with data needed in unstrip()
 478          * If the $state is already a valid strip state, it adds to the state
 479          *
 480          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 481          *  will be stripped in addition to other tags. This is important
 482          *  for section editing, where these comments cause confusion when
 483          *  counting the sections in the wikisource
 484          *
 485          * @param array dontstrip contains tags which should not be stripped;
 486          *  used to prevent stipping of <gallery> when saving (fixes bug 2700)
 487          *
 488          * @private
 489          */
 490         function strip( $text, &$state, $stripcomments = false , $dontstrip = array () ) {
 491                 wfProfileIn( __METHOD__ );
 492                 $render = ($this->mOutputType == OT_HTML);
 493
 494                 # Replace any instances of the placeholders
 495                 $uniq_prefix = $this->mUniqPrefix;
 496                 #$text = str_replace( $uniq_prefix, wfHtmlEscapeFirst( $uniq_prefix ), $text );
 497                 $commentState = array();
 498
 499                 $elements = array_merge(
 500                         array( 'nowiki', 'gallery' ),
 501                         array_keys( $this->mTagHooks ) );
 502                 global $wgRawHtml;
 503                 if( $wgRawHtml ) {
 504                         $elements[] = 'html';
 505                 }
 506                 if( $this->mOptions->getUseTeX() ) {
 507                         $elements[] = 'math';
 508                 }
 509
 510                 # Removing $dontstrip tags from $elements list (currently only 'gallery', fixing bug 2700)
 511                 foreach ( $elements AS $k => $v ) {
 512                         if ( !in_array ( $v , $dontstrip ) ) continue;
 513                         unset ( $elements[$k] );
 514                 }
 515
 516                 $matches = array();
 517                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 518
 519                 foreach( $matches as $marker => $data ) {
 520                         list( $element, $content, $params, $tag ) = $data;
 521                         if( $render ) {
 522                                 $tagName = strtolower( $element );
 523                                 wfProfileIn( __METHOD__."-render-$tagName" );
 524                                 switch( $tagName ) {
 525                                 case '!--':
 526                                         // Comment
 527                                         if( substr( $tag, -3 ) == '-->' ) {
 528                                                 $output = $tag;
 529                                         } else {
 530                                                 // Unclosed comment in input.
 531                                                 // Close it so later stripping can remove it
 532                                                 $output = "$tag-->";
 533                                         }
 534                                         break;
 535                                 case 'html':
 536                                         if( $wgRawHtml ) {
 537                                                 $output = $content;
 538                                                 break;
 539                                         }
 540                                         // Shouldn't happen otherwise. :)
 541                                 case 'nowiki':
 542                                         $output = wfEscapeHTMLTagsOnly( $content );
 543                                         break;
 544                                 case 'math':
 545                                         $output = MathRenderer::renderMath( $content );
 546                                         break;
 547                                 case 'gallery':
 548                                         $output = $this->renderImageGallery( $content, $params );
 549                                         break;
 550                                 default:
 551                                         if( isset( $this->mTagHooks[$tagName] ) ) {
 552                                                 $output = call_user_func_array( $this->mTagHooks[$tagName],
 553                                                         array( $content, $params, $this ) );
 554                                         } else {
 555                                                 throw new MWException( "Invalid call hook $element" );
 556                                         }
 557                                 }
 558                                 wfProfileOut( __METHOD__."-render-$tagName" );
 559                         } else {
 560                                 // Just stripping tags; keep the source
 561                                 $output = $tag;
 562                         }
 563
 564                         // Unstrip the output, because unstrip() is no longer recursive so
 565                         // it won't do it itself
 566                         $output = $this->unstrip( $output, $state );
 567
 568                         if( !$stripcomments && $element == '!--' ) {
 569                                 $commentState[$marker] = $output;
 570                         } elseif ( $element == 'html' || $element == 'nowiki' ) {
 571                                 $state['nowiki'][$marker] = $output;
 572                         } else {
 573                                 $state['general'][$marker] = $output;
 574                         }
 575                 }
 576
 577                 # Unstrip comments unless explicitly told otherwise.
 578                 # (The comments are always stripped prior to this point, so as to
 579                 # not invoke any extension tags / parser hooks contained within
 580                 # a comment.)
 581                 if ( !$stripcomments ) {
 582                         // Put them all back and forget them
 583                         $text = strtr( $text, $commentState );
 584                 }
 585
 586                 wfProfileOut( __METHOD__ );
 587                 return $text;
 588         }
 589
 590         /**
 591          * Restores pre, math, and other extensions removed by strip()
 592          *
 593          * always call unstripNoWiki() after this one
 594          * @private
 595          */
 596         function unstrip( $text, &$state ) {
 597                 if ( !isset( $state['general'] ) ) {
 598                         return $text;
 599                 }
 600
 601                 wfProfileIn( __METHOD__ );
 602                 # TODO: good candidate for FSS
 603                 $text = strtr( $text, $state['general'] );
 604                 wfProfileOut( __METHOD__ );
 605                 return $text;
 606         }
 607
 608         /**
 609          * Always call this after unstrip() to preserve the order
 610          *
 611          * @private
 612          */
 613         function unstripNoWiki( $text, &$state ) {
 614                 if ( !isset( $state['nowiki'] ) ) {
 615                         return $text;
 616                 }
 617
 618                 wfProfileIn( __METHOD__ );
 619                 # TODO: good candidate for FSS
 620                 $text = strtr( $text, $state['nowiki'] );
 621                 wfProfileOut( __METHOD__ );
 622
 623                 return $text;
 624         }
 625
 626         /**
 627          * Add an item to the strip state
 628          * Returns the unique tag which must be inserted into the stripped text
 629          * The tag will be replaced with the original text in unstrip()
 630          *
 631          * @private
 632          */
 633         function insertStripItem( $text, &$state ) {
 634                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 635                 if ( !$state ) {
 636                         $state = array();
 637                 }
 638                 $state['general'][$rnd] = $text;
 639                 return $rnd;
 640         }
 641
 642         /**
 643          * Interface with html tidy, used if $wgUseTidy = true.
 644          * If tidy isn't able to correct the markup, the original will be
 645          * returned in all its glory with a warning comment appended.
 646          *
 647          * Either the external tidy program or the in-process tidy extension
 648          * will be used depending on availability. Override the default
 649          * $wgTidyInternal setting to disable the internal if it's not working.
 650          *
 651          * @param string $text Hideous HTML input
 652          * @return string Corrected HTML output
 653          * @public
 654          * @static
 655          */
 656         function tidy( $text ) {
 657                 global $wgTidyInternal;
 658                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 659 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 660 '<head><title>test</title></head><body>'.$text.'</body></html>';
 661                 if( $wgTidyInternal ) {
 662                         $correctedtext = Parser::internalTidy( $wrappedtext );
 663                 } else {
 664                         $correctedtext = Parser::externalTidy( $wrappedtext );
 665                 }
 666                 if( is_null( $correctedtext ) ) {
 667                         wfDebug( "Tidy error detected!\n" );
 668                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 669                 }
 670                 return $correctedtext;
 671         }
 672
 673         /**
 674          * Spawn an external HTML tidy process and get corrected markup back from it.
 675          *
 676          * @private
 677          * @static
 678          */
 679         function externalTidy( $text ) {
 680                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 681                 $fname = 'Parser::externalTidy';
 682                 wfProfileIn( $fname );
 683
 684                 $cleansource = '';
 685                 $opts = ' -utf8';
 686
 687                 $descriptorspec = array(
 688                         0 => array('pipe', 'r'),
 689                         1 => array('pipe', 'w'),
 690                         2 => array('file', '/dev/null', 'a')
 691                 );
 692                 $pipes = array();
 693                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 694                 if (is_resource($process)) {
 695                         // Theoretically, this style of communication could cause a deadlock
 696                         // here. If the stdout buffer fills up, then writes to stdin could
 697                         // block. This doesn't appear to happen with tidy, because tidy only
 698                         // writes to stdout after it's finished reading from stdin. Search
 699                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 700                         fwrite($pipes[0], $text);
 701                         fclose($pipes[0]);
 702                         while (!feof($pipes[1])) {
 703                                 $cleansource .= fgets($pipes[1], 1024);
 704                         }
 705                         fclose($pipes[1]);
 706                         proc_close($process);
 707                 }
 708
 709                 wfProfileOut( $fname );
 710
 711                 if( $cleansource == '' && $text != '') {
 712                         // Some kind of error happened, so we couldn't get the corrected text.
 713                         // Just give up; we'll use the source text and append a warning.
 714                         return null;
 715                 } else {
 716                         return $cleansource;
 717                 }
 718         }
 719
 720         /**
 721          * Use the HTML tidy PECL extension to use the tidy library in-process,
 722          * saving the overhead of spawning a new process. Currently written to
 723          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 724          *
 725          * 'pear install tidy' should be able to compile the extension module.
 726          *
 727          * @private
 728          * @static
 729          */
 730         function internalTidy( $text ) {
 731                 global $wgTidyConf;
 732                 $fname = 'Parser::internalTidy';
 733                 wfProfileIn( $fname );
 734
 735                 tidy_load_config( $wgTidyConf );
 736                 tidy_set_encoding( 'utf8' );
 737                 tidy_parse_string( $text );
 738                 tidy_clean_repair();
 739                 if( tidy_get_status() == 2 ) {
 740                         // 2 is magic number for fatal error
 741                         // http://www.php.net/manual/en/function.tidy-get-status.php
 742                         $cleansource = null;
 743                 } else {
 744                         $cleansource = tidy_get_output();
 745                 }
 746                 wfProfileOut( $fname );
 747                 return $cleansource;
 748         }
 749
 750         /**
 751          * parse the wiki syntax used to render tables
 752          *
 753          * @private
 754          */
 755         function doTableStuff ( $t ) {
 756                 $fname = 'Parser::doTableStuff';
 757                 wfProfileIn( $fname );
 758
 759                 $t = explode ( "\n" , $t ) ;
 760                 $td = array () ; # Is currently a td tag open?
 761                 $ltd = array () ; # Was it TD or TH?
 762                 $tr = array () ; # Is currently a tr tag open?
 763                 $ltr = array () ; # tr attributes
 764                 $has_opened_tr = array(); # Did this table open a <tr> element?
 765                 $indent_level = 0; # indent level of the table
 766                 foreach ( $t AS $k => $x )
 767                 {
 768                         $x = trim ( $x ) ;
 769                         $fc = substr ( $x , 0 , 1 ) ;
 770                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 771                                 $indent_level = strlen( $matches[1] );
 772
 773                                 $attributes = $this->unstripForHTML( $matches[2] );
 774
 775                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 776                                         '<table' . Sanitizer::fixTagAttributes ( $attributes, 'table' ) . '>' ;
 777                                 array_push ( $td , false ) ;
 778                                 array_push ( $ltd , '' ) ;
 779                                 array_push ( $tr , false ) ;
 780                                 array_push ( $ltr , '' ) ;
 781                                 array_push ( $has_opened_tr, false );
 782                         }
 783                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 784                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 785                                 $z = "</table>" . substr ( $x , 2);
 786                                 $l = array_pop ( $ltd ) ;
 787                                 if ( !array_pop ( $has_opened_tr ) ) $z = "<tr><td></td></tr>" . $z ;
 788                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 789                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 790                                 array_pop ( $ltr ) ;
 791                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 792                         }
 793                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 794                                 $x = substr ( $x , 1 ) ;
 795                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 796                                 $z = '' ;
 797                                 $l = array_pop ( $ltd ) ;
 798                                 array_pop ( $has_opened_tr );
 799                                 array_push ( $has_opened_tr , true ) ;
 800                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 801                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 802                                 array_pop ( $ltr ) ;
 803                                 $t[$k] = $z ;
 804                                 array_push ( $tr , false ) ;
 805                                 array_push ( $td , false ) ;
 806                                 array_push ( $ltd , '' ) ;
 807                                 $attributes = $this->unstripForHTML( $x );
 808                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $attributes, 'tr' ) ) ;
 809                         }
 810                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 811                                 # $x is a table row
 812                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 813                                         $fc = '+' ;
 814                                         $x = substr ( $x , 1 ) ;
 815                                 }
 816                                 $after = substr ( $x , 1 ) ;
 817                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 818
 819                                 // Split up multiple cells on the same line.
 820                                 // FIXME: This can result in improper nesting of tags processed
 821                                 // by earlier parser steps, but should avoid splitting up eg
 822                                 // attribute values containing literal "||".
 823                                 $after = wfExplodeMarkup( '||', $after );
 824
 825                                 $t[$k] = '' ;
 826
 827                                 # Loop through each table cell
 828                                 foreach ( $after AS $theline )
 829                                 {
 830                                         $z = '' ;
 831                                         if ( $fc != '+' )
 832                                         {
 833                                                 $tra = array_pop ( $ltr ) ;
 834                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 835                                                 array_push ( $tr , true ) ;
 836                                                 array_push ( $ltr , '' ) ;
 837                                                 array_pop ( $has_opened_tr );
 838                                                 array_push ( $has_opened_tr , true ) ;
 839                                         }
 840
 841                                         $l = array_pop ( $ltd ) ;
 842                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 843                                         if ( $fc == '|' ) $l = 'td' ;
 844                                         else if ( $fc == '!' ) $l = 'th' ;
 845                                         else if ( $fc == '+' ) $l = 'caption' ;
 846                                         else $l = '' ;
 847                                         array_push ( $ltd , $l ) ;
 848
 849                                         # Cell parameters
 850                                         $y = explode ( '|' , $theline , 2 ) ;
 851                                         # Note that a '|' inside an invalid link should not
 852                                         # be mistaken as delimiting cell parameters
 853                                         if ( strpos( $y[0], '[[' ) !== false ) {
 854                                                 $y = array ($theline);
 855                                         }
 856                                         if ( count ( $y ) == 1 )
 857                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 858                                         else {
 859                                                 $attributes = $this->unstripForHTML( $y[0] );
 860                                                 $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($attributes, $l).">{$y[1]}" ;
 861                                         }
 862                                         $t[$k] .= $y ;
 863                                         array_push ( $td , true ) ;
 864                                 }
 865                         }
 866                 }
 867
 868                 # Closing open td, tr && table
 869                 while ( count ( $td ) > 0 )
 870                 {
 871                         $l = array_pop ( $ltd ) ;
 872                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 873                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 874                         if ( !array_pop ( $has_opened_tr ) ) $t[] = "<tr><td></td></tr>" ;
 875                         $t[] = '</table>' ;
 876                 }
 877
 878                 $t = implode ( "\n" , $t ) ;
 879                 # special case: don't return empty table
 880                 if($t == "<table>\n<tr><td></td></tr>\n</table>")
 881                         $t = '';
 882                 wfProfileOut( $fname );
 883                 return $t ;
 884         }
 885
 886         /**
 887          * Helper function for parse() that transforms wiki markup into
 888          * HTML. Only called for $mOutputType == OT_HTML.
 889          *
 890          * @private
 891          */
 892         function internalParse( $text ) {
 893                 $args = array();
 894                 $isMain = true;
 895                 $fname = 'Parser::internalParse';
 896                 wfProfileIn( $fname );
 897
 898                 # Hook to suspend the parser in this state
 899                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$x ) ) ) {
 900                         wfProfileOut( $fname );
 901                         return $text ;
 902                 }
 903
 904                 # Remove <noinclude> tags and <includeonly> sections
 905                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 906                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 907                 $text = preg_replace( '/<includeonly>.*?<\/includeonly>/s', '', $text );
 908
 909                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
 910
 911                 $text = $this->replaceVariables( $text, $args );
 912
 913                 // Tables need to come after variable replacement for things to work
 914                 // properly; putting them before other transformations should keep
 915                 // exciting things like link expansions from showing up in surprising
 916                 // places.
 917                 $text = $this->doTableStuff( $text );
 918
 919                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 920
 921                 $text = $this->stripToc( $text );
 922                 $this->stripNoGallery( $text );
 923                 $text = $this->doHeadings( $text );
 924                 if($this->mOptions->getUseDynamicDates()) {
 925                         $df =& DateFormatter::getInstance();
 926                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 927                 }
 928                 $text = $this->doAllQuotes( $text );
 929                 $text = $this->replaceInternalLinks( $text );
 930                 $text = $this->replaceExternalLinks( $text );
 931
 932                 # replaceInternalLinks may sometimes leave behind
 933                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 934                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 935
 936                 $text = $this->doMagicLinks( $text );
 937                 $text = $this->formatHeadings( $text, $isMain );
 938
 939                 wfProfileOut( $fname );
 940                 return $text;
 941         }
 942
 943         /**
 944          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 945          * magic external links.
 946          *
 947          * @private
 948          */
 949         function &doMagicLinks( &$text ) {
 950                 wfProfileIn( __METHOD__ );
 951                 $text = preg_replace_callback(
 952                         '!(?:                           # Start cases
 953                             <a.*?</a> |                 # Skip link text
 954                             <.*?> |                     # Skip stuff inside HTML elements
 955                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
 956                             ISBN\s+([0-9Xx-]+)          # ISBN, capture number as m[2]
 957                         )!x', array( &$this, 'magicLinkCallback' ), $text );
 958                 wfProfileOut( __METHOD__ );
 959                 return $text;
 960         }
 961
 962         function magicLinkCallback( $m ) {
 963                 if ( substr( $m[0], 0, 1 ) == '<' ) {
 964                         # Skip HTML element
 965                         return $m[0];
 966                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
 967                         $isbn = $m[2];
 968                         $num = strtr( $isbn, array(
 969                                 '-' => '',
 970                                 ' ' => '',
 971                                 'x' => 'X',
 972                         ));
 973                         $titleObj = Title::makeTitle( NS_SPECIAL, 'Booksources' );
 974                         $text = '<a href="' .
 975                                 $titleObj->escapeLocalUrl( "isbn=$num" ) .
 976                                 "\" class=\"internal\">ISBN $isbn</a>";
 977                 } else {
 978                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
 979                                 $keyword = 'RFC';
 980                                 $urlmsg = 'rfcurl';
 981                                 $id = $m[1];
 982                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
 983                                 $keyword = 'PMID';
 984                                 $urlmsg = 'pubmedurl';
 985                                 $id = $m[1];
 986                         } else {
 987                                 throw new MWException( __METHOD__.': unrecognised match type "' .
 988                                         substr($m[0], 0, 20 ) . '"' );
 989                         }
 990
 991                         $url = wfMsg( $urlmsg, $id);
 992                         $sk =& $this->mOptions->getSkin();
 993                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
 994                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
 995                 }
 996                 return $text;
 997         }
 998
 999         /**
1000          * Parse headers and return html
1001          *
1002          * @private
1003          */
1004         function doHeadings( $text ) {
1005                 $fname = 'Parser::doHeadings';
1006                 wfProfileIn( $fname );
1007                 for ( $i = 6; $i >= 1; --$i ) {
1008                         $h = str_repeat( '=', $i );
1009                         $text = preg_replace( "/^{$h}(.+){$h}\\s*$/m",
1010                           "<h{$i}>\\1</h{$i}>\\2", $text );
1011                 }
1012                 wfProfileOut( $fname );
1013                 return $text;
1014         }
1015
1016         /**
1017          * Replace single quotes with HTML markup
1018          * @private
1019          * @return string the altered text
1020          */
1021         function doAllQuotes( $text ) {
1022                 $fname = 'Parser::doAllQuotes';
1023                 wfProfileIn( $fname );
1024                 $outtext = '';
1025                 $lines = explode( "\n", $text );
1026                 foreach ( $lines as $line ) {
1027                         $outtext .= $this->doQuotes ( $line ) . "\n";
1028                 }
1029                 $outtext = substr($outtext, 0,-1);
1030                 wfProfileOut( $fname );
1031                 return $outtext;
1032         }
1033
1034         /**
1035          * Helper function for doAllQuotes()
1036          * @private
1037          */
1038         function doQuotes( $text ) {
1039                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1040                 if ( count( $arr ) == 1 )
1041                         return $text;
1042                 else
1043                 {
1044                         # First, do some preliminary work. This may shift some apostrophes from
1045                         # being mark-up to being text. It also counts the number of occurrences
1046                         # of bold and italics mark-ups.
1047                         $i = 0;
1048                         $numbold = 0;
1049                         $numitalics = 0;
1050                         foreach ( $arr as $r )
1051                         {
1052                                 if ( ( $i % 2 ) == 1 )
1053                                 {
1054                                         # If there are ever four apostrophes, assume the first is supposed to
1055                                         # be text, and the remaining three constitute mark-up for bold text.
1056                                         if ( strlen( $arr[$i] ) == 4 )
1057                                         {
1058                                                 $arr[$i-1] .= "'";
1059                                                 $arr[$i] = "'''";
1060                                         }
1061                                         # If there are more than 5 apostrophes in a row, assume they're all
1062                                         # text except for the last 5.
1063                                         else if ( strlen( $arr[$i] ) > 5 )
1064                                         {
1065                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1066                                                 $arr[$i] = "'''''";
1067                                         }
1068                                         # Count the number of occurrences of bold and italics mark-ups.
1069                                         # We are not counting sequences of five apostrophes.
1070                                         if ( strlen( $arr[$i] ) == 2 ) $numitalics++;  else
1071                                         if ( strlen( $arr[$i] ) == 3 ) $numbold++;     else
1072                                         if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1073                                 }
1074                                 $i++;
1075                         }
1076
1077                         # If there is an odd number of both bold and italics, it is likely
1078                         # that one of the bold ones was meant to be an apostrophe followed
1079                         # by italics. Which one we cannot know for certain, but it is more
1080                         # likely to be one that has a single-letter word before it.
1081                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1082                         {
1083                                 $i = 0;
1084                                 $firstsingleletterword = -1;
1085                                 $firstmultiletterword = -1;
1086                                 $firstspace = -1;
1087                                 foreach ( $arr as $r )
1088                                 {
1089                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1090                                         {
1091                                                 $x1 = substr ($arr[$i-1], -1);
1092                                                 $x2 = substr ($arr[$i-1], -2, 1);
1093                                                 if ($x1 == ' ') {
1094                                                         if ($firstspace == -1) $firstspace = $i;
1095                                                 } else if ($x2 == ' ') {
1096                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1097                                                 } else {
1098                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1099                                                 }
1100                                         }
1101                                         $i++;
1102                                 }
1103
1104                                 # If there is a single-letter word, use it!
1105                                 if ($firstsingleletterword > -1)
1106                                 {
1107                                         $arr [ $firstsingleletterword ] = "''";
1108                                         $arr [ $firstsingleletterword-1 ] .= "'";
1109                                 }
1110                                 # If not, but there's a multi-letter word, use that one.
1111                                 else if ($firstmultiletterword > -1)
1112                                 {
1113                                         $arr [ $firstmultiletterword ] = "''";
1114                                         $arr [ $firstmultiletterword-1 ] .= "'";
1115                                 }
1116                                 # ... otherwise use the first one that has neither.
1117                                 # (notice that it is possible for all three to be -1 if, for example,
1118                                 # there is only one pentuple-apostrophe in the line)
1119                                 else if ($firstspace > -1)
1120                                 {
1121                                         $arr [ $firstspace ] = "''";
1122                                         $arr [ $firstspace-1 ] .= "'";
1123                                 }
1124                         }
1125
1126                         # Now let's actually convert our apostrophic mush to HTML!
1127                         $output = '';
1128                         $buffer = '';
1129                         $state = '';
1130                         $i = 0;
1131                         foreach ($arr as $r)
1132                         {
1133                                 if (($i % 2) == 0)
1134                                 {
1135                                         if ($state == 'both')
1136                                                 $buffer .= $r;
1137                                         else
1138                                                 $output .= $r;
1139                                 }
1140                                 else
1141                                 {
1142                                         if (strlen ($r) == 2)
1143                                         {
1144                                                 if ($state == 'i')
1145                                                 { $output .= '</i>'; $state = ''; }
1146                                                 else if ($state == 'bi')
1147                                                 { $output .= '</i>'; $state = 'b'; }
1148                                                 else if ($state == 'ib')
1149                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1150                                                 else if ($state == 'both')
1151                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1152                                                 else # $state can be 'b' or ''
1153                                                 { $output .= '<i>'; $state .= 'i'; }
1154                                         }
1155                                         else if (strlen ($r) == 3)
1156                                         {
1157                                                 if ($state == 'b')
1158                                                 { $output .= '</b>'; $state = ''; }
1159                                                 else if ($state == 'bi')
1160                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1161                                                 else if ($state == 'ib')
1162                                                 { $output .= '</b>'; $state = 'i'; }
1163                                                 else if ($state == 'both')
1164                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1165                                                 else # $state can be 'i' or ''
1166                                                 { $output .= '<b>'; $state .= 'b'; }
1167                                         }
1168                                         else if (strlen ($r) == 5)
1169                                         {
1170                                                 if ($state == 'b')
1171                                                 { $output .= '</b><i>'; $state = 'i'; }
1172                                                 else if ($state == 'i')
1173                                                 { $output .= '</i><b>'; $state = 'b'; }
1174                                                 else if ($state == 'bi')
1175                                                 { $output .= '</i></b>'; $state = ''; }
1176                                                 else if ($state == 'ib')
1177                                                 { $output .= '</b></i>'; $state = ''; }
1178                                                 else if ($state == 'both')
1179                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1180                                                 else # ($state == '')
1181                                                 { $buffer = ''; $state = 'both'; }
1182                                         }
1183                                 }
1184                                 $i++;
1185                         }
1186                         # Now close all remaining tags.  Notice that the order is important.
1187                         if ($state == 'b' || $state == 'ib')
1188                                 $output .= '</b>';
1189                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1190                                 $output .= '</i>';
1191                         if ($state == 'bi')
1192                                 $output .= '</b>';
1193                         if ($state == 'both')
1194                                 $output .= '<b><i>'.$buffer.'</i></b>';
1195                         return $output;
1196                 }
1197         }
1198
1199         /**
1200          * Replace external links
1201          *
1202          * Note: this is all very hackish and the order of execution matters a lot.
1203          * Make sure to run maintenance/parserTests.php if you change this code.
1204          *
1205          * @private
1206          */
1207         function replaceExternalLinks( $text ) {
1208                 global $wgContLang;
1209                 $fname = 'Parser::replaceExternalLinks';
1210                 wfProfileIn( $fname );
1211
1212                 $sk =& $this->mOptions->getSkin();
1213
1214                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1215
1216                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1217
1218                 $i = 0;
1219                 while ( $i<count( $bits ) ) {
1220                         $url = $bits[$i++];
1221                         $protocol = $bits[$i++];
1222                         $text = $bits[$i++];
1223                         $trail = $bits[$i++];
1224
1225                         # The characters '<' and '>' (which were escaped by
1226                         # removeHTMLtags()) should not be included in
1227                         # URLs, per RFC 2396.
1228                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1229                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1230                                 $url = substr($url, 0, $m2[0][1]);
1231                         }
1232
1233                         # If the link text is an image URL, replace it with an <img> tag
1234                         # This happened by accident in the original parser, but some people used it extensively
1235                         $img = $this->maybeMakeExternalImage( $text );
1236                         if ( $img !== false ) {
1237                                 $text = $img;
1238                         }
1239
1240                         $dtrail = '';
1241
1242                         # Set linktype for CSS - if URL==text, link is essentially free
1243                         $linktype = ($text == $url) ? 'free' : 'text';
1244
1245                         # No link text, e.g. [http://domain.tld/some.link]
1246                         if ( $text == '' ) {
1247                                 # Autonumber if allowed. See bug #5918
1248                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1249                                         $text = '[' . ++$this->mAutonumber . ']';
1250                                         $linktype = 'autonumber';
1251                                 } else {
1252                                         # Otherwise just use the URL
1253                                         $text = htmlspecialchars( $url );
1254                                         $linktype = 'free';
1255                                 }
1256                         } else {
1257                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1258                                 # Check for trail
1259                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1260                         }
1261
1262                         $text = $wgContLang->markNoConversion($text);
1263
1264                         $url = Sanitizer::cleanUrl( $url );
1265
1266                         # Process the trail (i.e. everything after this link up until start of the next link),
1267                         # replacing any non-bracketed links
1268                         $trail = $this->replaceFreeExternalLinks( $trail );
1269
1270                         # Use the encoded URL
1271                         # This means that users can paste URLs directly into the text
1272                         # Funny characters like &ouml; aren't valid in URLs anyway
1273                         # This was changed in August 2004
1274                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1275
1276                         # Register link in the output object.
1277                         # Replace unnecessary URL escape codes with the referenced character
1278                         # This prevents spammers from hiding links from the filters
1279                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1280                         $this->mOutput->addExternalLink( $pasteurized );
1281                 }
1282
1283                 wfProfileOut( $fname );
1284                 return $s;
1285         }
1286
1287         /**
1288          * Replace anything that looks like a URL with a link
1289          * @private
1290          */
1291         function replaceFreeExternalLinks( $text ) {
1292                 global $wgContLang;
1293                 $fname = 'Parser::replaceFreeExternalLinks';
1294                 wfProfileIn( $fname );
1295
1296                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1297                 $s = array_shift( $bits );
1298                 $i = 0;
1299
1300                 $sk =& $this->mOptions->getSkin();
1301
1302                 while ( $i < count( $bits ) ){
1303                         $protocol = $bits[$i++];
1304                         $remainder = $bits[$i++];
1305
1306                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1307                                 # Found some characters after the protocol that look promising
1308                                 $url = $protocol . $m[1];
1309                                 $trail = $m[2];
1310
1311                                 # special case: handle urls as url args:
1312                                 # http://www.example.com/foo?=http://www.example.com/bar
1313                                 if(strlen($trail) == 0 &&
1314                                         isset($bits[$i]) &&
1315                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1316                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1317                                 {
1318                                         # add protocol, arg
1319                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1320                                         $i += 2;
1321                                         $trail = $m[2];
1322                                 }
1323
1324                                 # The characters '<' and '>' (which were escaped by
1325                                 # removeHTMLtags()) should not be included in
1326                                 # URLs, per RFC 2396.
1327                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1328                                         $trail = substr($url, $m2[0][1]) . $trail;
1329                                         $url = substr($url, 0, $m2[0][1]);
1330                                 }
1331
1332                                 # Move trailing punctuation to $trail
1333                                 $sep = ',;\.:!?';
1334                                 # If there is no left bracket, then consider right brackets fair game too
1335                                 if ( strpos( $url, '(' ) === false ) {
1336                                         $sep .= ')';
1337                                 }
1338
1339                                 $numSepChars = strspn( strrev( $url ), $sep );
1340                                 if ( $numSepChars ) {
1341                                         $trail = substr( $url, -$numSepChars ) . $trail;
1342                                         $url = substr( $url, 0, -$numSepChars );
1343                                 }
1344
1345                                 $url = Sanitizer::cleanUrl( $url );
1346
1347                                 # Is this an external image?
1348                                 $text = $this->maybeMakeExternalImage( $url );
1349                                 if ( $text === false ) {
1350                                         # Not an image, make a link
1351                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1352                                         # Register it in the output object...
1353                                         # Replace unnecessary URL escape codes with their equivalent characters
1354                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1355                                         $this->mOutput->addExternalLink( $pasteurized );
1356                                 }
1357                                 $s .= $text . $trail;
1358                         } else {
1359                                 $s .= $protocol . $remainder;
1360                         }
1361                 }
1362                 wfProfileOut( $fname );
1363                 return $s;
1364         }
1365
1366         /**
1367          * Replace unusual URL escape codes with their equivalent characters
1368          * @param string
1369          * @return string
1370          * @static
1371          * @fixme This can merge genuinely required bits in the path or query string,
1372          *        breaking legit URLs. A proper fix would treat the various parts of
1373          *        the URL differently; as a workaround, just use the output for
1374          *        statistical records, not for actual linking/output.
1375          */
1376         static function replaceUnusualEscapes( $url ) {
1377                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1378                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1379         }
1380
1381         /**
1382          * Callback function used in replaceUnusualEscapes().
1383          * Replaces unusual URL escape codes with their equivalent character
1384          * @static
1385          * @private
1386          */
1387         private static function replaceUnusualEscapesCallback( $matches ) {
1388                 $char = urldecode( $matches[0] );
1389                 $ord = ord( $char );
1390                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1391                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1392                         // No, shouldn't be escaped
1393                         return $char;
1394                 } else {
1395                         // Yes, leave it escaped
1396                         return $matches[0];
1397                 }
1398         }
1399
1400         /**
1401          * make an image if it's allowed, either through the global
1402          * option or through the exception
1403          * @private
1404          */
1405         function maybeMakeExternalImage( $url ) {
1406                 $sk =& $this->mOptions->getSkin();
1407                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1408                 $imagesexception = !empty($imagesfrom);
1409                 $text = false;
1410                 if ( $this->mOptions->getAllowExternalImages()
1411                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1412                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1413                                 # Image found
1414                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1415                         }
1416                 }
1417                 return $text;
1418         }
1419
1420         /**
1421          * Process [[ ]] wikilinks
1422          *
1423          * @private
1424          */
1425         function replaceInternalLinks( $s ) {
1426                 global $wgContLang;
1427                 static $fname = 'Parser::replaceInternalLinks' ;
1428
1429                 wfProfileIn( $fname );
1430
1431                 wfProfileIn( $fname.'-setup' );
1432                 static $tc = FALSE;
1433                 # the % is needed to support urlencoded titles as well
1434                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1435
1436                 $sk =& $this->mOptions->getSkin();
1437
1438                 #split the entire text string on occurences of [[
1439                 $a = explode( '[[', ' ' . $s );
1440                 #get the first element (all text up to first [[), and remove the space we added
1441                 $s = array_shift( $a );
1442                 $s = substr( $s, 1 );
1443
1444                 # Match a link having the form [[namespace:link|alternate]]trail
1445                 static $e1 = FALSE;
1446                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1447                 # Match cases where there is no "]]", which might still be images
1448                 static $e1_img = FALSE;
1449                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1450                 # Match the end of a line for a word that's not followed by whitespace,
1451                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1452                 $e2 = wfMsgForContent( 'linkprefix' );
1453
1454                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1455
1456                 if( is_null( $this->mTitle ) ) {
1457                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1458                 }
1459                 $nottalk = !$this->mTitle->isTalkPage();
1460
1461                 if ( $useLinkPrefixExtension ) {
1462                         if ( preg_match( $e2, $s, $m ) ) {
1463                                 $first_prefix = $m[2];
1464                         } else {
1465                                 $first_prefix = false;
1466                         }
1467                 } else {
1468                         $prefix = '';
1469                 }
1470
1471                 $selflink = $this->mTitle->getPrefixedText();
1472                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
1473                 $useSubpages = $this->areSubpagesAllowed();
1474                 wfProfileOut( $fname.'-setup' );
1475
1476                 # Loop for each link
1477                 for ($k = 0; isset( $a[$k] ); $k++) {
1478                         $line = $a[$k];
1479                         if ( $useLinkPrefixExtension ) {
1480                                 wfProfileIn( $fname.'-prefixhandling' );
1481                                 if ( preg_match( $e2, $s, $m ) ) {
1482                                         $prefix = $m[2];
1483                                         $s = $m[1];
1484                                 } else {
1485                                         $prefix='';
1486                                 }
1487                                 # first link
1488                                 if($first_prefix) {
1489                                         $prefix = $first_prefix;
1490                                         $first_prefix = false;
1491                                 }
1492                                 wfProfileOut( $fname.'-prefixhandling' );
1493                         }
1494
1495                         $might_be_img = false;
1496
1497                         wfProfileIn( "$fname-e1" );
1498                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1499                                 $text = $m[2];
1500                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1501                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1502                                 # the real problem is with the $e1 regex
1503                                 # See bug 1300.
1504                                 #
1505                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1506                                 # and no image is in sight. See bug 2095.
1507                                 #
1508                                 if( $text !== '' &&
1509                                         substr( $m[3], 0, 1 ) === ']' &&
1510                                         strpos($text, '[') !== false
1511                                 )
1512                                 {
1513                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1514                                         $m[3] = substr( $m[3], 1 );
1515                                 }
1516                                 # fix up urlencoded title texts
1517                                 if( strpos( $m[1], '%' ) !== false ) {
1518                                         # Should anchors '#' also be rejected?
1519                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1520                                 }
1521                                 $trail = $m[3];
1522                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1523                                 $might_be_img = true;
1524                                 $text = $m[2];
1525                                 if ( strpos( $m[1], '%' ) !== false ) {
1526                                        $m[1] = urldecode($m[1]);
1527                                 }
1528                                 $trail = "";
1529                         } else { # Invalid form; output directly
1530                                 $s .= $prefix . '[[' . $line ;
1531                                 wfProfileOut( "$fname-e1" );
1532                                 continue;
1533                         }
1534                         wfProfileOut( "$fname-e1" );
1535                         wfProfileIn( "$fname-misc" );
1536
1537                         # Don't allow internal links to pages containing
1538                         # PROTO: where PROTO is a valid URL protocol; these
1539                         # should be external links.
1540                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1541                                 $s .= $prefix . '[[' . $line ;
1542                                 continue;
1543                         }
1544
1545                         # Make subpage if necessary
1546                         if( $useSubpages ) {
1547                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1548                         } else {
1549                                 $link = $m[1];
1550                         }
1551
1552                         $noforce = (substr($m[1], 0, 1) != ':');
1553                         if (!$noforce) {
1554                                 # Strip off leading ':'
1555                                 $link = substr($link, 1);
1556                         }
1557
1558                         wfProfileOut( "$fname-misc" );
1559                         wfProfileIn( "$fname-title" );
1560                         $nt = Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1561                         if( !$nt ) {
1562                                 $s .= $prefix . '[[' . $line;
1563                                 wfProfileOut( "$fname-title" );
1564                                 continue;
1565                         }
1566
1567                         #check other language variants of the link
1568                         #if the article does not exist
1569                         if( $checkVariantLink
1570                             && $nt->getArticleID() == 0 ) {
1571                                 $wgContLang->findVariantLink($link, $nt);
1572                         }
1573
1574                         $ns = $nt->getNamespace();
1575                         $iw = $nt->getInterWiki();
1576                         wfProfileOut( "$fname-title" );
1577
1578                         if ($might_be_img) { # if this is actually an invalid link
1579                                 wfProfileIn( "$fname-might_be_img" );
1580                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1581                                         $found = false;
1582                                         while (isset ($a[$k+1]) ) {
1583                                                 #look at the next 'line' to see if we can close it there
1584                                                 $spliced = array_splice( $a, $k + 1, 1 );
1585                                                 $next_line = array_shift( $spliced );
1586                                                 $m = explode( ']]', $next_line, 3 );
1587                                                 if ( count( $m ) == 3 ) {
1588                                                         # the first ]] closes the inner link, the second the image
1589                                                         $found = true;
1590                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1591                                                         $trail = $m[2];
1592                                                         break;
1593                                                 } elseif ( count( $m ) == 2 ) {
1594                                                         #if there's exactly one ]] that's fine, we'll keep looking
1595                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1596                                                 } else {
1597                                                         #if $next_line is invalid too, we need look no further
1598                                                         $text .= '[[' . $next_line;
1599                                                         break;
1600                                                 }
1601                                         }
1602                                         if ( !$found ) {
1603                                                 # we couldn't find the end of this imageLink, so output it raw
1604                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1605                                                 $text = $this->replaceInternalLinks($text);
1606                                                 $s .= "{$prefix}[[$link|$text";
1607                                                 # note: no $trail, because without an end, there *is* no trail
1608                                                 wfProfileOut( "$fname-might_be_img" );
1609                                                 continue;
1610                                         }
1611                                 } else { #it's not an image, so output it raw
1612                                         $s .= "{$prefix}[[$link|$text";
1613                                         # note: no $trail, because without an end, there *is* no trail
1614                                         wfProfileOut( "$fname-might_be_img" );
1615                                         continue;
1616                                 }
1617                                 wfProfileOut( "$fname-might_be_img" );
1618                         }
1619
1620                         $wasblank = ( '' == $text );
1621                         if( $wasblank ) $text = $link;
1622
1623                         # Link not escaped by : , create the various objects
1624                         if( $noforce ) {
1625
1626                                 # Interwikis
1627                                 wfProfileIn( "$fname-interwiki" );
1628                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1629                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1630                                         $s = rtrim($s . "\n");
1631                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1632                                         wfProfileOut( "$fname-interwiki" );
1633                                         continue;
1634                                 }
1635                                 wfProfileOut( "$fname-interwiki" );
1636
1637                                 if ( $ns == NS_IMAGE ) {
1638                                         wfProfileIn( "$fname-image" );
1639                                         if ( !wfIsBadImage( $nt->getDBkey() ) ) {
1640                                                 # recursively parse links inside the image caption
1641                                                 # actually, this will parse them in any other parameters, too,
1642                                                 # but it might be hard to fix that, and it doesn't matter ATM
1643                                                 $text = $this->replaceExternalLinks($text);
1644                                                 $text = $this->replaceInternalLinks($text);
1645
1646                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1647                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1648                                                 $this->mOutput->addImage( $nt->getDBkey() );
1649
1650                                                 wfProfileOut( "$fname-image" );
1651                                                 continue;
1652                                         } else {
1653                                                 # We still need to record the image's presence on the page
1654                                                 $this->mOutput->addImage( $nt->getDBkey() );
1655                                         }
1656                                         wfProfileOut( "$fname-image" );
1657
1658                                 }
1659
1660                                 if ( $ns == NS_CATEGORY ) {
1661                                         wfProfileIn( "$fname-category" );
1662                                         $s = rtrim($s . "\n"); # bug 87
1663
1664                                         if ( $wasblank ) {
1665                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1666                                                         $sortkey = $this->mTitle->getText();
1667                                                 } else {
1668                                                         $sortkey = $this->mTitle->getPrefixedText();
1669                                                 }
1670                                         } else {
1671                                                 $sortkey = $text;
1672                                         }
1673                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1674                                         $sortkey = str_replace( "\n", '', $sortkey );
1675                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1676                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1677
1678                                         /**
1679                                          * Strip the whitespace Category links produce, see bug 87
1680                                          * @todo We might want to use trim($tmp, "\n") here.
1681                                          */
1682                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1683
1684                                         wfProfileOut( "$fname-category" );
1685                                         continue;
1686                                 }
1687                         }
1688
1689                         if( ( $nt->getPrefixedText() === $selflink ) &&
1690                             ( $nt->getFragment() === '' ) ) {
1691                                 # Self-links are handled specially; generally de-link and change to bold.
1692                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1693                                 continue;
1694                         }
1695
1696                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1697                         if( $ns == NS_MEDIA ) {
1698                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1699                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1700                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1701                                 $this->mOutput->addImage( $nt->getDBkey() );
1702                                 continue;
1703                         } elseif( $ns == NS_SPECIAL ) {
1704                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1705                                 continue;
1706                         } elseif( $ns == NS_IMAGE ) {
1707                                 $img = new Image( $nt );
1708                                 if( $img->exists() ) {
1709                                         // Force a blue link if the file exists; may be a remote
1710                                         // upload on the shared repository, and we want to see its
1711                                         // auto-generated page.
1712                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1713                                         continue;
1714                                 }
1715                         }
1716                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1717                 }
1718                 wfProfileOut( $fname );
1719                 return $s;
1720         }
1721
1722         /**
1723          * Make a link placeholder. The text returned can be later resolved to a real link with
1724          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1725          * parsing of interwiki links, and secondly to allow all existence checks and
1726          * article length checks (for stub links) to be bundled into a single query.
1727          *
1728          */
1729         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1730                 wfProfileIn( __METHOD__ );
1731                 if ( ! is_object($nt) ) {
1732                         # Fail gracefully
1733                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1734                 } else {
1735                         # Separate the link trail from the rest of the link
1736                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1737
1738                         if ( $nt->isExternal() ) {
1739                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1740                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1741                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1742                         } else {
1743                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1744                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1745                                 $this->mLinkHolders['queries'][] = $query;
1746                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1747                                 $this->mLinkHolders['titles'][] = $nt;
1748
1749                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1750                         }
1751                 }
1752                 wfProfileOut( __METHOD__ );
1753                 return $retVal;
1754         }
1755
1756         /**
1757          * Render a forced-blue link inline; protect against double expansion of
1758          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1759          * Since this little disaster has to split off the trail text to avoid
1760          * breaking URLs in the following text without breaking trails on the
1761          * wiki links, it's been made into a horrible function.
1762          *
1763          * @param Title $nt
1764          * @param string $text
1765          * @param string $query
1766          * @param string $trail
1767          * @param string $prefix
1768          * @return string HTML-wikitext mix oh yuck
1769          */
1770         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1771                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1772                 $sk =& $this->mOptions->getSkin();
1773                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1774                 return $this->armorLinks( $link ) . $trail;
1775         }
1776
1777         /**
1778          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1779          * going to go through further parsing steps before inline URL expansion.
1780          *
1781          * In particular this is important when using action=render, which causes
1782          * full URLs to be included.
1783          *
1784          * Oh man I hate our multi-layer parser!
1785          *
1786          * @param string more-or-less HTML
1787          * @return string less-or-more HTML with NOPARSE bits
1788          */
1789         function armorLinks( $text ) {
1790                 return preg_replace( "/\b(" . wfUrlProtocols() . ')/',
1791                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1792         }
1793
1794         /**
1795          * Return true if subpage links should be expanded on this page.
1796          * @return bool
1797          */
1798         function areSubpagesAllowed() {
1799                 # Some namespaces don't allow subpages
1800                 global $wgNamespacesWithSubpages;
1801                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1802         }
1803
1804         /**
1805          * Handle link to subpage if necessary
1806          * @param string $target the source of the link
1807          * @param string &$text the link text, modified as necessary
1808          * @return string the full name of the link
1809          * @private
1810          */
1811         function maybeDoSubpageLink($target, &$text) {
1812                 # Valid link forms:
1813                 # Foobar -- normal
1814                 # :Foobar -- override special treatment of prefix (images, language links)
1815                 # /Foobar -- convert to CurrentPage/Foobar
1816                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1817                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1818                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1819
1820                 $fname = 'Parser::maybeDoSubpageLink';
1821                 wfProfileIn( $fname );
1822                 $ret = $target; # default return value is no change
1823
1824                 # Some namespaces don't allow subpages,
1825                 # so only perform processing if subpages are allowed
1826                 if( $this->areSubpagesAllowed() ) {
1827                         # Look at the first character
1828                         if( $target != '' && $target{0} == '/' ) {
1829                                 # / at end means we don't want the slash to be shown
1830                                 if( substr( $target, -1, 1 ) == '/' ) {
1831                                         $target = substr( $target, 1, -1 );
1832                                         $noslash = $target;
1833                                 } else {
1834                                         $noslash = substr( $target, 1 );
1835                                 }
1836
1837                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1838                                 if( '' === $text ) {
1839                                         $text = $target;
1840                                 } # this might be changed for ugliness reasons
1841                         } else {
1842                                 # check for .. subpage backlinks
1843                                 $dotdotcount = 0;
1844                                 $nodotdot = $target;
1845                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1846                                         ++$dotdotcount;
1847                                         $nodotdot = substr( $nodotdot, 3 );
1848                                 }
1849                                 if($dotdotcount > 0) {
1850                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1851                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1852                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1853                                                 # / at the end means don't show full path
1854                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1855                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1856                                                         if( '' === $text ) {
1857                                                                 $text = $nodotdot;
1858                                                         }
1859                                                 }
1860                                                 $nodotdot = trim( $nodotdot );
1861                                                 if( $nodotdot != '' ) {
1862                                                         $ret .= '/' . $nodotdot;
1863                                                 }
1864                                         }
1865                                 }
1866                         }
1867                 }
1868
1869                 wfProfileOut( $fname );
1870                 return $ret;
1871         }
1872
1873         /**#@+
1874          * Used by doBlockLevels()
1875          * @private
1876          */
1877         /* private */ function closeParagraph() {
1878                 $result = '';
1879                 if ( '' != $this->mLastSection ) {
1880                         $result = '</' . $this->mLastSection  . ">\n";
1881                 }
1882                 $this->mInPre = false;
1883                 $this->mLastSection = '';
1884                 return $result;
1885         }
1886         # getCommon() returns the length of the longest common substring
1887         # of both arguments, starting at the beginning of both.
1888         #
1889         /* private */ function getCommon( $st1, $st2 ) {
1890                 $fl = strlen( $st1 );
1891                 $shorter = strlen( $st2 );
1892                 if ( $fl < $shorter ) { $shorter = $fl; }
1893
1894                 for ( $i = 0; $i < $shorter; ++$i ) {
1895                         if ( $st1{$i} != $st2{$i} ) { break; }
1896                 }
1897                 return $i;
1898         }
1899         # These next three functions open, continue, and close the list
1900         # element appropriate to the prefix character passed into them.
1901         #
1902         /* private */ function openList( $char ) {
1903                 $result = $this->closeParagraph();
1904
1905                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1906                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1907                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1908                 else if ( ';' == $char ) {
1909                         $result .= '<dl><dt>';
1910                         $this->mDTopen = true;
1911                 }
1912                 else { $result = '<!-- ERR 1 -->'; }
1913
1914                 return $result;
1915         }
1916
1917         /* private */ function nextItem( $char ) {
1918                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1919                 else if ( ':' == $char || ';' == $char ) {
1920                         $close = '</dd>';
1921                         if ( $this->mDTopen ) { $close = '</dt>'; }
1922                         if ( ';' == $char ) {
1923                                 $this->mDTopen = true;
1924                                 return $close . '<dt>';
1925                         } else {
1926                                 $this->mDTopen = false;
1927                                 return $close . '<dd>';
1928                         }
1929                 }
1930                 return '<!-- ERR 2 -->';
1931         }
1932
1933         /* private */ function closeList( $char ) {
1934                 if ( '*' == $char ) { $text = '</li></ul>'; }
1935                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1936                 else if ( ':' == $char ) {
1937                         if ( $this->mDTopen ) {
1938                                 $this->mDTopen = false;
1939                                 $text = '</dt></dl>';
1940                         } else {
1941                                 $text = '</dd></dl>';
1942                         }
1943                 }
1944                 else {  return '<!-- ERR 3 -->'; }
1945                 return $text."\n";
1946         }
1947         /**#@-*/
1948
1949         /**
1950          * Make lists from lines starting with ':', '*', '#', etc.
1951          *
1952          * @private
1953          * @return string the lists rendered as HTML
1954          */
1955         function doBlockLevels( $text, $linestart ) {
1956                 $fname = 'Parser::doBlockLevels';
1957                 wfProfileIn( $fname );
1958
1959                 # Parsing through the text line by line.  The main thing
1960                 # happening here is handling of block-level elements p, pre,
1961                 # and making lists from lines starting with * # : etc.
1962                 #
1963                 $textLines = explode( "\n", $text );
1964
1965                 $lastPrefix = $output = '';
1966                 $this->mDTopen = $inBlockElem = false;
1967                 $prefixLength = 0;
1968                 $paragraphStack = false;
1969
1970                 if ( !$linestart ) {
1971                         $output .= array_shift( $textLines );
1972                 }
1973                 foreach ( $textLines as $oLine ) {
1974                         $lastPrefixLength = strlen( $lastPrefix );
1975                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
1976                         $preOpenMatch = preg_match('/<pre/i', $oLine );
1977                         if ( !$this->mInPre ) {
1978                                 # Multiple prefixes may abut each other for nested lists.
1979                                 $prefixLength = strspn( $oLine, '*#:;' );
1980                                 $pref = substr( $oLine, 0, $prefixLength );
1981
1982                                 # eh?
1983                                 $pref2 = str_replace( ';', ':', $pref );
1984                                 $t = substr( $oLine, $prefixLength );
1985                                 $this->mInPre = !empty($preOpenMatch);
1986                         } else {
1987                                 # Don't interpret any other prefixes in preformatted text
1988                                 $prefixLength = 0;
1989                                 $pref = $pref2 = '';
1990                                 $t = $oLine;
1991                         }
1992
1993                         # List generation
1994                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
1995                                 # Same as the last item, so no need to deal with nesting or opening stuff
1996                                 $output .= $this->nextItem( substr( $pref, -1 ) );
1997                                 $paragraphStack = false;
1998
1999                                 if ( substr( $pref, -1 ) == ';') {
2000                                         # The one nasty exception: definition lists work like this:
2001                                         # ; title : definition text
2002                                         # So we check for : in the remainder text to split up the
2003                                         # title and definition, without b0rking links.
2004                                         $term = $t2 = '';
2005                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2006                                                 $t = $t2;
2007                                                 $output .= $term . $this->nextItem( ':' );
2008                                         }
2009                                 }
2010                         } elseif( $prefixLength || $lastPrefixLength ) {
2011                                 # Either open or close a level...
2012                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2013                                 $paragraphStack = false;
2014
2015                                 while( $commonPrefixLength < $lastPrefixLength ) {
2016                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2017                                         --$lastPrefixLength;
2018                                 }
2019                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2020                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2021                                 }
2022                                 while ( $prefixLength > $commonPrefixLength ) {
2023                                         $char = substr( $pref, $commonPrefixLength, 1 );
2024                                         $output .= $this->openList( $char );
2025
2026                                         if ( ';' == $char ) {
2027                                                 # FIXME: This is dupe of code above
2028                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2029                                                         $t = $t2;
2030                                                         $output .= $term . $this->nextItem( ':' );
2031                                                 }
2032                                         }
2033                                         ++$commonPrefixLength;
2034                                 }
2035                                 $lastPrefix = $pref2;
2036                         }
2037                         if( 0 == $prefixLength ) {
2038                                 wfProfileIn( "$fname-paragraph" );
2039                                 # No prefix (not in list)--go to paragraph mode
2040                                 // XXX: use a stack for nestable elements like span, table and div
2041                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/center|<\\/tr|<\\/td|<\\/th)/iS', $t );
2042                                 $closematch = preg_match(
2043                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2044                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<center)/iS', $t );
2045                                 if ( $openmatch or $closematch ) {
2046                                         $paragraphStack = false;
2047                                         # TODO bug 5718: paragraph closed
2048                                         $output .= $this->closeParagraph();
2049                                         if ( $preOpenMatch and !$preCloseMatch ) {
2050                                                 $this->mInPre = true;
2051                                         }
2052                                         if ( $closematch ) {
2053                                                 $inBlockElem = false;
2054                                         } else {
2055                                                 $inBlockElem = true;
2056                                         }
2057                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2058                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2059                                                 // pre
2060                                                 if ($this->mLastSection != 'pre') {
2061                                                         $paragraphStack = false;
2062                                                         $output .= $this->closeParagraph().'<pre>';
2063                                                         $this->mLastSection = 'pre';
2064                                                 }
2065                                                 $t = substr( $t, 1 );
2066                                         } else {
2067                                                 // paragraph
2068                                                 if ( '' == trim($t) ) {
2069                                                         if ( $paragraphStack ) {
2070                                                                 $output .= $paragraphStack.'<br />';
2071                                                                 $paragraphStack = false;
2072                                                                 $this->mLastSection = 'p';
2073                                                         } else {
2074                                                                 if ($this->mLastSection != 'p' ) {
2075                                                                         $output .= $this->closeParagraph();
2076                                                                         $this->mLastSection = '';
2077                                                                         $paragraphStack = '<p>';
2078                                                                 } else {
2079                                                                         $paragraphStack = '</p><p>';
2080                                                                 }
2081                                                         }
2082                                                 } else {
2083                                                         if ( $paragraphStack ) {
2084                                                                 $output .= $paragraphStack;
2085                                                                 $paragraphStack = false;
2086                                                                 $this->mLastSection = 'p';
2087                                                         } else if ($this->mLastSection != 'p') {
2088                                                                 $output .= $this->closeParagraph().'<p>';
2089                                                                 $this->mLastSection = 'p';
2090                                                         }
2091                                                 }
2092                                         }
2093                                 }
2094                                 wfProfileOut( "$fname-paragraph" );
2095                         }
2096                         // somewhere above we forget to get out of pre block (bug 785)
2097                         if($preCloseMatch && $this->mInPre) {
2098                                 $this->mInPre = false;
2099                         }
2100                         if ($paragraphStack === false) {
2101                                 $output .= $t."\n";
2102                         }
2103                 }
2104                 while ( $prefixLength ) {
2105                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2106                         --$prefixLength;
2107                 }
2108                 if ( '' != $this->mLastSection ) {
2109                         $output .= '</' . $this->mLastSection . '>';
2110                         $this->mLastSection = '';
2111                 }
2112
2113                 wfProfileOut( $fname );
2114                 return $output;
2115         }
2116
2117         /**
2118          * Split up a string on ':', ignoring any occurences inside tags
2119          * to prevent illegal overlapping.
2120          * @param string $str the string to split
2121          * @param string &$before set to everything before the ':'
2122          * @param string &$after set to everything after the ':'
2123          * return string the position of the ':', or false if none found
2124          */
2125         function findColonNoLinks($str, &$before, &$after) {
2126                 $fname = 'Parser::findColonNoLinks';
2127                 wfProfileIn( $fname );
2128
2129                 $pos = strpos( $str, ':' );
2130                 if( $pos === false ) {
2131                         // Nothing to find!
2132                         wfProfileOut( $fname );
2133                         return false;
2134                 }
2135
2136                 $lt = strpos( $str, '<' );
2137                 if( $lt === false || $lt > $pos ) {
2138                         // Easy; no tag nesting to worry about
2139                         $before = substr( $str, 0, $pos );
2140                         $after = substr( $str, $pos+1 );
2141                         wfProfileOut( $fname );
2142                         return $pos;
2143                 }
2144
2145                 // Ugly state machine to walk through avoiding tags.
2146                 $state = MW_COLON_STATE_TEXT;
2147                 $stack = 0;
2148                 $len = strlen( $str );
2149                 for( $i = 0; $i < $len; $i++ ) {
2150                         $c = $str{$i};
2151
2152                         switch( $state ) {
2153                         // (Using the number is a performance hack for common cases)
2154                         case 0: // MW_COLON_STATE_TEXT:
2155                                 switch( $c ) {
2156                                 case "<":
2157                                         // Could be either a <start> tag or an </end> tag
2158                                         $state = MW_COLON_STATE_TAGSTART;
2159                                         break;
2160                                 case ":":
2161                                         if( $stack == 0 ) {
2162                                                 // We found it!
2163                                                 $before = substr( $str, 0, $i );
2164                                                 $after = substr( $str, $i + 1 );
2165                                                 wfProfileOut( $fname );
2166                                                 return $i;
2167                                         }
2168                                         // Embedded in a tag; don't break it.
2169                                         break;
2170                                 default:
2171                                         // Skip ahead looking for something interesting
2172                                         $colon = strpos( $str, ':', $i );
2173                                         if( $colon === false ) {
2174                                                 // Nothing else interesting
2175                                                 wfProfileOut( $fname );
2176                                                 return false;
2177                                         }
2178                                         $lt = strpos( $str, '<', $i );
2179                                         if( $stack === 0 ) {
2180                                                 if( $lt === false || $colon < $lt ) {
2181                                                         // We found it!
2182                                                         $before = substr( $str, 0, $colon );
2183                                                         $after = substr( $str, $colon + 1 );
2184                                                         wfProfileOut( $fname );
2185                                                         return $i;
2186                                                 }
2187                                         }
2188                                         if( $lt === false ) {
2189                                                 // Nothing else interesting to find; abort!
2190                                                 // We're nested, but there's no close tags left. Abort!
2191                                                 break 2;
2192                                         }
2193                                         // Skip ahead to next tag start
2194                                         $i = $lt;
2195                                         $state = MW_COLON_STATE_TAGSTART;
2196                                 }
2197                                 break;
2198                         case 1: // MW_COLON_STATE_TAG:
2199                                 // In a <tag>
2200                                 switch( $c ) {
2201                                 case ">":
2202                                         $stack++;
2203                                         $state = MW_COLON_STATE_TEXT;
2204                                         break;
2205                                 case "/":
2206                                         // Slash may be followed by >?
2207                                         $state = MW_COLON_STATE_TAGSLASH;
2208                                         break;
2209                                 default:
2210                                         // ignore
2211                                 }
2212                                 break;
2213                         case 2: // MW_COLON_STATE_TAGSTART:
2214                                 switch( $c ) {
2215                                 case "/":
2216                                         $state = MW_COLON_STATE_CLOSETAG;
2217                                         break;
2218                                 case "!":
2219                                         $state = MW_COLON_STATE_COMMENT;
2220                                         break;
2221                                 case ">":
2222                                         // Illegal early close? This shouldn't happen D:
2223                                         $state = MW_COLON_STATE_TEXT;
2224                                         break;
2225                                 default:
2226                                         $state = MW_COLON_STATE_TAG;
2227                                 }
2228                                 break;
2229                         case 3: // MW_COLON_STATE_CLOSETAG:
2230                                 // In a </tag>
2231                                 if( $c == ">" ) {
2232                                         $stack--;
2233                                         if( $stack < 0 ) {
2234                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2235                                                 wfProfileOut( $fname );
2236                                                 return false;
2237                                         }
2238                                         $state = MW_COLON_STATE_TEXT;
2239                                 }
2240                                 break;
2241                         case MW_COLON_STATE_TAGSLASH:
2242                                 if( $c == ">" ) {
2243                                         // Yes, a self-closed tag <blah/>
2244                                         $state = MW_COLON_STATE_TEXT;
2245                                 } else {
2246                                         // Probably we're jumping the gun, and this is an attribute
2247                                         $state = MW_COLON_STATE_TAG;
2248                                 }
2249                                 break;
2250                         case 5: // MW_COLON_STATE_COMMENT:
2251                                 if( $c == "-" ) {
2252                                         $state = MW_COLON_STATE_COMMENTDASH;
2253                                 }
2254                                 break;
2255                         case MW_COLON_STATE_COMMENTDASH:
2256                                 if( $c == "-" ) {
2257                                         $state = MW_COLON_STATE_COMMENTDASHDASH;
2258                                 } else {
2259                                         $state = MW_COLON_STATE_COMMENT;
2260                                 }
2261                                 break;
2262                         case MW_COLON_STATE_COMMENTDASHDASH:
2263                                 if( $c == ">" ) {
2264                                         $state = MW_COLON_STATE_TEXT;
2265                                 } else {
2266                                         $state = MW_COLON_STATE_COMMENT;
2267                                 }
2268                                 break;
2269                         default:
2270                                 throw new MWException( "State machine error in $fname" );
2271                         }
2272                 }
2273                 if( $stack > 0 ) {
2274                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2275                         return false;
2276                 }
2277                 wfProfileOut( $fname );
2278                 return false;
2279         }
2280
2281         /**
2282          * Return value of a magic variable (like PAGENAME)
2283          *
2284          * @private
2285          */
2286         function getVariableValue( $index ) {
2287                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2288
2289                 /**
2290                  * Some of these require message or data lookups and can be
2291                  * expensive to check many times.
2292                  */
2293                 static $varCache = array();
2294                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) )
2295                         if ( isset( $varCache[$index] ) )
2296                                 return $varCache[$index];
2297
2298                 $ts = time();
2299                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2300
2301                 switch ( $index ) {
2302                         case 'currentmonth':
2303                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
2304                         case 'currentmonthname':
2305                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2306                         case 'currentmonthnamegen':
2307                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2308                         case 'currentmonthabbrev':
2309                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2310                         case 'currentday':
2311                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2312                         case 'currentday2':
2313                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2314                         case 'pagename':
2315                                 return $this->mTitle->getText();
2316                         case 'pagenamee':
2317                                 return $this->mTitle->getPartialURL();
2318                         case 'fullpagename':
2319                                 return $this->mTitle->getPrefixedText();
2320                         case 'fullpagenamee':
2321                                 return $this->mTitle->getPrefixedURL();
2322                         case 'subpagename':
2323                                 return $this->mTitle->getSubpageText();
2324                         case 'subpagenamee':
2325                                 return $this->mTitle->getSubpageUrlForm();
2326                         case 'basepagename':
2327                                 return $this->mTitle->getBaseText();
2328                         case 'basepagenamee':
2329                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2330                         case 'talkpagename':
2331                                 if( $this->mTitle->canTalk() ) {
2332                                         $talkPage = $this->mTitle->getTalkPage();
2333                                         return $talkPage->getPrefixedText();
2334                                 } else {
2335                                         return '';
2336                                 }
2337                         case 'talkpagenamee':
2338                                 if( $this->mTitle->canTalk() ) {
2339                                         $talkPage = $this->mTitle->getTalkPage();
2340                                         return $talkPage->getPrefixedUrl();
2341                                 } else {
2342                                         return '';
2343                                 }
2344                         case 'subjectpagename':
2345                                 $subjPage = $this->mTitle->getSubjectPage();
2346                                 return $subjPage->getPrefixedText();
2347                         case 'subjectpagenamee':
2348                                 $subjPage = $this->mTitle->getSubjectPage();
2349                                 return $subjPage->getPrefixedUrl();
2350                         case 'revisionid':
2351                                 return $this->mRevisionId;
2352                         case 'namespace':
2353                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2354                         case 'namespacee':
2355                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2356                         case 'talkspace':
2357                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2358                         case 'talkspacee':
2359                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2360                         case 'subjectspace':
2361                                 return $this->mTitle->getSubjectNsText();
2362                         case 'subjectspacee':
2363                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2364                         case 'currentdayname':
2365                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2366                         case 'currentyear':
2367                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2368                         case 'currenttime':
2369                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2370                         case 'currentweek':
2371                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2372                                 // int to remove the padding
2373                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2374                         case 'currentdow':
2375                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2376                         case 'numberofarticles':
2377                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
2378                         case 'numberoffiles':
2379                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
2380                         case 'numberofusers':
2381                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfUsers() );
2382                         case 'numberofpages':
2383                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfPages() );
2384                         case 'numberofadmins':
2385                                 return $varCache[$index]  = $wgContLang->formatNum( wfNumberOfAdmins() );
2386                         case 'currenttimestamp':
2387                                 return $varCache[$index] = wfTimestampNow();
2388                         case 'currentversion':
2389                                 global $wgVersion;
2390                                 return $wgVersion;
2391                         case 'sitename':
2392                                 return $wgSitename;
2393                         case 'server':
2394                                 return $wgServer;
2395                         case 'servername':
2396                                 return $wgServerName;
2397                         case 'scriptpath':
2398                                 return $wgScriptPath;
2399                         case 'directionmark':
2400                                 return $wgContLang->getDirMark();
2401                         case 'contentlanguage':
2402                                 global $wgContLanguageCode;
2403                                 return $wgContLanguageCode;
2404                         default:
2405                                 $ret = null;
2406                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2407                                         return $ret;
2408                                 else
2409                                         return null;
2410                 }
2411         }
2412
2413         /**
2414          * initialise the magic variables (like CURRENTMONTHNAME)
2415          *
2416          * @private
2417          */
2418         function initialiseVariables() {
2419                 $fname = 'Parser::initialiseVariables';
2420                 wfProfileIn( $fname );
2421                 $variableIDs = MagicWord::getVariableIDs();
2422
2423                 $this->mVariables = array();
2424                 foreach ( $variableIDs as $id ) {
2425                         $mw =& MagicWord::get( $id );
2426                         $mw->addToArray( $this->mVariables, $id );
2427                 }
2428                 wfProfileOut( $fname );
2429         }
2430
2431         /**
2432          * parse any parentheses in format ((title|part|part))
2433          * and call callbacks to get a replacement text for any found piece
2434          *
2435          * @param string $text The text to parse
2436          * @param array $callbacks rules in form:
2437          *     '{' => array(                            # opening parentheses
2438          *                                      'end' => '}',   # closing parentheses
2439          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2440          *                                                                3 => callback         # replacement callback to call if {{{..}}} is found
2441          *                                                                )
2442          *                                      )
2443          *                                      'min' => 2,     # Minimum parenthesis count in cb
2444          *                                      'max' => 3,     # Maximum parenthesis count in cb
2445          * @private
2446          */
2447         function replace_callback ($text, $callbacks) {
2448                 wfProfileIn( __METHOD__ );
2449                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2450                 $lastOpeningBrace = -1;                 # last not closed parentheses
2451
2452                 $validOpeningBraces = implode( '', array_keys( $callbacks ) );
2453
2454                 $i = 0;
2455                 while ( $i < strlen( $text ) ) {
2456                         # Find next opening brace, closing brace or pipe
2457                         if ( $lastOpeningBrace == -1 ) {
2458                                 $currentClosing = '';
2459                                 $search = $validOpeningBraces;
2460                         } else {
2461                                 $currentClosing = $openingBraceStack[$lastOpeningBrace]['braceEnd'];
2462                                 $search = $validOpeningBraces . '|' . $currentClosing;
2463                         }
2464                         $rule = null;
2465                         $i += strcspn( $text, $search, $i );
2466                         if ( $i < strlen( $text ) ) {
2467                                 if ( $text[$i] == '|' ) {
2468                                         $found = 'pipe';
2469                                 } elseif ( $text[$i] == $currentClosing ) {
2470                                         $found = 'close';
2471                                 } else {
2472                                         $found = 'open';
2473                                         $rule = $callbacks[$text[$i]];
2474                                 }
2475                         } else {
2476                                 # All done
2477                                 break;
2478                         }
2479
2480                         if ( $found == 'open' ) {
2481                                 # found opening brace, let's add it to parentheses stack
2482                                 $piece = array('brace' => $text[$i],
2483                                                            'braceEnd' => $rule['end'],
2484                                                            'title' => '',
2485                                                            'parts' => null);
2486
2487                                 # count opening brace characters
2488                                 $piece['count'] = strspn( $text, $piece['brace'], $i );
2489                                 $piece['startAt'] = $piece['partStart'] = $i + $piece['count'];
2490                                 $i += $piece['count'];
2491
2492                                 # we need to add to stack only if opening brace count is enough for one of the rules
2493                                 if ( $piece['count'] >= $rule['min'] ) {
2494                                         $lastOpeningBrace ++;
2495                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2496                                 }
2497                         } elseif ( $found == 'close' ) {
2498                                 # lets check if it is enough characters for closing brace
2499                                 $maxCount = $openingBraceStack[$lastOpeningBrace]['count'];
2500                                 $count = strspn( $text, $text[$i], $i, $maxCount );
2501
2502                                 # check for maximum matching characters (if there are 5 closing
2503                                 # characters, we will probably need only 3 - depending on the rules)
2504                                 $matchingCount = 0;
2505                                 $matchingCallback = null;
2506                                 $cbType = $callbacks[$openingBraceStack[$lastOpeningBrace]['brace']];
2507                                 if ( $count > $cbType['max'] ) {
2508                                         # The specified maximum exists in the callback array, unless the caller
2509                                         # has made an error
2510                                         $matchingCount = $cbType['max'];
2511                                 } else {
2512                                         # Count is less than the maximum
2513                                         # Skip any gaps in the callback array to find the true largest match
2514                                         # Need to use array_key_exists not isset because the callback can be null
2515                                         $matchingCount = $count;
2516                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $cbType['cb'] ) ) {
2517                                                 --$matchingCount;
2518                                         }
2519                                 }
2520
2521                                 if ($matchingCount <= 0) {
2522                                         $i += $count;
2523                                         continue;
2524                                 }
2525                                 $matchingCallback = $cbType['cb'][$matchingCount];
2526
2527                                 # let's set a title or last part (if '|' was found)
2528                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2529                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2530                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2531                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2532                                 } else {
2533                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2534                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2535                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2536                                 }
2537
2538                                 $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2539                                 $pieceEnd = $i + $matchingCount;
2540
2541                                 if( is_callable( $matchingCallback ) ) {
2542                                         $cbArgs = array (
2543                                                                          'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2544                                                                          'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2545                                                                          'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2546                                                                          'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == "\n")),
2547                                                                          );
2548                                         # finally we can call a user callback and replace piece of text
2549                                         $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2550                                         $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2551                                         $i = $pieceStart + strlen($replaceWith);
2552                                 } else {
2553                                         # null value for callback means that parentheses should be parsed, but not replaced
2554                                         $i += $matchingCount;
2555                                 }
2556
2557                                 # reset last opening parentheses, but keep it in case there are unused characters
2558                                 $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2559                                                            'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2560                                                            'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2561                                                            'title' => '',
2562                                                            'parts' => null,
2563                                                            'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2564                                 $openingBraceStack[$lastOpeningBrace--] = null;
2565
2566                                 if ($matchingCount < $piece['count']) {
2567                                         $piece['count'] -= $matchingCount;
2568                                         $piece['startAt'] -= $matchingCount;
2569                                         $piece['partStart'] = $piece['startAt'];
2570                                         # do we still qualify for any callback with remaining count?
2571                                         $currentCbList = $callbacks[$piece['brace']]['cb'];
2572                                         while ( $piece['count'] ) {
2573                                                 if ( array_key_exists( $piece['count'], $currentCbList ) ) {
2574                                                         $lastOpeningBrace++;
2575                                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2576                                                         break;
2577                                                 }
2578                                                 --$piece['count'];
2579                                         }
2580                                 }
2581                         } elseif ( $found == 'pipe' ) {
2582                                 # lets set a title if it is a first separator, or next part otherwise
2583                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2584                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2585                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2586                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2587                                         $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2588                                 } else {
2589                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2590                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2591                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2592                                 }
2593                                 $openingBraceStack[$lastOpeningBrace]['partStart'] = ++$i;
2594                         }
2595                 }
2596
2597                 wfProfileOut( __METHOD__ );
2598                 return $text;
2599         }
2600
2601         /**
2602          * Replace magic variables, templates, and template arguments
2603          * with the appropriate text. Templates are substituted recursively,
2604          * taking care to avoid infinite loops.
2605          *
2606          * Note that the substitution depends on value of $mOutputType:
2607          *  OT_WIKI: only {{subst:}} templates
2608          *  OT_MSG: only magic variables
2609          *  OT_HTML: all templates and magic variables
2610          *
2611          * @param string $tex The text to transform
2612          * @param array $args Key-value pairs representing template parameters to substitute
2613          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2614          * @private
2615          */
2616         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2617                 # Prevent too big inclusions
2618                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2619                         return $text;
2620                 }
2621
2622                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2623                 wfProfileIn( $fname );
2624
2625                 # This function is called recursively. To keep track of arguments we need a stack:
2626                 array_push( $this->mArgStack, $args );
2627
2628                 $braceCallbacks = array();
2629                 if ( !$argsOnly ) {
2630                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2631                 }
2632                 if ( $this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI ) {
2633                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2634                 }
2635                 if ( $braceCallbacks ) {
2636                         $callbacks = array(
2637                                 '{' => array(
2638                                         'end' => '}',
2639                                         'cb' => $braceCallbacks,
2640                                         'min' => $argsOnly ? 3 : 2,
2641                                         'max' => isset( $braceCallbacks[3] ) ? 3 : 2,
2642                                 ),
2643                                 '[' => array(
2644                                         'end' => ']',
2645                                         'cb' => array(2=>null),
2646                                         'min' => 2,
2647                                         'max' => 2,
2648                                 )
2649                         );
2650                         $text = $this->replace_callback ($text, $callbacks);
2651
2652                         array_pop( $this->mArgStack );
2653                 }
2654                 wfProfileOut( $fname );
2655                 return $text;
2656         }
2657
2658         /**
2659          * Replace magic variables
2660          * @private
2661          */
2662         function variableSubstitution( $matches ) {
2663                 $fname = 'Parser::variableSubstitution';
2664                 $varname = $matches[1];
2665                 wfProfileIn( $fname );
2666                 $skip = false;
2667                 if ( $this->mOutputType == OT_WIKI ) {
2668                         # Do only magic variables prefixed by SUBST
2669                         $mwSubst =& MagicWord::get( 'subst' );
2670                         if (!$mwSubst->matchStartAndRemove( $varname ))
2671                                 $skip = true;
2672                         # Note that if we don't substitute the variable below,
2673                         # we don't remove the {{subst:}} magic word, in case
2674                         # it is a template rather than a magic variable.
2675                 }
2676                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2677                         $id = $this->mVariables[$varname];
2678                         $text = $this->getVariableValue( $id );
2679                         $this->mOutput->mContainsOldMagic = true;
2680                 } else {
2681                         $text = $matches[0];
2682                 }
2683                 wfProfileOut( $fname );
2684                 return $text;
2685         }
2686
2687         # Split template arguments
2688         function getTemplateArgs( $argsString ) {
2689                 if ( $argsString === '' ) {
2690                         return array();
2691                 }
2692
2693                 $args = explode( '|', substr( $argsString, 1 ) );
2694
2695                 # If any of the arguments contains a '[[' but no ']]', it needs to be
2696                 # merged with the next arg because the '|' character between belongs
2697                 # to the link syntax and not the template parameter syntax.
2698                 $argc = count($args);
2699
2700                 for ( $i = 0; $i < $argc-1; $i++ ) {
2701                         if ( substr_count ( $args[$i], '[[' ) != substr_count ( $args[$i], ']]' ) ) {
2702                                 $args[$i] .= '|'.$args[$i+1];
2703                                 array_splice($args, $i+1, 1);
2704                                 $i--;
2705                                 $argc--;
2706                         }
2707                 }
2708
2709                 return $args;
2710         }
2711
2712         /**
2713          * Return the text of a template, after recursively
2714          * replacing any variables or templates within the template.
2715          *
2716          * @param array $piece The parts of the template
2717          *  $piece['text']: matched text
2718          *  $piece['title']: the title, i.e. the part before the |
2719          *  $piece['parts']: the parameter array
2720          * @return string the text of the template
2721          * @private
2722          */
2723         function braceSubstitution( $piece ) {
2724                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $action;
2725                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2726                 wfProfileIn( $fname );
2727                 wfProfileIn( __METHOD__.'-setup' );
2728
2729                 # Flags
2730                 $found = false;             # $text has been filled
2731                 $nowiki = false;            # wiki markup in $text should be escaped
2732                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2733                 $noargs = false;            # Don't replace triple-brace arguments in $text
2734                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2735                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2736                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2737
2738                 # Title object, where $text came from
2739                 $title = NULL;
2740
2741                 $linestart = '';
2742
2743                 # $part1 is the bit before the first |, and must contain only title characters
2744                 # $args is a list of arguments, starting from index 0, not including $part1
2745
2746                 $part1 = $piece['title'];
2747                 # If the third subpattern matched anything, it will start with |
2748
2749                 if (null == $piece['parts']) {
2750                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2751                         if ($replaceWith != $piece['text']) {
2752                                 $text = $replaceWith;
2753                                 $found = true;
2754                                 $noparse = true;
2755                                 $noargs = true;
2756                         }
2757                 }
2758
2759                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2760                 $argc = count( $args );
2761                 wfProfileOut( __METHOD__.'-setup' );
2762
2763                 # SUBST
2764                 wfProfileIn( __METHOD__.'-modifiers' );
2765                 if ( !$found ) {
2766                         $mwSubst =& MagicWord::get( 'subst' );
2767                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor ($this->mOutputType == OT_WIKI) ) {
2768                                 # One of two possibilities is true:
2769                                 # 1) Found SUBST but not in the PST phase
2770                                 # 2) Didn't find SUBST and in the PST phase
2771                                 # In either case, return without further processing
2772                                 $text = $piece['text'];
2773                                 $found = true;
2774                                 $noparse = true;
2775                                 $noargs = true;
2776                         }
2777                 }
2778
2779                 # MSG, MSGNW, INT and RAW
2780                 if ( !$found ) {
2781                         # Check for MSGNW:
2782                         $mwMsgnw =& MagicWord::get( 'msgnw' );
2783                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2784                                 $nowiki = true;
2785                         } else {
2786                                 # Remove obsolete MSG:
2787                                 $mwMsg =& MagicWord::get( 'msg' );
2788                                 $mwMsg->matchStartAndRemove( $part1 );
2789                         }
2790
2791                         # Check for RAW:
2792                         $mwRaw =& MagicWord::get( 'raw' );
2793                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2794                                 $forceRawInterwiki = true;
2795                         }
2796
2797                         # Check if it is an internal message
2798                         $mwInt =& MagicWord::get( 'int' );
2799                         if ( $mwInt->matchStartAndRemove( $part1 ) ) {
2800                                 $text = $linestart . wfMsgReal( $part1, $args, true );
2801                                 $found = true;
2802                         }
2803                 }
2804                 wfProfileOut( __METHOD__.'-modifiers' );
2805
2806                 # Parser functions
2807                 if ( !$found ) {
2808                         wfProfileIn( __METHOD__ . '-pfunc' );
2809
2810                         $colonPos = strpos( $part1, ':' );
2811                         if ( $colonPos !== false ) {
2812                                 # Case sensitive functions
2813                                 $function = substr( $part1, 0, $colonPos );
2814                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2815                                         $function = $this->mFunctionSynonyms[1][$function];
2816                                 } else {
2817                                         # Case insensitive functions
2818                                         $function = strtolower( $function );
2819                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2820                                                 $function = $this->mFunctionSynonyms[0][$function];
2821                                         } else {
2822                                                 $function = false;
2823                                         }
2824                                 }
2825                                 if ( $function ) {
2826                                         $funcArgs = array_map( 'trim', $args );
2827                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
2828                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
2829                                         $found = true;
2830
2831                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
2832                                         //$noargs = true;
2833                                         //$noparse = true;
2834
2835                                         if ( is_array( $result ) ) {
2836                                                 if ( isset( $result[0] ) ) {
2837                                                         $text = $linestart . $result[0];
2838                                                         unset( $result[0] );
2839                                                 }
2840
2841                                                 // Extract flags into the local scope
2842                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
2843                                                 extract( $result );
2844                                         } else {
2845                                                 $text = $linestart . $result;
2846                                         }
2847                                 }
2848                         }
2849                         wfProfileOut( __METHOD__ . '-pfunc' );
2850                 }
2851
2852                 # Template table test
2853
2854                 # Did we encounter this template already? If yes, it is in the cache
2855                 # and we need to check for loops.
2856                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
2857                         $found = true;
2858
2859                         # Infinite loop test
2860                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2861                                 $noparse = true;
2862                                 $noargs = true;
2863                                 $found = true;
2864                                 $text = $linestart .
2865                                         '{{' . $part1 . '}}' .
2866                                         '<!-- WARNING: template loop detected -->';
2867                                 wfDebug( __METHOD__.": template loop broken at '$part1'\n" );
2868                         } else {
2869                                 # set $text to cached message.
2870                                 $text = $linestart . $this->mTemplates[$piece['title']];
2871                         }
2872                 }
2873
2874                 # Load from database
2875                 $lastPathLevel = $this->mTemplatePath;
2876                 if ( !$found ) {
2877                         wfProfileIn( __METHOD__ . '-loadtpl' );
2878                         $ns = NS_TEMPLATE;
2879                         # declaring $subpage directly in the function call
2880                         # does not work correctly with references and breaks
2881                         # {{/subpage}}-style inclusions
2882                         $subpage = '';
2883                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2884                         if ($subpage !== '') {
2885                                 $ns = $this->mTitle->getNamespace();
2886                         }
2887                         $title = Title::newFromText( $part1, $ns );
2888
2889
2890                         if ( !is_null( $title ) ) {
2891                                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
2892                                 # Check for language variants if the template is not found
2893                                 if($checkVariantLink && $title->getArticleID() == 0){
2894                                         $wgContLang->findVariantLink($part1, $title);
2895                                 }
2896
2897                                 if ( !$title->isExternal() ) {
2898                                         if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->mOutputType != OT_WIKI ) {
2899                                                 $text = SpecialPage::capturePath( $title );
2900                                                 if ( is_string( $text ) ) {
2901                                                         $found = true;
2902                                                         $noparse = true;
2903                                                         $noargs = true;
2904                                                         $isHTML = true;
2905                                                         $this->disableCache();
2906                                                 }
2907                                         } else {
2908                                                 $articleContent = $this->fetchTemplate( $title );
2909                                                 if ( $articleContent !== false ) {
2910                                                         $found = true;
2911                                                         $text = $articleContent;
2912                                                         $replaceHeadings = true;
2913                                                 }
2914                                         }
2915
2916                                         # If the title is valid but undisplayable, make a link to it
2917                                         if ( $this->mOutputType == OT_HTML && !$found ) {
2918                                                 $text = '[['.$title->getPrefixedText().']]';
2919                                                 $found = true;
2920                                         }
2921                                 } elseif ( $title->isTrans() ) {
2922                                         // Interwiki transclusion
2923                                         if ( $this->mOutputType == OT_HTML && !$forceRawInterwiki ) {
2924                                                 $text = $this->interwikiTransclude( $title, 'render' );
2925                                                 $isHTML = true;
2926                                                 $noparse = true;
2927                                         } else {
2928                                                 $text = $this->interwikiTransclude( $title, 'raw' );
2929                                                 $replaceHeadings = true;
2930                                         }
2931                                         $found = true;
2932                                 }
2933
2934                                 # Template cache array insertion
2935                                 # Use the original $piece['title'] not the mangled $part1, so that
2936                                 # modifiers such as RAW: produce separate cache entries
2937                                 if( $found ) {
2938                                         if( $isHTML ) {
2939                                                 // A special page; don't store it in the template cache.
2940                                         } else {
2941                                                 $this->mTemplates[$piece['title']] = $text;
2942                                         }
2943                                         $text = $linestart . $text;
2944                                 }
2945                         }
2946                         wfProfileOut( __METHOD__ . '-loadtpl' );
2947                 }
2948
2949                 if ( $found && !$this->incrementIncludeSize( 'pre-expand', strlen( $text ) ) ) {
2950                         # Error, oversize inclusion
2951                         $text = $linestart .
2952                                 '{{' . $part1 . '}}' .
2953                                 '<!-- WARNING: template omitted, pre-expand include size too large -->';
2954                         $noparse = true;
2955                         $noargs = true;
2956                 }
2957
2958                 # Recursive parsing, escaping and link table handling
2959                 # Only for HTML output
2960                 if ( $nowiki && $found && $this->mOutputType == OT_HTML ) {
2961                         $text = wfEscapeWikiText( $text );
2962                 } elseif ( ($this->mOutputType == OT_HTML || $this->mOutputType == OT_WIKI) && $found ) {
2963                         if ( $noargs ) {
2964                                 $assocArgs = array();
2965                         } else {
2966                                 # Clean up argument array
2967                                 $assocArgs = array();
2968                                 $index = 1;
2969                                 foreach( $args as $arg ) {
2970                                         $eqpos = strpos( $arg, '=' );
2971                                         if ( $eqpos === false ) {
2972                                                 $assocArgs[$index++] = $arg;
2973                                         } else {
2974                                                 $name = trim( substr( $arg, 0, $eqpos ) );
2975                                                 $value = trim( substr( $arg, $eqpos+1 ) );
2976                                                 if ( $value === false ) {
2977                                                         $value = '';
2978                                                 }
2979                                                 if ( $name !== false ) {
2980                                                         $assocArgs[$name] = $value;
2981                                                 }
2982                                         }
2983                                 }
2984
2985                                 # Add a new element to the templace recursion path
2986                                 $this->mTemplatePath[$part1] = 1;
2987                         }
2988
2989                         if ( !$noparse ) {
2990                                 # If there are any <onlyinclude> tags, only include them
2991                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
2992                                         preg_match_all( '/<onlyinclude>(.*?)\n?<\/onlyinclude>/s', $text, $m );
2993                                         $text = '';
2994                                         foreach ($m[1] as $piece)
2995                                                 $text .= $piece;
2996                                 }
2997                                 # Remove <noinclude> sections and <includeonly> tags
2998                                 $text = preg_replace( '/<noinclude>.*?<\/noinclude>/s', '', $text );
2999                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
3000
3001                                 if( $this->mOutputType == OT_HTML ) {
3002                                         # Strip <nowiki>, <pre>, etc.
3003                                         $text = $this->strip( $text, $this->mStripState );
3004                                         $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
3005                                 }
3006                                 $text = $this->replaceVariables( $text, $assocArgs );
3007
3008                                 # If the template begins with a table or block-level
3009                                 # element, it should be treated as beginning a new line.
3010                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) {
3011                                         $text = "\n" . $text;
3012                                 }
3013                         } elseif ( !$noargs ) {
3014                                 # $noparse and !$noargs
3015                                 # Just replace the arguments, not any double-brace items
3016                                 # This is used for rendered interwiki transclusion
3017                                 $text = $this->replaceVariables( $text, $assocArgs, true );
3018                         }
3019                 }
3020                 # Prune lower levels off the recursion check path
3021                 $this->mTemplatePath = $lastPathLevel;
3022
3023                 if ( $found && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3024                         # Error, oversize inclusion
3025                         $text = $linestart .
3026                                 '{{' . $part1 . '}}' .
3027                                 '<!-- WARNING: template omitted, post-expand include size too large -->';
3028                         $noparse = true;
3029                         $noargs = true;
3030                 }
3031
3032                 if ( !$found ) {
3033                         wfProfileOut( $fname );
3034                         return $piece['text'];
3035                 } else {
3036                         wfProfileIn( __METHOD__ . '-placeholders' );
3037                         if ( $isHTML ) {
3038                                 # Replace raw HTML by a placeholder
3039                                 # Add a blank line preceding, to prevent it from mucking up
3040                                 # immediately preceding headings
3041                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
3042                         } else {
3043                                 # replace ==section headers==
3044                                 # XXX this needs to go away once we have a better parser.
3045                                 if ( $this->mOutputType != OT_WIKI && $replaceHeadings ) {
3046                                         if( !is_null( $title ) )
3047                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
3048                                         else
3049                                                 $encodedname = base64_encode("");
3050                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
3051                                                 PREG_SPLIT_DELIM_CAPTURE);
3052                                         $text = '';
3053                                         $nsec = 0;
3054                                         for( $i = 0; $i < count($m); $i += 2 ) {
3055                                                 $text .= $m[$i];
3056                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
3057                                                 $hl = $m[$i + 1];
3058                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
3059                                                         $text .= $hl;
3060                                                         continue;
3061                                                 }
3062                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
3063                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
3064                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
3065
3066                                                 $nsec++;
3067                                         }
3068                                 }
3069                         }
3070                         wfProfileOut( __METHOD__ . '-placeholders' );
3071                 }
3072
3073                 # Prune lower levels off the recursion check path
3074                 $this->mTemplatePath = $lastPathLevel;
3075
3076                 if ( !$found ) {
3077                         wfProfileOut( $fname );
3078                         return $piece['text'];
3079                 } else {
3080                         wfProfileOut( $fname );
3081                         return $text;
3082                 }
3083         }
3084
3085         /**
3086          * Fetch the unparsed text of a template and register a reference to it.
3087          */
3088         function fetchTemplate( $title ) {
3089                 $text = false;
3090                 // Loop to fetch the article, with up to 1 redirect
3091                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3092                         $rev = Revision::newFromTitle( $title );
3093                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
3094                         if ( !$rev ) {
3095                                 break;
3096                         }
3097                         $text = $rev->getText();
3098                         if ( $text === false ) {
3099                                 break;
3100                         }
3101                         // Redirect?
3102                         $title = Title::newFromRedirect( $text );
3103                 }
3104                 return $text;
3105         }
3106
3107         /**
3108          * Transclude an interwiki link.
3109          */
3110         function interwikiTransclude( $title, $action ) {
3111                 global $wgEnableScaryTranscluding, $wgCanonicalNamespaceNames;
3112
3113                 if (!$wgEnableScaryTranscluding)
3114                         return wfMsg('scarytranscludedisabled');
3115
3116                 // The namespace will actually only be 0 or 10, depending on whether there was a leading :
3117                 // But we'll handle it generally anyway
3118                 if ( $title->getNamespace() ) {
3119                         // Use the canonical namespace, which should work anywhere
3120                         $articleName = $wgCanonicalNamespaceNames[$title->getNamespace()] . ':' . $title->getDBkey();
3121                 } else {
3122                         $articleName = $title->getDBkey();
3123                 }
3124
3125                 $url = str_replace('$1', urlencode($articleName), Title::getInterwikiLink($title->getInterwiki()));
3126                 $url .= "?action=$action";
3127                 if (strlen($url) > 255)
3128                         return wfMsg('scarytranscludetoolong');
3129                 return $this->fetchScaryTemplateMaybeFromCache($url);
3130         }
3131
3132         function fetchScaryTemplateMaybeFromCache($url) {
3133                 global $wgTranscludeCacheExpiry;
3134                 $dbr =& wfGetDB(DB_SLAVE);
3135                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3136                                 array('tc_url' => $url));
3137                 if ($obj) {
3138                         $time = $obj->tc_time;
3139                         $text = $obj->tc_contents;
3140                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3141                                 return $text;
3142                         }
3143                 }
3144
3145                 $text = Http::get($url);
3146                 if (!$text)
3147                         return wfMsg('scarytranscludefailed', $url);
3148
3149                 $dbw =& wfGetDB(DB_MASTER);
3150                 $dbw->replace('transcache', array('tc_url'), array(
3151                         'tc_url' => $url,
3152                         'tc_time' => time(),
3153                         'tc_contents' => $text));
3154                 return $text;
3155         }
3156
3157
3158         /**
3159          * Triple brace replacement -- used for template arguments
3160          * @private
3161          */
3162         function argSubstitution( $matches ) {
3163                 $arg = trim( $matches['title'] );
3164                 $text = $matches['text'];
3165                 $inputArgs = end( $this->mArgStack );
3166
3167                 if ( array_key_exists( $arg, $inputArgs ) ) {
3168                         $text = $inputArgs[$arg];
3169                 } else if ($this->mOutputType == OT_HTML && null != $matches['parts'] && count($matches['parts']) > 0) {
3170                         $text = $matches['parts'][0];
3171                 }
3172                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3173                         $text = $matches['text'] .
3174                                 '<!-- WARNING: argument omitted, expansion size too large -->';
3175                 }
3176
3177                 return $text;
3178         }
3179
3180         /**
3181          * Increment an include size counter
3182          *
3183          * @param string $type The type of expansion
3184          * @param integer $size The size of the text
3185          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3186          */
3187         function incrementIncludeSize( $type, $size ) {
3188                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3189                         return false;
3190                 } else {
3191                         $this->mIncludeSizes[$type] += $size;
3192                         return true;
3193                 }
3194         }
3195
3196         /**
3197          * Detect __NOGALLERY__ magic word and set a placeholder
3198          */
3199         function stripNoGallery( &$text ) {
3200                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3201                 # do not add TOC
3202                 $mw = MagicWord::get( 'nogallery' );
3203                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3204         }
3205
3206         /**
3207          * Detect __TOC__ magic word and set a placeholder
3208          */
3209         function stripToc( $text ) {
3210                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3211                 # do not add TOC
3212                 $mw = MagicWord::get( 'notoc' );
3213                 if( $mw->matchAndRemove( $text ) ) {
3214                         $this->mShowToc = false;
3215                 }
3216
3217                 $mw = MagicWord::get( 'toc' );
3218                 if( $mw->match( $text ) ) {
3219                         $this->mShowToc = true;
3220                         $this->mForceTocPosition = true;
3221
3222                         // Set a placeholder. At the end we'll fill it in with the TOC.
3223                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3224
3225                         // Only keep the first one.
3226                         $text = $mw->replace( '', $text );
3227                 }
3228                 return $text;
3229         }
3230
3231         /**
3232          * This function accomplishes several tasks:
3233          * 1) Auto-number headings if that option is enabled
3234          * 2) Add an [edit] link to sections for logged in users who have enabled the option
3235          * 3) Add a Table of contents on the top for users who have enabled the option
3236          * 4) Auto-anchor headings
3237          *
3238          * It loops through all headlines, collects the necessary data, then splits up the
3239          * string and re-inserts the newly formatted headlines.
3240          *
3241          * @param string $text
3242          * @param boolean $isMain
3243          * @private
3244          */
3245         function formatHeadings( $text, $isMain=true ) {
3246                 global $wgMaxTocLevel, $wgContLang;
3247
3248                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3249                 if( !$this->mTitle->userCanEdit() ) {
3250                         $showEditLink = 0;
3251                 } else {
3252                         $showEditLink = $this->mOptions->getEditSection();
3253                 }
3254
3255                 # Inhibit editsection links if requested in the page
3256                 $esw =& MagicWord::get( 'noeditsection' );
3257                 if( $esw->matchAndRemove( $text ) ) {
3258                         $showEditLink = 0;
3259                 }
3260
3261                 # Get all headlines for numbering them and adding funky stuff like [edit]
3262                 # links - this is for later, but we need the number of headlines right now
3263                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
3264
3265                 # if there are fewer than 4 headlines in the article, do not show TOC
3266                 # unless it's been explicitly enabled.
3267                 $enoughToc = $this->mShowToc &&
3268                         (($numMatches >= 4) || $this->mForceTocPosition);
3269
3270                 # Allow user to stipulate that a page should have a "new section"
3271                 # link added via __NEWSECTIONLINK__
3272                 $mw =& MagicWord::get( 'newsectionlink' );
3273                 if( $mw->matchAndRemove( $text ) )
3274                         $this->mOutput->setNewSection( true );
3275
3276                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3277                 # override above conditions and always show TOC above first header
3278                 $mw =& MagicWord::get( 'forcetoc' );
3279                 if ($mw->matchAndRemove( $text ) ) {
3280                         $this->mShowToc = true;
3281                         $enoughToc = true;
3282                 }
3283
3284                 # Never ever show TOC if no headers
3285                 if( $numMatches < 1 ) {
3286                         $enoughToc = false;
3287                 }
3288
3289                 # We need this to perform operations on the HTML
3290                 $sk =& $this->mOptions->getSkin();
3291
3292                 # headline counter
3293                 $headlineCount = 0;
3294                 $sectionCount = 0; # headlineCount excluding template sections
3295
3296                 # Ugh .. the TOC should have neat indentation levels which can be
3297                 # passed to the skin functions. These are determined here
3298                 $toc = '';
3299                 $full = '';
3300                 $head = array();
3301                 $sublevelCount = array();
3302                 $levelCount = array();
3303                 $toclevel = 0;
3304                 $level = 0;
3305                 $prevlevel = 0;
3306                 $toclevel = 0;
3307                 $prevtoclevel = 0;
3308
3309                 foreach( $matches[3] as $headline ) {
3310                         $istemplate = 0;
3311                         $templatetitle = '';
3312                         $templatesection = 0;
3313                         $numbering = '';
3314
3315                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3316                                 $istemplate = 1;
3317                                 $templatetitle = base64_decode($mat[1]);
3318                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3319                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3320                         }
3321
3322                         if( $toclevel ) {
3323                                 $prevlevel = $level;
3324                                 $prevtoclevel = $toclevel;
3325                         }
3326                         $level = $matches[1][$headlineCount];
3327
3328                         if( $doNumberHeadings || $enoughToc ) {
3329
3330                                 if ( $level > $prevlevel ) {
3331                                         # Increase TOC level
3332                                         $toclevel++;
3333                                         $sublevelCount[$toclevel] = 0;
3334                                         if( $toclevel<$wgMaxTocLevel ) {
3335                                                 $toc .= $sk->tocIndent();
3336                                         }
3337                                 }
3338                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3339                                         # Decrease TOC level, find level to jump to
3340
3341                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3342                                                 # Can only go down to level 1
3343                                                 $toclevel = 1;
3344                                         } else {
3345                                                 for ($i = $toclevel; $i > 0; $i--) {
3346                                                         if ( $levelCount[$i] == $level ) {
3347                                                                 # Found last matching level
3348                                                                 $toclevel = $i;
3349                                                                 break;
3350                                                         }
3351                                                         elseif ( $levelCount[$i] < $level ) {
3352                                                                 # Found first matching level below current level
3353                                                                 $toclevel = $i + 1;
3354                                                                 break;
3355                                                         }
3356                                                 }
3357                                         }
3358                                         if( $toclevel<$wgMaxTocLevel ) {
3359                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3360                                         }
3361                                 }
3362                                 else {
3363                                         # No change in level, end TOC line
3364                                         if( $toclevel<$wgMaxTocLevel ) {
3365                                                 $toc .= $sk->tocLineEnd();
3366                                         }
3367                                 }
3368
3369                                 $levelCount[$toclevel] = $level;
3370
3371                                 # count number of headlines for each level
3372                                 @$sublevelCount[$toclevel]++;
3373                                 $dot = 0;
3374                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3375                                         if( !empty( $sublevelCount[$i] ) ) {
3376                                                 if( $dot ) {
3377                                                         $numbering .= '.';
3378                                                 }
3379                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3380                                                 $dot = 1;
3381                                         }
3382                                 }
3383                         }
3384
3385                         # The canonized header is a version of the header text safe to use for links
3386                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3387                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
3388                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
3389
3390                         # Remove link placeholders by the link text.
3391                         #     <!--LINK number-->
3392                         # turns into
3393                         #     link text with suffix
3394                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3395                                                             "\$this->mLinkHolders['texts'][\$1]",
3396                                                             $canonized_headline );
3397                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3398                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3399                                                             $canonized_headline );
3400
3401                         # strip out HTML
3402                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3403                         $tocline = trim( $canonized_headline );
3404                         # Save headline for section edit hint before it's escaped
3405                         $headline_hint = trim( $canonized_headline );
3406                         $canonized_headline = Sanitizer::escapeId( $tocline );
3407                         $refers[$headlineCount] = $canonized_headline;
3408
3409                         # count how many in assoc. array so we can track dupes in anchors
3410                         @$refers[$canonized_headline]++;
3411                         $refcount[$headlineCount]=$refers[$canonized_headline];
3412
3413                         # Don't number the heading if it is the only one (looks silly)
3414                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3415                                 # the two are different if the line contains a link
3416                                 $headline=$numbering . ' ' . $headline;
3417                         }
3418
3419                         # Create the anchor for linking from the TOC to the section
3420                         $anchor = $canonized_headline;
3421                         if($refcount[$headlineCount] > 1 ) {
3422                                 $anchor .= '_' . $refcount[$headlineCount];
3423                         }
3424                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3425                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3426                         }
3427                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3428                                 if ( empty( $head[$headlineCount] ) ) {
3429                                         $head[$headlineCount] = '';
3430                                 }
3431                                 if( $istemplate )
3432                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
3433                                 else
3434                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3435                         }
3436
3437                         # give headline the correct <h#> tag
3438                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount] .$headline.'</h'.$level.'>';
3439
3440                         $headlineCount++;
3441                         if( !$istemplate )
3442                                 $sectionCount++;
3443                 }
3444
3445                 if( $enoughToc ) {
3446                         if( $toclevel<$wgMaxTocLevel ) {
3447                                 $toc .= $sk->tocUnindent( $toclevel - 1 );
3448                         }
3449                         $toc = $sk->tocList( $toc );
3450                 }
3451
3452                 # split up and insert constructed headlines
3453
3454                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3455                 $i = 0;
3456
3457                 foreach( $blocks as $block ) {
3458                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3459                                 # This is the [edit] link that appears for the top block of text when
3460                                 # section editing is enabled
3461
3462                                 # Disabled because it broke block formatting
3463                                 # For example, a bullet point in the top line
3464                                 # $full .= $sk->editSectionLink(0);
3465                         }
3466                         $full .= $block;
3467                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3468                                 # Top anchor now in skin
3469                                 $full = $full.$toc;
3470                         }
3471
3472                         if( !empty( $head[$i] ) ) {
3473                                 $full .= $head[$i];
3474                         }
3475                         $i++;
3476                 }
3477                 if( $this->mForceTocPosition ) {
3478                         return str_replace( '<!--MWTOC-->', $toc, $full );
3479                 } else {
3480                         return $full;
3481                 }
3482         }
3483
3484         /**
3485          * Transform wiki markup when saving a page by doing \r\n -> \n
3486          * conversion, substitting signatures, {{subst:}} templates, etc.
3487          *
3488          * @param string $text the text to transform
3489          * @param Title &$title the Title object for the current article
3490          * @param User &$user the User object describing the current user
3491          * @param ParserOptions $options parsing options
3492          * @param bool $clearState whether to clear the parser state first
3493          * @return string the altered wiki markup
3494          * @public
3495          */
3496         function preSaveTransform( $text, &$title, &$user, $options, $clearState = true ) {
3497                 $this->mOptions = $options;
3498                 $this->mTitle =& $title;
3499                 $this->mOutputType = OT_WIKI;
3500
3501                 if ( $clearState ) {
3502                         $this->clearState();
3503                 }
3504
3505                 $stripState = false;
3506                 $pairs = array(
3507                         "\r\n" => "\n",
3508                 );
3509                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3510                 $text = $this->strip( $text, $stripState, true, array( 'gallery' ) );
3511                 $text = $this->pstPass2( $text, $stripState, $user );
3512                 $text = $this->unstrip( $text, $stripState );
3513                 $text = $this->unstripNoWiki( $text, $stripState );
3514                 return $text;
3515         }
3516
3517         /**
3518          * Pre-save transform helper function
3519          * @private
3520          */
3521         function pstPass2( $text, &$stripState, &$user ) {
3522                 global $wgContLang, $wgLocaltimezone;
3523
3524                 /* Note: This is the timestamp saved as hardcoded wikitext to
3525                  * the database, we use $wgContLang here in order to give
3526                  * everyone the same signature and use the default one rather
3527                  * than the one selected in each user's preferences.
3528                  */
3529                 if ( isset( $wgLocaltimezone ) ) {
3530                         $oldtz = getenv( 'TZ' );
3531                         putenv( 'TZ='.$wgLocaltimezone );
3532                 }
3533                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3534                   ' (' . date( 'T' ) . ')';
3535                 if ( isset( $wgLocaltimezone ) ) {
3536                         putenv( 'TZ='.$oldtz );
3537                 }
3538
3539                 # Variable replacement
3540                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3541                 $text = $this->replaceVariables( $text );
3542
3543                 # Strip out <nowiki> etc. added via replaceVariables
3544                 $text = $this->strip( $text, $stripState, false, array( 'gallery' ) );
3545
3546                 # Signatures
3547                 $sigText = $this->getUserSig( $user );
3548                 $text = strtr( $text, array(
3549                         '~~~~~' => $d,
3550                         '~~~~' => "$sigText $d",
3551                         '~~~' => $sigText
3552                 ) );
3553
3554                 # Context links: [[|name]] and [[name (context)|]]
3555                 #
3556                 global $wgLegalTitleChars;
3557                 $tc = "[$wgLegalTitleChars]";
3558
3559                 $namespacechar = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3560                 $conpat = "/^{$tc}+?( \\({$tc}+\\)|)$/";
3561
3562                 $p1 = "/\[\[(:?$namespacechar+:|:|)({$tc}+?)( \\({$tc}+\\)|)\\|]]/";    # [[ns:page (context)|]]
3563                 $p2 = "/\[\[\\|({$tc}+)]]/";                                            # [[|page]]
3564
3565                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3566
3567                 $t = $this->mTitle->getText();
3568                 if ( preg_match( $conpat, $t, $m ) && '' != $m[1] ) {
3569                         $text = preg_replace( $p2, "[[\\1{$m[1]}|\\1]]", $text );
3570                 } else {
3571                         # if $m[1] is empty, don't bother duplicating the title
3572                         $text = preg_replace( $p2, '[[\\1]]', $text );
3573                 }
3574
3575                 # Trim trailing whitespace
3576                 # __END__ tag allows for trailing
3577                 # whitespace to be deliberately included
3578                 $text = rtrim( $text );
3579                 $mw =& MagicWord::get( 'end' );
3580                 $mw->matchAndRemove( $text );
3581
3582                 return $text;
3583         }
3584
3585         /**
3586          * Fetch the user's signature text, if any, and normalize to
3587          * validated, ready-to-insert wikitext.
3588          *
3589          * @param User $user
3590          * @return string
3591          * @private
3592          */
3593         function getUserSig( &$user ) {
3594                 $username = $user->getName();
3595                 $nickname = $user->getOption( 'nickname' );
3596                 $nickname = $nickname === '' ? $username : $nickname;
3597
3598                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3599                         # Sig. might contain markup; validate this
3600                         if( $this->validateSig( $nickname ) !== false ) {
3601                                 # Validated; clean up (if needed) and return it
3602                                 return $this->cleanSig( $nickname, true );
3603                         } else {
3604                                 # Failed to validate; fall back to the default
3605                                 $nickname = $username;
3606                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3607                         }
3608                 }
3609
3610                 // Make sure nickname doesnt get a sig in a sig
3611                 $nickname = $this->cleanSigInSig( $nickname );
3612
3613                 # If we're still here, make it a link to the user page
3614                 $userpage = $user->getUserPage();
3615                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3616         }
3617
3618         /**
3619          * Check that the user's signature contains no bad XML
3620          *
3621          * @param string $text
3622          * @return mixed An expanded string, or false if invalid.
3623          */
3624         function validateSig( $text ) {
3625                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3626         }
3627
3628         /**
3629          * Clean up signature text
3630          *
3631          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3632          * 2) Substitute all transclusions
3633          *
3634          * @param string $text
3635          * @param $parsing Whether we're cleaning (preferences save) or parsing
3636          * @return string Signature text
3637          */
3638         function cleanSig( $text, $parsing = false ) {
3639                 global $wgTitle;
3640                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
3641
3642                 $substWord = MagicWord::get( 'subst' );
3643                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3644                 $substText = '{{' . $substWord->getSynonym( 0 );
3645
3646                 $text = preg_replace( $substRegex, $substText, $text );
3647                 $text = $this->cleanSigInSig( $text );
3648                 $text = $this->replaceVariables( $text );
3649
3650                 $this->clearState();
3651                 return $text;
3652         }
3653
3654         /**
3655          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3656          * @param string $text
3657          * @return string Signature text with /~{3,5}/ removed
3658          */
3659         function cleanSigInSig( $text ) {
3660                 $text = preg_replace( '/~{3,5}/', '', $text );
3661                 return $text;
3662         }
3663
3664         /**
3665          * Set up some variables which are usually set up in parse()
3666          * so that an external function can call some class members with confidence
3667          * @public
3668          */
3669         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3670                 $this->mTitle =& $title;
3671                 $this->mOptions = $options;
3672                 $this->mOutputType = $outputType;
3673                 if ( $clearState ) {
3674                         $this->clearState();
3675                 }
3676         }
3677
3678         /**
3679          * Transform a MediaWiki message by replacing magic variables.
3680          *
3681          * @param string $text the text to transform
3682          * @param ParserOptions $options  options
3683          * @return string the text with variables substituted
3684          * @public
3685          */
3686         function transformMsg( $text, $options ) {
3687                 global $wgTitle;
3688                 static $executing = false;
3689
3690                 $fname = "Parser::transformMsg";
3691
3692                 # Guard against infinite recursion
3693                 if ( $executing ) {
3694                         return $text;
3695                 }
3696                 $executing = true;
3697
3698                 wfProfileIn($fname);
3699
3700                 $this->mTitle = $wgTitle;
3701                 $this->mOptions = $options;
3702                 $this->mOutputType = OT_MSG;
3703                 $this->clearState();
3704                 $text = $this->replaceVariables( $text );
3705
3706                 $executing = false;
3707                 wfProfileOut($fname);
3708                 return $text;
3709         }
3710
3711         /**
3712          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3713          * The callback should have the following form:
3714          *    function myParserHook( $text, $params, &$parser ) { ... }
3715          *
3716          * Transform and return $text. Use $parser for any required context, e.g. use
3717          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3718          *
3719          * @public
3720          *
3721          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3722          * @param mixed $callback The callback function (and object) to use for the tag
3723          *
3724          * @return The old value of the mTagHooks array associated with the hook
3725          */
3726         function setHook( $tag, $callback ) {
3727                 $tag = strtolower( $tag );
3728                 $oldVal = @$this->mTagHooks[$tag];
3729                 $this->mTagHooks[$tag] = $callback;
3730
3731                 return $oldVal;
3732         }
3733
3734         /**
3735          * Create a function, e.g. {{sum:1|2|3}}
3736          * The callback function should have the form:
3737          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3738          *
3739          * The callback may either return the text result of the function, or an array with the text
3740          * in element 0, and a number of flags in the other elements. The names of the flags are
3741          * specified in the keys. Valid flags are:
3742          *   found                     The text returned is valid, stop processing the template. This
3743          *                             is on by default.
3744          *   nowiki                    Wiki markup in the return value should be escaped
3745          *   noparse                   Unsafe HTML tags should not be stripped, etc.
3746          *   noargs                    Don't replace triple-brace arguments in the return value
3747          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3748          *
3749          * @public
3750          *
3751          * @param string $id The magic word ID
3752          * @param mixed $callback The callback function (and object) to use
3753          * @param integer $flags a combination of the following flags:
3754          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
3755          *
3756          * @return The old callback function for this name, if any
3757          */
3758         function setFunctionHook( $id, $callback, $flags = 0 ) {
3759                 $oldVal = @$this->mFunctionHooks[$id];
3760                 $this->mFunctionHooks[$id] = $callback;
3761
3762                 # Add to function cache
3763                 $mw = MagicWord::get( $id );
3764                 if ( !$mw ) {
3765                         throw new MWException( 'The calling convention to Parser::setFunctionHook() has changed, ' .
3766                                 'it is now required to pass a MagicWord ID as the first parameter.' );
3767                 }
3768
3769                 $synonyms = $mw->getSynonyms();
3770                 $sensitive = intval( $mw->isCaseSensitive() );
3771
3772                 foreach ( $synonyms as $syn ) {
3773                         # Case
3774                         if ( !$sensitive ) {
3775                                 $syn = strtolower( $syn );
3776                         }
3777                         # Add leading hash
3778                         if ( !( $flags & SFH_NO_HASH ) ) {
3779                                 $syn = '#' . $syn;
3780                         }
3781                         # Remove trailing colon
3782                         if ( substr( $syn, -1, 1 ) == ':' ) {
3783                                 $syn = substr( $syn, 0, -1 );
3784                         }
3785                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
3786                 }
3787                 return $oldVal;
3788         }
3789
3790         /**
3791          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3792          * Placeholders created in Skin::makeLinkObj()
3793          * Returns an array of links found, indexed by PDBK:
3794          *  0 - broken
3795          *  1 - normal link
3796          *  2 - stub
3797          * $options is a bit field, RLH_FOR_UPDATE to select for update
3798          */
3799         function replaceLinkHolders( &$text, $options = 0 ) {
3800                 global $wgUser;
3801                 global $wgOutputReplace;
3802
3803                 $fname = 'Parser::replaceLinkHolders';
3804                 wfProfileIn( $fname );
3805
3806                 $pdbks = array();
3807                 $colours = array();
3808                 $sk =& $this->mOptions->getSkin();
3809                 $linkCache =& LinkCache::singleton();
3810
3811                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3812                         wfProfileIn( $fname.'-check' );
3813                         $dbr =& wfGetDB( DB_SLAVE );
3814                         $page = $dbr->tableName( 'page' );
3815                         $threshold = $wgUser->getOption('stubthreshold');
3816
3817                         # Sort by namespace
3818                         asort( $this->mLinkHolders['namespaces'] );
3819
3820                         # Generate query
3821                         $query = false;
3822                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3823                                 # Make title object
3824                                 $title = $this->mLinkHolders['titles'][$key];
3825
3826                                 # Skip invalid entries.
3827                                 # Result will be ugly, but prevents crash.
3828                                 if ( is_null( $title ) ) {
3829                                         continue;
3830                                 }
3831                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3832
3833                                 # Check if it's a static known link, e.g. interwiki
3834                                 if ( $title->isAlwaysKnown() ) {
3835                                         $colours[$pdbk] = 1;
3836                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3837                                         $colours[$pdbk] = 1;
3838                                         $this->mOutput->addLink( $title, $id );
3839                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3840                                         $colours[$pdbk] = 0;
3841                                 } else {
3842                                         # Not in the link cache, add it to the query
3843                                         if ( !isset( $current ) ) {
3844                                                 $current = $ns;
3845                                                 $query =  "SELECT page_id, page_namespace, page_title";
3846                                                 if ( $threshold > 0 ) {
3847                                                         $query .= ', page_len, page_is_redirect';
3848                                                 }
3849                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
3850                                         } elseif ( $current != $ns ) {
3851                                                 $current = $ns;
3852                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
3853                                         } else {
3854                                                 $query .= ', ';
3855                                         }
3856
3857                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3858                                 }
3859                         }
3860                         if ( $query ) {
3861                                 $query .= '))';
3862                                 if ( $options & RLH_FOR_UPDATE ) {
3863                                         $query .= ' FOR UPDATE';
3864                                 }
3865
3866                                 $res = $dbr->query( $query, $fname );
3867
3868                                 # Fetch data and form into an associative array
3869                                 # non-existent = broken
3870                                 # 1 = known
3871                                 # 2 = stub
3872                                 while ( $s = $dbr->fetchObject($res) ) {
3873                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
3874                                         $pdbk = $title->getPrefixedDBkey();
3875                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
3876                                         $this->mOutput->addLink( $title, $s->page_id );
3877
3878                                         if ( $threshold >  0 ) {
3879                                                 $size = $s->page_len;
3880                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
3881                                                         $colours[$pdbk] = 1;
3882                                                 } else {
3883                                                         $colours[$pdbk] = 2;
3884                                                 }
3885                                         } else {
3886                                                 $colours[$pdbk] = 1;
3887                                         }
3888                                 }
3889                         }
3890                         wfProfileOut( $fname.'-check' );
3891
3892                         # Construct search and replace arrays
3893                         wfProfileIn( $fname.'-construct' );
3894                         $wgOutputReplace = array();
3895                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3896                                 $pdbk = $pdbks[$key];
3897                                 $searchkey = "<!--LINK $key-->";
3898                                 $title = $this->mLinkHolders['titles'][$key];
3899                                 if ( empty( $colours[$pdbk] ) ) {
3900                                         $linkCache->addBadLinkObj( $title );
3901                                         $colours[$pdbk] = 0;
3902                                         $this->mOutput->addLink( $title, 0 );
3903                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
3904                                                                         $this->mLinkHolders['texts'][$key],
3905                                                                         $this->mLinkHolders['queries'][$key] );
3906                                 } elseif ( $colours[$pdbk] == 1 ) {
3907                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
3908                                                                         $this->mLinkHolders['texts'][$key],
3909                                                                         $this->mLinkHolders['queries'][$key] );
3910                                 } elseif ( $colours[$pdbk] == 2 ) {
3911                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
3912                                                                         $this->mLinkHolders['texts'][$key],
3913                                                                         $this->mLinkHolders['queries'][$key] );
3914                                 }
3915                         }
3916                         wfProfileOut( $fname.'-construct' );
3917
3918                         # Do the thing
3919                         wfProfileIn( $fname.'-replace' );
3920
3921                         $text = preg_replace_callback(
3922                                 '/(<!--LINK .*?-->)/',
3923                                 "wfOutputReplaceMatches",
3924                                 $text);
3925
3926                         wfProfileOut( $fname.'-replace' );
3927                 }
3928
3929                 # Now process interwiki link holders
3930                 # This is quite a bit simpler than internal links
3931                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
3932                         wfProfileIn( $fname.'-interwiki' );
3933                         # Make interwiki link HTML
3934                         $wgOutputReplace = array();
3935                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
3936                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
3937                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
3938                         }
3939
3940                         $text = preg_replace_callback(
3941                                 '/<!--IWLINK (.*?)-->/',
3942                                 "wfOutputReplaceMatches",
3943                                 $text );
3944                         wfProfileOut( $fname.'-interwiki' );
3945                 }
3946
3947                 wfProfileOut( $fname );
3948                 return $colours;
3949         }
3950
3951         /**
3952          * Replace <!--LINK--> link placeholders with plain text of links
3953          * (not HTML-formatted).
3954          * @param string $text
3955          * @return string
3956          */
3957         function replaceLinkHoldersText( $text ) {
3958                 $fname = 'Parser::replaceLinkHoldersText';
3959                 wfProfileIn( $fname );
3960
3961                 $text = preg_replace_callback(
3962                         '/<!--(LINK|IWLINK) (.*?)-->/',
3963                         array( &$this, 'replaceLinkHoldersTextCallback' ),
3964                         $text );
3965
3966                 wfProfileOut( $fname );
3967                 return $text;
3968         }
3969
3970         /**
3971          * @param array $matches
3972          * @return string
3973          * @private
3974          */
3975         function replaceLinkHoldersTextCallback( $matches ) {
3976                 $type = $matches[1];
3977                 $key  = $matches[2];
3978                 if( $type == 'LINK' ) {
3979                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
3980                                 return $this->mLinkHolders['texts'][$key];
3981                         }
3982                 } elseif( $type == 'IWLINK' ) {
3983                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
3984                                 return $this->mInterwikiLinkHolders['texts'][$key];
3985                         }
3986                 }
3987                 return $matches[0];
3988         }
3989
3990         /**
3991          * Tag hook handler for 'pre'.
3992          */
3993         function renderPreTag( $text, $attribs, $parser ) {
3994                 // Backwards-compatibility hack
3995                 $content = preg_replace( '!<nowiki>(.*?)</nowiki>!is', '\\1', $text );
3996
3997                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
3998                 return wfOpenElement( 'pre', $attribs ) .
3999                         wfEscapeHTMLTagsOnly( $content ) .
4000                         '</pre>';
4001         }
4002
4003         /**
4004          * Renders an image gallery from a text with one line per image.
4005          * text labels may be given by using |-style alternative text. E.g.
4006          *   Image:one.jpg|The number "1"
4007          *   Image:tree.jpg|A tree
4008          * given as text will return the HTML of a gallery with two images,
4009          * labeled 'The number "1"' and
4010          * 'A tree'.
4011          */
4012         function renderImageGallery( $text, $params ) {
4013                 $ig = new ImageGallery();
4014                 $ig->setShowBytes( false );
4015                 $ig->setShowFilename( false );
4016                 $ig->setParsing();
4017                 $ig->useSkin( $this->mOptions->getSkin() );
4018
4019                 if( isset( $params['caption'] ) )
4020                         $ig->setCaption( $params['caption'] );
4021
4022                 $lines = explode( "\n", $text );
4023                 foreach ( $lines as $line ) {
4024                         # match lines like these:
4025                         # Image:someimage.jpg|This is some image
4026                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4027                         # Skip empty lines
4028                         if ( count( $matches ) == 0 ) {
4029                                 continue;
4030                         }
4031                         $tp = Title::newFromText( $matches[1] );
4032                         $nt =& $tp;
4033                         if( is_null( $nt ) ) {
4034                                 # Bogus title. Ignore these so we don't bomb out later.
4035                                 continue;
4036                         }
4037                         if ( isset( $matches[3] ) ) {
4038                                 $label = $matches[3];
4039                         } else {
4040                                 $label = '';
4041                         }
4042
4043                         $pout = $this->parse( $label,
4044                                 $this->mTitle,
4045                                 $this->mOptions,
4046                                 false, // Strip whitespace...?
4047                                 false  // Don't clear state!
4048                         );
4049                         $html = $pout->getText();
4050
4051                         $ig->add( new Image( $nt ), $html );
4052
4053                         # Only add real images (bug #5586)
4054                         if ( $nt->getNamespace() == NS_IMAGE ) {
4055                                 $this->mOutput->addImage( $nt->getDBkey() );
4056                         }
4057                 }
4058                 return $ig->toHTML();
4059         }
4060
4061         /**
4062          * Parse image options text and use it to make an image
4063          */
4064         function makeImage( &$nt, $options ) {
4065                 global $wgUseImageResize;
4066
4067                 $align = '';
4068
4069                 # Check if the options text is of the form "options|alt text"
4070                 # Options are:
4071                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4072                 #  * left               no resizing, just left align. label is used for alt= only
4073                 #  * right              same, but right aligned
4074                 #  * none               same, but not aligned
4075                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4076                 #  * center             center the image
4077                 #  * framed             Keep original image size, no magnify-button.
4078
4079                 $part = explode( '|', $options);
4080
4081                 $mwThumb  =& MagicWord::get( 'img_thumbnail' );
4082                 $mwManualThumb =& MagicWord::get( 'img_manualthumb' );
4083                 $mwLeft   =& MagicWord::get( 'img_left' );
4084                 $mwRight  =& MagicWord::get( 'img_right' );
4085                 $mwNone   =& MagicWord::get( 'img_none' );
4086                 $mwWidth  =& MagicWord::get( 'img_width' );
4087                 $mwCenter =& MagicWord::get( 'img_center' );
4088                 $mwFramed =& MagicWord::get( 'img_framed' );
4089                 $caption = '';
4090
4091                 $width = $height = $framed = $thumb = false;
4092                 $manual_thumb = '' ;
4093
4094                 foreach( $part as $key => $val ) {
4095                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
4096                                 $thumb=true;
4097                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
4098                                 # use manually specified thumbnail
4099                                 $thumb=true;
4100                                 $manual_thumb = $match;
4101                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
4102                                 # remember to set an alignment, don't render immediately
4103                                 $align = 'right';
4104                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
4105                                 # remember to set an alignment, don't render immediately
4106                                 $align = 'left';
4107                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
4108                                 # remember to set an alignment, don't render immediately
4109                                 $align = 'center';
4110                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
4111                                 # remember to set an alignment, don't render immediately
4112                                 $align = 'none';
4113                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
4114                                 wfDebug( "img_width match: $match\n" );
4115                                 # $match is the image width in pixels
4116                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
4117                                         $width = intval( $m[1] );
4118                                         $height = intval( $m[2] );
4119                                 } else {
4120                                         $width = intval($match);
4121                                 }
4122                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
4123                                 $framed=true;
4124                         } else {
4125                                 $caption = $val;
4126                         }
4127                 }
4128                 # Strip bad stuff out of the alt text
4129                 $alt = $this->replaceLinkHoldersText( $caption );
4130
4131                 # make sure there are no placeholders in thumbnail attributes
4132                 # that are later expanded to html- so expand them now and
4133                 # remove the tags
4134                 $alt = $this->unstrip($alt, $this->mStripState);
4135                 $alt = Sanitizer::stripAllTags( $alt );
4136
4137                 # Linker does the rest
4138                 $sk =& $this->mOptions->getSkin();
4139                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb );
4140         }
4141
4142         /**
4143          * Set a flag in the output object indicating that the content is dynamic and
4144          * shouldn't be cached.
4145          */
4146         function disableCache() {
4147                 wfDebug( "Parser output marked as uncacheable.\n" );
4148                 $this->mOutput->mCacheTime = -1;
4149         }
4150
4151         /**#@+
4152          * Callback from the Sanitizer for expanding items found in HTML attribute
4153          * values, so they can be safely tested and escaped.
4154          * @param string $text
4155          * @param array $args
4156          * @return string
4157          * @private
4158          */
4159         function attributeStripCallback( &$text, $args ) {
4160                 $text = $this->replaceVariables( $text, $args );
4161                 $text = $this->unstripForHTML( $text );
4162                 return $text;
4163         }
4164
4165         function unstripForHTML( $text ) {
4166                 $text = $this->unstrip( $text, $this->mStripState );
4167                 $text = $this->unstripNoWiki( $text, $this->mStripState );
4168                 return $text;
4169         }
4170         /**#@-*/
4171
4172         /**#@+
4173          * Accessor/mutator
4174          */
4175         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4176         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4177         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4178         /**#@-*/
4179
4180         /**#@+
4181          * Accessor
4182          */
4183         function getTags() { return array_keys( $this->mTagHooks ); }
4184         /**#@-*/
4185
4186
4187         /**
4188          * Break wikitext input into sections, and either pull or replace
4189          * some particular section's text.
4190          *
4191          * External callers should use the getSection and replaceSection methods.
4192          *
4193          * @param $text Page wikitext
4194          * @param $section Numbered section. 0 pulls the text before the first
4195          *                 heading; other numbers will pull the given section
4196          *                 along with its lower-level subsections.
4197          * @param $mode One of "get" or "replace"
4198          * @param $newtext Replacement text for section data.
4199          * @return string for "get", the extracted section text.
4200          *                for "replace", the whole page with the section replaced.
4201          */
4202         private function extractSections( $text, $section, $mode, $newtext='' ) {
4203                 # strip NOWIKI etc. to avoid confusion (true-parameter causes HTML
4204                 # comments to be stripped as well)
4205                 $striparray = array();
4206
4207                 $oldOutputType = $this->mOutputType;
4208                 $oldOptions = $this->mOptions;
4209                 $this->mOptions = new ParserOptions();
4210                 $this->mOutputType = OT_WIKI;
4211
4212                 $striptext = $this->strip( $text, $striparray, true );
4213
4214                 $this->mOutputType = $oldOutputType;
4215                 $this->mOptions = $oldOptions;
4216
4217                 # now that we can be sure that no pseudo-sections are in the source,
4218                 # split it up by section
4219                 $uniq = preg_quote( $this->uniqPrefix(), '/' );
4220                 $comment = "(?:$uniq-!--.*?QINU)";
4221                 $secs = preg_split(
4222                 /*
4223                         "/
4224                         ^(
4225                         (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4226                         (?:
4227                                 (=+) # Should this be limited to 6?
4228                                 .+?  # Section title...
4229                                 \\2  # Ending = count must match start
4230                         |
4231                                 ^
4232                                 <h([1-6])\b.*?>
4233                                 .*?
4234                                 <\/h\\3\s*>
4235                         )
4236                         (?:$comment|<\/?noinclude>|\s+)* # Trailing whitespace ok
4237                         )$
4238                         /mix",
4239                 */
4240                         "/
4241                         (
4242                                 ^
4243                                 (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4244                                 (=+) # Should this be limited to 6?
4245                                 .+?  # Section title...
4246                                 \\2  # Ending = count must match start
4247                                 (?:$comment|<\/?noinclude>|[ \\t]+)* # Trailing whitespace ok
4248                                 $
4249                         |
4250                                 <h([1-6])\b.*?>
4251                                 .*?
4252                                 <\/h\\3\s*>
4253                         )
4254                         /mix",
4255                         $striptext, -1,
4256                         PREG_SPLIT_DELIM_CAPTURE);
4257
4258                 if( $mode == "get" ) {
4259                         if( $section == 0 ) {
4260                                 // "Section 0" returns the content before any other section.
4261                                 $rv = $secs[0];
4262                         } else {
4263                                 $rv = "";
4264                         }
4265                 } elseif( $mode == "replace" ) {
4266                         if( $section == 0 ) {
4267                                 $rv = $newtext . "\n\n";
4268                                 $remainder = true;
4269                         } else {
4270                                 $rv = $secs[0];
4271                                 $remainder = false;
4272                         }
4273                 }
4274                 $count = 0;
4275                 $sectionLevel = 0;
4276                 for( $index = 1; $index < count( $secs ); ) {
4277                         $headerLine = $secs[$index++];
4278                         if( $secs[$index] ) {
4279                                 // A wiki header
4280                                 $headerLevel = strlen( $secs[$index++] );
4281                         } else {
4282                                 // An HTML header
4283                                 $index++;
4284                                 $headerLevel = intval( $secs[$index++] );
4285                         }
4286                         $content = $secs[$index++];
4287
4288                         $count++;
4289                         if( $mode == "get" ) {
4290                                 if( $count == $section ) {
4291                                         $rv = $headerLine . $content;
4292                                         $sectionLevel = $headerLevel;
4293                                 } elseif( $count > $section ) {
4294                                         if( $sectionLevel && $headerLevel > $sectionLevel ) {
4295                                                 $rv .= $headerLine . $content;
4296                                         } else {
4297                                                 // Broke out to a higher-level section
4298                                                 break;
4299                                         }
4300                                 }
4301                         } elseif( $mode == "replace" ) {
4302                                 if( $count < $section ) {
4303                                         $rv .= $headerLine . $content;
4304                                 } elseif( $count == $section ) {
4305                                         $rv .= $newtext . "\n\n";
4306                                         $sectionLevel = $headerLevel;
4307                                 } elseif( $count > $section ) {
4308                                         if( $headerLevel <= $sectionLevel ) {
4309                                                 // Passed the section's sub-parts.
4310                                                 $remainder = true;
4311                                         }
4312                                         if( $remainder ) {
4313                                                 $rv .= $headerLine . $content;
4314                                         }
4315                                 }
4316                         }
4317                 }
4318                 # reinsert stripped tags
4319                 $rv = $this->unstrip( $rv, $striparray );
4320                 $rv = $this->unstripNoWiki( $rv, $striparray );
4321                 $rv = trim( $rv );
4322                 return $rv;
4323         }
4324
4325         /**
4326          * This function returns the text of a section, specified by a number ($section).
4327          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4328          * the first section before any such heading (section 0).
4329          *
4330          * If a section contains subsections, these are also returned.
4331          *
4332          * @param $text String: text to look in
4333          * @param $section Integer: section number
4334          * @return string text of the requested section
4335          */
4336         function getSection( $text, $section ) {
4337                 return $this->extractSections( $text, $section, "get" );
4338         }
4339
4340         function replaceSection( $oldtext, $section, $text ) {
4341                 return $this->extractSections( $oldtext, $section, "replace", $text );
4342         }
4343
4344 }
4345
4346 /**
4347  * @todo document
4348  * @package MediaWiki
4349  */
4350 class ParserOutput
4351 {
4352         var $mText,             # The output text
4353                 $mLanguageLinks,    # List of the full text of language links, in the order they appear
4354                 $mCategories,       # Map of category names to sort keys
4355                 $mContainsOldMagic, # Boolean variable indicating if the input contained variables like {{CURRENTDAY}}
4356                 $mCacheTime,        # Time when this object was generated, or -1 for uncacheable. Used in ParserCache.
4357                 $mVersion,          # Compatibility check
4358                 $mTitleText,        # title text of the chosen language variant
4359                 $mLinks,            # 2-D map of NS/DBK to ID for the links in the document. ID=zero for broken.
4360                 $mTemplates,        # 2-D map of NS/DBK to ID for the template references. ID=zero for broken.
4361                 $mImages,           # DB keys of the images used, in the array key only
4362                 $mExternalLinks,    # External link URLs, in the key only
4363                 $mHTMLtitle,            # Display HTML title
4364                 $mSubtitle,                     # Additional subtitle
4365                 $mNewSection,           # Show a new section link?
4366                 $mNoGallery;            # No gallery on category page? (__NOGALLERY__)
4367
4368         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
4369                 $containsOldMagic = false, $titletext = '' )
4370         {
4371                 $this->mText = $text;
4372                 $this->mLanguageLinks = $languageLinks;
4373                 $this->mCategories = $categoryLinks;
4374                 $this->mContainsOldMagic = $containsOldMagic;
4375                 $this->mCacheTime = '';
4376                 $this->mVersion = MW_PARSER_VERSION;
4377                 $this->mTitleText = $titletext;
4378                 $this->mLinks = array();
4379                 $this->mTemplates = array();
4380                 $this->mImages = array();
4381                 $this->mExternalLinks = array();
4382                 $this->mHTMLtitle = "" ;
4383                 $this->mSubtitle = "" ;
4384                 $this->mNewSection = false;
4385                 $this->mNoGallery = false;
4386         }
4387
4388         function getText()                   { return $this->mText; }
4389         function &getLanguageLinks()          { return $this->mLanguageLinks; }
4390         function getCategoryLinks()          { return array_keys( $this->mCategories ); }
4391         function &getCategories()            { return $this->mCategories; }
4392         function getCacheTime()              { return $this->mCacheTime; }
4393         function getTitleText()              { return $this->mTitleText; }
4394         function &getLinks()                 { return $this->mLinks; }
4395         function &getTemplates()             { return $this->mTemplates; }
4396         function &getImages()                { return $this->mImages; }
4397         function &getExternalLinks()         { return $this->mExternalLinks; }
4398         function getNoGallery()              { return $this->mNoGallery; }
4399         function getSubtitle()               { return $this->mSubtitle; }
4400
4401         function containsOldMagic()          { return $this->mContainsOldMagic; }
4402         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
4403         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
4404         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategories, $cl ); }
4405         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
4406         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
4407         function setTitleText( $t )          { return wfSetVar($this->mTitleText, $t); }
4408         function setSubtitle( $st )          { return wfSetVar( $this->mSubtitle, $st ); }
4409
4410         function addCategory( $c, $sort )    { $this->mCategories[$c] = $sort; }
4411         function addImage( $name )           { $this->mImages[$name] = 1; }
4412         function addLanguageLink( $t )       { $this->mLanguageLinks[] = $t; }
4413         function addExternalLink( $url )     { $this->mExternalLinks[$url] = 1; }
4414
4415         function setNewSection( $value ) {
4416                 $this->mNewSection = (bool)$value;
4417         }
4418         function getNewSection() {
4419                 return (bool)$this->mNewSection;
4420         }
4421
4422         function addLink( $title, $id ) {
4423                 $ns = $title->getNamespace();
4424                 $dbk = $title->getDBkey();
4425                 if ( !isset( $this->mLinks[$ns] ) ) {
4426                         $this->mLinks[$ns] = array();
4427                 }
4428                 $this->mLinks[$ns][$dbk] = $id;
4429         }
4430
4431         function addTemplate( $title, $id ) {
4432                 $ns = $title->getNamespace();
4433                 $dbk = $title->getDBkey();
4434                 if ( !isset( $this->mTemplates[$ns] ) ) {
4435                         $this->mTemplates[$ns] = array();
4436                 }
4437                 $this->mTemplates[$ns][$dbk] = $id;
4438         }
4439
4440         /**
4441          * Return true if this cached output object predates the global or
4442          * per-article cache invalidation timestamps, or if it comes from
4443          * an incompatible older version.
4444          *
4445          * @param string $touched the affected article's last touched timestamp
4446          * @return bool
4447          * @public
4448          */
4449         function expired( $touched ) {
4450                 global $wgCacheEpoch;
4451                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
4452                        $this->getCacheTime() < $touched ||
4453                        $this->getCacheTime() <= $wgCacheEpoch ||
4454                        !isset( $this->mVersion ) ||
4455                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
4456         }
4457 }
4458
4459 /**
4460  * Set options of the Parser
4461  * @todo document
4462  * @package MediaWiki
4463  */
4464 class ParserOptions
4465 {
4466         # All variables are private
4467         var $mUseTeX;                    # Use texvc to expand <math> tags
4468         var $mUseDynamicDates;           # Use DateFormatter to format dates
4469         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
4470         var $mAllowExternalImages;       # Allow external images inline
4471         var $mAllowExternalImagesFrom;   # If not, any exception?
4472         var $mSkin;                      # Reference to the preferred skin
4473         var $mDateFormat;                # Date format index
4474         var $mEditSection;               # Create "edit section" links
4475         var $mNumberHeadings;            # Automatically number headings
4476         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
4477         var $mTidy;                      # Ask for tidy cleanup
4478         var $mInterfaceMessage;          # Which lang to call for PLURAL and GRAMMAR
4479         var $mMaxIncludeSize;            # Maximum size of template expansions, in bytes
4480
4481         var $mUser;                      # Stored user object, just used to initialise the skin
4482
4483         function getUseTeX()                        { return $this->mUseTeX; }
4484         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
4485         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
4486         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
4487         function getAllowExternalImagesFrom()       { return $this->mAllowExternalImagesFrom; }
4488         function getEditSection()                   { return $this->mEditSection; }
4489         function getNumberHeadings()                { return $this->mNumberHeadings; }
4490         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
4491         function getTidy()                          { return $this->mTidy; }
4492         function getInterfaceMessage()              { return $this->mInterfaceMessage; }
4493         function getMaxIncludeSize()                { return $this->mMaxIncludeSize; }
4494
4495         function &getSkin() {
4496                 if ( !isset( $this->mSkin ) ) {
4497                         $this->mSkin = $this->mUser->getSkin();
4498                 }
4499                 return $this->mSkin;
4500         }
4501
4502         function getDateFormat() {
4503                 if ( !isset( $this->mDateFormat ) ) {
4504                         $this->mDateFormat = $this->mUser->getDatePreference();
4505                 }
4506                 return $this->mDateFormat;
4507         }
4508
4509         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
4510         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
4511         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
4512         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
4513         function setAllowExternalImagesFrom( $x )   { return wfSetVar( $this->mAllowExternalImagesFrom, $x ); }
4514         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
4515         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
4516         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
4517         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
4518         function setTidy( $x )                      { return wfSetVar( $this->mTidy, $x); }
4519         function setSkin( &$x ) { $this->mSkin =& $x; }
4520         function setInterfaceMessage( $x )          { return wfSetVar( $this->mInterfaceMessage, $x); }
4521         function setMaxIncludeSize( $x )            { return wfSetVar( $this->mMaxIncludeSize, $x ); }
4522
4523         function ParserOptions( $user = null ) {
4524                 $this->initialiseFromUser( $user );
4525         }
4526
4527         /**
4528          * Get parser options
4529          * @static
4530          */
4531         static function newFromUser( $user ) {
4532                 return new ParserOptions( $user );
4533         }
4534
4535         /** Get user options */
4536         function initialiseFromUser( $userInput ) {
4537                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages;
4538                 global $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion, $wgMaxArticleSize;
4539                 $fname = 'ParserOptions::initialiseFromUser';
4540                 wfProfileIn( $fname );
4541                 if ( !$userInput ) {
4542                         global $wgUser;
4543                         if ( isset( $wgUser ) ) {
4544                                 $user = $wgUser;
4545                         } else {
4546                                 $user = new User;
4547                                 $user->setLoaded( true );
4548                         }
4549                 } else {
4550                         $user =& $userInput;
4551                 }
4552
4553                 $this->mUser = $user;
4554
4555                 $this->mUseTeX = $wgUseTeX;
4556                 $this->mUseDynamicDates = $wgUseDynamicDates;
4557                 $this->mInterwikiMagic = $wgInterwikiMagic;
4558                 $this->mAllowExternalImages = $wgAllowExternalImages;
4559                 $this->mAllowExternalImagesFrom = $wgAllowExternalImagesFrom;
4560                 $this->mSkin = null; # Deferred
4561                 $this->mDateFormat = null; # Deferred
4562                 $this->mEditSection = true;
4563                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
4564                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
4565                 $this->mTidy = false;
4566                 $this->mInterfaceMessage = false;
4567                 $this->mMaxIncludeSize = $wgMaxArticleSize * 1024;
4568                 wfProfileOut( $fname );
4569         }
4570 }
4571
4572 /**
4573  * Callback function used by Parser::replaceLinkHolders()
4574  * to substitute link placeholders.
4575  */
4576 function &wfOutputReplaceMatches( $matches ) {
4577         global $wgOutputReplace;
4578         return $wgOutputReplace[$matches[1]];
4579 }
4580
4581 /**
4582  * Return the total number of articles
4583  */
4584 function wfNumberOfArticles() {
4585         global $wgNumberOfArticles;
4586
4587         wfLoadSiteStats();
4588         return $wgNumberOfArticles;
4589 }
4590
4591 /**
4592  * Return the number of files
4593  */
4594 function wfNumberOfFiles() {
4595         $fname = 'wfNumberOfFiles';
4596
4597         wfProfileIn( $fname );
4598         $dbr =& wfGetDB( DB_SLAVE );
4599         $numImages = $dbr->selectField('site_stats', 'ss_images', array(), $fname );
4600         wfProfileOut( $fname );
4601
4602         return $numImages;
4603 }
4604
4605 /**
4606  * Return the number of user accounts
4607  * @return integer
4608  */
4609 function wfNumberOfUsers() {
4610         wfProfileIn( 'wfNumberOfUsers' );
4611         $dbr =& wfGetDB( DB_SLAVE );
4612         $count = $dbr->selectField( 'site_stats', 'ss_users', array(), 'wfNumberOfUsers' );
4613         wfProfileOut( 'wfNumberOfUsers' );
4614         return (int)$count;
4615 }
4616
4617 /**
4618  * Return the total number of pages
4619  * @return integer
4620  */
4621 function wfNumberOfPages() {
4622         wfProfileIn( 'wfNumberOfPages' );
4623         $dbr =& wfGetDB( DB_SLAVE );
4624         $count = $dbr->selectField( 'site_stats', 'ss_total_pages', array(), 'wfNumberOfPages' );
4625         wfProfileOut( 'wfNumberOfPages' );
4626         return (int)$count;
4627 }
4628
4629 /**
4630  * Return the total number of admins
4631  *
4632  * @return integer
4633  */
4634 function wfNumberOfAdmins() {
4635         static $admins = -1;
4636         wfProfileIn( 'wfNumberOfAdmins' );
4637         if( $admins == -1 ) {
4638                 $dbr =& wfGetDB( DB_SLAVE );
4639                 $admins = $dbr->selectField( 'user_groups', 'COUNT(*)', array( 'ug_group' => 'sysop' ), 'wfNumberOfAdmins' );
4640         }
4641         wfProfileOut( 'wfNumberOfAdmins' );
4642         return (int)$admins;
4643 }
4644
4645 /**
4646  * Count the number of pages in a particular namespace
4647  *
4648  * @param $ns Namespace
4649  * @return integer
4650  */
4651 function wfPagesInNs( $ns ) {
4652         static $pageCount = array();
4653         wfProfileIn( 'wfPagesInNs' );
4654         if( !isset( $pageCount[$ns] ) ) {
4655                 $dbr =& wfGetDB( DB_SLAVE );
4656                 $pageCount[$ns] = $dbr->selectField( 'page', 'COUNT(*)', array( 'page_namespace' => $ns ), 'wfPagesInNs' );
4657         }
4658         wfProfileOut( 'wfPagesInNs' );
4659         return (int)$pageCount[$ns];
4660 }
4661
4662 /**
4663  * Get various statistics from the database
4664  * @private
4665  */
4666 function wfLoadSiteStats() {
4667         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
4668         $fname = 'wfLoadSiteStats';
4669
4670         if ( -1 != $wgNumberOfArticles ) return;
4671         $dbr =& wfGetDB( DB_SLAVE );
4672         $s = $dbr->selectRow( 'site_stats',
4673                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
4674                 array( 'ss_row_id' => 1 ), $fname
4675         );
4676
4677         if ( $s === false ) {
4678                 return;
4679         } else {
4680                 $wgTotalViews = $s->ss_total_views;
4681                 $wgTotalEdits = $s->ss_total_edits;
4682                 $wgNumberOfArticles = $s->ss_good_articles;
4683         }
4684 }
4685
4686 /**
4687  * Escape html tags
4688  * Basically replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
4689  *
4690  * @param $in String: text that might contain HTML tags.
4691  * @return string Escaped string
4692  */
4693 function wfEscapeHTMLTagsOnly( $in ) {
4694         return str_replace(
4695                 array( '"', '>', '<' ),
4696                 array( '&quot;', '&gt;', '&lt;' ),
4697                 $in );
4698 }
4699
4700 ?>