includes/Parser.php

   1 <?php
   2 /**
   3  * File for Parser and related classes
   4  *
   5  * @addtogroup Parser
   6  */
   7
   8 /**
   9  * Update this version number when the ParserOutput format
  10  * changes in an incompatible way, so the parser cache
  11  * can automatically discard old data.
  12  */
  13 define( 'MW_PARSER_VERSION', '1.6.1' );
  14
  15 define( 'RLH_FOR_UPDATE', 1 );
  16
  17 # Allowed values for $mOutputType
  18 define( 'OT_HTML', 1 );
  19 define( 'OT_WIKI', 2 );
  20 define( 'OT_MSG' , 3 );
  21 define( 'OT_PREPROCESS', 4 );
  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 four 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  * preprocess()
  72  *   removes HTML comments and expands templates
  73  *
  74  * Globals used:
  75  *    objects:   $wgLang, $wgContLang
  76  *
  77  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  78  *
  79  * settings:
  80  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  81  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  82  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  83  *  $wgMaxArticleSize*
  84  *
  85  *  * only within ParserOptions
  86  * </pre>
  87  *
  88  */
  89 class Parser
  90 {
  91         /**#@+
  92          * @private
  93          */
  94         # Persistent:
  95         var $mTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables;
  96
  97         # Cleared with clearState():
  98         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
  99         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 100         var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
 101         var $mIncludeSizes, $mDefaultSort;
 102         var $mTemplates,        // cache of already loaded templates, avoids
 103                                 // multiple SQL queries for the same string
 104             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 105                                 // in this path. Used for loop detection.
 106
 107         # Temporary
 108         # These are variables reset at least once per parse regardless of $clearState
 109         var $mOptions,      // ParserOptions object
 110                 $mTitle,        // Title context, used for self-link rendering and similar things
 111                 $mOutputType,   // Output type, one of the OT_xxx constants
 112                 $ot,            // Shortcut alias, see setOutputType()
 113                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 114                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 115                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 116
 117         /**#@-*/
 118
 119         /**
 120          * Constructor
 121          *
 122          * @public
 123          */
 124         function Parser() {
 125                 $this->mTagHooks = array();
 126                 $this->mFunctionHooks = array();
 127                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 128                 $this->mFirstCall = true;
 129         }
 130
 131         /**
 132          * Do various kinds of initialisation on the first call of the parser
 133          */
 134         function firstCallInit() {
 135                 if ( !$this->mFirstCall ) {
 136                         return;
 137                 }
 138
 139                 wfProfileIn( __METHOD__ );
 140                 global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
 141
 142                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 143
 144                 $this->setFunctionHook( 'int', array( 'CoreParserFunctions', 'intFunction' ), SFH_NO_HASH );
 145                 $this->setFunctionHook( 'ns', array( 'CoreParserFunctions', 'ns' ), SFH_NO_HASH );
 146                 $this->setFunctionHook( 'urlencode', array( 'CoreParserFunctions', 'urlencode' ), SFH_NO_HASH );
 147                 $this->setFunctionHook( 'lcfirst', array( 'CoreParserFunctions', 'lcfirst' ), SFH_NO_HASH );
 148                 $this->setFunctionHook( 'ucfirst', array( 'CoreParserFunctions', 'ucfirst' ), SFH_NO_HASH );
 149                 $this->setFunctionHook( 'lc', array( 'CoreParserFunctions', 'lc' ), SFH_NO_HASH );
 150                 $this->setFunctionHook( 'uc', array( 'CoreParserFunctions', 'uc' ), SFH_NO_HASH );
 151                 $this->setFunctionHook( 'localurl', array( 'CoreParserFunctions', 'localurl' ), SFH_NO_HASH );
 152                 $this->setFunctionHook( 'localurle', array( 'CoreParserFunctions', 'localurle' ), SFH_NO_HASH );
 153                 $this->setFunctionHook( 'fullurl', array( 'CoreParserFunctions', 'fullurl' ), SFH_NO_HASH );
 154                 $this->setFunctionHook( 'fullurle', array( 'CoreParserFunctions', 'fullurle' ), SFH_NO_HASH );
 155                 $this->setFunctionHook( 'formatnum', array( 'CoreParserFunctions', 'formatnum' ), SFH_NO_HASH );
 156                 $this->setFunctionHook( 'grammar', array( 'CoreParserFunctions', 'grammar' ), SFH_NO_HASH );
 157                 $this->setFunctionHook( 'plural', array( 'CoreParserFunctions', 'plural' ), SFH_NO_HASH );
 158                 $this->setFunctionHook( 'numberofpages', array( 'CoreParserFunctions', 'numberofpages' ), SFH_NO_HASH );
 159                 $this->setFunctionHook( 'numberofusers', array( 'CoreParserFunctions', 'numberofusers' ), SFH_NO_HASH );
 160                 $this->setFunctionHook( 'numberofarticles', array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 161                 $this->setFunctionHook( 'numberoffiles', array( 'CoreParserFunctions', 'numberoffiles' ), SFH_NO_HASH );
 162                 $this->setFunctionHook( 'numberofadmins', array( 'CoreParserFunctions', 'numberofadmins' ), SFH_NO_HASH );
 163                 $this->setFunctionHook( 'language', array( 'CoreParserFunctions', 'language' ), SFH_NO_HASH );
 164                 $this->setFunctionHook( 'padleft', array( 'CoreParserFunctions', 'padleft' ), SFH_NO_HASH );
 165                 $this->setFunctionHook( 'padright', array( 'CoreParserFunctions', 'padright' ), SFH_NO_HASH );
 166                 $this->setFunctionHook( 'anchorencode', array( 'CoreParserFunctions', 'anchorencode' ), SFH_NO_HASH );
 167                 $this->setFunctionHook( 'special', array( 'CoreParserFunctions', 'special' ) );
 168                 $this->setFunctionHook( 'defaultsort', array( 'CoreParserFunctions', 'defaultsort' ), SFH_NO_HASH );
 169
 170                 if ( $wgAllowDisplayTitle ) {
 171                         $this->setFunctionHook( 'displaytitle', array( 'CoreParserFunctions', 'displaytitle' ), SFH_NO_HASH );
 172                 }
 173                 if ( $wgAllowSlowParserFunctions ) {
 174                         $this->setFunctionHook( 'pagesinnamespace', array( 'CoreParserFunctions', 'pagesinnamespace' ), SFH_NO_HASH );
 175                 }
 176
 177                 $this->initialiseVariables();
 178                 $this->mFirstCall = false;
 179                 wfProfileOut( __METHOD__ );
 180         }
 181
 182         /**
 183          * Clear Parser state
 184          *
 185          * @private
 186          */
 187         function clearState() {
 188                 wfProfileIn( __METHOD__ );
 189                 if ( $this->mFirstCall ) {
 190                         $this->firstCallInit();
 191                 }
 192                 $this->mOutput = new ParserOutput;
 193                 $this->mAutonumber = 0;
 194                 $this->mLastSection = '';
 195                 $this->mDTopen = false;
 196                 $this->mIncludeCount = array();
 197                 $this->mStripState = new StripState;
 198                 $this->mArgStack = array();
 199                 $this->mInPre = false;
 200                 $this->mInterwikiLinkHolders = array(
 201                         'texts' => array(),
 202                         'titles' => array()
 203                 );
 204                 $this->mLinkHolders = array(
 205                         'namespaces' => array(),
 206                         'dbkeys' => array(),
 207                         'queries' => array(),
 208                         'texts' => array(),
 209                         'titles' => array()
 210                 );
 211                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 212
 213                 /**
 214                  * Prefix for temporary replacement strings for the multipass parser.
 215                  * \x07 should never appear in input as it's disallowed in XML.
 216                  * Using it at the front also gives us a little extra robustness
 217                  * since it shouldn't match when butted up against identifier-like
 218                  * string constructs.
 219                  */
 220                 $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 221
 222                 # Clear these on every parse, bug 4549
 223                 $this->mTemplates = array();
 224                 $this->mTemplatePath = array();
 225
 226                 $this->mShowToc = true;
 227                 $this->mForceTocPosition = false;
 228                 $this->mIncludeSizes = array(
 229                         'pre-expand' => 0,
 230                         'post-expand' => 0,
 231                         'arg' => 0
 232                 );
 233                 $this->mDefaultSort = false;
 234
 235                 wfRunHooks( 'ParserClearState', array( &$this ) );
 236                 wfProfileOut( __METHOD__ );
 237         }
 238
 239         function setOutputType( $ot ) {
 240                 $this->mOutputType = $ot;
 241                 // Shortcut alias
 242                 $this->ot = array(
 243                         'html' => $ot == OT_HTML,
 244                         'wiki' => $ot == OT_WIKI,
 245                         'msg' => $ot == OT_MSG,
 246                         'pre' => $ot == OT_PREPROCESS,
 247                 );
 248         }
 249
 250         /**
 251          * Accessor for mUniqPrefix.
 252          *
 253          * @public
 254          */
 255         function uniqPrefix() {
 256                 return $this->mUniqPrefix;
 257         }
 258
 259         /**
 260          * Convert wikitext to HTML
 261          * Do not call this function recursively.
 262          *
 263          * @param string $text Text we want to parse
 264          * @param Title &$title A title object
 265          * @param array $options
 266          * @param boolean $linestart
 267          * @param boolean $clearState
 268          * @param int $revid number to pass in {{REVISIONID}}
 269          * @return ParserOutput a ParserOutput
 270          */
 271         public function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 272                 /**
 273                  * First pass--just handle <nowiki> sections, pass the rest off
 274                  * to internalParse() which does all the real work.
 275                  */
 276
 277                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 278                 $fname = 'Parser::parse-' . wfGetCaller();
 279                 wfProfileIn( __METHOD__ );
 280                 wfProfileIn( $fname );
 281
 282                 if ( $clearState ) {
 283                         $this->clearState();
 284                 }
 285
 286                 $this->mOptions = $options;
 287                 $this->mTitle =& $title;
 288                 $oldRevisionId = $this->mRevisionId;
 289                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 290                 if( $revid !== null ) {
 291                         $this->mRevisionId = $revid;
 292                         $this->mRevisionTimestamp = null;
 293                 }
 294                 $this->setOutputType( OT_HTML );
 295                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 296                 $text = $this->strip( $text, $this->mStripState );
 297                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 298                 $text = $this->internalParse( $text );
 299                 $text = $this->mStripState->unstripGeneral( $text );
 300
 301                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 302                 $fixtags = array(
 303                         # french spaces, last one Guillemet-left
 304                         # only if there is something before the space
 305                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 306                         # french spaces, Guillemet-right
 307                         '/(\\302\\253) /' => '\\1&nbsp;',
 308                 );
 309                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 310
 311                 # only once and last
 312                 $text = $this->doBlockLevels( $text, $linestart );
 313
 314                 $this->replaceLinkHolders( $text );
 315
 316                 # the position of the parserConvert() call should not be changed. it
 317                 # assumes that the links are all replaced and the only thing left
 318                 # is the <nowiki> mark.
 319                 # Side-effects: this calls $this->mOutput->setTitleText()
 320                 $text = $wgContLang->parserConvert( $text, $this );
 321
 322                 $text = $this->mStripState->unstripNoWiki( $text );
 323
 324                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 325
 326                 $text = Sanitizer::normalizeCharReferences( $text );
 327
 328                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 329                         $text = Parser::tidy($text);
 330                 } else {
 331                         # attempt to sanitize at least some nesting problems
 332                         # (bug #2702 and quite a few others)
 333                         $tidyregs = array(
 334                                 # ''Something [http://www.cool.com cool''] -->
 335                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 336                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 337                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 338                                 # fix up an anchor inside another anchor, only
 339                                 # at least for a single single nested link (bug 3695)
 340                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 341                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 342                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 343                                 # contains a div, so fix it up here; replace
 344                                 # div with escaped text
 345                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 346                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 347                                 # remove empty italic or bold tag pairs, some
 348                                 # introduced by rules above
 349                                 '/<([bi])><\/\\1>/' => '',
 350                         );
 351
 352                         $text = preg_replace(
 353                                 array_keys( $tidyregs ),
 354                                 array_values( $tidyregs ),
 355                                 $text );
 356                 }
 357
 358                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 359
 360                 # Information on include size limits, for the benefit of users who try to skirt them
 361                 if ( max( $this->mIncludeSizes ) > 1000 ) {
 362                         $max = $this->mOptions->getMaxIncludeSize();
 363                         $text .= "<!-- \n" .
 364                                 "Pre-expand include size: {$this->mIncludeSizes['pre-expand']} bytes\n" .
 365                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']} bytes\n" .
 366                                 "Template argument size: {$this->mIncludeSizes['arg']} bytes\n" .
 367                                 "Maximum: $max bytes\n" .
 368                                 "-->\n";
 369                 }
 370                 $this->mOutput->setText( $text );
 371                 $this->mRevisionId = $oldRevisionId;
 372                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 373                 wfProfileOut( $fname );
 374                 wfProfileOut( __METHOD__ );
 375
 376                 return $this->mOutput;
 377         }
 378
 379         /**
 380          * Recursive parser entry point that can be called from an extension tag
 381          * hook.
 382          */
 383         function recursiveTagParse( $text ) {
 384                 wfProfileIn( __METHOD__ );
 385                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 386                 $text = $this->strip( $text, $this->mStripState );
 387                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 388                 $text = $this->internalParse( $text );
 389                 wfProfileOut( __METHOD__ );
 390                 return $text;
 391         }
 392
 393         /**
 394          * Expand templates and variables in the text, producing valid, static wikitext.
 395          * Also removes comments.
 396          */
 397         function preprocess( $text, $title, $options ) {
 398                 wfProfileIn( __METHOD__ );
 399                 $this->clearState();
 400                 $this->setOutputType( OT_PREPROCESS );
 401                 $this->mOptions = $options;
 402                 $this->mTitle = $title;
 403                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 404                 $text = $this->strip( $text, $this->mStripState );
 405                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 406                 if ( $this->mOptions->getRemoveComments() ) {
 407                         $text = Sanitizer::removeHTMLcomments( $text );
 408                 }
 409                 $text = $this->replaceVariables( $text );
 410                 $text = $this->mStripState->unstripBoth( $text );
 411                 wfProfileOut( __METHOD__ );
 412                 return $text;
 413         }
 414
 415         /**
 416          * Get a random string
 417          *
 418          * @private
 419          * @static
 420          */
 421         function getRandomString() {
 422                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 423         }
 424
 425         function &getTitle() { return $this->mTitle; }
 426         function getOptions() { return $this->mOptions; }
 427
 428         function getFunctionLang() {
 429                 global $wgLang, $wgContLang;
 430                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 431         }
 432
 433         /**
 434          * Replaces all occurrences of HTML-style comments and the given tags
 435          * in the text with a random marker and returns teh next text. The output
 436          * parameter $matches will be an associative array filled with data in
 437          * the form:
 438          *   'UNIQ-xxxxx' => array(
 439          *     'element',
 440          *     'tag content',
 441          *     array( 'param' => 'x' ),
 442          *     '<element param="x">tag content</element>' ) )
 443          *
 444          * @param $elements list of element names. Comments are always extracted.
 445          * @param $text Source text string.
 446          * @param $uniq_prefix
 447          *
 448          * @private
 449          * @static
 450          */
 451         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 452                 static $n = 1;
 453                 $stripped = '';
 454                 $matches = array();
 455
 456                 $taglist = implode( '|', $elements );
 457                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 458
 459                 while ( '' != $text ) {
 460                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 461                         $stripped .= $p[0];
 462                         if( count( $p ) < 5 ) {
 463                                 break;
 464                         }
 465                         if( count( $p ) > 5 ) {
 466                                 // comment
 467                                 $element    = $p[4];
 468                                 $attributes = '';
 469                                 $close      = '';
 470                                 $inside     = $p[5];
 471                         } else {
 472                                 // tag
 473                                 $element    = $p[1];
 474                                 $attributes = $p[2];
 475                                 $close      = $p[3];
 476                                 $inside     = $p[4];
 477                         }
 478
 479                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . '-QINU';
 480                         $stripped .= $marker;
 481
 482                         if ( $close === '/>' ) {
 483                                 // Empty element tag, <tag />
 484                                 $content = null;
 485                                 $text = $inside;
 486                                 $tail = null;
 487                         } else {
 488                                 if( $element == '!--' ) {
 489                                         $end = '/(-->)/';
 490                                 } else {
 491                                         $end = "/(<\\/$element\\s*>)/i";
 492                                 }
 493                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 494                                 $content = $q[0];
 495                                 if( count( $q ) < 3 ) {
 496                                         # No end tag -- let it run out to the end of the text.
 497                                         $tail = '';
 498                                         $text = '';
 499                                 } else {
 500                                         $tail = $q[1];
 501                                         $text = $q[2];
 502                                 }
 503                         }
 504
 505                         $matches[$marker] = array( $element,
 506                                 $content,
 507                                 Sanitizer::decodeTagAttributes( $attributes ),
 508                                 "<$element$attributes$close$content$tail" );
 509                 }
 510                 return $stripped;
 511         }
 512
 513         /**
 514          * Strips and renders nowiki, pre, math, hiero
 515          * If $render is set, performs necessary rendering operations on plugins
 516          * Returns the text, and fills an array with data needed in unstrip()
 517          *
 518          * @param StripState $state
 519          *
 520          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 521          *  will be stripped in addition to other tags. This is important
 522          *  for section editing, where these comments cause confusion when
 523          *  counting the sections in the wikisource
 524          *
 525          * @param array dontstrip contains tags which should not be stripped;
 526          *  used to prevent stipping of <gallery> when saving (fixes bug 2700)
 527          *
 528          * @private
 529          */
 530         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 531                 global $wgContLang;
 532                 wfProfileIn( __METHOD__ );
 533                 $render = ($this->mOutputType == OT_HTML);
 534
 535                 $uniq_prefix = $this->mUniqPrefix;
 536                 $commentState = new ReplacementArray;
 537                 $nowikiItems = array();
 538                 $generalItems = array();
 539
 540                 $elements = array_merge(
 541                         array( 'nowiki', 'gallery' ),
 542                         array_keys( $this->mTagHooks ) );
 543                 global $wgRawHtml;
 544                 if( $wgRawHtml ) {
 545                         $elements[] = 'html';
 546                 }
 547                 if( $this->mOptions->getUseTeX() ) {
 548                         $elements[] = 'math';
 549                 }
 550
 551                 # Removing $dontstrip tags from $elements list (currently only 'gallery', fixing bug 2700)
 552                 foreach ( $elements AS $k => $v ) {
 553                         if ( !in_array ( $v , $dontstrip ) ) continue;
 554                         unset ( $elements[$k] );
 555                 }
 556
 557                 $matches = array();
 558                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 559
 560                 foreach( $matches as $marker => $data ) {
 561                         list( $element, $content, $params, $tag ) = $data;
 562                         if( $render ) {
 563                                 $tagName = strtolower( $element );
 564                                 wfProfileIn( __METHOD__."-render-$tagName" );
 565                                 switch( $tagName ) {
 566                                 case '!--':
 567                                         // Comment
 568                                         if( substr( $tag, -3 ) == '-->' ) {
 569                                                 $output = $tag;
 570                                         } else {
 571                                                 // Unclosed comment in input.
 572                                                 // Close it so later stripping can remove it
 573                                                 $output = "$tag-->";
 574                                         }
 575                                         break;
 576                                 case 'html':
 577                                         if( $wgRawHtml ) {
 578                                                 $output = $content;
 579                                                 break;
 580                                         }
 581                                         // Shouldn't happen otherwise. :)
 582                                 case 'nowiki':
 583                                         $output = Xml::escapeTagsOnly( $content );
 584                                         break;
 585                                 case 'math':
 586                                         $output = $wgContLang->armourMath( MathRenderer::renderMath( $content ) );
 587                                         break;
 588                                 case 'gallery':
 589                                         $output = $this->renderImageGallery( $content, $params );
 590                                         break;
 591                                 default:
 592                                         if( isset( $this->mTagHooks[$tagName] ) ) {
 593                                                 $output = call_user_func_array( $this->mTagHooks[$tagName],
 594                                                         array( $content, $params, $this ) );
 595                                         } else {
 596                                                 throw new MWException( "Invalid call hook $element" );
 597                                         }
 598                                 }
 599                                 wfProfileOut( __METHOD__."-render-$tagName" );
 600                         } else {
 601                                 // Just stripping tags; keep the source
 602                                 $output = $tag;
 603                         }
 604
 605                         // Unstrip the output, to support recursive strip() calls
 606                         $output = $state->unstripBoth( $output );
 607
 608                         if( !$stripcomments && $element == '!--' ) {
 609                                 $commentState->setPair( $marker, $output );
 610                         } elseif ( $element == 'html' || $element == 'nowiki' ) {
 611                                 $nowikiItems[$marker] = $output;
 612                         } else {
 613                                 $generalItems[$marker] = $output;
 614                         }
 615                 }
 616                 # Add the new items to the state
 617                 # We do this after the loop instead of during it to avoid slowing
 618                 # down the recursive unstrip
 619                 $state->nowiki->mergeArray( $nowikiItems );
 620                 $state->general->mergeArray( $generalItems );
 621
 622                 # Unstrip comments unless explicitly told otherwise.
 623                 # (The comments are always stripped prior to this point, so as to
 624                 # not invoke any extension tags / parser hooks contained within
 625                 # a comment.)
 626                 if ( !$stripcomments ) {
 627                         // Put them all back and forget them
 628                         $text = $commentState->replace( $text );
 629                 }
 630
 631                 wfProfileOut( __METHOD__ );
 632                 return $text;
 633         }
 634
 635         /**
 636          * Restores pre, math, and other extensions removed by strip()
 637          *
 638          * always call unstripNoWiki() after this one
 639          * @private
 640          * @deprecated use $this->mStripState->unstrip()
 641          */
 642         function unstrip( $text, $state ) {
 643                 return $state->unstripGeneral( $text );
 644         }
 645
 646         /**
 647          * Always call this after unstrip() to preserve the order
 648          *
 649          * @private
 650          * @deprecated use $this->mStripState->unstrip()
 651          */
 652         function unstripNoWiki( $text, $state ) {
 653                 return $state->unstripNoWiki( $text );
 654         }
 655
 656         /**
 657          * @deprecated use $this->mStripState->unstripBoth()
 658          */
 659         function unstripForHTML( $text ) {
 660                 return $this->mStripState->unstripBoth( $text );
 661         }
 662
 663         /**
 664          * Add an item to the strip state
 665          * Returns the unique tag which must be inserted into the stripped text
 666          * The tag will be replaced with the original text in unstrip()
 667          *
 668          * @private
 669          */
 670         function insertStripItem( $text, &$state ) {
 671                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 672                 $state->general->setPair( $rnd, $text );
 673                 return $rnd;
 674         }
 675
 676         /**
 677          * Interface with html tidy, used if $wgUseTidy = true.
 678          * If tidy isn't able to correct the markup, the original will be
 679          * returned in all its glory with a warning comment appended.
 680          *
 681          * Either the external tidy program or the in-process tidy extension
 682          * will be used depending on availability. Override the default
 683          * $wgTidyInternal setting to disable the internal if it's not working.
 684          *
 685          * @param string $text Hideous HTML input
 686          * @return string Corrected HTML output
 687          * @public
 688          * @static
 689          */
 690         function tidy( $text ) {
 691                 global $wgTidyInternal;
 692                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 693 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 694 '<head><title>test</title></head><body>'.$text.'</body></html>';
 695                 if( $wgTidyInternal ) {
 696                         $correctedtext = Parser::internalTidy( $wrappedtext );
 697                 } else {
 698                         $correctedtext = Parser::externalTidy( $wrappedtext );
 699                 }
 700                 if( is_null( $correctedtext ) ) {
 701                         wfDebug( "Tidy error detected!\n" );
 702                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 703                 }
 704                 return $correctedtext;
 705         }
 706
 707         /**
 708          * Spawn an external HTML tidy process and get corrected markup back from it.
 709          *
 710          * @private
 711          * @static
 712          */
 713         function externalTidy( $text ) {
 714                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 715                 $fname = 'Parser::externalTidy';
 716                 wfProfileIn( $fname );
 717
 718                 $cleansource = '';
 719                 $opts = ' -utf8';
 720
 721                 $descriptorspec = array(
 722                         0 => array('pipe', 'r'),
 723                         1 => array('pipe', 'w'),
 724                         2 => array('file', '/dev/null', 'a')
 725                 );
 726                 $pipes = array();
 727                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 728                 if (is_resource($process)) {
 729                         // Theoretically, this style of communication could cause a deadlock
 730                         // here. If the stdout buffer fills up, then writes to stdin could
 731                         // block. This doesn't appear to happen with tidy, because tidy only
 732                         // writes to stdout after it's finished reading from stdin. Search
 733                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 734                         fwrite($pipes[0], $text);
 735                         fclose($pipes[0]);
 736                         while (!feof($pipes[1])) {
 737                                 $cleansource .= fgets($pipes[1], 1024);
 738                         }
 739                         fclose($pipes[1]);
 740                         proc_close($process);
 741                 }
 742
 743                 wfProfileOut( $fname );
 744
 745                 if( $cleansource == '' && $text != '') {
 746                         // Some kind of error happened, so we couldn't get the corrected text.
 747                         // Just give up; we'll use the source text and append a warning.
 748                         return null;
 749                 } else {
 750                         return $cleansource;
 751                 }
 752         }
 753
 754         /**
 755          * Use the HTML tidy PECL extension to use the tidy library in-process,
 756          * saving the overhead of spawning a new process. Currently written to
 757          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 758          *
 759          * 'pear install tidy' should be able to compile the extension module.
 760          *
 761          * @private
 762          * @static
 763          */
 764         function internalTidy( $text ) {
 765                 global $wgTidyConf;
 766                 $fname = 'Parser::internalTidy';
 767                 wfProfileIn( $fname );
 768
 769                 tidy_load_config( $wgTidyConf );
 770                 tidy_set_encoding( 'utf8' );
 771                 tidy_parse_string( $text );
 772                 tidy_clean_repair();
 773                 if( tidy_get_status() == 2 ) {
 774                         // 2 is magic number for fatal error
 775                         // http://www.php.net/manual/en/function.tidy-get-status.php
 776                         $cleansource = null;
 777                 } else {
 778                         $cleansource = tidy_get_output();
 779                 }
 780                 wfProfileOut( $fname );
 781                 return $cleansource;
 782         }
 783
 784         /**
 785          * parse the wiki syntax used to render tables
 786          *
 787          * @private
 788          */
 789         function doTableStuff ( $text ) {
 790                 $fname = 'Parser::doTableStuff';
 791                 wfProfileIn( $fname );
 792
 793                 $lines = explode ( "\n" , $text );
 794                 $td_history = array (); // Is currently a td tag open?
 795                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 796                 $tr_history = array (); // Is currently a tr tag open?
 797                 $tr_attributes = array (); // history of tr attributes
 798                 $has_opened_tr = array(); // Did this table open a <tr> element?
 799                 $indent_level = 0; // indent level of the table
 800                 foreach ( $lines as $key => $line )
 801                 {
 802                         $line = trim ( $line );
 803
 804                         if( $line == '' ) { // empty line, go to next line
 805                                 continue;
 806                         }
 807                         $first_character = $line{0};
 808                         $matches = array();
 809
 810                         if ( preg_match( '/^(:*)\{\|(.*)$/' , $line , $matches ) ) {
 811                                 // First check if we are starting a new table
 812                                 $indent_level = strlen( $matches[1] );
 813
 814                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 815                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 816
 817                                 $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 818                                 array_push ( $td_history , false );
 819                                 array_push ( $last_tag_history , '' );
 820                                 array_push ( $tr_history , false );
 821                                 array_push ( $tr_attributes , '' );
 822                                 array_push ( $has_opened_tr , false );
 823                         } else if ( count ( $td_history ) == 0 ) {
 824                                 // Don't do any of the following
 825                                 continue;
 826                         } else if ( substr ( $line , 0 , 2 ) == '|}' ) {
 827                                 // We are ending a table
 828                                 $line = '</table>' . substr ( $line , 2 );
 829                                 $last_tag = array_pop ( $last_tag_history );
 830
 831                                 if ( !array_pop ( $has_opened_tr ) ) {
 832                                         $line = "<tr><td></td></tr>{$line}";
 833                                 }
 834
 835                                 if ( array_pop ( $tr_history ) ) {
 836                                         $line = "</tr>{$line}";
 837                                 }
 838
 839                                 if ( array_pop ( $td_history ) ) {
 840                                         $line = "</{$last_tag}>{$line}";
 841                                 }
 842                                 array_pop ( $tr_attributes );
 843                                 $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );
 844                         } else if ( substr ( $line , 0 , 2 ) == '|-' ) {
 845                                 // Now we have a table row
 846                                 $line = preg_replace( '#^\|-+#', '', $line );
 847
 848                                 // Whats after the tag is now only attributes
 849                                 $attributes = $this->mStripState->unstripBoth( $line );
 850                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 851                                 array_pop ( $tr_attributes );
 852                                 array_push ( $tr_attributes , $attributes );
 853
 854                                 $line = '';
 855                                 $last_tag = array_pop ( $last_tag_history );
 856                                 array_pop ( $has_opened_tr );
 857                                 array_push ( $has_opened_tr , true );
 858
 859                                 if ( array_pop ( $tr_history ) ) {
 860                                         $line = '</tr>';
 861                                 }
 862
 863                                 if ( array_pop ( $td_history ) ) {
 864                                         $line = "</{$last_tag}>{$line}";
 865                                 }
 866
 867                                 $lines[$key] = $line;
 868                                 array_push ( $tr_history , false );
 869                                 array_push ( $td_history , false );
 870                                 array_push ( $last_tag_history , '' );
 871                         }
 872                         else if ( $first_character == '|' || $first_character == '!' || substr ( $line , 0 , 2 )  == '|+' ) {
 873                                 // This might be cell elements, td, th or captions
 874                                 if ( substr ( $line , 0 , 2 ) == '|+' ) {
 875                                         $first_character = '+';
 876                                         $line = substr ( $line , 1 );
 877                                 }
 878
 879                                 $line = substr ( $line , 1 );
 880
 881                                 if ( $first_character == '!' ) {
 882                                         $line = str_replace ( '!!' , '||' , $line );
 883                                 }
 884
 885                                 // Split up multiple cells on the same line.
 886                                 // FIXME : This can result in improper nesting of tags processed
 887                                 // by earlier parser steps, but should avoid splitting up eg
 888                                 // attribute values containing literal "||".
 889                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 890
 891                                 $lines[$key] = '';
 892
 893                                 // Loop through each table cell
 894                                 foreach ( $cells as $cell )
 895                                 {
 896                                         $previous = '';
 897                                         if ( $first_character != '+' )
 898                                         {
 899                                                 $tr_after = array_pop ( $tr_attributes );
 900                                                 if ( !array_pop ( $tr_history ) ) {
 901                                                         $previous = "<tr{$tr_after}>\n";
 902                                                 }
 903                                                 array_push ( $tr_history , true );
 904                                                 array_push ( $tr_attributes , '' );
 905                                                 array_pop ( $has_opened_tr );
 906                                                 array_push ( $has_opened_tr , true );
 907                                         }
 908
 909                                         $last_tag = array_pop ( $last_tag_history );
 910
 911                                         if ( array_pop ( $td_history ) ) {
 912                                                 $previous = "</{$last_tag}>{$previous}";
 913                                         }
 914
 915                                         if ( $first_character == '|' ) {
 916                                                 $last_tag = 'td';
 917                                         } else if ( $first_character == '!' ) {
 918                                                 $last_tag = 'th';
 919                                         } else if ( $first_character == '+' ) {
 920                                                 $last_tag = 'caption';
 921                                         } else {
 922                                                 $last_tag = '';
 923                                         }
 924
 925                                         array_push ( $last_tag_history , $last_tag );
 926
 927                                         // A cell could contain both parameters and data
 928                                         $cell_data = explode ( '|' , $cell , 2 );
 929
 930                                         // Bug 553: Note that a '|' inside an invalid link should not
 931                                         // be mistaken as delimiting cell parameters
 932                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 933                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 934                                         } else if ( count ( $cell_data ) == 1 )
 935                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 936                                         else {
 937                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 938                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 939                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 940                                         }
 941
 942                                         $lines[$key] .= $cell;
 943                                         array_push ( $td_history , true );
 944                                 }
 945                         }
 946                 }
 947
 948                 // Closing open td, tr && table
 949                 while ( count ( $td_history ) > 0 )
 950                 {
 951                         if ( array_pop ( $td_history ) ) {
 952                                 $lines[] = '</td>' ;
 953                         }
 954                         if ( array_pop ( $tr_history ) ) {
 955                                 $lines[] = '</tr>' ;
 956                         }
 957                         if ( !array_pop ( $has_opened_tr ) ) {
 958                                 $lines[] = "<tr><td></td></tr>" ;
 959                         }
 960
 961                         $lines[] = '</table>' ;
 962                 }
 963
 964                 $output = implode ( "\n" , $lines ) ;
 965
 966                 // special case: don't return empty table
 967                 if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {
 968                         $output = '';
 969                 }
 970
 971                 wfProfileOut( $fname );
 972
 973                 return $output;
 974         }
 975
 976         /**
 977          * Helper function for parse() that transforms wiki markup into
 978          * HTML. Only called for $mOutputType == OT_HTML.
 979          *
 980          * @private
 981          */
 982         function internalParse( $text ) {
 983                 $args = array();
 984                 $isMain = true;
 985                 $fname = 'Parser::internalParse';
 986                 wfProfileIn( $fname );
 987
 988                 # Hook to suspend the parser in this state
 989                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 990                         wfProfileOut( $fname );
 991                         return $text ;
 992                 }
 993
 994                 # Remove <noinclude> tags and <includeonly> sections
 995                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 996                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 997                 $text = StringUtils::delimiterReplace( '<includeonly>', '</includeonly>', '', $text );
 998
 999                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
1000
1001                 $text = $this->replaceVariables( $text, $args );
1002
1003                 // Tables need to come after variable replacement for things to work
1004                 // properly; putting them before other transformations should keep
1005                 // exciting things like link expansions from showing up in surprising
1006                 // places.
1007                 $text = $this->doTableStuff( $text );
1008
1009                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1010
1011                 $text = $this->stripToc( $text );
1012                 $this->stripNoGallery( $text );
1013                 $text = $this->doHeadings( $text );
1014                 if($this->mOptions->getUseDynamicDates()) {
1015                         $df =& DateFormatter::getInstance();
1016                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1017                 }
1018                 $text = $this->doAllQuotes( $text );
1019                 $text = $this->replaceInternalLinks( $text );
1020                 $text = $this->replaceExternalLinks( $text );
1021
1022                 # replaceInternalLinks may sometimes leave behind
1023                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1024                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
1025
1026                 $text = $this->doMagicLinks( $text );
1027                 $text = $this->formatHeadings( $text, $isMain );
1028
1029                 wfProfileOut( $fname );
1030                 return $text;
1031         }
1032
1033         /**
1034          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1035          * magic external links.
1036          *
1037          * @private
1038          */
1039         function &doMagicLinks( &$text ) {
1040                 wfProfileIn( __METHOD__ );
1041                 $text = preg_replace_callback(
1042                         '!(?:                           # Start cases
1043                             <a.*?</a> |                 # Skip link text
1044                             <.*?> |                     # Skip stuff inside HTML elements
1045                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
1046                             ISBN\s+(\b                  # ISBN, capture number as m[2]
1047                                      (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1048                                      (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1049                                      [0-9Xx]                 # check digit
1050                                    \b)
1051                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1052                 wfProfileOut( __METHOD__ );
1053                 return $text;
1054         }
1055
1056         function magicLinkCallback( $m ) {
1057                 if ( substr( $m[0], 0, 1 ) == '<' ) {
1058                         # Skip HTML element
1059                         return $m[0];
1060                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
1061                         $isbn = $m[2];
1062                         $num = strtr( $isbn, array(
1063                                 '-' => '',
1064                                 ' ' => '',
1065                                 'x' => 'X',
1066                         ));
1067                         $titleObj = SpecialPage::getTitleFor( 'Booksources' );
1068                         $text = '<a href="' .
1069                                 $titleObj->escapeLocalUrl( "isbn=$num" ) .
1070                                 "\" class=\"internal\">ISBN $isbn</a>";
1071                 } else {
1072                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
1073                                 $keyword = 'RFC';
1074                                 $urlmsg = 'rfcurl';
1075                                 $id = $m[1];
1076                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
1077                                 $keyword = 'PMID';
1078                                 $urlmsg = 'pubmedurl';
1079                                 $id = $m[1];
1080                         } else {
1081                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1082                                         substr($m[0], 0, 20 ) . '"' );
1083                         }
1084
1085                         $url = wfMsg( $urlmsg, $id);
1086                         $sk =& $this->mOptions->getSkin();
1087                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1088                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1089                 }
1090                 return $text;
1091         }
1092
1093         /**
1094          * Parse headers and return html
1095          *
1096          * @private
1097          */
1098         function doHeadings( $text ) {
1099                 $fname = 'Parser::doHeadings';
1100                 wfProfileIn( $fname );
1101                 for ( $i = 6; $i >= 1; --$i ) {
1102                         $h = str_repeat( '=', $i );
1103                         $text = preg_replace( "/^{$h}(.+){$h}\\s*$/m",
1104                           "<h{$i}>\\1</h{$i}>\\2", $text );
1105                 }
1106                 wfProfileOut( $fname );
1107                 return $text;
1108         }
1109
1110         /**
1111          * Replace single quotes with HTML markup
1112          * @private
1113          * @return string the altered text
1114          */
1115         function doAllQuotes( $text ) {
1116                 $fname = 'Parser::doAllQuotes';
1117                 wfProfileIn( $fname );
1118                 $outtext = '';
1119                 $lines = explode( "\n", $text );
1120                 foreach ( $lines as $line ) {
1121                         $outtext .= $this->doQuotes ( $line ) . "\n";
1122                 }
1123                 $outtext = substr($outtext, 0,-1);
1124                 wfProfileOut( $fname );
1125                 return $outtext;
1126         }
1127
1128         /**
1129          * Helper function for doAllQuotes()
1130          * @private
1131          */
1132         function doQuotes( $text ) {
1133                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1134                 if ( count( $arr ) == 1 )
1135                         return $text;
1136                 else
1137                 {
1138                         # First, do some preliminary work. This may shift some apostrophes from
1139                         # being mark-up to being text. It also counts the number of occurrences
1140                         # of bold and italics mark-ups.
1141                         $i = 0;
1142                         $numbold = 0;
1143                         $numitalics = 0;
1144                         foreach ( $arr as $r )
1145                         {
1146                                 if ( ( $i % 2 ) == 1 )
1147                                 {
1148                                         # If there are ever four apostrophes, assume the first is supposed to
1149                                         # be text, and the remaining three constitute mark-up for bold text.
1150                                         if ( strlen( $arr[$i] ) == 4 )
1151                                         {
1152                                                 $arr[$i-1] .= "'";
1153                                                 $arr[$i] = "'''";
1154                                         }
1155                                         # If there are more than 5 apostrophes in a row, assume they're all
1156                                         # text except for the last 5.
1157                                         else if ( strlen( $arr[$i] ) > 5 )
1158                                         {
1159                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1160                                                 $arr[$i] = "'''''";
1161                                         }
1162                                         # Count the number of occurrences of bold and italics mark-ups.
1163                                         # We are not counting sequences of five apostrophes.
1164                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1165                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1166                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1167                                 }
1168                                 $i++;
1169                         }
1170
1171                         # If there is an odd number of both bold and italics, it is likely
1172                         # that one of the bold ones was meant to be an apostrophe followed
1173                         # by italics. Which one we cannot know for certain, but it is more
1174                         # likely to be one that has a single-letter word before it.
1175                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1176                         {
1177                                 $i = 0;
1178                                 $firstsingleletterword = -1;
1179                                 $firstmultiletterword = -1;
1180                                 $firstspace = -1;
1181                                 foreach ( $arr as $r )
1182                                 {
1183                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1184                                         {
1185                                                 $x1 = substr ($arr[$i-1], -1);
1186                                                 $x2 = substr ($arr[$i-1], -2, 1);
1187                                                 if ($x1 == ' ') {
1188                                                         if ($firstspace == -1) $firstspace = $i;
1189                                                 } else if ($x2 == ' ') {
1190                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1191                                                 } else {
1192                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1193                                                 }
1194                                         }
1195                                         $i++;
1196                                 }
1197
1198                                 # If there is a single-letter word, use it!
1199                                 if ($firstsingleletterword > -1)
1200                                 {
1201                                         $arr [ $firstsingleletterword ] = "''";
1202                                         $arr [ $firstsingleletterword-1 ] .= "'";
1203                                 }
1204                                 # If not, but there's a multi-letter word, use that one.
1205                                 else if ($firstmultiletterword > -1)
1206                                 {
1207                                         $arr [ $firstmultiletterword ] = "''";
1208                                         $arr [ $firstmultiletterword-1 ] .= "'";
1209                                 }
1210                                 # ... otherwise use the first one that has neither.
1211                                 # (notice that it is possible for all three to be -1 if, for example,
1212                                 # there is only one pentuple-apostrophe in the line)
1213                                 else if ($firstspace > -1)
1214                                 {
1215                                         $arr [ $firstspace ] = "''";
1216                                         $arr [ $firstspace-1 ] .= "'";
1217                                 }
1218                         }
1219
1220                         # Now let's actually convert our apostrophic mush to HTML!
1221                         $output = '';
1222                         $buffer = '';
1223                         $state = '';
1224                         $i = 0;
1225                         foreach ($arr as $r)
1226                         {
1227                                 if (($i % 2) == 0)
1228                                 {
1229                                         if ($state == 'both')
1230                                                 $buffer .= $r;
1231                                         else
1232                                                 $output .= $r;
1233                                 }
1234                                 else
1235                                 {
1236                                         if (strlen ($r) == 2)
1237                                         {
1238                                                 if ($state == 'i')
1239                                                 { $output .= '</i>'; $state = ''; }
1240                                                 else if ($state == 'bi')
1241                                                 { $output .= '</i>'; $state = 'b'; }
1242                                                 else if ($state == 'ib')
1243                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1244                                                 else if ($state == 'both')
1245                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1246                                                 else # $state can be 'b' or ''
1247                                                 { $output .= '<i>'; $state .= 'i'; }
1248                                         }
1249                                         else if (strlen ($r) == 3)
1250                                         {
1251                                                 if ($state == 'b')
1252                                                 { $output .= '</b>'; $state = ''; }
1253                                                 else if ($state == 'bi')
1254                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1255                                                 else if ($state == 'ib')
1256                                                 { $output .= '</b>'; $state = 'i'; }
1257                                                 else if ($state == 'both')
1258                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1259                                                 else # $state can be 'i' or ''
1260                                                 { $output .= '<b>'; $state .= 'b'; }
1261                                         }
1262                                         else if (strlen ($r) == 5)
1263                                         {
1264                                                 if ($state == 'b')
1265                                                 { $output .= '</b><i>'; $state = 'i'; }
1266                                                 else if ($state == 'i')
1267                                                 { $output .= '</i><b>'; $state = 'b'; }
1268                                                 else if ($state == 'bi')
1269                                                 { $output .= '</i></b>'; $state = ''; }
1270                                                 else if ($state == 'ib')
1271                                                 { $output .= '</b></i>'; $state = ''; }
1272                                                 else if ($state == 'both')
1273                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1274                                                 else # ($state == '')
1275                                                 { $buffer = ''; $state = 'both'; }
1276                                         }
1277                                 }
1278                                 $i++;
1279                         }
1280                         # Now close all remaining tags.  Notice that the order is important.
1281                         if ($state == 'b' || $state == 'ib')
1282                                 $output .= '</b>';
1283                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1284                                 $output .= '</i>';
1285                         if ($state == 'bi')
1286                                 $output .= '</b>';
1287                         if ($state == 'both')
1288                                 $output .= '<b><i>'.$buffer.'</i></b>';
1289                         return $output;
1290                 }
1291         }
1292
1293         /**
1294          * Replace external links
1295          *
1296          * Note: this is all very hackish and the order of execution matters a lot.
1297          * Make sure to run maintenance/parserTests.php if you change this code.
1298          *
1299          * @private
1300          */
1301         function replaceExternalLinks( $text ) {
1302                 global $wgContLang;
1303                 $fname = 'Parser::replaceExternalLinks';
1304                 wfProfileIn( $fname );
1305
1306                 $sk =& $this->mOptions->getSkin();
1307
1308                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1309
1310                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1311
1312                 $i = 0;
1313                 while ( $i<count( $bits ) ) {
1314                         $url = $bits[$i++];
1315                         $protocol = $bits[$i++];
1316                         $text = $bits[$i++];
1317                         $trail = $bits[$i++];
1318
1319                         # The characters '<' and '>' (which were escaped by
1320                         # removeHTMLtags()) should not be included in
1321                         # URLs, per RFC 2396.
1322                         $m2 = array();
1323                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1324                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1325                                 $url = substr($url, 0, $m2[0][1]);
1326                         }
1327
1328                         # If the link text is an image URL, replace it with an <img> tag
1329                         # This happened by accident in the original parser, but some people used it extensively
1330                         $img = $this->maybeMakeExternalImage( $text );
1331                         if ( $img !== false ) {
1332                                 $text = $img;
1333                         }
1334
1335                         $dtrail = '';
1336
1337                         # Set linktype for CSS - if URL==text, link is essentially free
1338                         $linktype = ($text == $url) ? 'free' : 'text';
1339
1340                         # No link text, e.g. [http://domain.tld/some.link]
1341                         if ( $text == '' ) {
1342                                 # Autonumber if allowed. See bug #5918
1343                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1344                                         $text = '[' . ++$this->mAutonumber . ']';
1345                                         $linktype = 'autonumber';
1346                                 } else {
1347                                         # Otherwise just use the URL
1348                                         $text = htmlspecialchars( $url );
1349                                         $linktype = 'free';
1350                                 }
1351                         } else {
1352                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1353                                 # Check for trail
1354                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1355                         }
1356
1357                         $text = $wgContLang->markNoConversion($text);
1358
1359                         $url = Sanitizer::cleanUrl( $url );
1360
1361                         # Process the trail (i.e. everything after this link up until start of the next link),
1362                         # replacing any non-bracketed links
1363                         $trail = $this->replaceFreeExternalLinks( $trail );
1364
1365                         # Use the encoded URL
1366                         # This means that users can paste URLs directly into the text
1367                         # Funny characters like &ouml; aren't valid in URLs anyway
1368                         # This was changed in August 2004
1369                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1370
1371                         # Register link in the output object.
1372                         # Replace unnecessary URL escape codes with the referenced character
1373                         # This prevents spammers from hiding links from the filters
1374                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1375                         $this->mOutput->addExternalLink( $pasteurized );
1376                 }
1377
1378                 wfProfileOut( $fname );
1379                 return $s;
1380         }
1381
1382         /**
1383          * Replace anything that looks like a URL with a link
1384          * @private
1385          */
1386         function replaceFreeExternalLinks( $text ) {
1387                 global $wgContLang;
1388                 $fname = 'Parser::replaceFreeExternalLinks';
1389                 wfProfileIn( $fname );
1390
1391                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1392                 $s = array_shift( $bits );
1393                 $i = 0;
1394
1395                 $sk =& $this->mOptions->getSkin();
1396
1397                 while ( $i < count( $bits ) ){
1398                         $protocol = $bits[$i++];
1399                         $remainder = $bits[$i++];
1400
1401                         $m = array();
1402                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1403                                 # Found some characters after the protocol that look promising
1404                                 $url = $protocol . $m[1];
1405                                 $trail = $m[2];
1406
1407                                 # special case: handle urls as url args:
1408                                 # http://www.example.com/foo?=http://www.example.com/bar
1409                                 if(strlen($trail) == 0 &&
1410                                         isset($bits[$i]) &&
1411                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1412                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1413                                 {
1414                                         # add protocol, arg
1415                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1416                                         $i += 2;
1417                                         $trail = $m[2];
1418                                 }
1419
1420                                 # The characters '<' and '>' (which were escaped by
1421                                 # removeHTMLtags()) should not be included in
1422                                 # URLs, per RFC 2396.
1423                                 $m2 = array();
1424                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1425                                         $trail = substr($url, $m2[0][1]) . $trail;
1426                                         $url = substr($url, 0, $m2[0][1]);
1427                                 }
1428
1429                                 # Move trailing punctuation to $trail
1430                                 $sep = ',;\.:!?';
1431                                 # If there is no left bracket, then consider right brackets fair game too
1432                                 if ( strpos( $url, '(' ) === false ) {
1433                                         $sep .= ')';
1434                                 }
1435
1436                                 $numSepChars = strspn( strrev( $url ), $sep );
1437                                 if ( $numSepChars ) {
1438                                         $trail = substr( $url, -$numSepChars ) . $trail;
1439                                         $url = substr( $url, 0, -$numSepChars );
1440                                 }
1441
1442                                 $url = Sanitizer::cleanUrl( $url );
1443
1444                                 # Is this an external image?
1445                                 $text = $this->maybeMakeExternalImage( $url );
1446                                 if ( $text === false ) {
1447                                         # Not an image, make a link
1448                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1449                                         # Register it in the output object...
1450                                         # Replace unnecessary URL escape codes with their equivalent characters
1451                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1452                                         $this->mOutput->addExternalLink( $pasteurized );
1453                                 }
1454                                 $s .= $text . $trail;
1455                         } else {
1456                                 $s .= $protocol . $remainder;
1457                         }
1458                 }
1459                 wfProfileOut( $fname );
1460                 return $s;
1461         }
1462
1463         /**
1464          * Replace unusual URL escape codes with their equivalent characters
1465          * @param string
1466          * @return string
1467          * @static
1468          * @fixme This can merge genuinely required bits in the path or query string,
1469          *        breaking legit URLs. A proper fix would treat the various parts of
1470          *        the URL differently; as a workaround, just use the output for
1471          *        statistical records, not for actual linking/output.
1472          */
1473         static function replaceUnusualEscapes( $url ) {
1474                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1475                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1476         }
1477
1478         /**
1479          * Callback function used in replaceUnusualEscapes().
1480          * Replaces unusual URL escape codes with their equivalent character
1481          * @static
1482          * @private
1483          */
1484         private static function replaceUnusualEscapesCallback( $matches ) {
1485                 $char = urldecode( $matches[0] );
1486                 $ord = ord( $char );
1487                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1488                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1489                         // No, shouldn't be escaped
1490                         return $char;
1491                 } else {
1492                         // Yes, leave it escaped
1493                         return $matches[0];
1494                 }
1495         }
1496
1497         /**
1498          * make an image if it's allowed, either through the global
1499          * option or through the exception
1500          * @private
1501          */
1502         function maybeMakeExternalImage( $url ) {
1503                 $sk =& $this->mOptions->getSkin();
1504                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1505                 $imagesexception = !empty($imagesfrom);
1506                 $text = false;
1507                 if ( $this->mOptions->getAllowExternalImages()
1508                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1509                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1510                                 # Image found
1511                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1512                         }
1513                 }
1514                 return $text;
1515         }
1516
1517         /**
1518          * Process [[ ]] wikilinks
1519          *
1520          * @private
1521          */
1522         function replaceInternalLinks( $s ) {
1523                 global $wgContLang;
1524                 static $fname = 'Parser::replaceInternalLinks' ;
1525
1526                 wfProfileIn( $fname );
1527
1528                 wfProfileIn( $fname.'-setup' );
1529                 static $tc = FALSE;
1530                 # the % is needed to support urlencoded titles as well
1531                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1532
1533                 $sk =& $this->mOptions->getSkin();
1534
1535                 #split the entire text string on occurences of [[
1536                 $a = explode( '[[', ' ' . $s );
1537                 #get the first element (all text up to first [[), and remove the space we added
1538                 $s = array_shift( $a );
1539                 $s = substr( $s, 1 );
1540
1541                 # Match a link having the form [[namespace:link|alternate]]trail
1542                 static $e1 = FALSE;
1543                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1544                 # Match cases where there is no "]]", which might still be images
1545                 static $e1_img = FALSE;
1546                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1547                 # Match the end of a line for a word that's not followed by whitespace,
1548                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1549                 $e2 = wfMsgForContent( 'linkprefix' );
1550
1551                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1552                 if( is_null( $this->mTitle ) ) {
1553                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1554                 }
1555                 $nottalk = !$this->mTitle->isTalkPage();
1556
1557                 if ( $useLinkPrefixExtension ) {
1558                         $m = array();
1559                         if ( preg_match( $e2, $s, $m ) ) {
1560                                 $first_prefix = $m[2];
1561                         } else {
1562                                 $first_prefix = false;
1563                         }
1564                 } else {
1565                         $prefix = '';
1566                 }
1567
1568                 if($wgContLang->hasVariants()) {
1569                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1570                 } else {
1571                         $selflink = array($this->mTitle->getPrefixedText());
1572                 }
1573                 $useSubpages = $this->areSubpagesAllowed();
1574                 wfProfileOut( $fname.'-setup' );
1575
1576                 # Loop for each link
1577                 for ($k = 0; isset( $a[$k] ); $k++) {
1578                         $line = $a[$k];
1579                         if ( $useLinkPrefixExtension ) {
1580                                 wfProfileIn( $fname.'-prefixhandling' );
1581                                 if ( preg_match( $e2, $s, $m ) ) {
1582                                         $prefix = $m[2];
1583                                         $s = $m[1];
1584                                 } else {
1585                                         $prefix='';
1586                                 }
1587                                 # first link
1588                                 if($first_prefix) {
1589                                         $prefix = $first_prefix;
1590                                         $first_prefix = false;
1591                                 }
1592                                 wfProfileOut( $fname.'-prefixhandling' );
1593                         }
1594
1595                         $might_be_img = false;
1596
1597                         wfProfileIn( "$fname-e1" );
1598                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1599                                 $text = $m[2];
1600                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1601                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1602                                 # the real problem is with the $e1 regex
1603                                 # See bug 1300.
1604                                 #
1605                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1606                                 # and no image is in sight. See bug 2095.
1607                                 #
1608                                 if( $text !== '' &&
1609                                         substr( $m[3], 0, 1 ) === ']' &&
1610                                         strpos($text, '[') !== false
1611                                 )
1612                                 {
1613                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1614                                         $m[3] = substr( $m[3], 1 );
1615                                 }
1616                                 # fix up urlencoded title texts
1617                                 if( strpos( $m[1], '%' ) !== false ) {
1618                                         # Should anchors '#' also be rejected?
1619                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1620                                 }
1621                                 $trail = $m[3];
1622                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1623                                 $might_be_img = true;
1624                                 $text = $m[2];
1625                                 if ( strpos( $m[1], '%' ) !== false ) {
1626                                        $m[1] = urldecode($m[1]);
1627                                 }
1628                                 $trail = "";
1629                         } else { # Invalid form; output directly
1630                                 $s .= $prefix . '[[' . $line ;
1631                                 wfProfileOut( "$fname-e1" );
1632                                 continue;
1633                         }
1634                         wfProfileOut( "$fname-e1" );
1635                         wfProfileIn( "$fname-misc" );
1636
1637                         # Don't allow internal links to pages containing
1638                         # PROTO: where PROTO is a valid URL protocol; these
1639                         # should be external links.
1640                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1641                                 $s .= $prefix . '[[' . $line ;
1642                                 continue;
1643                         }
1644
1645                         # Make subpage if necessary
1646                         if( $useSubpages ) {
1647                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1648                         } else {
1649                                 $link = $m[1];
1650                         }
1651
1652                         $noforce = (substr($m[1], 0, 1) != ':');
1653                         if (!$noforce) {
1654                                 # Strip off leading ':'
1655                                 $link = substr($link, 1);
1656                         }
1657
1658                         wfProfileOut( "$fname-misc" );
1659                         wfProfileIn( "$fname-title" );
1660                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1661                         if( !$nt ) {
1662                                 $s .= $prefix . '[[' . $line;
1663                                 wfProfileOut( "$fname-title" );
1664                                 continue;
1665                         }
1666
1667                         $ns = $nt->getNamespace();
1668                         $iw = $nt->getInterWiki();
1669                         wfProfileOut( "$fname-title" );
1670
1671                         if ($might_be_img) { # if this is actually an invalid link
1672                                 wfProfileIn( "$fname-might_be_img" );
1673                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1674                                         $found = false;
1675                                         while (isset ($a[$k+1]) ) {
1676                                                 #look at the next 'line' to see if we can close it there
1677                                                 $spliced = array_splice( $a, $k + 1, 1 );
1678                                                 $next_line = array_shift( $spliced );
1679                                                 $m = explode( ']]', $next_line, 3 );
1680                                                 if ( count( $m ) == 3 ) {
1681                                                         # the first ]] closes the inner link, the second the image
1682                                                         $found = true;
1683                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1684                                                         $trail = $m[2];
1685                                                         break;
1686                                                 } elseif ( count( $m ) == 2 ) {
1687                                                         #if there's exactly one ]] that's fine, we'll keep looking
1688                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1689                                                 } else {
1690                                                         #if $next_line is invalid too, we need look no further
1691                                                         $text .= '[[' . $next_line;
1692                                                         break;
1693                                                 }
1694                                         }
1695                                         if ( !$found ) {
1696                                                 # we couldn't find the end of this imageLink, so output it raw
1697                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1698                                                 $text = $this->replaceInternalLinks($text);
1699                                                 $s .= "{$prefix}[[$link|$text";
1700                                                 # note: no $trail, because without an end, there *is* no trail
1701                                                 wfProfileOut( "$fname-might_be_img" );
1702                                                 continue;
1703                                         }
1704                                 } else { #it's not an image, so output it raw
1705                                         $s .= "{$prefix}[[$link|$text";
1706                                         # note: no $trail, because without an end, there *is* no trail
1707                                         wfProfileOut( "$fname-might_be_img" );
1708                                         continue;
1709                                 }
1710                                 wfProfileOut( "$fname-might_be_img" );
1711                         }
1712
1713                         $wasblank = ( '' == $text );
1714                         if( $wasblank ) $text = $link;
1715
1716                         # Link not escaped by : , create the various objects
1717                         if( $noforce ) {
1718
1719                                 # Interwikis
1720                                 wfProfileIn( "$fname-interwiki" );
1721                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1722                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1723                                         $s = rtrim($s . "\n");
1724                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1725                                         wfProfileOut( "$fname-interwiki" );
1726                                         continue;
1727                                 }
1728                                 wfProfileOut( "$fname-interwiki" );
1729
1730                                 if ( $ns == NS_IMAGE ) {
1731                                         wfProfileIn( "$fname-image" );
1732                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1733                                                 # recursively parse links inside the image caption
1734                                                 # actually, this will parse them in any other parameters, too,
1735                                                 # but it might be hard to fix that, and it doesn't matter ATM
1736                                                 $text = $this->replaceExternalLinks($text);
1737                                                 $text = $this->replaceInternalLinks($text);
1738
1739                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1740                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1741                                                 $this->mOutput->addImage( $nt->getDBkey() );
1742
1743                                                 wfProfileOut( "$fname-image" );
1744                                                 continue;
1745                                         } else {
1746                                                 # We still need to record the image's presence on the page
1747                                                 $this->mOutput->addImage( $nt->getDBkey() );
1748                                         }
1749                                         wfProfileOut( "$fname-image" );
1750
1751                                 }
1752
1753                                 if ( $ns == NS_CATEGORY ) {
1754                                         wfProfileIn( "$fname-category" );
1755                                         $s = rtrim($s . "\n"); # bug 87
1756
1757                                         if ( $wasblank ) {
1758                                                 $sortkey = $this->getDefaultSort();
1759                                         } else {
1760                                                 $sortkey = $text;
1761                                         }
1762                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1763                                         $sortkey = str_replace( "\n", '', $sortkey );
1764                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1765                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1766
1767                                         /**
1768                                          * Strip the whitespace Category links produce, see bug 87
1769                                          * @todo We might want to use trim($tmp, "\n") here.
1770                                          */
1771                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1772
1773                                         wfProfileOut( "$fname-category" );
1774                                         continue;
1775                                 }
1776                         }
1777
1778                         # Self-link checking
1779                         if( $nt->getFragment() === '' ) {
1780                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1781                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1782                                         continue;
1783                                 }
1784                         }
1785
1786                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1787                         if( $ns == NS_MEDIA ) {
1788                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1789                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1790                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1791                                 $this->mOutput->addImage( $nt->getDBkey() );
1792                                 continue;
1793                         } elseif( $ns == NS_SPECIAL ) {
1794                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1795                                 continue;
1796                         } elseif( $ns == NS_IMAGE ) {
1797                                 $img = new Image( $nt );
1798                                 if( $img->exists() ) {
1799                                         // Force a blue link if the file exists; may be a remote
1800                                         // upload on the shared repository, and we want to see its
1801                                         // auto-generated page.
1802                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1803                                         $this->mOutput->addLink( $nt );
1804                                         continue;
1805                                 }
1806                         }
1807                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1808                 }
1809                 wfProfileOut( $fname );
1810                 return $s;
1811         }
1812
1813         /**
1814          * Make a link placeholder. The text returned can be later resolved to a real link with
1815          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1816          * parsing of interwiki links, and secondly to allow all existence checks and
1817          * article length checks (for stub links) to be bundled into a single query.
1818          *
1819          */
1820         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1821                 wfProfileIn( __METHOD__ );
1822                 if ( ! is_object($nt) ) {
1823                         # Fail gracefully
1824                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1825                 } else {
1826                         # Separate the link trail from the rest of the link
1827                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1828
1829                         if ( $nt->isExternal() ) {
1830                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1831                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1832                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1833                         } else {
1834                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1835                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1836                                 $this->mLinkHolders['queries'][] = $query;
1837                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1838                                 $this->mLinkHolders['titles'][] = $nt;
1839
1840                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1841                         }
1842                 }
1843                 wfProfileOut( __METHOD__ );
1844                 return $retVal;
1845         }
1846
1847         /**
1848          * Render a forced-blue link inline; protect against double expansion of
1849          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1850          * Since this little disaster has to split off the trail text to avoid
1851          * breaking URLs in the following text without breaking trails on the
1852          * wiki links, it's been made into a horrible function.
1853          *
1854          * @param Title $nt
1855          * @param string $text
1856          * @param string $query
1857          * @param string $trail
1858          * @param string $prefix
1859          * @return string HTML-wikitext mix oh yuck
1860          */
1861         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1862                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1863                 $sk =& $this->mOptions->getSkin();
1864                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1865                 return $this->armorLinks( $link ) . $trail;
1866         }
1867
1868         /**
1869          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1870          * going to go through further parsing steps before inline URL expansion.
1871          *
1872          * In particular this is important when using action=render, which causes
1873          * full URLs to be included.
1874          *
1875          * Oh man I hate our multi-layer parser!
1876          *
1877          * @param string more-or-less HTML
1878          * @return string less-or-more HTML with NOPARSE bits
1879          */
1880         function armorLinks( $text ) {
1881                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1882                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1883         }
1884
1885         /**
1886          * Return true if subpage links should be expanded on this page.
1887          * @return bool
1888          */
1889         function areSubpagesAllowed() {
1890                 # Some namespaces don't allow subpages
1891                 global $wgNamespacesWithSubpages;
1892                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1893         }
1894
1895         /**
1896          * Handle link to subpage if necessary
1897          * @param string $target the source of the link
1898          * @param string &$text the link text, modified as necessary
1899          * @return string the full name of the link
1900          * @private
1901          */
1902         function maybeDoSubpageLink($target, &$text) {
1903                 # Valid link forms:
1904                 # Foobar -- normal
1905                 # :Foobar -- override special treatment of prefix (images, language links)
1906                 # /Foobar -- convert to CurrentPage/Foobar
1907                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1908                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1909                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1910
1911                 $fname = 'Parser::maybeDoSubpageLink';
1912                 wfProfileIn( $fname );
1913                 $ret = $target; # default return value is no change
1914
1915                 # bug 7425
1916                 $target = trim( $target );
1917
1918                 # Some namespaces don't allow subpages,
1919                 # so only perform processing if subpages are allowed
1920                 if( $this->areSubpagesAllowed() ) {
1921                         # Look at the first character
1922                         if( $target != '' && $target{0} == '/' ) {
1923                                 # / at end means we don't want the slash to be shown
1924                                 if( substr( $target, -1, 1 ) == '/' ) {
1925                                         $target = substr( $target, 1, -1 );
1926                                         $noslash = $target;
1927                                 } else {
1928                                         $noslash = substr( $target, 1 );
1929                                 }
1930
1931                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1932                                 if( '' === $text ) {
1933                                         $text = $target;
1934                                 } # this might be changed for ugliness reasons
1935                         } else {
1936                                 # check for .. subpage backlinks
1937                                 $dotdotcount = 0;
1938                                 $nodotdot = $target;
1939                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1940                                         ++$dotdotcount;
1941                                         $nodotdot = substr( $nodotdot, 3 );
1942                                 }
1943                                 if($dotdotcount > 0) {
1944                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1945                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1946                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1947                                                 # / at the end means don't show full path
1948                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1949                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1950                                                         if( '' === $text ) {
1951                                                                 $text = $nodotdot;
1952                                                         }
1953                                                 }
1954                                                 $nodotdot = trim( $nodotdot );
1955                                                 if( $nodotdot != '' ) {
1956                                                         $ret .= '/' . $nodotdot;
1957                                                 }
1958                                         }
1959                                 }
1960                         }
1961                 }
1962
1963                 wfProfileOut( $fname );
1964                 return $ret;
1965         }
1966
1967         /**#@+
1968          * Used by doBlockLevels()
1969          * @private
1970          */
1971         /* private */ function closeParagraph() {
1972                 $result = '';
1973                 if ( '' != $this->mLastSection ) {
1974                         $result = '</' . $this->mLastSection  . ">\n";
1975                 }
1976                 $this->mInPre = false;
1977                 $this->mLastSection = '';
1978                 return $result;
1979         }
1980         # getCommon() returns the length of the longest common substring
1981         # of both arguments, starting at the beginning of both.
1982         #
1983         /* private */ function getCommon( $st1, $st2 ) {
1984                 $fl = strlen( $st1 );
1985                 $shorter = strlen( $st2 );
1986                 if ( $fl < $shorter ) { $shorter = $fl; }
1987
1988                 for ( $i = 0; $i < $shorter; ++$i ) {
1989                         if ( $st1{$i} != $st2{$i} ) { break; }
1990                 }
1991                 return $i;
1992         }
1993         # These next three functions open, continue, and close the list
1994         # element appropriate to the prefix character passed into them.
1995         #
1996         /* private */ function openList( $char ) {
1997                 $result = $this->closeParagraph();
1998
1999                 if ( '*' == $char ) { $result .= '<ul><li>'; }
2000                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
2001                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
2002                 else if ( ';' == $char ) {
2003                         $result .= '<dl><dt>';
2004                         $this->mDTopen = true;
2005                 }
2006                 else { $result = '<!-- ERR 1 -->'; }
2007
2008                 return $result;
2009         }
2010
2011         /* private */ function nextItem( $char ) {
2012                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
2013                 else if ( ':' == $char || ';' == $char ) {
2014                         $close = '</dd>';
2015                         if ( $this->mDTopen ) { $close = '</dt>'; }
2016                         if ( ';' == $char ) {
2017                                 $this->mDTopen = true;
2018                                 return $close . '<dt>';
2019                         } else {
2020                                 $this->mDTopen = false;
2021                                 return $close . '<dd>';
2022                         }
2023                 }
2024                 return '<!-- ERR 2 -->';
2025         }
2026
2027         /* private */ function closeList( $char ) {
2028                 if ( '*' == $char ) { $text = '</li></ul>'; }
2029                 else if ( '#' == $char ) { $text = '</li></ol>'; }
2030                 else if ( ':' == $char ) {
2031                         if ( $this->mDTopen ) {
2032                                 $this->mDTopen = false;
2033                                 $text = '</dt></dl>';
2034                         } else {
2035                                 $text = '</dd></dl>';
2036                         }
2037                 }
2038                 else {  return '<!-- ERR 3 -->'; }
2039                 return $text."\n";
2040         }
2041         /**#@-*/
2042
2043         /**
2044          * Make lists from lines starting with ':', '*', '#', etc.
2045          *
2046          * @private
2047          * @return string the lists rendered as HTML
2048          */
2049         function doBlockLevels( $text, $linestart ) {
2050                 $fname = 'Parser::doBlockLevels';
2051                 wfProfileIn( $fname );
2052
2053                 # Parsing through the text line by line.  The main thing
2054                 # happening here is handling of block-level elements p, pre,
2055                 # and making lists from lines starting with * # : etc.
2056                 #
2057                 $textLines = explode( "\n", $text );
2058
2059                 $lastPrefix = $output = '';
2060                 $this->mDTopen = $inBlockElem = false;
2061                 $prefixLength = 0;
2062                 $paragraphStack = false;
2063
2064                 if ( !$linestart ) {
2065                         $output .= array_shift( $textLines );
2066                 }
2067                 foreach ( $textLines as $oLine ) {
2068                         $lastPrefixLength = strlen( $lastPrefix );
2069                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2070                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2071                         if ( !$this->mInPre ) {
2072                                 # Multiple prefixes may abut each other for nested lists.
2073                                 $prefixLength = strspn( $oLine, '*#:;' );
2074                                 $pref = substr( $oLine, 0, $prefixLength );
2075
2076                                 # eh?
2077                                 $pref2 = str_replace( ';', ':', $pref );
2078                                 $t = substr( $oLine, $prefixLength );
2079                                 $this->mInPre = !empty($preOpenMatch);
2080                         } else {
2081                                 # Don't interpret any other prefixes in preformatted text
2082                                 $prefixLength = 0;
2083                                 $pref = $pref2 = '';
2084                                 $t = $oLine;
2085                         }
2086
2087                         # List generation
2088                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2089                                 # Same as the last item, so no need to deal with nesting or opening stuff
2090                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2091                                 $paragraphStack = false;
2092
2093                                 if ( substr( $pref, -1 ) == ';') {
2094                                         # The one nasty exception: definition lists work like this:
2095                                         # ; title : definition text
2096                                         # So we check for : in the remainder text to split up the
2097                                         # title and definition, without b0rking links.
2098                                         $term = $t2 = '';
2099                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2100                                                 $t = $t2;
2101                                                 $output .= $term . $this->nextItem( ':' );
2102                                         }
2103                                 }
2104                         } elseif( $prefixLength || $lastPrefixLength ) {
2105                                 # Either open or close a level...
2106                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2107                                 $paragraphStack = false;
2108
2109                                 while( $commonPrefixLength < $lastPrefixLength ) {
2110                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2111                                         --$lastPrefixLength;
2112                                 }
2113                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2114                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2115                                 }
2116                                 while ( $prefixLength > $commonPrefixLength ) {
2117                                         $char = substr( $pref, $commonPrefixLength, 1 );
2118                                         $output .= $this->openList( $char );
2119
2120                                         if ( ';' == $char ) {
2121                                                 # FIXME: This is dupe of code above
2122                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2123                                                         $t = $t2;
2124                                                         $output .= $term . $this->nextItem( ':' );
2125                                                 }
2126                                         }
2127                                         ++$commonPrefixLength;
2128                                 }
2129                                 $lastPrefix = $pref2;
2130                         }
2131                         if( 0 == $prefixLength ) {
2132                                 wfProfileIn( "$fname-paragraph" );
2133                                 # No prefix (not in list)--go to paragraph mode
2134                                 // XXX: use a stack for nestable elements like span, table and div
2135                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2136                                 $closematch = preg_match(
2137                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2138                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2139                                 if ( $openmatch or $closematch ) {
2140                                         $paragraphStack = false;
2141                                         # TODO bug 5718: paragraph closed
2142                                         $output .= $this->closeParagraph();
2143                                         if ( $preOpenMatch and !$preCloseMatch ) {
2144                                                 $this->mInPre = true;
2145                                         }
2146                                         if ( $closematch ) {
2147                                                 $inBlockElem = false;
2148                                         } else {
2149                                                 $inBlockElem = true;
2150                                         }
2151                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2152                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2153                                                 // pre
2154                                                 if ($this->mLastSection != 'pre') {
2155                                                         $paragraphStack = false;
2156                                                         $output .= $this->closeParagraph().'<pre>';
2157                                                         $this->mLastSection = 'pre';
2158                                                 }
2159                                                 $t = substr( $t, 1 );
2160                                         } else {
2161                                                 // paragraph
2162                                                 if ( '' == trim($t) ) {
2163                                                         if ( $paragraphStack ) {
2164                                                                 $output .= $paragraphStack.'<br />';
2165                                                                 $paragraphStack = false;
2166                                                                 $this->mLastSection = 'p';
2167                                                         } else {
2168                                                                 if ($this->mLastSection != 'p' ) {
2169                                                                         $output .= $this->closeParagraph();
2170                                                                         $this->mLastSection = '';
2171                                                                         $paragraphStack = '<p>';
2172                                                                 } else {
2173                                                                         $paragraphStack = '</p><p>';
2174                                                                 }
2175                                                         }
2176                                                 } else {
2177                                                         if ( $paragraphStack ) {
2178                                                                 $output .= $paragraphStack;
2179                                                                 $paragraphStack = false;
2180                                                                 $this->mLastSection = 'p';
2181                                                         } else if ($this->mLastSection != 'p') {
2182                                                                 $output .= $this->closeParagraph().'<p>';
2183                                                                 $this->mLastSection = 'p';
2184                                                         }
2185                                                 }
2186                                         }
2187                                 }
2188                                 wfProfileOut( "$fname-paragraph" );
2189                         }
2190                         // somewhere above we forget to get out of pre block (bug 785)
2191                         if($preCloseMatch && $this->mInPre) {
2192                                 $this->mInPre = false;
2193                         }
2194                         if ($paragraphStack === false) {
2195                                 $output .= $t."\n";
2196                         }
2197                 }
2198                 while ( $prefixLength ) {
2199                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2200                         --$prefixLength;
2201                 }
2202                 if ( '' != $this->mLastSection ) {
2203                         $output .= '</' . $this->mLastSection . '>';
2204                         $this->mLastSection = '';
2205                 }
2206
2207                 wfProfileOut( $fname );
2208                 return $output;
2209         }
2210
2211         /**
2212          * Split up a string on ':', ignoring any occurences inside tags
2213          * to prevent illegal overlapping.
2214          * @param string $str the string to split
2215          * @param string &$before set to everything before the ':'
2216          * @param string &$after set to everything after the ':'
2217          * return string the position of the ':', or false if none found
2218          */
2219         function findColonNoLinks($str, &$before, &$after) {
2220                 $fname = 'Parser::findColonNoLinks';
2221                 wfProfileIn( $fname );
2222
2223                 $pos = strpos( $str, ':' );
2224                 if( $pos === false ) {
2225                         // Nothing to find!
2226                         wfProfileOut( $fname );
2227                         return false;
2228                 }
2229
2230                 $lt = strpos( $str, '<' );
2231                 if( $lt === false || $lt > $pos ) {
2232                         // Easy; no tag nesting to worry about
2233                         $before = substr( $str, 0, $pos );
2234                         $after = substr( $str, $pos+1 );
2235                         wfProfileOut( $fname );
2236                         return $pos;
2237                 }
2238
2239                 // Ugly state machine to walk through avoiding tags.
2240                 $state = MW_COLON_STATE_TEXT;
2241                 $stack = 0;
2242                 $len = strlen( $str );
2243                 for( $i = 0; $i < $len; $i++ ) {
2244                         $c = $str{$i};
2245
2246                         switch( $state ) {
2247                         // (Using the number is a performance hack for common cases)
2248                         case 0: // MW_COLON_STATE_TEXT:
2249                                 switch( $c ) {
2250                                 case "<":
2251                                         // Could be either a <start> tag or an </end> tag
2252                                         $state = MW_COLON_STATE_TAGSTART;
2253                                         break;
2254                                 case ":":
2255                                         if( $stack == 0 ) {
2256                                                 // We found it!
2257                                                 $before = substr( $str, 0, $i );
2258                                                 $after = substr( $str, $i + 1 );
2259                                                 wfProfileOut( $fname );
2260                                                 return $i;
2261                                         }
2262                                         // Embedded in a tag; don't break it.
2263                                         break;
2264                                 default:
2265                                         // Skip ahead looking for something interesting
2266                                         $colon = strpos( $str, ':', $i );
2267                                         if( $colon === false ) {
2268                                                 // Nothing else interesting
2269                                                 wfProfileOut( $fname );
2270                                                 return false;
2271                                         }
2272                                         $lt = strpos( $str, '<', $i );
2273                                         if( $stack === 0 ) {
2274                                                 if( $lt === false || $colon < $lt ) {
2275                                                         // We found it!
2276                                                         $before = substr( $str, 0, $colon );
2277                                                         $after = substr( $str, $colon + 1 );
2278                                                         wfProfileOut( $fname );
2279                                                         return $i;
2280                                                 }
2281                                         }
2282                                         if( $lt === false ) {
2283                                                 // Nothing else interesting to find; abort!
2284                                                 // We're nested, but there's no close tags left. Abort!
2285                                                 break 2;
2286                                         }
2287                                         // Skip ahead to next tag start
2288                                         $i = $lt;
2289                                         $state = MW_COLON_STATE_TAGSTART;
2290                                 }
2291                                 break;
2292                         case 1: // MW_COLON_STATE_TAG:
2293                                 // In a <tag>
2294                                 switch( $c ) {
2295                                 case ">":
2296                                         $stack++;
2297                                         $state = MW_COLON_STATE_TEXT;
2298                                         break;
2299                                 case "/":
2300                                         // Slash may be followed by >?
2301                                         $state = MW_COLON_STATE_TAGSLASH;
2302                                         break;
2303                                 default:
2304                                         // ignore
2305                                 }
2306                                 break;
2307                         case 2: // MW_COLON_STATE_TAGSTART:
2308                                 switch( $c ) {
2309                                 case "/":
2310                                         $state = MW_COLON_STATE_CLOSETAG;
2311                                         break;
2312                                 case "!":
2313                                         $state = MW_COLON_STATE_COMMENT;
2314                                         break;
2315                                 case ">":
2316                                         // Illegal early close? This shouldn't happen D:
2317                                         $state = MW_COLON_STATE_TEXT;
2318                                         break;
2319                                 default:
2320                                         $state = MW_COLON_STATE_TAG;
2321                                 }
2322                                 break;
2323                         case 3: // MW_COLON_STATE_CLOSETAG:
2324                                 // In a </tag>
2325                                 if( $c == ">" ) {
2326                                         $stack--;
2327                                         if( $stack < 0 ) {
2328                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2329                                                 wfProfileOut( $fname );
2330                                                 return false;
2331                                         }
2332                                         $state = MW_COLON_STATE_TEXT;
2333                                 }
2334                                 break;
2335                         case MW_COLON_STATE_TAGSLASH:
2336                                 if( $c == ">" ) {
2337                                         // Yes, a self-closed tag <blah/>
2338                                         $state = MW_COLON_STATE_TEXT;
2339                                 } else {
2340                                         // Probably we're jumping the gun, and this is an attribute
2341                                         $state = MW_COLON_STATE_TAG;
2342                                 }
2343                                 break;
2344                         case 5: // MW_COLON_STATE_COMMENT:
2345                                 if( $c == "-" ) {
2346                                         $state = MW_COLON_STATE_COMMENTDASH;
2347                                 }
2348                                 break;
2349                         case MW_COLON_STATE_COMMENTDASH:
2350                                 if( $c == "-" ) {
2351                                         $state = MW_COLON_STATE_COMMENTDASHDASH;
2352                                 } else {
2353                                         $state = MW_COLON_STATE_COMMENT;
2354                                 }
2355                                 break;
2356                         case MW_COLON_STATE_COMMENTDASHDASH:
2357                                 if( $c == ">" ) {
2358                                         $state = MW_COLON_STATE_TEXT;
2359                                 } else {
2360                                         $state = MW_COLON_STATE_COMMENT;
2361                                 }
2362                                 break;
2363                         default:
2364                                 throw new MWException( "State machine error in $fname" );
2365                         }
2366                 }
2367                 if( $stack > 0 ) {
2368                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2369                         return false;
2370                 }
2371                 wfProfileOut( $fname );
2372                 return false;
2373         }
2374
2375         /**
2376          * Return value of a magic variable (like PAGENAME)
2377          *
2378          * @private
2379          */
2380         function getVariableValue( $index ) {
2381                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2382
2383                 /**
2384                  * Some of these require message or data lookups and can be
2385                  * expensive to check many times.
2386                  */
2387                 static $varCache = array();
2388                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) ) {
2389                         if ( isset( $varCache[$index] ) ) {
2390                                 return $varCache[$index];
2391                         }
2392                 }
2393
2394                 $ts = time();
2395                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2396
2397                 # Use the time zone
2398                 global $wgLocaltimezone;
2399                 if ( isset( $wgLocaltimezone ) ) {
2400                         $oldtz = getenv( 'TZ' );
2401                         putenv( 'TZ='.$wgLocaltimezone );
2402                 }
2403                 $localTimestamp = date( 'YmdHis', $ts );
2404                 $localMonth = date( 'm', $ts );
2405                 $localMonthName = date( 'n', $ts );
2406                 $localDay = date( 'j', $ts );
2407                 $localDay2 = date( 'd', $ts );
2408                 $localDayOfWeek = date( 'w', $ts );
2409                 $localWeek = date( 'W', $ts );
2410                 $localYear = date( 'Y', $ts );
2411                 $localHour = date( 'H', $ts );
2412                 if ( isset( $wgLocaltimezone ) ) {
2413                         putenv( 'TZ='.$oldtz );
2414                 }
2415
2416                 switch ( $index ) {
2417                         case 'currentmonth':
2418                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
2419                         case 'currentmonthname':
2420                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2421                         case 'currentmonthnamegen':
2422                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2423                         case 'currentmonthabbrev':
2424                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2425                         case 'currentday':
2426                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2427                         case 'currentday2':
2428                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2429                         case 'localmonth':
2430                                 return $varCache[$index] = $wgContLang->formatNum( $localMonth );
2431                         case 'localmonthname':
2432                                 return $varCache[$index] = $wgContLang->getMonthName( $localMonthName );
2433                         case 'localmonthnamegen':
2434                                 return $varCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2435                         case 'localmonthabbrev':
2436                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2437                         case 'localday':
2438                                 return $varCache[$index] = $wgContLang->formatNum( $localDay );
2439                         case 'localday2':
2440                                 return $varCache[$index] = $wgContLang->formatNum( $localDay2 );
2441                         case 'pagename':
2442                                 return $this->mTitle->getText();
2443                         case 'pagenamee':
2444                                 return $this->mTitle->getPartialURL();
2445                         case 'fullpagename':
2446                                 return $this->mTitle->getPrefixedText();
2447                         case 'fullpagenamee':
2448                                 return $this->mTitle->getPrefixedURL();
2449                         case 'subpagename':
2450                                 return $this->mTitle->getSubpageText();
2451                         case 'subpagenamee':
2452                                 return $this->mTitle->getSubpageUrlForm();
2453                         case 'basepagename':
2454                                 return $this->mTitle->getBaseText();
2455                         case 'basepagenamee':
2456                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2457                         case 'talkpagename':
2458                                 if( $this->mTitle->canTalk() ) {
2459                                         $talkPage = $this->mTitle->getTalkPage();
2460                                         return $talkPage->getPrefixedText();
2461                                 } else {
2462                                         return '';
2463                                 }
2464                         case 'talkpagenamee':
2465                                 if( $this->mTitle->canTalk() ) {
2466                                         $talkPage = $this->mTitle->getTalkPage();
2467                                         return $talkPage->getPrefixedUrl();
2468                                 } else {
2469                                         return '';
2470                                 }
2471                         case 'subjectpagename':
2472                                 $subjPage = $this->mTitle->getSubjectPage();
2473                                 return $subjPage->getPrefixedText();
2474                         case 'subjectpagenamee':
2475                                 $subjPage = $this->mTitle->getSubjectPage();
2476                                 return $subjPage->getPrefixedUrl();
2477                         case 'revisionid':
2478                                 return $this->mRevisionId;
2479                         case 'revisionday':
2480                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2481                         case 'revisionday2':
2482                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2483                         case 'revisionmonth':
2484                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2485                         case 'revisionyear':
2486                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2487                         case 'revisiontimestamp':
2488                                 return $this->getRevisionTimestamp();
2489                         case 'namespace':
2490                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2491                         case 'namespacee':
2492                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2493                         case 'talkspace':
2494                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2495                         case 'talkspacee':
2496                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2497                         case 'subjectspace':
2498                                 return $this->mTitle->getSubjectNsText();
2499                         case 'subjectspacee':
2500                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2501                         case 'currentdayname':
2502                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2503                         case 'currentyear':
2504                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2505                         case 'currenttime':
2506                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2507                         case 'currenthour':
2508                                 return $varCache[$index] = $wgContLang->formatNum( date( 'H', $ts ), true );
2509                         case 'currentweek':
2510                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2511                                 // int to remove the padding
2512                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2513                         case 'currentdow':
2514                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2515                         case 'localdayname':
2516                                 return $varCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2517                         case 'localyear':
2518                                 return $varCache[$index] = $wgContLang->formatNum( $localYear, true );
2519                         case 'localtime':
2520                                 return $varCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2521                         case 'localhour':
2522                                 return $varCache[$index] = $wgContLang->formatNum( $localHour, true );
2523                         case 'localweek':
2524                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2525                                 // int to remove the padding
2526                                 return $varCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2527                         case 'localdow':
2528                                 return $varCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2529                         case 'numberofarticles':
2530                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2531                         case 'numberoffiles':
2532                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2533                         case 'numberofusers':
2534                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2535                         case 'numberofpages':
2536                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2537                         case 'numberofadmins':
2538                                 return $varCache[$index]  = $wgContLang->formatNum( SiteStats::admins() );
2539                         case 'currenttimestamp':
2540                                 return $varCache[$index] = wfTimestampNow();
2541                         case 'localtimestamp':
2542                                 return $varCache[$index] = $localTimestamp;
2543                         case 'currentversion':
2544                                 return $varCache[$index] = SpecialVersion::getVersion();
2545                         case 'sitename':
2546                                 return $wgSitename;
2547                         case 'server':
2548                                 return $wgServer;
2549                         case 'servername':
2550                                 return $wgServerName;
2551                         case 'scriptpath':
2552                                 return $wgScriptPath;
2553                         case 'directionmark':
2554                                 return $wgContLang->getDirMark();
2555                         case 'contentlanguage':
2556                                 global $wgContLanguageCode;
2557                                 return $wgContLanguageCode;
2558                         default:
2559                                 $ret = null;
2560                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2561                                         return $ret;
2562                                 else
2563                                         return null;
2564                 }
2565         }
2566
2567         /**
2568          * initialise the magic variables (like CURRENTMONTHNAME)
2569          *
2570          * @private
2571          */
2572         function initialiseVariables() {
2573                 $fname = 'Parser::initialiseVariables';
2574                 wfProfileIn( $fname );
2575                 $variableIDs = MagicWord::getVariableIDs();
2576
2577                 $this->mVariables = array();
2578                 foreach ( $variableIDs as $id ) {
2579                         $mw =& MagicWord::get( $id );
2580                         $mw->addToArray( $this->mVariables, $id );
2581                 }
2582                 wfProfileOut( $fname );
2583         }
2584
2585         /**
2586          * parse any parentheses in format ((title|part|part))
2587          * and call callbacks to get a replacement text for any found piece
2588          *
2589          * @param string $text The text to parse
2590          * @param array $callbacks rules in form:
2591          *     '{' => array(                            # opening parentheses
2592          *                                      'end' => '}',   # closing parentheses
2593          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2594          *                                                                3 => callback         # replacement callback to call if {{{..}}} is found
2595          *                                                                )
2596          *                                      )
2597          *                                      'min' => 2,     # Minimum parenthesis count in cb
2598          *                                      'max' => 3,     # Maximum parenthesis count in cb
2599          * @private
2600          */
2601         function replace_callback ($text, $callbacks) {
2602                 wfProfileIn( __METHOD__ );
2603                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2604                 $lastOpeningBrace = -1;                 # last not closed parentheses
2605
2606                 $validOpeningBraces = implode( '', array_keys( $callbacks ) );
2607
2608                 $i = 0;
2609                 while ( $i < strlen( $text ) ) {
2610                         # Find next opening brace, closing brace or pipe
2611                         if ( $lastOpeningBrace == -1 ) {
2612                                 $currentClosing = '';
2613                                 $search = $validOpeningBraces;
2614                         } else {
2615                                 $currentClosing = $openingBraceStack[$lastOpeningBrace]['braceEnd'];
2616                                 $search = $validOpeningBraces . '|' . $currentClosing;
2617                         }
2618                         $rule = null;
2619                         $i += strcspn( $text, $search, $i );
2620                         if ( $i < strlen( $text ) ) {
2621                                 if ( $text[$i] == '|' ) {
2622                                         $found = 'pipe';
2623                                 } elseif ( $text[$i] == $currentClosing ) {
2624                                         $found = 'close';
2625                                 } elseif ( isset( $callbacks[$text[$i]] ) ) {
2626                                         $found = 'open';
2627                                         $rule = $callbacks[$text[$i]];
2628                                 } else {
2629                                         # Some versions of PHP have a strcspn which stops on null characters
2630                                         # Ignore and continue
2631                                         ++$i;
2632                                         continue;
2633                                 }
2634                         } else {
2635                                 # All done
2636                                 break;
2637                         }
2638
2639                         if ( $found == 'open' ) {
2640                                 # found opening brace, let's add it to parentheses stack
2641                                 $piece = array('brace' => $text[$i],
2642                                                            'braceEnd' => $rule['end'],
2643                                                            'title' => '',
2644                                                            'parts' => null);
2645
2646                                 # count opening brace characters
2647                                 $piece['count'] = strspn( $text, $piece['brace'], $i );
2648                                 $piece['startAt'] = $piece['partStart'] = $i + $piece['count'];
2649                                 $i += $piece['count'];
2650
2651                                 # we need to add to stack only if opening brace count is enough for one of the rules
2652                                 if ( $piece['count'] >= $rule['min'] ) {
2653                                         $lastOpeningBrace ++;
2654                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2655                                 }
2656                         } elseif ( $found == 'close' ) {
2657                                 # lets check if it is enough characters for closing brace
2658                                 $maxCount = $openingBraceStack[$lastOpeningBrace]['count'];
2659                                 $count = strspn( $text, $text[$i], $i, $maxCount );
2660
2661                                 # check for maximum matching characters (if there are 5 closing
2662                                 # characters, we will probably need only 3 - depending on the rules)
2663                                 $matchingCount = 0;
2664                                 $matchingCallback = null;
2665                                 $cbType = $callbacks[$openingBraceStack[$lastOpeningBrace]['brace']];
2666                                 if ( $count > $cbType['max'] ) {
2667                                         # The specified maximum exists in the callback array, unless the caller
2668                                         # has made an error
2669                                         $matchingCount = $cbType['max'];
2670                                 } else {
2671                                         # Count is less than the maximum
2672                                         # Skip any gaps in the callback array to find the true largest match
2673                                         # Need to use array_key_exists not isset because the callback can be null
2674                                         $matchingCount = $count;
2675                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $cbType['cb'] ) ) {
2676                                                 --$matchingCount;
2677                                         }
2678                                 }
2679
2680                                 if ($matchingCount <= 0) {
2681                                         $i += $count;
2682                                         continue;
2683                                 }
2684                                 $matchingCallback = $cbType['cb'][$matchingCount];
2685
2686                                 # let's set a title or last part (if '|' was found)
2687                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2688                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2689                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2690                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2691                                 } else {
2692                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2693                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2694                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2695                                 }
2696
2697                                 $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2698                                 $pieceEnd = $i + $matchingCount;
2699
2700                                 if( is_callable( $matchingCallback ) ) {
2701                                         $cbArgs = array (
2702                                                                          'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2703                                                                          'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2704                                                                          'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2705                                                                          'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == "\n")),
2706                                                                          );
2707                                         # finally we can call a user callback and replace piece of text
2708                                         $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2709                                         $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2710                                         $i = $pieceStart + strlen($replaceWith);
2711                                 } else {
2712                                         # null value for callback means that parentheses should be parsed, but not replaced
2713                                         $i += $matchingCount;
2714                                 }
2715
2716                                 # reset last opening parentheses, but keep it in case there are unused characters
2717                                 $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2718                                                            'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2719                                                            'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2720                                                            'title' => '',
2721                                                            'parts' => null,
2722                                                            'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2723                                 $openingBraceStack[$lastOpeningBrace--] = null;
2724
2725                                 if ($matchingCount < $piece['count']) {
2726                                         $piece['count'] -= $matchingCount;
2727                                         $piece['startAt'] -= $matchingCount;
2728                                         $piece['partStart'] = $piece['startAt'];
2729                                         # do we still qualify for any callback with remaining count?
2730                                         $currentCbList = $callbacks[$piece['brace']]['cb'];
2731                                         while ( $piece['count'] ) {
2732                                                 if ( array_key_exists( $piece['count'], $currentCbList ) ) {
2733                                                         $lastOpeningBrace++;
2734                                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2735                                                         break;
2736                                                 }
2737                                                 --$piece['count'];
2738                                         }
2739                                 }
2740                         } elseif ( $found == 'pipe' ) {
2741                                 # lets set a title if it is a first separator, or next part otherwise
2742                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2743                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2744                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2745                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2746                                         $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2747                                 } else {
2748                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2749                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2750                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2751                                 }
2752                                 $openingBraceStack[$lastOpeningBrace]['partStart'] = ++$i;
2753                         }
2754                 }
2755
2756                 wfProfileOut( __METHOD__ );
2757                 return $text;
2758         }
2759
2760         /**
2761          * Replace magic variables, templates, and template arguments
2762          * with the appropriate text. Templates are substituted recursively,
2763          * taking care to avoid infinite loops.
2764          *
2765          * Note that the substitution depends on value of $mOutputType:
2766          *  OT_WIKI: only {{subst:}} templates
2767          *  OT_MSG: only magic variables
2768          *  OT_HTML: all templates and magic variables
2769          *
2770          * @param string $tex The text to transform
2771          * @param array $args Key-value pairs representing template parameters to substitute
2772          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2773          * @private
2774          */
2775         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2776                 # Prevent too big inclusions
2777                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2778                         return $text;
2779                 }
2780
2781                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2782                 wfProfileIn( $fname );
2783
2784                 # This function is called recursively. To keep track of arguments we need a stack:
2785                 array_push( $this->mArgStack, $args );
2786
2787                 $braceCallbacks = array();
2788                 if ( !$argsOnly ) {
2789                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2790                 }
2791                 if ( $this->mOutputType != OT_MSG ) {
2792                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2793                 }
2794                 if ( $braceCallbacks ) {
2795                         $callbacks = array(
2796                                 '{' => array(
2797                                         'end' => '}',
2798                                         'cb' => $braceCallbacks,
2799                                         'min' => $argsOnly ? 3 : 2,
2800                                         'max' => isset( $braceCallbacks[3] ) ? 3 : 2,
2801                                 ),
2802                                 '[' => array(
2803                                         'end' => ']',
2804                                         'cb' => array(2=>null),
2805                                         'min' => 2,
2806                                         'max' => 2,
2807                                 )
2808                         );
2809                         $text = $this->replace_callback ($text, $callbacks);
2810
2811                         array_pop( $this->mArgStack );
2812                 }
2813                 wfProfileOut( $fname );
2814                 return $text;
2815         }
2816
2817         /**
2818          * Replace magic variables
2819          * @private
2820          */
2821         function variableSubstitution( $matches ) {
2822                 global $wgContLang;
2823                 $fname = 'Parser::variableSubstitution';
2824                 $varname = $wgContLang->lc($matches[1]);
2825                 wfProfileIn( $fname );
2826                 $skip = false;
2827                 if ( $this->mOutputType == OT_WIKI ) {
2828                         # Do only magic variables prefixed by SUBST
2829                         $mwSubst =& MagicWord::get( 'subst' );
2830                         if (!$mwSubst->matchStartAndRemove( $varname ))
2831                                 $skip = true;
2832                         # Note that if we don't substitute the variable below,
2833                         # we don't remove the {{subst:}} magic word, in case
2834                         # it is a template rather than a magic variable.
2835                 }
2836                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2837                         $id = $this->mVariables[$varname];
2838                         # Now check if we did really match, case sensitive or not
2839                         $mw =& MagicWord::get( $id );
2840                         if ($mw->match($matches[1])) {
2841                                 $text = $this->getVariableValue( $id );
2842                                 $this->mOutput->mContainsOldMagic = true;
2843                         } else {
2844                                 $text = $matches[0];
2845                         }
2846                 } else {
2847                         $text = $matches[0];
2848                 }
2849                 wfProfileOut( $fname );
2850                 return $text;
2851         }
2852
2853
2854         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2855         static function createAssocArgs( $args ) {
2856                 $assocArgs = array();
2857                 $index = 1;
2858                 foreach( $args as $arg ) {
2859                         $eqpos = strpos( $arg, '=' );
2860                         if ( $eqpos === false ) {
2861                                 $assocArgs[$index++] = $arg;
2862                         } else {
2863                                 $name = trim( substr( $arg, 0, $eqpos ) );
2864                                 $value = trim( substr( $arg, $eqpos+1 ) );
2865                                 if ( $value === false ) {
2866                                         $value = '';
2867                                 }
2868                                 if ( $name !== false ) {
2869                                         $assocArgs[$name] = $value;
2870                                 }
2871                         }
2872                 }
2873
2874                 return $assocArgs;
2875         }
2876
2877         /**
2878          * Return the text of a template, after recursively
2879          * replacing any variables or templates within the template.
2880          *
2881          * @param array $piece The parts of the template
2882          *  $piece['text']: matched text
2883          *  $piece['title']: the title, i.e. the part before the |
2884          *  $piece['parts']: the parameter array
2885          * @return string the text of the template
2886          * @private
2887          */
2888         function braceSubstitution( $piece ) {
2889                 global $wgContLang, $wgLang, $wgAllowDisplayTitle;
2890                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2891                 wfProfileIn( $fname );
2892                 wfProfileIn( __METHOD__.'-setup' );
2893
2894                 # Flags
2895                 $found = false;             # $text has been filled
2896                 $nowiki = false;            # wiki markup in $text should be escaped
2897                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2898                 $noargs = false;            # Don't replace triple-brace arguments in $text
2899                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2900                 $headingOffset = 0;         # Skip headings when number, to account for those that weren't transcluded.
2901                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2902                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2903
2904                 # Title object, where $text came from
2905                 $title = NULL;
2906
2907                 $linestart = '';
2908
2909
2910                 # $part1 is the bit before the first |, and must contain only title characters
2911                 # $args is a list of arguments, starting from index 0, not including $part1
2912
2913                 $titleText = $part1 = $piece['title'];
2914                 # If the third subpattern matched anything, it will start with |
2915
2916                 if (null == $piece['parts']) {
2917                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2918                         if ($replaceWith != $piece['text']) {
2919                                 $text = $replaceWith;
2920                                 $found = true;
2921                                 $noparse = true;
2922                                 $noargs = true;
2923                         }
2924                 }
2925
2926                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2927                 wfProfileOut( __METHOD__.'-setup' );
2928
2929                 # SUBST
2930                 wfProfileIn( __METHOD__.'-modifiers' );
2931                 if ( !$found ) {
2932                         $mwSubst =& MagicWord::get( 'subst' );
2933                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2934                                 # One of two possibilities is true:
2935                                 # 1) Found SUBST but not in the PST phase
2936                                 # 2) Didn't find SUBST and in the PST phase
2937                                 # In either case, return without further processing
2938                                 $text = $piece['text'];
2939                                 $found = true;
2940                                 $noparse = true;
2941                                 $noargs = true;
2942                         }
2943                 }
2944
2945                 # MSG, MSGNW and RAW
2946                 if ( !$found ) {
2947                         # Check for MSGNW:
2948                         $mwMsgnw =& MagicWord::get( 'msgnw' );
2949                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2950                                 $nowiki = true;
2951                         } else {
2952                                 # Remove obsolete MSG:
2953                                 $mwMsg =& MagicWord::get( 'msg' );
2954                                 $mwMsg->matchStartAndRemove( $part1 );
2955                         }
2956
2957                         # Check for RAW:
2958                         $mwRaw =& MagicWord::get( 'raw' );
2959                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2960                                 $forceRawInterwiki = true;
2961                         }
2962                 }
2963                 wfProfileOut( __METHOD__.'-modifiers' );
2964
2965                 //save path level before recursing into functions & templates.
2966                 $lastPathLevel = $this->mTemplatePath;
2967
2968                 # Parser functions
2969                 if ( !$found ) {
2970                         wfProfileIn( __METHOD__ . '-pfunc' );
2971
2972                         $colonPos = strpos( $part1, ':' );
2973                         if ( $colonPos !== false ) {
2974                                 # Case sensitive functions
2975                                 $function = substr( $part1, 0, $colonPos );
2976                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2977                                         $function = $this->mFunctionSynonyms[1][$function];
2978                                 } else {
2979                                         # Case insensitive functions
2980                                         $function = strtolower( $function );
2981                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2982                                                 $function = $this->mFunctionSynonyms[0][$function];
2983                                         } else {
2984                                                 $function = false;
2985                                         }
2986                                 }
2987                                 if ( $function ) {
2988                                         $funcArgs = array_map( 'trim', $args );
2989                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
2990                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
2991                                         $found = true;
2992
2993                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
2994                                         //$noargs = true;
2995                                         //$noparse = true;
2996
2997                                         if ( is_array( $result ) ) {
2998                                                 if ( isset( $result[0] ) ) {
2999                                                         $text = $linestart . $result[0];
3000                                                         unset( $result[0] );
3001                                                 }
3002
3003                                                 // Extract flags into the local scope
3004                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
3005                                                 extract( $result );
3006                                         } else {
3007                                                 $text = $linestart . $result;
3008                                         }
3009                                 }
3010                         }
3011                         wfProfileOut( __METHOD__ . '-pfunc' );
3012                 }
3013
3014                 # Template table test
3015
3016                 # Did we encounter this template already? If yes, it is in the cache
3017                 # and we need to check for loops.
3018                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
3019                         $found = true;
3020
3021                         # Infinite loop test
3022                         if ( isset( $this->mTemplatePath[$part1] ) ) {
3023                                 $noparse = true;
3024                                 $noargs = true;
3025                                 $found = true;
3026                                 $text = $linestart .
3027                                         "[[$part1]]<!-- WARNING: template loop detected -->";
3028                                 wfDebug( __METHOD__.": template loop broken at '$part1'\n" );
3029                         } else {
3030                                 # set $text to cached message.
3031                                 $text = $linestart . $this->mTemplates[$piece['title']];
3032                         }
3033                 }
3034
3035                 # Load from database
3036                 if ( !$found ) {
3037                         wfProfileIn( __METHOD__ . '-loadtpl' );
3038                         $ns = NS_TEMPLATE;
3039                         # declaring $subpage directly in the function call
3040                         # does not work correctly with references and breaks
3041                         # {{/subpage}}-style inclusions
3042                         $subpage = '';
3043                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3044                         if ($subpage !== '') {
3045                                 $ns = $this->mTitle->getNamespace();
3046                         }
3047                         $title = Title::newFromText( $part1, $ns );
3048
3049
3050                         if ( !is_null( $title ) ) {
3051                                 $titleText = $title->getPrefixedText();
3052                                 # Check for language variants if the template is not found
3053                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
3054                                         $wgContLang->findVariantLink($part1, $title);
3055                                 }
3056
3057                                 if ( !$title->isExternal() ) {
3058                                         if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
3059                                                 $text = SpecialPage::capturePath( $title );
3060                                                 if ( is_string( $text ) ) {
3061                                                         $found = true;
3062                                                         $noparse = true;
3063                                                         $noargs = true;
3064                                                         $isHTML = true;
3065                                                         $this->disableCache();
3066                                                 }
3067                                         } else {
3068                                                 $articleContent = $this->fetchTemplate( $title );
3069                                                 if ( $articleContent !== false ) {
3070                                                         $found = true;
3071                                                         $text = $articleContent;
3072                                                         $replaceHeadings = true;
3073                                                 }
3074                                         }
3075
3076                                         # If the title is valid but undisplayable, make a link to it
3077                                         if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3078                                                 $text = "[[:$titleText]]";
3079                                                 $found = true;
3080                                         }
3081                                 } elseif ( $title->isTrans() ) {
3082                                         // Interwiki transclusion
3083                                         if ( $this->ot['html'] && !$forceRawInterwiki ) {
3084                                                 $text = $this->interwikiTransclude( $title, 'render' );
3085                                                 $isHTML = true;
3086                                                 $noparse = true;
3087                                         } else {
3088                                                 $text = $this->interwikiTransclude( $title, 'raw' );
3089                                                 $replaceHeadings = true;
3090                                         }
3091                                         $found = true;
3092                                 }
3093
3094                                 # Template cache array insertion
3095                                 # Use the original $piece['title'] not the mangled $part1, so that
3096                                 # modifiers such as RAW: produce separate cache entries
3097                                 if( $found ) {
3098                                         if( $isHTML ) {
3099                                                 // A special page; don't store it in the template cache.
3100                                         } else {
3101                                                 $this->mTemplates[$piece['title']] = $text;
3102                                         }
3103                                         $text = $linestart . $text;
3104                                 }
3105                         }
3106                         wfProfileOut( __METHOD__ . '-loadtpl' );
3107                 }
3108
3109                 if ( $found && !$this->incrementIncludeSize( 'pre-expand', strlen( $text ) ) ) {
3110                         # Error, oversize inclusion
3111                         $text = $linestart .
3112                                 "[[$titleText]]<!-- WARNING: template omitted, pre-expand include size too large -->";
3113                         $noparse = true;
3114                         $noargs = true;
3115                 }
3116
3117                 # Recursive parsing, escaping and link table handling
3118                 # Only for HTML output
3119                 if ( $nowiki && $found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3120                         $text = wfEscapeWikiText( $text );
3121                 } elseif ( !$this->ot['msg'] && $found ) {
3122                         if ( $noargs ) {
3123                                 $assocArgs = array();
3124                         } else {
3125                                 # Clean up argument array
3126                                 $assocArgs = self::createAssocArgs($args);
3127                                 # Add a new element to the templace recursion path
3128                                 $this->mTemplatePath[$part1] = 1;
3129                         }
3130
3131                         if ( !$noparse ) {
3132                                 # If there are any <onlyinclude> tags, only include them
3133                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
3134                                         $replacer = new OnlyIncludeReplacer;
3135                                         StringUtils::delimiterReplaceCallback( '<onlyinclude>', '</onlyinclude>',
3136                                                 array( &$replacer, 'replace' ), $text );
3137                                         $text = $replacer->output;
3138                                 }
3139                                 # Remove <noinclude> sections and <includeonly> tags
3140                                 $text = StringUtils::delimiterReplace( '<noinclude>', '</noinclude>', '', $text );
3141                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
3142
3143                                 if( $this->ot['html'] || $this->ot['pre'] ) {
3144                                         # Strip <nowiki>, <pre>, etc.
3145                                         $text = $this->strip( $text, $this->mStripState );
3146                                         if ( $this->ot['html'] ) {
3147                                                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
3148                                         } elseif ( $this->ot['pre'] && $this->mOptions->getRemoveComments() ) {
3149                                                 $text = Sanitizer::removeHTMLcomments( $text );
3150                                         }
3151                                 }
3152                                 $text = $this->replaceVariables( $text, $assocArgs );
3153
3154                                 # If the template begins with a table or block-level
3155                                 # element, it should be treated as beginning a new line.
3156                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) /*}*/{
3157                                         $text = "\n" . $text;
3158                                 }
3159                         } elseif ( !$noargs ) {
3160                                 # $noparse and !$noargs
3161                                 # Just replace the arguments, not any double-brace items
3162                                 # This is used for rendered interwiki transclusion
3163                                 $text = $this->replaceVariables( $text, $assocArgs, true );
3164                         }
3165                 }
3166                 # Prune lower levels off the recursion check path
3167                 $this->mTemplatePath = $lastPathLevel;
3168
3169                 if ( $found && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3170                         # Error, oversize inclusion
3171                         $text = $linestart .
3172                                 "[[$titleText]]<!-- WARNING: template omitted, post-expand include size too large -->";
3173                         $noparse = true;
3174                         $noargs = true;
3175                 }
3176
3177                 if ( !$found ) {
3178                         wfProfileOut( $fname );
3179                         return $piece['text'];
3180                 } else {
3181                         wfProfileIn( __METHOD__ . '-placeholders' );
3182                         if ( $isHTML ) {
3183                                 # Replace raw HTML by a placeholder
3184                                 # Add a blank line preceding, to prevent it from mucking up
3185                                 # immediately preceding headings
3186                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
3187                         } else {
3188                                 # replace ==section headers==
3189                                 # XXX this needs to go away once we have a better parser.
3190                                 if ( !$this->ot['wiki'] && !$this->ot['pre'] && $replaceHeadings ) {
3191                                         if( !is_null( $title ) )
3192                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
3193                                         else
3194                                                 $encodedname = base64_encode("");
3195                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
3196                                                 PREG_SPLIT_DELIM_CAPTURE);
3197                                         $text = '';
3198                                         $nsec = $headingOffset;
3199                                         for( $i = 0; $i < count($m); $i += 2 ) {
3200                                                 $text .= $m[$i];
3201                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
3202                                                 $hl = $m[$i + 1];
3203                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
3204                                                         $text .= $hl;
3205                                                         continue;
3206                                                 }
3207                                                 $m2 = array();
3208                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
3209                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
3210                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
3211
3212                                                 $nsec++;
3213                                         }
3214                                 }
3215                         }
3216                         wfProfileOut( __METHOD__ . '-placeholders' );
3217                 }
3218
3219                 # Prune lower levels off the recursion check path
3220                 $this->mTemplatePath = $lastPathLevel;
3221
3222                 if ( !$found ) {
3223                         wfProfileOut( $fname );
3224                         return $piece['text'];
3225                 } else {
3226                         wfProfileOut( $fname );
3227                         return $text;
3228                 }
3229         }
3230
3231         /**
3232          * Fetch the unparsed text of a template and register a reference to it.
3233          */
3234         function fetchTemplate( $title ) {
3235                 $text = false;
3236                 // Loop to fetch the article, with up to 1 redirect
3237                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3238                         $rev = Revision::newFromTitle( $title );
3239                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
3240                         if ( $rev ) {
3241                                 $text = $rev->getText();
3242                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3243                                 global $wgLang;
3244                                 $message = $wgLang->lcfirst( $title->getText() );
3245                                 $text = wfMsgForContentNoTrans( $message );
3246                                 if( wfEmptyMsg( $message, $text ) ) {
3247                                         $text = false;
3248                                         break;
3249                                 }
3250                         } else {
3251                                 break;
3252                         }
3253                         if ( $text === false ) {
3254                                 break;
3255                         }
3256                         // Redirect?
3257                         $title = Title::newFromRedirect( $text );
3258                 }
3259                 return $text;
3260         }
3261
3262         /**
3263          * Transclude an interwiki link.
3264          */
3265         function interwikiTransclude( $title, $action ) {
3266                 global $wgEnableScaryTranscluding;
3267
3268                 if (!$wgEnableScaryTranscluding)
3269                         return wfMsg('scarytranscludedisabled');
3270
3271                 $url = $title->getFullUrl( "action=$action" );
3272
3273                 if (strlen($url) > 255)
3274                         return wfMsg('scarytranscludetoolong');
3275                 return $this->fetchScaryTemplateMaybeFromCache($url);
3276         }
3277
3278         function fetchScaryTemplateMaybeFromCache($url) {
3279                 global $wgTranscludeCacheExpiry;
3280                 $dbr =& wfGetDB(DB_SLAVE);
3281                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3282                                 array('tc_url' => $url));
3283                 if ($obj) {
3284                         $time = $obj->tc_time;
3285                         $text = $obj->tc_contents;
3286                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3287                                 return $text;
3288                         }
3289                 }
3290
3291                 $text = Http::get($url);
3292                 if (!$text)
3293                         return wfMsg('scarytranscludefailed', $url);
3294
3295                 $dbw =& wfGetDB(DB_MASTER);
3296                 $dbw->replace('transcache', array('tc_url'), array(
3297                         'tc_url' => $url,
3298                         'tc_time' => time(),
3299                         'tc_contents' => $text));
3300                 return $text;
3301         }
3302
3303
3304         /**
3305          * Triple brace replacement -- used for template arguments
3306          * @private
3307          */
3308         function argSubstitution( $matches ) {
3309                 $arg = trim( $matches['title'] );
3310                 $text = $matches['text'];
3311                 $inputArgs = end( $this->mArgStack );
3312
3313                 if ( array_key_exists( $arg, $inputArgs ) ) {
3314                         $text = $inputArgs[$arg];
3315                 } else if (($this->mOutputType == OT_HTML || $this->mOutputType == OT_PREPROCESS ) &&
3316                 null != $matches['parts'] && count($matches['parts']) > 0) {
3317                         $text = $matches['parts'][0];
3318                 }
3319                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3320                         $text = $matches['text'] .
3321                                 '<!-- WARNING: argument omitted, expansion size too large -->';
3322                 }
3323
3324                 return $text;
3325         }
3326
3327         /**
3328          * Increment an include size counter
3329          *
3330          * @param string $type The type of expansion
3331          * @param integer $size The size of the text
3332          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3333          */
3334         function incrementIncludeSize( $type, $size ) {
3335                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3336                         return false;
3337                 } else {
3338                         $this->mIncludeSizes[$type] += $size;
3339                         return true;
3340                 }
3341         }
3342
3343         /**
3344          * Detect __NOGALLERY__ magic word and set a placeholder
3345          */
3346         function stripNoGallery( &$text ) {
3347                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3348                 # do not add TOC
3349                 $mw = MagicWord::get( 'nogallery' );
3350                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3351         }
3352
3353         /**
3354          * Detect __TOC__ magic word and set a placeholder
3355          */
3356         function stripToc( $text ) {
3357                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3358                 # do not add TOC
3359                 $mw = MagicWord::get( 'notoc' );
3360                 if( $mw->matchAndRemove( $text ) ) {
3361                         $this->mShowToc = false;
3362                 }
3363
3364                 $mw = MagicWord::get( 'toc' );
3365                 if( $mw->match( $text ) ) {
3366                         $this->mShowToc = true;
3367                         $this->mForceTocPosition = true;
3368
3369                         // Set a placeholder. At the end we'll fill it in with the TOC.
3370                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3371
3372                         // Only keep the first one.
3373                         $text = $mw->replace( '', $text );
3374                 }
3375                 return $text;
3376         }
3377
3378         /**
3379          * This function accomplishes several tasks:
3380          * 1) Auto-number headings if that option is enabled
3381          * 2) Add an [edit] link to sections for logged in users who have enabled the option
3382          * 3) Add a Table of contents on the top for users who have enabled the option
3383          * 4) Auto-anchor headings
3384          *
3385          * It loops through all headlines, collects the necessary data, then splits up the
3386          * string and re-inserts the newly formatted headlines.
3387          *
3388          * @param string $text
3389          * @param boolean $isMain
3390          * @private
3391          */
3392         function formatHeadings( $text, $isMain=true ) {
3393                 global $wgMaxTocLevel, $wgContLang;
3394
3395                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3396                 if( !$this->mTitle->quickUserCan( 'edit' ) ) {
3397                         $showEditLink = 0;
3398                 } else {
3399                         $showEditLink = $this->mOptions->getEditSection();
3400                 }
3401
3402                 # Inhibit editsection links if requested in the page
3403                 $esw =& MagicWord::get( 'noeditsection' );
3404                 if( $esw->matchAndRemove( $text ) ) {
3405                         $showEditLink = 0;
3406                 }
3407
3408                 # Get all headlines for numbering them and adding funky stuff like [edit]
3409                 # links - this is for later, but we need the number of headlines right now
3410                 $matches = array();
3411                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3412
3413                 # if there are fewer than 4 headlines in the article, do not show TOC
3414                 # unless it's been explicitly enabled.
3415                 $enoughToc = $this->mShowToc &&
3416                         (($numMatches >= 4) || $this->mForceTocPosition);
3417
3418                 # Allow user to stipulate that a page should have a "new section"
3419                 # link added via __NEWSECTIONLINK__
3420                 $mw =& MagicWord::get( 'newsectionlink' );
3421                 if( $mw->matchAndRemove( $text ) )
3422                         $this->mOutput->setNewSection( true );
3423
3424                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3425                 # override above conditions and always show TOC above first header
3426                 $mw =& MagicWord::get( 'forcetoc' );
3427                 if ($mw->matchAndRemove( $text ) ) {
3428                         $this->mShowToc = true;
3429                         $enoughToc = true;
3430                 }
3431
3432                 # Never ever show TOC if no headers
3433                 if( $numMatches < 1 ) {
3434                         $enoughToc = false;
3435                 }
3436
3437                 # We need this to perform operations on the HTML
3438                 $sk =& $this->mOptions->getSkin();
3439
3440                 # headline counter
3441                 $headlineCount = 0;
3442                 $sectionCount = 0; # headlineCount excluding template sections
3443
3444                 # Ugh .. the TOC should have neat indentation levels which can be
3445                 # passed to the skin functions. These are determined here
3446                 $toc = '';
3447                 $full = '';
3448                 $head = array();
3449                 $sublevelCount = array();
3450                 $levelCount = array();
3451                 $toclevel = 0;
3452                 $level = 0;
3453                 $prevlevel = 0;
3454                 $toclevel = 0;
3455                 $prevtoclevel = 0;
3456
3457                 foreach( $matches[3] as $headline ) {
3458                         $istemplate = 0;
3459                         $templatetitle = '';
3460                         $templatesection = 0;
3461                         $numbering = '';
3462                         $mat = array();
3463                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3464                                 $istemplate = 1;
3465                                 $templatetitle = base64_decode($mat[1]);
3466                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3467                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3468                         }
3469
3470                         if( $toclevel ) {
3471                                 $prevlevel = $level;
3472                                 $prevtoclevel = $toclevel;
3473                         }
3474                         $level = $matches[1][$headlineCount];
3475
3476                         if( $doNumberHeadings || $enoughToc ) {
3477
3478                                 if ( $level > $prevlevel ) {
3479                                         # Increase TOC level
3480                                         $toclevel++;
3481                                         $sublevelCount[$toclevel] = 0;
3482                                         if( $toclevel<$wgMaxTocLevel ) {
3483                                                 $toc .= $sk->tocIndent();
3484                                         }
3485                                 }
3486                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3487                                         # Decrease TOC level, find level to jump to
3488
3489                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3490                                                 # Can only go down to level 1
3491                                                 $toclevel = 1;
3492                                         } else {
3493                                                 for ($i = $toclevel; $i > 0; $i--) {
3494                                                         if ( $levelCount[$i] == $level ) {
3495                                                                 # Found last matching level
3496                                                                 $toclevel = $i;
3497                                                                 break;
3498                                                         }
3499                                                         elseif ( $levelCount[$i] < $level ) {
3500                                                                 # Found first matching level below current level
3501                                                                 $toclevel = $i + 1;
3502                                                                 break;
3503                                                         }
3504                                                 }
3505                                         }
3506                                         if( $toclevel<$wgMaxTocLevel ) {
3507                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3508                                         }
3509                                 }
3510                                 else {
3511                                         # No change in level, end TOC line
3512                                         if( $toclevel<$wgMaxTocLevel ) {
3513                                                 $toc .= $sk->tocLineEnd();
3514                                         }
3515                                 }
3516
3517                                 $levelCount[$toclevel] = $level;
3518
3519                                 # count number of headlines for each level
3520                                 @$sublevelCount[$toclevel]++;
3521                                 $dot = 0;
3522                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3523                                         if( !empty( $sublevelCount[$i] ) ) {
3524                                                 if( $dot ) {
3525                                                         $numbering .= '.';
3526                                                 }
3527                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3528                                                 $dot = 1;
3529                                         }
3530                                 }
3531                         }
3532
3533                         # The canonized header is a version of the header text safe to use for links
3534                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3535                         $canonized_headline = $this->mStripState->unstripBoth( $headline );
3536
3537                         # Remove link placeholders by the link text.
3538                         #     <!--LINK number-->
3539                         # turns into
3540                         #     link text with suffix
3541                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3542                                                             "\$this->mLinkHolders['texts'][\$1]",
3543                                                             $canonized_headline );
3544                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3545                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3546                                                             $canonized_headline );
3547
3548                         # strip out HTML
3549                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3550                         $tocline = trim( $canonized_headline );
3551                         # Save headline for section edit hint before it's escaped
3552                         $headline_hint = trim( $canonized_headline );
3553                         $canonized_headline = Sanitizer::escapeId( $tocline );
3554                         $refers[$headlineCount] = $canonized_headline;
3555
3556                         # count how many in assoc. array so we can track dupes in anchors
3557                         isset( $refers[$canonized_headline] ) ? $refers[$canonized_headline]++ : $refers[$canonized_headline] = 1;
3558                         $refcount[$headlineCount]=$refers[$canonized_headline];
3559
3560                         # Don't number the heading if it is the only one (looks silly)
3561                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3562                                 # the two are different if the line contains a link
3563                                 $headline=$numbering . ' ' . $headline;
3564                         }
3565
3566                         # Create the anchor for linking from the TOC to the section
3567                         $anchor = $canonized_headline;
3568                         if($refcount[$headlineCount] > 1 ) {
3569                                 $anchor .= '_' . $refcount[$headlineCount];
3570                         }
3571                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3572                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3573                         }
3574                         # give headline the correct <h#> tag
3575                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3576                                 if( $istemplate )
3577                                         $editlink = $sk->editSectionLinkForOther($templatetitle, $templatesection);
3578                                 else
3579                                         $editlink = $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3580                         } else {
3581                                 $editlink = '';
3582                         }
3583                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
3584
3585                         $headlineCount++;
3586                         if( !$istemplate )
3587                                 $sectionCount++;
3588                 }
3589
3590                 if( $enoughToc ) {
3591                         if( $toclevel<$wgMaxTocLevel ) {
3592                                 $toc .= $sk->tocUnindent( $toclevel - 1 );
3593                         }
3594                         $toc = $sk->tocList( $toc );
3595                 }
3596
3597                 # split up and insert constructed headlines
3598
3599                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3600                 $i = 0;
3601
3602                 foreach( $blocks as $block ) {
3603                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3604                                 # This is the [edit] link that appears for the top block of text when
3605                                 # section editing is enabled
3606
3607                                 # Disabled because it broke block formatting
3608                                 # For example, a bullet point in the top line
3609                                 # $full .= $sk->editSectionLink(0);
3610                         }
3611                         $full .= $block;
3612                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3613                                 # Top anchor now in skin
3614                                 $full = $full.$toc;
3615                         }
3616
3617                         if( !empty( $head[$i] ) ) {
3618                                 $full .= $head[$i];
3619                         }
3620                         $i++;
3621                 }
3622                 if( $this->mForceTocPosition ) {
3623                         return str_replace( '<!--MWTOC-->', $toc, $full );
3624                 } else {
3625                         return $full;
3626                 }
3627         }
3628
3629         /**
3630          * Transform wiki markup when saving a page by doing \r\n -> \n
3631          * conversion, substitting signatures, {{subst:}} templates, etc.
3632          *
3633          * @param string $text the text to transform
3634          * @param Title &$title the Title object for the current article
3635          * @param User &$user the User object describing the current user
3636          * @param ParserOptions $options parsing options
3637          * @param bool $clearState whether to clear the parser state first
3638          * @return string the altered wiki markup
3639          * @public
3640          */
3641         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3642                 $this->mOptions = $options;
3643                 $this->mTitle =& $title;
3644                 $this->setOutputType( OT_WIKI );
3645
3646                 if ( $clearState ) {
3647                         $this->clearState();
3648                 }
3649
3650                 $stripState = new StripState;
3651                 $pairs = array(
3652                         "\r\n" => "\n",
3653                 );
3654                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3655                 $text = $this->strip( $text, $stripState, true, array( 'gallery' ) );
3656                 $text = $this->pstPass2( $text, $stripState, $user );
3657                 $text = $stripState->unstripBoth( $text );
3658                 return $text;
3659         }
3660
3661         /**
3662          * Pre-save transform helper function
3663          * @private
3664          */
3665         function pstPass2( $text, &$stripState, $user ) {
3666                 global $wgContLang, $wgLocaltimezone;
3667
3668                 /* Note: This is the timestamp saved as hardcoded wikitext to
3669                  * the database, we use $wgContLang here in order to give
3670                  * everyone the same signature and use the default one rather
3671                  * than the one selected in each user's preferences.
3672                  */
3673                 if ( isset( $wgLocaltimezone ) ) {
3674                         $oldtz = getenv( 'TZ' );
3675                         putenv( 'TZ='.$wgLocaltimezone );
3676                 }
3677                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3678                   ' (' . date( 'T' ) . ')';
3679                 if ( isset( $wgLocaltimezone ) ) {
3680                         putenv( 'TZ='.$oldtz );
3681                 }
3682
3683                 # Variable replacement
3684                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3685                 $text = $this->replaceVariables( $text );
3686
3687                 # Strip out <nowiki> etc. added via replaceVariables
3688                 $text = $this->strip( $text, $stripState, false, array( 'gallery' ) );
3689
3690                 # Signatures
3691                 $sigText = $this->getUserSig( $user );
3692                 $text = strtr( $text, array(
3693                         '~~~~~' => $d,
3694                         '~~~~' => "$sigText $d",
3695                         '~~~' => $sigText
3696                 ) );
3697
3698                 # Context links: [[|name]] and [[name (context)|]]
3699                 #
3700                 global $wgLegalTitleChars;
3701                 $tc = "[$wgLegalTitleChars]";
3702                 $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3703
3704                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3705                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3706                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3707
3708                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3709                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3710                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3711
3712                 $t = $this->mTitle->getText();
3713                 $m = array();
3714                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3715                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3716                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3717                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3718                 } else {
3719                         # if there's no context, don't bother duplicating the title
3720                         $text = preg_replace( $p2, '[[\\1]]', $text );
3721                 }
3722
3723                 # Trim trailing whitespace
3724                 $text = rtrim( $text );
3725
3726                 return $text;
3727         }
3728
3729         /**
3730          * Fetch the user's signature text, if any, and normalize to
3731          * validated, ready-to-insert wikitext.
3732          *
3733          * @param User $user
3734          * @return string
3735          * @private
3736          */
3737         function getUserSig( &$user ) {
3738                 $username = $user->getName();
3739                 $nickname = $user->getOption( 'nickname' );
3740                 $nickname = $nickname === '' ? $username : $nickname;
3741
3742                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3743                         # Sig. might contain markup; validate this
3744                         if( $this->validateSig( $nickname ) !== false ) {
3745                                 # Validated; clean up (if needed) and return it
3746                                 return $this->cleanSig( $nickname, true );
3747                         } else {
3748                                 # Failed to validate; fall back to the default
3749                                 $nickname = $username;
3750                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3751                         }
3752                 }
3753
3754                 // Make sure nickname doesnt get a sig in a sig
3755                 $nickname = $this->cleanSigInSig( $nickname );
3756
3757                 # If we're still here, make it a link to the user page
3758                 $userpage = $user->getUserPage();
3759                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3760         }
3761
3762         /**
3763          * Check that the user's signature contains no bad XML
3764          *
3765          * @param string $text
3766          * @return mixed An expanded string, or false if invalid.
3767          */
3768         function validateSig( $text ) {
3769                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3770         }
3771
3772         /**
3773          * Clean up signature text
3774          *
3775          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3776          * 2) Substitute all transclusions
3777          *
3778          * @param string $text
3779          * @param $parsing Whether we're cleaning (preferences save) or parsing
3780          * @return string Signature text
3781          */
3782         function cleanSig( $text, $parsing = false ) {
3783                 global $wgTitle;
3784                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
3785
3786                 $substWord = MagicWord::get( 'subst' );
3787                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3788                 $substText = '{{' . $substWord->getSynonym( 0 );
3789
3790                 $text = preg_replace( $substRegex, $substText, $text );
3791                 $text = $this->cleanSigInSig( $text );
3792                 $text = $this->replaceVariables( $text );
3793
3794                 $this->clearState();
3795                 return $text;
3796         }
3797
3798         /**
3799          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3800          * @param string $text
3801          * @return string Signature text with /~{3,5}/ removed
3802          */
3803         function cleanSigInSig( $text ) {
3804                 $text = preg_replace( '/~{3,5}/', '', $text );
3805                 return $text;
3806         }
3807
3808         /**
3809          * Set up some variables which are usually set up in parse()
3810          * so that an external function can call some class members with confidence
3811          * @public
3812          */
3813         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3814                 $this->mTitle =& $title;
3815                 $this->mOptions = $options;
3816                 $this->setOutputType( $outputType );
3817                 if ( $clearState ) {
3818                         $this->clearState();
3819                 }
3820         }
3821
3822         /**
3823          * Transform a MediaWiki message by replacing magic variables.
3824          *
3825          * @param string $text the text to transform
3826          * @param ParserOptions $options  options
3827          * @return string the text with variables substituted
3828          * @public
3829          */
3830         function transformMsg( $text, $options ) {
3831                 global $wgTitle;
3832                 static $executing = false;
3833
3834                 $fname = "Parser::transformMsg";
3835
3836                 # Guard against infinite recursion
3837                 if ( $executing ) {
3838                         return $text;
3839                 }
3840                 $executing = true;
3841
3842                 wfProfileIn($fname);
3843
3844                 if ( $wgTitle && !( $wgTitle instanceof FakeTitle ) ) {
3845                         $this->mTitle = $wgTitle;
3846                 } else {
3847                         $this->mTitle = Title::newFromText('msg');
3848                 }
3849                 $this->mOptions = $options;
3850                 $this->setOutputType( OT_MSG );
3851                 $this->clearState();
3852                 $text = $this->replaceVariables( $text );
3853
3854                 $executing = false;
3855                 wfProfileOut($fname);
3856                 return $text;
3857         }
3858
3859         /**
3860          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3861          * The callback should have the following form:
3862          *    function myParserHook( $text, $params, &$parser ) { ... }
3863          *
3864          * Transform and return $text. Use $parser for any required context, e.g. use
3865          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3866          *
3867          * @public
3868          *
3869          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3870          * @param mixed $callback The callback function (and object) to use for the tag
3871          *
3872          * @return The old value of the mTagHooks array associated with the hook
3873          */
3874         function setHook( $tag, $callback ) {
3875                 $tag = strtolower( $tag );
3876                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
3877                 $this->mTagHooks[$tag] = $callback;
3878
3879                 return $oldVal;
3880         }
3881
3882         /**
3883          * Create a function, e.g. {{sum:1|2|3}}
3884          * The callback function should have the form:
3885          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3886          *
3887          * The callback may either return the text result of the function, or an array with the text
3888          * in element 0, and a number of flags in the other elements. The names of the flags are
3889          * specified in the keys. Valid flags are:
3890          *   found                     The text returned is valid, stop processing the template. This
3891          *                             is on by default.
3892          *   nowiki                    Wiki markup in the return value should be escaped
3893          *   noparse                   Unsafe HTML tags should not be stripped, etc.
3894          *   noargs                    Don't replace triple-brace arguments in the return value
3895          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3896          *
3897          * @public
3898          *
3899          * @param string $id The magic word ID
3900          * @param mixed $callback The callback function (and object) to use
3901          * @param integer $flags a combination of the following flags:
3902          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
3903          *
3904          * @return The old callback function for this name, if any
3905          */
3906         function setFunctionHook( $id, $callback, $flags = 0 ) {
3907                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id] : null;
3908                 $this->mFunctionHooks[$id] = $callback;
3909
3910                 # Add to function cache
3911                 $mw = MagicWord::get( $id );
3912                 if( !$mw )
3913                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
3914
3915                 $synonyms = $mw->getSynonyms();
3916                 $sensitive = intval( $mw->isCaseSensitive() );
3917
3918                 foreach ( $synonyms as $syn ) {
3919                         # Case
3920                         if ( !$sensitive ) {
3921                                 $syn = strtolower( $syn );
3922                         }
3923                         # Add leading hash
3924                         if ( !( $flags & SFH_NO_HASH ) ) {
3925                                 $syn = '#' . $syn;
3926                         }
3927                         # Remove trailing colon
3928                         if ( substr( $syn, -1, 1 ) == ':' ) {
3929                                 $syn = substr( $syn, 0, -1 );
3930                         }
3931                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
3932                 }
3933                 return $oldVal;
3934         }
3935
3936         /**
3937          * Get all registered function hook identifiers
3938          *
3939          * @return array
3940          */
3941         function getFunctionHooks() {
3942                 return array_keys( $this->mFunctionHooks );
3943         }
3944
3945         /**
3946          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3947          * Placeholders created in Skin::makeLinkObj()
3948          * Returns an array of links found, indexed by PDBK:
3949          *  0 - broken
3950          *  1 - normal link
3951          *  2 - stub
3952          * $options is a bit field, RLH_FOR_UPDATE to select for update
3953          */
3954         function replaceLinkHolders( &$text, $options = 0 ) {
3955                 global $wgUser;
3956                 global $wgContLang;
3957
3958                 $fname = 'Parser::replaceLinkHolders';
3959                 wfProfileIn( $fname );
3960
3961                 $pdbks = array();
3962                 $colours = array();
3963                 $sk =& $this->mOptions->getSkin();
3964                 $linkCache =& LinkCache::singleton();
3965
3966                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3967                         wfProfileIn( $fname.'-check' );
3968                         $dbr =& wfGetDB( DB_SLAVE );
3969                         $page = $dbr->tableName( 'page' );
3970                         $threshold = $wgUser->getOption('stubthreshold');
3971
3972                         # Sort by namespace
3973                         asort( $this->mLinkHolders['namespaces'] );
3974
3975                         # Generate query
3976                         $query = false;
3977                         $current = null;
3978                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3979                                 # Make title object
3980                                 $title = $this->mLinkHolders['titles'][$key];
3981
3982                                 # Skip invalid entries.
3983                                 # Result will be ugly, but prevents crash.
3984                                 if ( is_null( $title ) ) {
3985                                         continue;
3986                                 }
3987                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3988
3989                                 # Check if it's a static known link, e.g. interwiki
3990                                 if ( $title->isAlwaysKnown() ) {
3991                                         $colours[$pdbk] = 1;
3992                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3993                                         $colours[$pdbk] = 1;
3994                                         $this->mOutput->addLink( $title, $id );
3995                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3996                                         $colours[$pdbk] = 0;
3997                                 } else {
3998                                         # Not in the link cache, add it to the query
3999                                         if ( !isset( $current ) ) {
4000                                                 $current = $ns;
4001                                                 $query =  "SELECT page_id, page_namespace, page_title";
4002                                                 if ( $threshold > 0 ) {
4003                                                         $query .= ', page_len, page_is_redirect';
4004                                                 }
4005                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
4006                                         } elseif ( $current != $ns ) {
4007                                                 $current = $ns;
4008                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
4009                                         } else {
4010                                                 $query .= ', ';
4011                                         }
4012
4013                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
4014                                 }
4015                         }
4016                         if ( $query ) {
4017                                 $query .= '))';
4018                                 if ( $options & RLH_FOR_UPDATE ) {
4019                                         $query .= ' FOR UPDATE';
4020                                 }
4021
4022                                 $res = $dbr->query( $query, $fname );
4023
4024                                 # Fetch data and form into an associative array
4025                                 # non-existent = broken
4026                                 # 1 = known
4027                                 # 2 = stub
4028                                 while ( $s = $dbr->fetchObject($res) ) {
4029                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4030                                         $pdbk = $title->getPrefixedDBkey();
4031                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
4032                                         $this->mOutput->addLink( $title, $s->page_id );
4033
4034                                         if ( $threshold >  0 ) {
4035                                                 $size = $s->page_len;
4036                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
4037                                                         $colours[$pdbk] = 1;
4038                                                 } else {
4039                                                         $colours[$pdbk] = 2;
4040                                                 }
4041                                         } else {
4042                                                 $colours[$pdbk] = 1;
4043                                         }
4044                                 }
4045                         }
4046                         wfProfileOut( $fname.'-check' );
4047
4048                         # Do a second query for different language variants of links and categories
4049                         if($wgContLang->hasVariants()){
4050                                 $linkBatch = new LinkBatch();
4051                                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
4052                                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
4053                                 $varCategories = array(); // category replacements oldDBkey => newDBkey
4054
4055                                 $categories = $this->mOutput->getCategoryLinks();
4056
4057                                 // Add variants of links to link batch
4058                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4059                                         $title = $this->mLinkHolders['titles'][$key];
4060                                         if ( is_null( $title ) )
4061                                                 continue;
4062
4063                                         $pdbk = $title->getPrefixedDBkey();
4064                                         $titleText = $title->getText();
4065
4066                                         // generate all variants of the link title text
4067                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
4068
4069                                         // if link was not found (in first query), add all variants to query
4070                                         if ( !isset($colours[$pdbk]) ){
4071                                                 foreach($allTextVariants as $textVariant){
4072                                                         if($textVariant != $titleText){
4073                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
4074                                                                 if(is_null($variantTitle)) continue;
4075                                                                 $linkBatch->addObj( $variantTitle );
4076                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4077                                                         }
4078                                                 }
4079                                         }
4080                                 }
4081
4082                                 // process categories, check if a category exists in some variant
4083                                 foreach( $categories as $category){
4084                                         $variants = $wgContLang->convertLinkToAllVariants($category);
4085                                         foreach($variants as $variant){
4086                                                 if($variant != $category){
4087                                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
4088                                                         if(is_null($variantTitle)) continue;
4089                                                         $linkBatch->addObj( $variantTitle );
4090                                                         $categoryMap[$variant] = $category;
4091                                                 }
4092                                         }
4093                                 }
4094
4095
4096                                 if(!$linkBatch->isEmpty()){
4097                                         // construct query
4098                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4099
4100                                         $variantQuery =  "SELECT page_id, page_namespace, page_title";
4101                                         if ( $threshold > 0 ) {
4102                                                 $variantQuery .= ', page_len, page_is_redirect';
4103                                         }
4104
4105                                         $variantQuery .= " FROM $page WHERE $titleClause";
4106                                         if ( $options & RLH_FOR_UPDATE ) {
4107                                                 $variantQuery .= ' FOR UPDATE';
4108                                         }
4109
4110                                         $varRes = $dbr->query( $variantQuery, $fname );
4111
4112                                         // for each found variants, figure out link holders and replace
4113                                         while ( $s = $dbr->fetchObject($varRes) ) {
4114
4115                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4116                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4117                                                 $vardbk = $variantTitle->getDBkey();
4118
4119                                                 $holderKeys = array();
4120                                                 if(isset($variantMap[$varPdbk])){
4121                                                         $holderKeys = $variantMap[$varPdbk];
4122                                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle );
4123                                                         $this->mOutput->addLink( $variantTitle, $s->page_id );
4124                                                 }
4125
4126                                                 // loop over link holders
4127                                                 foreach($holderKeys as $key){
4128                                                         $title = $this->mLinkHolders['titles'][$key];
4129                                                         if ( is_null( $title ) ) continue;
4130
4131                                                         $pdbk = $title->getPrefixedDBkey();
4132
4133                                                         if(!isset($colours[$pdbk])){
4134                                                                 // found link in some of the variants, replace the link holder data
4135                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4136                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4137
4138                                                                 // set pdbk and colour
4139                                                                 $pdbks[$key] = $varPdbk;
4140                                                                 if ( $threshold >  0 ) {
4141                                                                         $size = $s->page_len;
4142                                                                         if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
4143                                                                                 $colours[$varPdbk] = 1;
4144                                                                         } else {
4145                                                                                 $colours[$varPdbk] = 2;
4146                                                                         }
4147                                                                 }
4148                                                                 else {
4149                                                                         $colours[$varPdbk] = 1;
4150                                                                 }
4151                                                         }
4152                                                 }
4153
4154                                                 // check if the object is a variant of a category
4155                                                 if(isset($categoryMap[$vardbk])){
4156                                                         $oldkey = $categoryMap[$vardbk];
4157                                                         if($oldkey != $vardbk)
4158                                                                 $varCategories[$oldkey]=$vardbk;
4159                                                 }
4160                                         }
4161
4162                                         // rebuild the categories in original order (if there are replacements)
4163                                         if(count($varCategories)>0){
4164                                                 $newCats = array();
4165                                                 $originalCats = $this->mOutput->getCategories();
4166                                                 foreach($originalCats as $cat => $sortkey){
4167                                                         // make the replacement
4168                                                         if( array_key_exists($cat,$varCategories) )
4169                                                                 $newCats[$varCategories[$cat]] = $sortkey;
4170                                                         else $newCats[$cat] = $sortkey;
4171                                                 }
4172                                                 $this->mOutput->setCategoryLinks($newCats);
4173                                         }
4174                                 }
4175                         }
4176
4177                         # Construct search and replace arrays
4178                         wfProfileIn( $fname.'-construct' );
4179                         $replacePairs = array();
4180                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4181                                 $pdbk = $pdbks[$key];
4182                                 $searchkey = "<!--LINK $key-->";
4183                                 $title = $this->mLinkHolders['titles'][$key];
4184                                 if ( empty( $colours[$pdbk] ) ) {
4185                                         $linkCache->addBadLinkObj( $title );
4186                                         $colours[$pdbk] = 0;
4187                                         $this->mOutput->addLink( $title, 0 );
4188                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4189                                                                         $this->mLinkHolders['texts'][$key],
4190                                                                         $this->mLinkHolders['queries'][$key] );
4191                                 } elseif ( $colours[$pdbk] == 1 ) {
4192                                         $replacePairs[$searchkey] = $sk->makeKnownLinkObj( $title,
4193                                                                         $this->mLinkHolders['texts'][$key],
4194                                                                         $this->mLinkHolders['queries'][$key] );
4195                                 } elseif ( $colours[$pdbk] == 2 ) {
4196                                         $replacePairs[$searchkey] = $sk->makeStubLinkObj( $title,
4197                                                                         $this->mLinkHolders['texts'][$key],
4198                                                                         $this->mLinkHolders['queries'][$key] );
4199                                 }
4200                         }
4201                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4202                         wfProfileOut( $fname.'-construct' );
4203
4204                         # Do the thing
4205                         wfProfileIn( $fname.'-replace' );
4206                         $text = preg_replace_callback(
4207                                 '/(<!--LINK .*?-->)/',
4208                                 $replacer->cb(),
4209                                 $text);
4210
4211                         wfProfileOut( $fname.'-replace' );
4212                 }
4213
4214                 # Now process interwiki link holders
4215                 # This is quite a bit simpler than internal links
4216                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4217                         wfProfileIn( $fname.'-interwiki' );
4218                         # Make interwiki link HTML
4219                         $replacePairs = array();
4220                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4221                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4222                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4223                         }
4224                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4225
4226                         $text = preg_replace_callback(
4227                                 '/<!--IWLINK (.*?)-->/',
4228                                 $replacer->cb(),
4229                                 $text );
4230                         wfProfileOut( $fname.'-interwiki' );
4231                 }
4232
4233                 wfProfileOut( $fname );
4234                 return $colours;
4235         }
4236
4237         /**
4238          * Replace <!--LINK--> link placeholders with plain text of links
4239          * (not HTML-formatted).
4240          * @param string $text
4241          * @return string
4242          */
4243         function replaceLinkHoldersText( $text ) {
4244                 $fname = 'Parser::replaceLinkHoldersText';
4245                 wfProfileIn( $fname );
4246
4247                 $text = preg_replace_callback(
4248                         '/<!--(LINK|IWLINK) (.*?)-->/',
4249                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4250                         $text );
4251
4252                 wfProfileOut( $fname );
4253                 return $text;
4254         }
4255
4256         /**
4257          * @param array $matches
4258          * @return string
4259          * @private
4260          */
4261         function replaceLinkHoldersTextCallback( $matches ) {
4262                 $type = $matches[1];
4263                 $key  = $matches[2];
4264                 if( $type == 'LINK' ) {
4265                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4266                                 return $this->mLinkHolders['texts'][$key];
4267                         }
4268                 } elseif( $type == 'IWLINK' ) {
4269                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4270                                 return $this->mInterwikiLinkHolders['texts'][$key];
4271                         }
4272                 }
4273                 return $matches[0];
4274         }
4275
4276         /**
4277          * Tag hook handler for 'pre'.
4278          */
4279         function renderPreTag( $text, $attribs ) {
4280                 // Backwards-compatibility hack
4281                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4282
4283                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4284                 return wfOpenElement( 'pre', $attribs ) .
4285                         Xml::escapeTagsOnly( $content ) .
4286                         '</pre>';
4287         }
4288
4289         /**
4290          * Renders an image gallery from a text with one line per image.
4291          * text labels may be given by using |-style alternative text. E.g.
4292          *   Image:one.jpg|The number "1"
4293          *   Image:tree.jpg|A tree
4294          * given as text will return the HTML of a gallery with two images,
4295          * labeled 'The number "1"' and
4296          * 'A tree'.
4297          */
4298         function renderImageGallery( $text, $params ) {
4299                 $ig = new ImageGallery();
4300                 $ig->setShowBytes( false );
4301                 $ig->setShowFilename( false );
4302                 $ig->setParsing();
4303                 $ig->useSkin( $this->mOptions->getSkin() );
4304
4305                 if( isset( $params['caption'] ) ) {
4306                         $caption = $params['caption'];
4307                         $caption = htmlspecialchars( $caption );
4308                         $caption = $this->replaceInternalLinks( $caption );
4309                         $ig->setCaptionHtml( $caption );
4310                 }
4311
4312                 $lines = explode( "\n", $text );
4313                 foreach ( $lines as $line ) {
4314                         # match lines like these:
4315                         # Image:someimage.jpg|This is some image
4316                         $matches = array();
4317                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4318                         # Skip empty lines
4319                         if ( count( $matches ) == 0 ) {
4320                                 continue;
4321                         }
4322                         $tp = Title::newFromText( $matches[1] );
4323                         $nt =& $tp;
4324                         if( is_null( $nt ) ) {
4325                                 # Bogus title. Ignore these so we don't bomb out later.
4326                                 continue;
4327                         }
4328                         if ( isset( $matches[3] ) ) {
4329                                 $label = $matches[3];
4330                         } else {
4331                                 $label = '';
4332                         }
4333
4334                         $pout = $this->parse( $label,
4335                                 $this->mTitle,
4336                                 $this->mOptions,
4337                                 false, // Strip whitespace...?
4338                                 false  // Don't clear state!
4339                         );
4340                         $html = $pout->getText();
4341
4342                         $ig->add( new Image( $nt ), $html );
4343
4344                         # Only add real images (bug #5586)
4345                         if ( $nt->getNamespace() == NS_IMAGE ) {
4346                                 $this->mOutput->addImage( $nt->getDBkey() );
4347                         }
4348                 }
4349                 return $ig->toHTML();
4350         }
4351
4352         /**
4353          * Parse image options text and use it to make an image
4354          */
4355         function makeImage( $nt, $options ) {
4356                 global $wgUseImageResize, $wgDjvuRenderer;
4357
4358                 $align = '';
4359
4360                 # Check if the options text is of the form "options|alt text"
4361                 # Options are:
4362                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4363                 #  * left               no resizing, just left align. label is used for alt= only
4364                 #  * right              same, but right aligned
4365                 #  * none               same, but not aligned
4366                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4367                 #  * center             center the image
4368                 #  * framed             Keep original image size, no magnify-button.
4369
4370                 $part = explode( '|', $options);
4371
4372                 $mwThumb  =& MagicWord::get( 'img_thumbnail' );
4373                 $mwManualThumb =& MagicWord::get( 'img_manualthumb' );
4374                 $mwLeft   =& MagicWord::get( 'img_left' );
4375                 $mwRight  =& MagicWord::get( 'img_right' );
4376                 $mwNone   =& MagicWord::get( 'img_none' );
4377                 $mwWidth  =& MagicWord::get( 'img_width' );
4378                 $mwCenter =& MagicWord::get( 'img_center' );
4379                 $mwFramed =& MagicWord::get( 'img_framed' );
4380                 $mwPage   =& MagicWord::get( 'img_page' );
4381                 $caption = '';
4382
4383                 $width = $height = $framed = $thumb = false;
4384                 $page = null;
4385                 $manual_thumb = '' ;
4386
4387                 foreach( $part as $val ) {
4388                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
4389                                 $thumb=true;
4390                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
4391                                 # use manually specified thumbnail
4392                                 $thumb=true;
4393                                 $manual_thumb = $match;
4394                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
4395                                 # remember to set an alignment, don't render immediately
4396                                 $align = 'right';
4397                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
4398                                 # remember to set an alignment, don't render immediately
4399                                 $align = 'left';
4400                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
4401                                 # remember to set an alignment, don't render immediately
4402                                 $align = 'center';
4403                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
4404                                 # remember to set an alignment, don't render immediately
4405                                 $align = 'none';
4406                         } elseif ( isset( $wgDjvuRenderer ) && $wgDjvuRenderer
4407                                    && ! is_null( $match = $mwPage->matchVariableStartToEnd($val) ) ) {
4408                                 # Select a page in a multipage document
4409                                 $page = $match;
4410                         } elseif ( $wgUseImageResize && !$width && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
4411                                 wfDebug( "img_width match: $match\n" );
4412                                 # $match is the image width in pixels
4413                                 $m = array();
4414                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
4415                                         $width = intval( $m[1] );
4416                                         $height = intval( $m[2] );
4417                                 } else {
4418                                         $width = intval($match);
4419                                 }
4420                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
4421                                 $framed=true;
4422                         } else {
4423                                 $caption = $val;
4424                         }
4425                 }
4426                 # Strip bad stuff out of the alt text
4427                 $alt = $this->replaceLinkHoldersText( $caption );
4428
4429                 # make sure there are no placeholders in thumbnail attributes
4430                 # that are later expanded to html- so expand them now and
4431                 # remove the tags
4432                 $alt = $this->mStripState->unstripBoth( $alt );
4433                 $alt = Sanitizer::stripAllTags( $alt );
4434
4435                 # Linker does the rest
4436                 $sk =& $this->mOptions->getSkin();
4437                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb, $page );
4438         }
4439
4440         /**
4441          * Set a flag in the output object indicating that the content is dynamic and
4442          * shouldn't be cached.
4443          */
4444         function disableCache() {
4445                 wfDebug( "Parser output marked as uncacheable.\n" );
4446                 $this->mOutput->mCacheTime = -1;
4447         }
4448
4449         /**#@+
4450          * Callback from the Sanitizer for expanding items found in HTML attribute
4451          * values, so they can be safely tested and escaped.
4452          * @param string $text
4453          * @param array $args
4454          * @return string
4455          * @private
4456          */
4457         function attributeStripCallback( &$text, $args ) {
4458                 $text = $this->replaceVariables( $text, $args );
4459                 $text = $this->mStripState->unstripBoth( $text );
4460                 return $text;
4461         }
4462
4463         /**#@-*/
4464
4465         /**#@+
4466          * Accessor/mutator
4467          */
4468         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4469         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4470         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4471         /**#@-*/
4472
4473         /**#@+
4474          * Accessor
4475          */
4476         function getTags() { return array_keys( $this->mTagHooks ); }
4477         /**#@-*/
4478
4479
4480         /**
4481          * Break wikitext input into sections, and either pull or replace
4482          * some particular section's text.
4483          *
4484          * External callers should use the getSection and replaceSection methods.
4485          *
4486          * @param $text Page wikitext
4487          * @param $section Numbered section. 0 pulls the text before the first
4488          *                 heading; other numbers will pull the given section
4489          *                 along with its lower-level subsections.
4490          * @param $mode One of "get" or "replace"
4491          * @param $newtext Replacement text for section data.
4492          * @return string for "get", the extracted section text.
4493          *                for "replace", the whole page with the section replaced.
4494          */
4495         private function extractSections( $text, $section, $mode, $newtext='' ) {
4496                 # strip NOWIKI etc. to avoid confusion (true-parameter causes HTML
4497                 # comments to be stripped as well)
4498                 $stripState = new StripState;
4499
4500                 $oldOutputType = $this->mOutputType;
4501                 $oldOptions = $this->mOptions;
4502                 $this->mOptions = new ParserOptions();
4503                 $this->setOutputType( OT_WIKI );
4504
4505                 $striptext = $this->strip( $text, $stripState, true );
4506
4507                 $this->setOutputType( $oldOutputType );
4508                 $this->mOptions = $oldOptions;
4509
4510                 # now that we can be sure that no pseudo-sections are in the source,
4511                 # split it up by section
4512                 $uniq = preg_quote( $this->uniqPrefix(), '/' );
4513                 $comment = "(?:$uniq-!--.*?QINU)";
4514                 $secs = preg_split(
4515                 /*
4516                         "/
4517                         ^(
4518                         (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4519                         (?:
4520                                 (=+) # Should this be limited to 6?
4521                                 .+?  # Section title...
4522                                 \\2  # Ending = count must match start
4523                         |
4524                                 ^
4525                                 <h([1-6])\b.*?>
4526                                 .*?
4527                                 <\/h\\3\s*>
4528                         )
4529                         (?:$comment|<\/?noinclude>|\s+)* # Trailing whitespace ok
4530                         )$
4531                         /mix",
4532                 */
4533                         "/
4534                         (
4535                                 ^
4536                                 (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4537                                 (=+) # Should this be limited to 6?
4538                                 .+?  # Section title...
4539                                 \\2  # Ending = count must match start
4540                                 (?:$comment|<\/?noinclude>|[ \\t]+)* # Trailing whitespace ok
4541                                 $
4542                         |
4543                                 <h([1-6])\b.*?>
4544                                 .*?
4545                                 <\/h\\3\s*>
4546                         )
4547                         /mix",
4548                         $striptext, -1,
4549                         PREG_SPLIT_DELIM_CAPTURE);
4550
4551                 if( $mode == "get" ) {
4552                         if( $section == 0 ) {
4553                                 // "Section 0" returns the content before any other section.
4554                                 $rv = $secs[0];
4555                         } else {
4556                                 $rv = "";
4557                         }
4558                 } elseif( $mode == "replace" ) {
4559                         if( $section == 0 ) {
4560                                 $rv = $newtext . "\n\n";
4561                                 $remainder = true;
4562                         } else {
4563                                 $rv = $secs[0];
4564                                 $remainder = false;
4565                         }
4566                 }
4567                 $count = 0;
4568                 $sectionLevel = 0;
4569                 for( $index = 1; $index < count( $secs ); ) {
4570                         $headerLine = $secs[$index++];
4571                         if( $secs[$index] ) {
4572                                 // A wiki header
4573                                 $headerLevel = strlen( $secs[$index++] );
4574                         } else {
4575                                 // An HTML header
4576                                 $index++;
4577                                 $headerLevel = intval( $secs[$index++] );
4578                         }
4579                         $content = $secs[$index++];
4580
4581                         $count++;
4582                         if( $mode == "get" ) {
4583                                 if( $count == $section ) {
4584                                         $rv = $headerLine . $content;
4585                                         $sectionLevel = $headerLevel;
4586                                 } elseif( $count > $section ) {
4587                                         if( $sectionLevel && $headerLevel > $sectionLevel ) {
4588                                                 $rv .= $headerLine . $content;
4589                                         } else {
4590                                                 // Broke out to a higher-level section
4591                                                 break;
4592                                         }
4593                                 }
4594                         } elseif( $mode == "replace" ) {
4595                                 if( $count < $section ) {
4596                                         $rv .= $headerLine . $content;
4597                                 } elseif( $count == $section ) {
4598                                         $rv .= $newtext . "\n\n";
4599                                         $sectionLevel = $headerLevel;
4600                                 } elseif( $count > $section ) {
4601                                         if( $headerLevel <= $sectionLevel ) {
4602                                                 // Passed the section's sub-parts.
4603                                                 $remainder = true;
4604                                         }
4605                                         if( $remainder ) {
4606                                                 $rv .= $headerLine . $content;
4607                                         }
4608                                 }
4609                         }
4610                 }
4611                 # reinsert stripped tags
4612                 $rv = trim( $stripState->unstripBoth( $rv ) );
4613                 return $rv;
4614         }
4615
4616         /**
4617          * This function returns the text of a section, specified by a number ($section).
4618          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4619          * the first section before any such heading (section 0).
4620          *
4621          * If a section contains subsections, these are also returned.
4622          *
4623          * @param $text String: text to look in
4624          * @param $section Integer: section number
4625          * @return string text of the requested section
4626          */
4627         function getSection( $text, $section ) {
4628                 return $this->extractSections( $text, $section, "get" );
4629         }
4630
4631         function replaceSection( $oldtext, $section, $text ) {
4632                 return $this->extractSections( $oldtext, $section, "replace", $text );
4633         }
4634
4635         /**
4636          * Get the timestamp associated with the current revision, adjusted for
4637          * the default server-local timestamp
4638          */
4639         function getRevisionTimestamp() {
4640                 if ( is_null( $this->mRevisionTimestamp ) ) {
4641                         wfProfileIn( __METHOD__ );
4642                         global $wgContLang;
4643                         $dbr =& wfGetDB( DB_SLAVE );
4644                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4645                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4646
4647                         // Normalize timestamp to internal MW format for timezone processing.
4648                         // This has the added side-effect of replacing a null value with
4649                         // the current time, which gives us more sensible behavior for
4650                         // previews.
4651                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4652
4653                         // The cryptic '' timezone parameter tells to use the site-default
4654                         // timezone offset instead of the user settings.
4655                         //
4656                         // Since this value will be saved into the parser cache, served
4657                         // to other users, and potentially even used inside links and such,
4658                         // it needs to be consistent for all visitors.
4659                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4660
4661                         wfProfileOut( __METHOD__ );
4662                 }
4663                 return $this->mRevisionTimestamp;
4664         }
4665
4666         /**
4667          * Mutator for $mDefaultSort
4668          *
4669          * @param $sort New value
4670          */
4671         public function setDefaultSort( $sort ) {
4672                 $this->mDefaultSort = $sort;
4673         }
4674
4675         /**
4676          * Accessor for $mDefaultSort
4677          * Will use the title/prefixed title if none is set
4678          *
4679          * @return string
4680          */
4681         public function getDefaultSort() {
4682                 if( $this->mDefaultSort !== false ) {
4683                         return $this->mDefaultSort;
4684                 } else {
4685                         return $this->mTitle->getNamespace() == NS_CATEGORY
4686                                         ? $this->mTitle->getText()
4687                                         : $this->mTitle->getPrefixedText();
4688                 }
4689         }
4690
4691 }
4692
4693 class OnlyIncludeReplacer {
4694         var $output = '';
4695
4696         function replace( $matches ) {
4697                 if ( substr( $matches[1], -1 ) == "\n" ) {
4698                         $this->output .= substr( $matches[1], 0, -1 );
4699                 } else {
4700                         $this->output .= $matches[1];
4701                 }
4702         }
4703 }
4704
4705 class StripState {
4706         var $general, $nowiki;
4707
4708         function __construct() {
4709                 $this->general = new ReplacementArray;
4710                 $this->nowiki = new ReplacementArray;
4711         }
4712
4713         function unstripGeneral( $text ) {
4714                 wfProfileIn( __METHOD__ );
4715                 $text = $this->general->replace( $text );
4716                 wfProfileOut( __METHOD__ );
4717                 return $text;
4718         }
4719
4720         function unstripNoWiki( $text ) {
4721                 wfProfileIn( __METHOD__ );
4722                 $text = $this->nowiki->replace( $text );
4723                 wfProfileOut( __METHOD__ );
4724                 return $text;
4725         }
4726
4727         function unstripBoth( $text ) {
4728                 wfProfileIn( __METHOD__ );
4729                 $text = $this->general->replace( $text );
4730                 $text = $this->nowiki->replace( $text );
4731                 wfProfileOut( __METHOD__ );
4732                 return $text;
4733         }
4734 }
4735
4736 ?>