includes/Parser.php

   1 <?php
   2 /**
   3  * File for Parser and related classes
   4  *
   5  * @package MediaWiki
   6  * @subpackage Parser
   7  */
   8
   9 /**
  10  * Update this version number when the ParserOutput format
  11  * changes in an incompatible way, so the parser cache
  12  * can automatically discard old data.
  13  */
  14 define( 'MW_PARSER_VERSION', '1.6.1' );
  15
  16 define( 'RLH_FOR_UPDATE', 1 );
  17
  18 # Allowed values for $mOutputType
  19 define( 'OT_HTML', 1 );
  20 define( 'OT_WIKI', 2 );
  21 define( 'OT_MSG' , 3 );
  22 define( 'OT_PREPROCESS', 4 );
  23
  24 # Flags for setFunctionHook
  25 define( 'SFH_NO_HASH', 1 );
  26
  27 # string parameter for extractTags which will cause it
  28 # to strip HTML comments in addition to regular
  29 # <XML>-style tags. This should not be anything we
  30 # may want to use in wikisyntax
  31 define( 'STRIP_COMMENTS', 'HTMLCommentStrip' );
  32
  33 # Constants needed for external link processing
  34 define( 'HTTP_PROTOCOLS', 'http:\/\/|https:\/\/' );
  35 # Everything except bracket, space, or control characters
  36 define( 'EXT_LINK_URL_CLASS', '[^][<>"\\x00-\\x20\\x7F]' );
  37 # Including space, but excluding newlines
  38 define( 'EXT_LINK_TEXT_CLASS', '[^\]\\x0a\\x0d]' );
  39 define( 'EXT_IMAGE_FNAME_CLASS', '[A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]' );
  40 define( 'EXT_IMAGE_EXTENSIONS', 'gif|png|jpg|jpeg' );
  41 define( 'EXT_LINK_BRACKETED',  '/\[(\b(' . wfUrlProtocols() . ')'.
  42         EXT_LINK_URL_CLASS.'+) *('.EXT_LINK_TEXT_CLASS.'*?)\]/S' );
  43 define( 'EXT_IMAGE_REGEX',
  44         '/^('.HTTP_PROTOCOLS.')'.  # Protocol
  45         '('.EXT_LINK_URL_CLASS.'+)\\/'.  # Hostname and path
  46         '('.EXT_IMAGE_FNAME_CLASS.'+)\\.((?i)'.EXT_IMAGE_EXTENSIONS.')$/S' # Filename
  47 );
  48
  49 // State constants for the definition list colon extraction
  50 define( 'MW_COLON_STATE_TEXT', 0 );
  51 define( 'MW_COLON_STATE_TAG', 1 );
  52 define( 'MW_COLON_STATE_TAGSTART', 2 );
  53 define( 'MW_COLON_STATE_CLOSETAG', 3 );
  54 define( 'MW_COLON_STATE_TAGSLASH', 4 );
  55 define( 'MW_COLON_STATE_COMMENT', 5 );
  56 define( 'MW_COLON_STATE_COMMENTDASH', 6 );
  57 define( 'MW_COLON_STATE_COMMENTDASHDASH', 7 );
  58
  59 /**
  60  * PHP Parser
  61  *
  62  * Processes wiki markup
  63  *
  64  * <pre>
  65  * There are four main entry points into the Parser class:
  66  * parse()
  67  *   produces HTML output
  68  * preSaveTransform().
  69  *   produces altered wiki markup.
  70  * transformMsg()
  71  *   performs brace substitution on MediaWiki messages
  72  * preprocess()
  73  *   removes HTML comments and expands templates
  74  *
  75  * Globals used:
  76  *    objects:   $wgLang, $wgContLang
  77  *
  78  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  79  *
  80  * settings:
  81  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  82  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  83  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  84  *  $wgMaxArticleSize*
  85  *
  86  *  * only within ParserOptions
  87  * </pre>
  88  *
  89  * @package MediaWiki
  90  */
  91 class Parser
  92 {
  93         /**#@+
  94          * @private
  95          */
  96         # Persistent:
  97         var $mTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables;
  98
  99         # Cleared with clearState():
 100         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 101         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 102         var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
 103         var $mIncludeSizes, $mDefaultSort;
 104         var $mTemplates,        // cache of already loaded templates, avoids
 105                                 // multiple SQL queries for the same string
 106             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 107                                 // in this path. Used for loop detection.
 108
 109         # Temporary
 110         # These are variables reset at least once per parse regardless of $clearState
 111         var $mOptions,      // ParserOptions object
 112                 $mTitle,        // Title context, used for self-link rendering and similar things
 113                 $mOutputType,   // Output type, one of the OT_xxx constants
 114                 $ot,            // Shortcut alias, see setOutputType()
 115                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 116                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 117                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 118
 119         /**#@-*/
 120
 121         /**
 122          * Constructor
 123          *
 124          * @public
 125          */
 126         function Parser() {
 127                 $this->mTagHooks = array();
 128                 $this->mFunctionHooks = array();
 129                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 130                 $this->mFirstCall = true;
 131         }
 132
 133         /**
 134          * Do various kinds of initialisation on the first call of the parser
 135          */
 136         function firstCallInit() {
 137                 if ( !$this->mFirstCall ) {
 138                         return;
 139                 }
 140
 141                 wfProfileIn( __METHOD__ );
 142                 global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
 143
 144                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 145
 146                 $this->setFunctionHook( 'int', array( 'CoreParserFunctions', 'intFunction' ), SFH_NO_HASH );
 147                 $this->setFunctionHook( 'ns', array( 'CoreParserFunctions', 'ns' ), SFH_NO_HASH );
 148                 $this->setFunctionHook( 'urlencode', array( 'CoreParserFunctions', 'urlencode' ), SFH_NO_HASH );
 149                 $this->setFunctionHook( 'lcfirst', array( 'CoreParserFunctions', 'lcfirst' ), SFH_NO_HASH );
 150                 $this->setFunctionHook( 'ucfirst', array( 'CoreParserFunctions', 'ucfirst' ), SFH_NO_HASH );
 151                 $this->setFunctionHook( 'lc', array( 'CoreParserFunctions', 'lc' ), SFH_NO_HASH );
 152                 $this->setFunctionHook( 'uc', array( 'CoreParserFunctions', 'uc' ), SFH_NO_HASH );
 153                 $this->setFunctionHook( 'localurl', array( 'CoreParserFunctions', 'localurl' ), SFH_NO_HASH );
 154                 $this->setFunctionHook( 'localurle', array( 'CoreParserFunctions', 'localurle' ), SFH_NO_HASH );
 155                 $this->setFunctionHook( 'fullurl', array( 'CoreParserFunctions', 'fullurl' ), SFH_NO_HASH );
 156                 $this->setFunctionHook( 'fullurle', array( 'CoreParserFunctions', 'fullurle' ), SFH_NO_HASH );
 157                 $this->setFunctionHook( 'formatnum', array( 'CoreParserFunctions', 'formatnum' ), SFH_NO_HASH );
 158                 $this->setFunctionHook( 'grammar', array( 'CoreParserFunctions', 'grammar' ), SFH_NO_HASH );
 159                 $this->setFunctionHook( 'plural', array( 'CoreParserFunctions', 'plural' ), SFH_NO_HASH );
 160                 $this->setFunctionHook( 'numberofpages', array( 'CoreParserFunctions', 'numberofpages' ), SFH_NO_HASH );
 161                 $this->setFunctionHook( 'numberofusers', array( 'CoreParserFunctions', 'numberofusers' ), SFH_NO_HASH );
 162                 $this->setFunctionHook( 'numberofarticles', array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 163                 $this->setFunctionHook( 'numberoffiles', array( 'CoreParserFunctions', 'numberoffiles' ), SFH_NO_HASH );
 164                 $this->setFunctionHook( 'numberofadmins', array( 'CoreParserFunctions', 'numberofadmins' ), SFH_NO_HASH );
 165                 $this->setFunctionHook( 'language', array( 'CoreParserFunctions', 'language' ), SFH_NO_HASH );
 166                 $this->setFunctionHook( 'padleft', array( 'CoreParserFunctions', 'padleft' ), SFH_NO_HASH );
 167                 $this->setFunctionHook( 'padright', array( 'CoreParserFunctions', 'padright' ), SFH_NO_HASH );
 168                 $this->setFunctionHook( 'anchorencode', array( 'CoreParserFunctions', 'anchorencode' ), SFH_NO_HASH );
 169                 $this->setFunctionHook( 'special', array( 'CoreParserFunctions', 'special' ) );
 170                 $this->setFunctionHook( 'defaultsort', array( 'CoreParserFunctions', 'defaultsort' ), SFH_NO_HASH );
 171
 172                 if ( $wgAllowDisplayTitle ) {
 173                         $this->setFunctionHook( 'displaytitle', array( 'CoreParserFunctions', 'displaytitle' ), SFH_NO_HASH );
 174                 }
 175                 if ( $wgAllowSlowParserFunctions ) {
 176                         $this->setFunctionHook( 'pagesinnamespace', array( 'CoreParserFunctions', 'pagesinnamespace' ), SFH_NO_HASH );
 177                 }
 178
 179                 $this->initialiseVariables();
 180                 $this->mFirstCall = false;
 181                 wfProfileOut( __METHOD__ );
 182         }
 183
 184         /**
 185          * Clear Parser state
 186          *
 187          * @private
 188          */
 189         function clearState() {
 190                 wfProfileIn( __METHOD__ );
 191                 if ( $this->mFirstCall ) {
 192                         $this->firstCallInit();
 193                 }
 194                 $this->mOutput = new ParserOutput;
 195                 $this->mAutonumber = 0;
 196                 $this->mLastSection = '';
 197                 $this->mDTopen = false;
 198                 $this->mIncludeCount = array();
 199                 $this->mStripState = new StripState;
 200                 $this->mArgStack = array();
 201                 $this->mInPre = false;
 202                 $this->mInterwikiLinkHolders = array(
 203                         'texts' => array(),
 204                         'titles' => array()
 205                 );
 206                 $this->mLinkHolders = array(
 207                         'namespaces' => array(),
 208                         'dbkeys' => array(),
 209                         'queries' => array(),
 210                         'texts' => array(),
 211                         'titles' => array()
 212                 );
 213                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 214
 215                 /**
 216                  * Prefix for temporary replacement strings for the multipass parser.
 217                  * \x07 should never appear in input as it's disallowed in XML.
 218                  * Using it at the front also gives us a little extra robustness
 219                  * since it shouldn't match when butted up against identifier-like
 220                  * string constructs.
 221                  */
 222                 $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 223
 224                 # Clear these on every parse, bug 4549
 225                 $this->mTemplates = array();
 226                 $this->mTemplatePath = array();
 227
 228                 $this->mShowToc = true;
 229                 $this->mForceTocPosition = false;
 230                 $this->mIncludeSizes = array(
 231                         'pre-expand' => 0,
 232                         'post-expand' => 0,
 233                         'arg' => 0
 234                 );
 235                 $this->mDefaultSort = false;
 236
 237                 wfRunHooks( 'ParserClearState', array( &$this ) );
 238                 wfProfileOut( __METHOD__ );
 239         }
 240
 241         function setOutputType( $ot ) {
 242                 $this->mOutputType = $ot;
 243                 // Shortcut alias
 244                 $this->ot = array(
 245                         'html' => $ot == OT_HTML,
 246                         'wiki' => $ot == OT_WIKI,
 247                         'msg' => $ot == OT_MSG,
 248                         'pre' => $ot == OT_PREPROCESS,
 249                 );
 250         }
 251
 252         /**
 253          * Accessor for mUniqPrefix.
 254          *
 255          * @public
 256          */
 257         function uniqPrefix() {
 258                 return $this->mUniqPrefix;
 259         }
 260
 261         /**
 262          * Convert wikitext to HTML
 263          * Do not call this function recursively.
 264          *
 265          * @param string $text Text we want to parse
 266          * @param Title &$title A title object
 267          * @param array $options
 268          * @param boolean $linestart
 269          * @param boolean $clearState
 270          * @param int $revid number to pass in {{REVISIONID}}
 271          * @return ParserOutput a ParserOutput
 272          */
 273         public function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 274                 /**
 275                  * First pass--just handle <nowiki> sections, pass the rest off
 276                  * to internalParse() which does all the real work.
 277                  */
 278
 279                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 280                 $fname = 'Parser::parse-' . wfGetCaller();
 281                 wfProfileIn( __METHOD__ );
 282                 wfProfileIn( $fname );
 283
 284                 if ( $clearState ) {
 285                         $this->clearState();
 286                 }
 287
 288                 $this->mOptions = $options;
 289                 $this->mTitle =& $title;
 290                 $oldRevisionId = $this->mRevisionId;
 291                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 292                 if( $revid !== null ) {
 293                         $this->mRevisionId = $revid;
 294                         $this->mRevisionTimestamp = null;
 295                 }
 296                 $this->setOutputType( OT_HTML );
 297                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 298                 $text = $this->strip( $text, $this->mStripState );
 299                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 300                 $text = $this->internalParse( $text );
 301                 $text = $this->mStripState->unstripGeneral( $text );
 302
 303                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 304                 $fixtags = array(
 305                         # french spaces, last one Guillemet-left
 306                         # only if there is something before the space
 307                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 308                         # french spaces, Guillemet-right
 309                         '/(\\302\\253) /' => '\\1&nbsp;',
 310                 );
 311                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 312
 313                 # only once and last
 314                 $text = $this->doBlockLevels( $text, $linestart );
 315
 316                 $this->replaceLinkHolders( $text );
 317
 318                 # the position of the parserConvert() call should not be changed. it
 319                 # assumes that the links are all replaced and the only thing left
 320                 # is the <nowiki> mark.
 321                 # Side-effects: this calls $this->mOutput->setTitleText()
 322                 $text = $wgContLang->parserConvert( $text, $this );
 323
 324                 $text = $this->mStripState->unstripNoWiki( $text );
 325
 326                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 327
 328                 $text = Sanitizer::normalizeCharReferences( $text );
 329
 330                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 331                         $text = Parser::tidy($text);
 332                 } else {
 333                         # attempt to sanitize at least some nesting problems
 334                         # (bug #2702 and quite a few others)
 335                         $tidyregs = array(
 336                                 # ''Something [http://www.cool.com cool''] -->
 337                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 338                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 339                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 340                                 # fix up an anchor inside another anchor, only
 341                                 # at least for a single single nested link (bug 3695)
 342                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 343                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 344                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 345                                 # contains a div, so fix it up here; replace
 346                                 # div with escaped text
 347                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 348                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 349                                 # remove empty italic or bold tag pairs, some
 350                                 # introduced by rules above
 351                                 '/<([bi])><\/\\1>/' => '',
 352                         );
 353
 354                         $text = preg_replace(
 355                                 array_keys( $tidyregs ),
 356                                 array_values( $tidyregs ),
 357                                 $text );
 358                 }
 359
 360                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 361
 362                 # Information on include size limits, for the benefit of users who try to skirt them
 363                 if ( max( $this->mIncludeSizes ) > 1000 ) {
 364                         $max = $this->mOptions->getMaxIncludeSize();
 365                         $text .= "<!-- \n" .
 366                                 "Pre-expand include size: {$this->mIncludeSizes['pre-expand']} bytes\n" .
 367                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']} bytes\n" .
 368                                 "Template argument size: {$this->mIncludeSizes['arg']} bytes\n" .
 369                                 "Maximum: $max bytes\n" .
 370                                 "-->\n";
 371                 }
 372                 $this->mOutput->setText( $text );
 373                 $this->mRevisionId = $oldRevisionId;
 374                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 375                 wfProfileOut( $fname );
 376                 wfProfileOut( __METHOD__ );
 377
 378                 return $this->mOutput;
 379         }
 380
 381         /**
 382          * Recursive parser entry point that can be called from an extension tag
 383          * hook.
 384          */
 385         function recursiveTagParse( $text ) {
 386                 wfProfileIn( __METHOD__ );
 387                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 388                 $text = $this->strip( $text, $this->mStripState );
 389                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 390                 $text = $this->internalParse( $text );
 391                 wfProfileOut( __METHOD__ );
 392                 return $text;
 393         }
 394
 395         /**
 396          * Expand templates and variables in the text, producing valid, static wikitext.
 397          * Also removes comments.
 398          */
 399         function preprocess( $text, $title, $options ) {
 400                 wfProfileIn( __METHOD__ );
 401                 $this->clearState();
 402                 $this->setOutputType( OT_PREPROCESS );
 403                 $this->mOptions = $options;
 404                 $this->mTitle = $title;
 405                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 406                 $text = $this->strip( $text, $this->mStripState );
 407                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 408                 if ( $this->mOptions->getRemoveComments() ) {
 409                         $text = Sanitizer::removeHTMLcomments( $text );
 410                 }
 411                 $text = $this->replaceVariables( $text );
 412                 $text = $this->mStripState->unstripBoth( $text );
 413                 wfProfileOut( __METHOD__ );
 414                 return $text;
 415         }
 416
 417         /**
 418          * Get a random string
 419          *
 420          * @private
 421          * @static
 422          */
 423         function getRandomString() {
 424                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 425         }
 426
 427         function &getTitle() { return $this->mTitle; }
 428         function getOptions() { return $this->mOptions; }
 429
 430         function getFunctionLang() {
 431                 global $wgLang, $wgContLang;
 432                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 433         }
 434
 435         /**
 436          * Replaces all occurrences of HTML-style comments and the given tags
 437          * in the text with a random marker and returns teh next text. The output
 438          * parameter $matches will be an associative array filled with data in
 439          * the form:
 440          *   'UNIQ-xxxxx' => array(
 441          *     'element',
 442          *     'tag content',
 443          *     array( 'param' => 'x' ),
 444          *     '<element param="x">tag content</element>' ) )
 445          *
 446          * @param $elements list of element names. Comments are always extracted.
 447          * @param $text Source text string.
 448          * @param $uniq_prefix
 449          *
 450          * @private
 451          * @static
 452          */
 453         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 454                 static $n = 1;
 455                 $stripped = '';
 456                 $matches = array();
 457
 458                 $taglist = implode( '|', $elements );
 459                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 460
 461                 while ( '' != $text ) {
 462                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 463                         $stripped .= $p[0];
 464                         if( count( $p ) < 5 ) {
 465                                 break;
 466                         }
 467                         if( count( $p ) > 5 ) {
 468                                 // comment
 469                                 $element    = $p[4];
 470                                 $attributes = '';
 471                                 $close      = '';
 472                                 $inside     = $p[5];
 473                         } else {
 474                                 // tag
 475                                 $element    = $p[1];
 476                                 $attributes = $p[2];
 477                                 $close      = $p[3];
 478                                 $inside     = $p[4];
 479                         }
 480
 481                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . '-QINU';
 482                         $stripped .= $marker;
 483
 484                         if ( $close === '/>' ) {
 485                                 // Empty element tag, <tag />
 486                                 $content = null;
 487                                 $text = $inside;
 488                                 $tail = null;
 489                         } else {
 490                                 if( $element == '!--' ) {
 491                                         $end = '/(-->)/';
 492                                 } else {
 493                                         $end = "/(<\\/$element\\s*>)/i";
 494                                 }
 495                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 496                                 $content = $q[0];
 497                                 if( count( $q ) < 3 ) {
 498                                         # No end tag -- let it run out to the end of the text.
 499                                         $tail = '';
 500                                         $text = '';
 501                                 } else {
 502                                         $tail = $q[1];
 503                                         $text = $q[2];
 504                                 }
 505                         }
 506
 507                         $matches[$marker] = array( $element,
 508                                 $content,
 509                                 Sanitizer::decodeTagAttributes( $attributes ),
 510                                 "<$element$attributes$close$content$tail" );
 511                 }
 512                 return $stripped;
 513         }
 514
 515         /**
 516          * Strips and renders nowiki, pre, math, hiero
 517          * If $render is set, performs necessary rendering operations on plugins
 518          * Returns the text, and fills an array with data needed in unstrip()
 519          *
 520          * @param StripState $state
 521          *
 522          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 523          *  will be stripped in addition to other tags. This is important
 524          *  for section editing, where these comments cause confusion when
 525          *  counting the sections in the wikisource
 526          *
 527          * @param array dontstrip contains tags which should not be stripped;
 528          *  used to prevent stipping of <gallery> when saving (fixes bug 2700)
 529          *
 530          * @private
 531          */
 532         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 533                 global $wgContLang;
 534                 wfProfileIn( __METHOD__ );
 535                 $render = ($this->mOutputType == OT_HTML);
 536
 537                 $uniq_prefix = $this->mUniqPrefix;
 538                 $commentState = new ReplacementArray;
 539                 $nowikiItems = array();
 540                 $generalItems = array();
 541
 542                 $elements = array_merge(
 543                         array( 'nowiki', 'gallery' ),
 544                         array_keys( $this->mTagHooks ) );
 545                 global $wgRawHtml;
 546                 if( $wgRawHtml ) {
 547                         $elements[] = 'html';
 548                 }
 549                 if( $this->mOptions->getUseTeX() ) {
 550                         $elements[] = 'math';
 551                 }
 552
 553                 # Removing $dontstrip tags from $elements list (currently only 'gallery', fixing bug 2700)
 554                 foreach ( $elements AS $k => $v ) {
 555                         if ( !in_array ( $v , $dontstrip ) ) continue;
 556                         unset ( $elements[$k] );
 557                 }
 558
 559                 $matches = array();
 560                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 561
 562                 foreach( $matches as $marker => $data ) {
 563                         list( $element, $content, $params, $tag ) = $data;
 564                         if( $render ) {
 565                                 $tagName = strtolower( $element );
 566                                 wfProfileIn( __METHOD__."-render-$tagName" );
 567                                 switch( $tagName ) {
 568                                 case '!--':
 569                                         // Comment
 570                                         if( substr( $tag, -3 ) == '-->' ) {
 571                                                 $output = $tag;
 572                                         } else {
 573                                                 // Unclosed comment in input.
 574                                                 // Close it so later stripping can remove it
 575                                                 $output = "$tag-->";
 576                                         }
 577                                         break;
 578                                 case 'html':
 579                                         if( $wgRawHtml ) {
 580                                                 $output = $content;
 581                                                 break;
 582                                         }
 583                                         // Shouldn't happen otherwise. :)
 584                                 case 'nowiki':
 585                                         $output = Xml::escapeTagsOnly( $content );
 586                                         break;
 587                                 case 'math':
 588                                         $output = $wgContLang->armourMath( MathRenderer::renderMath( $content ) );
 589                                         break;
 590                                 case 'gallery':
 591                                         $output = $this->renderImageGallery( $content, $params );
 592                                         break;
 593                                 default:
 594                                         if( isset( $this->mTagHooks[$tagName] ) ) {
 595                                                 $output = call_user_func_array( $this->mTagHooks[$tagName],
 596                                                         array( $content, $params, $this ) );
 597                                         } else {
 598                                                 throw new MWException( "Invalid call hook $element" );
 599                                         }
 600                                 }
 601                                 wfProfileOut( __METHOD__."-render-$tagName" );
 602                         } else {
 603                                 // Just stripping tags; keep the source
 604                                 $output = $tag;
 605                         }
 606
 607                         // Unstrip the output, to support recursive strip() calls
 608                         $output = $state->unstripBoth( $output );
 609
 610                         if( !$stripcomments && $element == '!--' ) {
 611                                 $commentState->setPair( $marker, $output );
 612                         } elseif ( $element == 'html' || $element == 'nowiki' ) {
 613                                 $nowikiItems[$marker] = $output;
 614                         } else {
 615                                 $generalItems[$marker] = $output;
 616                         }
 617                 }
 618                 # Add the new items to the state
 619                 # We do this after the loop instead of during it to avoid slowing
 620                 # down the recursive unstrip
 621                 $state->nowiki->mergeArray( $nowikiItems );
 622                 $state->general->mergeArray( $generalItems );
 623
 624                 # Unstrip comments unless explicitly told otherwise.
 625                 # (The comments are always stripped prior to this point, so as to
 626                 # not invoke any extension tags / parser hooks contained within
 627                 # a comment.)
 628                 if ( !$stripcomments ) {
 629                         // Put them all back and forget them
 630                         $text = $commentState->replace( $text );
 631                 }
 632
 633                 wfProfileOut( __METHOD__ );
 634                 return $text;
 635         }
 636
 637         /**
 638          * Restores pre, math, and other extensions removed by strip()
 639          *
 640          * always call unstripNoWiki() after this one
 641          * @private
 642          * @deprecated use $this->mStripState->unstrip()
 643          */
 644         function unstrip( $text, $state ) {
 645                 return $state->unstripGeneral( $text );
 646         }
 647
 648         /**
 649          * Always call this after unstrip() to preserve the order
 650          *
 651          * @private
 652          * @deprecated use $this->mStripState->unstrip()
 653          */
 654         function unstripNoWiki( $text, $state ) {
 655                 return $state->unstripNoWiki( $text );
 656         }
 657
 658         /**
 659          * @deprecated use $this->mStripState->unstripBoth()
 660          */
 661         function unstripForHTML( $text ) {
 662                 return $this->mStripState->unstripBoth( $text );
 663         }
 664
 665         /**
 666          * Add an item to the strip state
 667          * Returns the unique tag which must be inserted into the stripped text
 668          * The tag will be replaced with the original text in unstrip()
 669          *
 670          * @private
 671          */
 672         function insertStripItem( $text, &$state ) {
 673                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 674                 $state->general->setPair( $rnd, $text );
 675                 return $rnd;
 676         }
 677
 678         /**
 679          * Interface with html tidy, used if $wgUseTidy = true.
 680          * If tidy isn't able to correct the markup, the original will be
 681          * returned in all its glory with a warning comment appended.
 682          *
 683          * Either the external tidy program or the in-process tidy extension
 684          * will be used depending on availability. Override the default
 685          * $wgTidyInternal setting to disable the internal if it's not working.
 686          *
 687          * @param string $text Hideous HTML input
 688          * @return string Corrected HTML output
 689          * @public
 690          * @static
 691          */
 692         function tidy( $text ) {
 693                 global $wgTidyInternal;
 694                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 695 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 696 '<head><title>test</title></head><body>'.$text.'</body></html>';
 697                 if( $wgTidyInternal ) {
 698                         $correctedtext = Parser::internalTidy( $wrappedtext );
 699                 } else {
 700                         $correctedtext = Parser::externalTidy( $wrappedtext );
 701                 }
 702                 if( is_null( $correctedtext ) ) {
 703                         wfDebug( "Tidy error detected!\n" );
 704                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 705                 }
 706                 return $correctedtext;
 707         }
 708
 709         /**
 710          * Spawn an external HTML tidy process and get corrected markup back from it.
 711          *
 712          * @private
 713          * @static
 714          */
 715         function externalTidy( $text ) {
 716                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 717                 $fname = 'Parser::externalTidy';
 718                 wfProfileIn( $fname );
 719
 720                 $cleansource = '';
 721                 $opts = ' -utf8';
 722
 723                 $descriptorspec = array(
 724                         0 => array('pipe', 'r'),
 725                         1 => array('pipe', 'w'),
 726                         2 => array('file', '/dev/null', 'a')
 727                 );
 728                 $pipes = array();
 729                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 730                 if (is_resource($process)) {
 731                         // Theoretically, this style of communication could cause a deadlock
 732                         // here. If the stdout buffer fills up, then writes to stdin could
 733                         // block. This doesn't appear to happen with tidy, because tidy only
 734                         // writes to stdout after it's finished reading from stdin. Search
 735                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 736                         fwrite($pipes[0], $text);
 737                         fclose($pipes[0]);
 738                         while (!feof($pipes[1])) {
 739                                 $cleansource .= fgets($pipes[1], 1024);
 740                         }
 741                         fclose($pipes[1]);
 742                         proc_close($process);
 743                 }
 744
 745                 wfProfileOut( $fname );
 746
 747                 if( $cleansource == '' && $text != '') {
 748                         // Some kind of error happened, so we couldn't get the corrected text.
 749                         // Just give up; we'll use the source text and append a warning.
 750                         return null;
 751                 } else {
 752                         return $cleansource;
 753                 }
 754         }
 755
 756         /**
 757          * Use the HTML tidy PECL extension to use the tidy library in-process,
 758          * saving the overhead of spawning a new process. Currently written to
 759          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 760          *
 761          * 'pear install tidy' should be able to compile the extension module.
 762          *
 763          * @private
 764          * @static
 765          */
 766         function internalTidy( $text ) {
 767                 global $wgTidyConf;
 768                 $fname = 'Parser::internalTidy';
 769                 wfProfileIn( $fname );
 770
 771                 tidy_load_config( $wgTidyConf );
 772                 tidy_set_encoding( 'utf8' );
 773                 tidy_parse_string( $text );
 774                 tidy_clean_repair();
 775                 if( tidy_get_status() == 2 ) {
 776                         // 2 is magic number for fatal error
 777                         // http://www.php.net/manual/en/function.tidy-get-status.php
 778                         $cleansource = null;
 779                 } else {
 780                         $cleansource = tidy_get_output();
 781                 }
 782                 wfProfileOut( $fname );
 783                 return $cleansource;
 784         }
 785
 786         /**
 787          * parse the wiki syntax used to render tables
 788          *
 789          * @private
 790          */
 791         function doTableStuff ( $text ) {
 792                 $fname = 'Parser::doTableStuff';
 793                 wfProfileIn( $fname );
 794
 795                 $lines = explode ( "\n" , $text );
 796                 $td_history = array (); // Is currently a td tag open?
 797                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 798                 $tr_history = array (); // Is currently a tr tag open?
 799                 $tr_attributes = array (); // history of tr attributes
 800                 $has_opened_tr = array(); // Did this table open a <tr> element?
 801                 $indent_level = 0; // indent level of the table
 802                 foreach ( $lines as $key => $line )
 803                 {
 804                         $line = trim ( $line );
 805
 806                         if( $line == '' ) { // empty line, go to next line
 807                                 continue;
 808                         }
 809                         $first_character = $line{0};
 810                         $matches = array();
 811
 812                         if ( preg_match( '/^(:*)\{\|(.*)$/' , $line , $matches ) ) {
 813                                 // First check if we are starting a new table
 814                                 $indent_level = strlen( $matches[1] );
 815
 816                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 817                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 818
 819                                 $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 820                                 array_push ( $td_history , false );
 821                                 array_push ( $last_tag_history , '' );
 822                                 array_push ( $tr_history , false );
 823                                 array_push ( $tr_attributes , '' );
 824                                 array_push ( $has_opened_tr , false );
 825                         } else if ( count ( $td_history ) == 0 ) {
 826                                 // Don't do any of the following
 827                                 continue;
 828                         } else if ( substr ( $line , 0 , 2 ) == '|}' ) {
 829                                 // We are ending a table
 830                                 $line = '</table>' . substr ( $line , 2 );
 831                                 $last_tag = array_pop ( $last_tag_history );
 832
 833                                 if ( !array_pop ( $has_opened_tr ) ) {
 834                                         $line = "<tr><td></td></tr>{$line}";
 835                                 }
 836
 837                                 if ( array_pop ( $tr_history ) ) {
 838                                         $line = "</tr>{$line}";
 839                                 }
 840
 841                                 if ( array_pop ( $td_history ) ) {
 842                                         $line = "</{$last_tag}>{$line}";
 843                                 }
 844                                 array_pop ( $tr_attributes );
 845                                 $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );
 846                         } else if ( substr ( $line , 0 , 2 ) == '|-' ) {
 847                                 // Now we have a table row
 848                                 $line = preg_replace( '#^\|-+#', '', $line );
 849
 850                                 // Whats after the tag is now only attributes
 851                                 $attributes = $this->mStripState->unstripBoth( $line );
 852                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 853                                 array_pop ( $tr_attributes );
 854                                 array_push ( $tr_attributes , $attributes );
 855
 856                                 $line = '';
 857                                 $last_tag = array_pop ( $last_tag_history );
 858                                 array_pop ( $has_opened_tr );
 859                                 array_push ( $has_opened_tr , true );
 860
 861                                 if ( array_pop ( $tr_history ) ) {
 862                                         $line = '</tr>';
 863                                 }
 864
 865                                 if ( array_pop ( $td_history ) ) {
 866                                         $line = "</{$last_tag}>{$line}";
 867                                 }
 868
 869                                 $lines[$key] = $line;
 870                                 array_push ( $tr_history , false );
 871                                 array_push ( $td_history , false );
 872                                 array_push ( $last_tag_history , '' );
 873                         }
 874                         else if ( $first_character == '|' || $first_character == '!' || substr ( $line , 0 , 2 )  == '|+' ) {
 875                                 // This might be cell elements, td, th or captions
 876                                 if ( substr ( $line , 0 , 2 ) == '|+' ) {
 877                                         $first_character = '+';
 878                                         $line = substr ( $line , 1 );
 879                                 }
 880
 881                                 $line = substr ( $line , 1 );
 882
 883                                 if ( $first_character == '!' ) {
 884                                         $line = str_replace ( '!!' , '||' , $line );
 885                                 }
 886
 887                                 // Split up multiple cells on the same line.
 888                                 // FIXME : This can result in improper nesting of tags processed
 889                                 // by earlier parser steps, but should avoid splitting up eg
 890                                 // attribute values containing literal "||".
 891                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 892
 893                                 $lines[$key] = '';
 894
 895                                 // Loop through each table cell
 896                                 foreach ( $cells as $cell )
 897                                 {
 898                                         $previous = '';
 899                                         if ( $first_character != '+' )
 900                                         {
 901                                                 $tr_after = array_pop ( $tr_attributes );
 902                                                 if ( !array_pop ( $tr_history ) ) {
 903                                                         $previous = "<tr{$tr_after}>\n";
 904                                                 }
 905                                                 array_push ( $tr_history , true );
 906                                                 array_push ( $tr_attributes , '' );
 907                                                 array_pop ( $has_opened_tr );
 908                                                 array_push ( $has_opened_tr , true );
 909                                         }
 910
 911                                         $last_tag = array_pop ( $last_tag_history );
 912
 913                                         if ( array_pop ( $td_history ) ) {
 914                                                 $previous = "</{$last_tag}>{$previous}";
 915                                         }
 916
 917                                         if ( $first_character == '|' ) {
 918                                                 $last_tag = 'td';
 919                                         } else if ( $first_character == '!' ) {
 920                                                 $last_tag = 'th';
 921                                         } else if ( $first_character == '+' ) {
 922                                                 $last_tag = 'caption';
 923                                         } else {
 924                                                 $last_tag = '';
 925                                         }
 926
 927                                         array_push ( $last_tag_history , $last_tag );
 928
 929                                         // A cell could contain both parameters and data
 930                                         $cell_data = explode ( '|' , $cell , 2 );
 931
 932                                         // Bug 553: Note that a '|' inside an invalid link should not
 933                                         // be mistaken as delimiting cell parameters
 934                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 935                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 936                                         } else if ( count ( $cell_data ) == 1 )
 937                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 938                                         else {
 939                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 940                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 941                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 942                                         }
 943
 944                                         $lines[$key] .= $cell;
 945                                         array_push ( $td_history , true );
 946                                 }
 947                         }
 948                 }
 949
 950                 // Closing open td, tr && table
 951                 while ( count ( $td_history ) > 0 )
 952                 {
 953                         if ( array_pop ( $td_history ) ) {
 954                                 $lines[] = '</td>' ;
 955                         }
 956                         if ( array_pop ( $tr_history ) ) {
 957                                 $lines[] = '</tr>' ;
 958                         }
 959                         if ( !array_pop ( $has_opened_tr ) ) {
 960                                 $lines[] = "<tr><td></td></tr>" ;
 961                         }
 962
 963                         $lines[] = '</table>' ;
 964                 }
 965
 966                 $output = implode ( "\n" , $lines ) ;
 967
 968                 // special case: don't return empty table
 969                 if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {
 970                         $output = '';
 971                 }
 972
 973                 wfProfileOut( $fname );
 974
 975                 return $output;
 976         }
 977
 978         /**
 979          * Helper function for parse() that transforms wiki markup into
 980          * HTML. Only called for $mOutputType == OT_HTML.
 981          *
 982          * @private
 983          */
 984         function internalParse( $text ) {
 985                 $args = array();
 986                 $isMain = true;
 987                 $fname = 'Parser::internalParse';
 988                 wfProfileIn( $fname );
 989
 990                 # Hook to suspend the parser in this state
 991                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 992                         wfProfileOut( $fname );
 993                         return $text ;
 994                 }
 995
 996                 # Remove <noinclude> tags and <includeonly> sections
 997                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 998                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 999                 $text = StringUtils::delimiterReplace( '<includeonly>', '</includeonly>', '', $text );
1000
1001                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
1002
1003                 $text = $this->replaceVariables( $text, $args );
1004
1005                 // Tables need to come after variable replacement for things to work
1006                 // properly; putting them before other transformations should keep
1007                 // exciting things like link expansions from showing up in surprising
1008                 // places.
1009                 $text = $this->doTableStuff( $text );
1010
1011                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1012
1013                 $text = $this->stripToc( $text );
1014                 $this->stripNoGallery( $text );
1015                 $text = $this->doHeadings( $text );
1016                 if($this->mOptions->getUseDynamicDates()) {
1017                         $df =& DateFormatter::getInstance();
1018                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1019                 }
1020                 $text = $this->doAllQuotes( $text );
1021                 $text = $this->replaceInternalLinks( $text );
1022                 $text = $this->replaceExternalLinks( $text );
1023
1024                 # replaceInternalLinks may sometimes leave behind
1025                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1026                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
1027
1028                 $text = $this->doMagicLinks( $text );
1029                 $text = $this->formatHeadings( $text, $isMain );
1030
1031                 wfProfileOut( $fname );
1032                 return $text;
1033         }
1034
1035         /**
1036          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1037          * magic external links.
1038          *
1039          * @private
1040          */
1041         function &doMagicLinks( &$text ) {
1042                 wfProfileIn( __METHOD__ );
1043                 $text = preg_replace_callback(
1044                         '!(?:                           # Start cases
1045                             <a.*?</a> |                 # Skip link text
1046                             <.*?> |                     # Skip stuff inside HTML elements
1047                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
1048                             ISBN\s+(\b                  # ISBN, capture number as m[2]
1049                                      (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1050                                      (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1051                                      [0-9Xx]                 # check digit
1052                                    \b)
1053                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1054                 wfProfileOut( __METHOD__ );
1055                 return $text;
1056         }
1057
1058         function magicLinkCallback( $m ) {
1059                 if ( substr( $m[0], 0, 1 ) == '<' ) {
1060                         # Skip HTML element
1061                         return $m[0];
1062                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
1063                         $isbn = $m[2];
1064                         $num = strtr( $isbn, array(
1065                                 '-' => '',
1066                                 ' ' => '',
1067                                 'x' => 'X',
1068                         ));
1069                         $titleObj = SpecialPage::getTitleFor( 'Booksources' );
1070                         $text = '<a href="' .
1071                                 $titleObj->escapeLocalUrl( "isbn=$num" ) .
1072                                 "\" class=\"internal\">ISBN $isbn</a>";
1073                 } else {
1074                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
1075                                 $keyword = 'RFC';
1076                                 $urlmsg = 'rfcurl';
1077                                 $id = $m[1];
1078                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
1079                                 $keyword = 'PMID';
1080                                 $urlmsg = 'pubmedurl';
1081                                 $id = $m[1];
1082                         } else {
1083                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1084                                         substr($m[0], 0, 20 ) . '"' );
1085                         }
1086
1087                         $url = wfMsg( $urlmsg, $id);
1088                         $sk =& $this->mOptions->getSkin();
1089                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1090                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1091                 }
1092                 return $text;
1093         }
1094
1095         /**
1096          * Parse headers and return html
1097          *
1098          * @private
1099          */
1100         function doHeadings( $text ) {
1101                 $fname = 'Parser::doHeadings';
1102                 wfProfileIn( $fname );
1103                 for ( $i = 6; $i >= 1; --$i ) {
1104                         $h = str_repeat( '=', $i );
1105                         $text = preg_replace( "/^{$h}(.+){$h}\\s*$/m",
1106                           "<h{$i}>\\1</h{$i}>\\2", $text );
1107                 }
1108                 wfProfileOut( $fname );
1109                 return $text;
1110         }
1111
1112         /**
1113          * Replace single quotes with HTML markup
1114          * @private
1115          * @return string the altered text
1116          */
1117         function doAllQuotes( $text ) {
1118                 $fname = 'Parser::doAllQuotes';
1119                 wfProfileIn( $fname );
1120                 $outtext = '';
1121                 $lines = explode( "\n", $text );
1122                 foreach ( $lines as $line ) {
1123                         $outtext .= $this->doQuotes ( $line ) . "\n";
1124                 }
1125                 $outtext = substr($outtext, 0,-1);
1126                 wfProfileOut( $fname );
1127                 return $outtext;
1128         }
1129
1130         /**
1131          * Helper function for doAllQuotes()
1132          * @private
1133          */
1134         function doQuotes( $text ) {
1135                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1136                 if ( count( $arr ) == 1 )
1137                         return $text;
1138                 else
1139                 {
1140                         # First, do some preliminary work. This may shift some apostrophes from
1141                         # being mark-up to being text. It also counts the number of occurrences
1142                         # of bold and italics mark-ups.
1143                         $i = 0;
1144                         $numbold = 0;
1145                         $numitalics = 0;
1146                         foreach ( $arr as $r )
1147                         {
1148                                 if ( ( $i % 2 ) == 1 )
1149                                 {
1150                                         # If there are ever four apostrophes, assume the first is supposed to
1151                                         # be text, and the remaining three constitute mark-up for bold text.
1152                                         if ( strlen( $arr[$i] ) == 4 )
1153                                         {
1154                                                 $arr[$i-1] .= "'";
1155                                                 $arr[$i] = "'''";
1156                                         }
1157                                         # If there are more than 5 apostrophes in a row, assume they're all
1158                                         # text except for the last 5.
1159                                         else if ( strlen( $arr[$i] ) > 5 )
1160                                         {
1161                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1162                                                 $arr[$i] = "'''''";
1163                                         }
1164                                         # Count the number of occurrences of bold and italics mark-ups.
1165                                         # We are not counting sequences of five apostrophes.
1166                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1167                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1168                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1169                                 }
1170                                 $i++;
1171                         }
1172
1173                         # If there is an odd number of both bold and italics, it is likely
1174                         # that one of the bold ones was meant to be an apostrophe followed
1175                         # by italics. Which one we cannot know for certain, but it is more
1176                         # likely to be one that has a single-letter word before it.
1177                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1178                         {
1179                                 $i = 0;
1180                                 $firstsingleletterword = -1;
1181                                 $firstmultiletterword = -1;
1182                                 $firstspace = -1;
1183                                 foreach ( $arr as $r )
1184                                 {
1185                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1186                                         {
1187                                                 $x1 = substr ($arr[$i-1], -1);
1188                                                 $x2 = substr ($arr[$i-1], -2, 1);
1189                                                 if ($x1 == ' ') {
1190                                                         if ($firstspace == -1) $firstspace = $i;
1191                                                 } else if ($x2 == ' ') {
1192                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1193                                                 } else {
1194                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1195                                                 }
1196                                         }
1197                                         $i++;
1198                                 }
1199
1200                                 # If there is a single-letter word, use it!
1201                                 if ($firstsingleletterword > -1)
1202                                 {
1203                                         $arr [ $firstsingleletterword ] = "''";
1204                                         $arr [ $firstsingleletterword-1 ] .= "'";
1205                                 }
1206                                 # If not, but there's a multi-letter word, use that one.
1207                                 else if ($firstmultiletterword > -1)
1208                                 {
1209                                         $arr [ $firstmultiletterword ] = "''";
1210                                         $arr [ $firstmultiletterword-1 ] .= "'";
1211                                 }
1212                                 # ... otherwise use the first one that has neither.
1213                                 # (notice that it is possible for all three to be -1 if, for example,
1214                                 # there is only one pentuple-apostrophe in the line)
1215                                 else if ($firstspace > -1)
1216                                 {
1217                                         $arr [ $firstspace ] = "''";
1218                                         $arr [ $firstspace-1 ] .= "'";
1219                                 }
1220                         }
1221
1222                         # Now let's actually convert our apostrophic mush to HTML!
1223                         $output = '';
1224                         $buffer = '';
1225                         $state = '';
1226                         $i = 0;
1227                         foreach ($arr as $r)
1228                         {
1229                                 if (($i % 2) == 0)
1230                                 {
1231                                         if ($state == 'both')
1232                                                 $buffer .= $r;
1233                                         else
1234                                                 $output .= $r;
1235                                 }
1236                                 else
1237                                 {
1238                                         if (strlen ($r) == 2)
1239                                         {
1240                                                 if ($state == 'i')
1241                                                 { $output .= '</i>'; $state = ''; }
1242                                                 else if ($state == 'bi')
1243                                                 { $output .= '</i>'; $state = 'b'; }
1244                                                 else if ($state == 'ib')
1245                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1246                                                 else if ($state == 'both')
1247                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1248                                                 else # $state can be 'b' or ''
1249                                                 { $output .= '<i>'; $state .= 'i'; }
1250                                         }
1251                                         else if (strlen ($r) == 3)
1252                                         {
1253                                                 if ($state == 'b')
1254                                                 { $output .= '</b>'; $state = ''; }
1255                                                 else if ($state == 'bi')
1256                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1257                                                 else if ($state == 'ib')
1258                                                 { $output .= '</b>'; $state = 'i'; }
1259                                                 else if ($state == 'both')
1260                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1261                                                 else # $state can be 'i' or ''
1262                                                 { $output .= '<b>'; $state .= 'b'; }
1263                                         }
1264                                         else if (strlen ($r) == 5)
1265                                         {
1266                                                 if ($state == 'b')
1267                                                 { $output .= '</b><i>'; $state = 'i'; }
1268                                                 else if ($state == 'i')
1269                                                 { $output .= '</i><b>'; $state = 'b'; }
1270                                                 else if ($state == 'bi')
1271                                                 { $output .= '</i></b>'; $state = ''; }
1272                                                 else if ($state == 'ib')
1273                                                 { $output .= '</b></i>'; $state = ''; }
1274                                                 else if ($state == 'both')
1275                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1276                                                 else # ($state == '')
1277                                                 { $buffer = ''; $state = 'both'; }
1278                                         }
1279                                 }
1280                                 $i++;
1281                         }
1282                         # Now close all remaining tags.  Notice that the order is important.
1283                         if ($state == 'b' || $state == 'ib')
1284                                 $output .= '</b>';
1285                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1286                                 $output .= '</i>';
1287                         if ($state == 'bi')
1288                                 $output .= '</b>';
1289                         if ($state == 'both')
1290                                 $output .= '<b><i>'.$buffer.'</i></b>';
1291                         return $output;
1292                 }
1293         }
1294
1295         /**
1296          * Replace external links
1297          *
1298          * Note: this is all very hackish and the order of execution matters a lot.
1299          * Make sure to run maintenance/parserTests.php if you change this code.
1300          *
1301          * @private
1302          */
1303         function replaceExternalLinks( $text ) {
1304                 global $wgContLang;
1305                 $fname = 'Parser::replaceExternalLinks';
1306                 wfProfileIn( $fname );
1307
1308                 $sk =& $this->mOptions->getSkin();
1309
1310                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1311
1312                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1313
1314                 $i = 0;
1315                 while ( $i<count( $bits ) ) {
1316                         $url = $bits[$i++];
1317                         $protocol = $bits[$i++];
1318                         $text = $bits[$i++];
1319                         $trail = $bits[$i++];
1320
1321                         # The characters '<' and '>' (which were escaped by
1322                         # removeHTMLtags()) should not be included in
1323                         # URLs, per RFC 2396.
1324                         $m2 = array();
1325                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1326                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1327                                 $url = substr($url, 0, $m2[0][1]);
1328                         }
1329
1330                         # If the link text is an image URL, replace it with an <img> tag
1331                         # This happened by accident in the original parser, but some people used it extensively
1332                         $img = $this->maybeMakeExternalImage( $text );
1333                         if ( $img !== false ) {
1334                                 $text = $img;
1335                         }
1336
1337                         $dtrail = '';
1338
1339                         # Set linktype for CSS - if URL==text, link is essentially free
1340                         $linktype = ($text == $url) ? 'free' : 'text';
1341
1342                         # No link text, e.g. [http://domain.tld/some.link]
1343                         if ( $text == '' ) {
1344                                 # Autonumber if allowed. See bug #5918
1345                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1346                                         $text = '[' . ++$this->mAutonumber . ']';
1347                                         $linktype = 'autonumber';
1348                                 } else {
1349                                         # Otherwise just use the URL
1350                                         $text = htmlspecialchars( $url );
1351                                         $linktype = 'free';
1352                                 }
1353                         } else {
1354                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1355                                 # Check for trail
1356                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1357                         }
1358
1359                         $text = $wgContLang->markNoConversion($text);
1360
1361                         $url = Sanitizer::cleanUrl( $url );
1362
1363                         # Process the trail (i.e. everything after this link up until start of the next link),
1364                         # replacing any non-bracketed links
1365                         $trail = $this->replaceFreeExternalLinks( $trail );
1366
1367                         # Use the encoded URL
1368                         # This means that users can paste URLs directly into the text
1369                         # Funny characters like &ouml; aren't valid in URLs anyway
1370                         # This was changed in August 2004
1371                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1372
1373                         # Register link in the output object.
1374                         # Replace unnecessary URL escape codes with the referenced character
1375                         # This prevents spammers from hiding links from the filters
1376                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1377                         $this->mOutput->addExternalLink( $pasteurized );
1378                 }
1379
1380                 wfProfileOut( $fname );
1381                 return $s;
1382         }
1383
1384         /**
1385          * Replace anything that looks like a URL with a link
1386          * @private
1387          */
1388         function replaceFreeExternalLinks( $text ) {
1389                 global $wgContLang;
1390                 $fname = 'Parser::replaceFreeExternalLinks';
1391                 wfProfileIn( $fname );
1392
1393                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1394                 $s = array_shift( $bits );
1395                 $i = 0;
1396
1397                 $sk =& $this->mOptions->getSkin();
1398
1399                 while ( $i < count( $bits ) ){
1400                         $protocol = $bits[$i++];
1401                         $remainder = $bits[$i++];
1402
1403                         $m = array();
1404                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1405                                 # Found some characters after the protocol that look promising
1406                                 $url = $protocol . $m[1];
1407                                 $trail = $m[2];
1408
1409                                 # special case: handle urls as url args:
1410                                 # http://www.example.com/foo?=http://www.example.com/bar
1411                                 if(strlen($trail) == 0 &&
1412                                         isset($bits[$i]) &&
1413                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1414                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1415                                 {
1416                                         # add protocol, arg
1417                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1418                                         $i += 2;
1419                                         $trail = $m[2];
1420                                 }
1421
1422                                 # The characters '<' and '>' (which were escaped by
1423                                 # removeHTMLtags()) should not be included in
1424                                 # URLs, per RFC 2396.
1425                                 $m2 = array();
1426                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1427                                         $trail = substr($url, $m2[0][1]) . $trail;
1428                                         $url = substr($url, 0, $m2[0][1]);
1429                                 }
1430
1431                                 # Move trailing punctuation to $trail
1432                                 $sep = ',;\.:!?';
1433                                 # If there is no left bracket, then consider right brackets fair game too
1434                                 if ( strpos( $url, '(' ) === false ) {
1435                                         $sep .= ')';
1436                                 }
1437
1438                                 $numSepChars = strspn( strrev( $url ), $sep );
1439                                 if ( $numSepChars ) {
1440                                         $trail = substr( $url, -$numSepChars ) . $trail;
1441                                         $url = substr( $url, 0, -$numSepChars );
1442                                 }
1443
1444                                 $url = Sanitizer::cleanUrl( $url );
1445
1446                                 # Is this an external image?
1447                                 $text = $this->maybeMakeExternalImage( $url );
1448                                 if ( $text === false ) {
1449                                         # Not an image, make a link
1450                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1451                                         # Register it in the output object...
1452                                         # Replace unnecessary URL escape codes with their equivalent characters
1453                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1454                                         $this->mOutput->addExternalLink( $pasteurized );
1455                                 }
1456                                 $s .= $text . $trail;
1457                         } else {
1458                                 $s .= $protocol . $remainder;
1459                         }
1460                 }
1461                 wfProfileOut( $fname );
1462                 return $s;
1463         }
1464
1465         /**
1466          * Replace unusual URL escape codes with their equivalent characters
1467          * @param string
1468          * @return string
1469          * @static
1470          * @fixme This can merge genuinely required bits in the path or query string,
1471          *        breaking legit URLs. A proper fix would treat the various parts of
1472          *        the URL differently; as a workaround, just use the output for
1473          *        statistical records, not for actual linking/output.
1474          */
1475         static function replaceUnusualEscapes( $url ) {
1476                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1477                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1478         }
1479
1480         /**
1481          * Callback function used in replaceUnusualEscapes().
1482          * Replaces unusual URL escape codes with their equivalent character
1483          * @static
1484          * @private
1485          */
1486         private static function replaceUnusualEscapesCallback( $matches ) {
1487                 $char = urldecode( $matches[0] );
1488                 $ord = ord( $char );
1489                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1490                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1491                         // No, shouldn't be escaped
1492                         return $char;
1493                 } else {
1494                         // Yes, leave it escaped
1495                         return $matches[0];
1496                 }
1497         }
1498
1499         /**
1500          * make an image if it's allowed, either through the global
1501          * option or through the exception
1502          * @private
1503          */
1504         function maybeMakeExternalImage( $url ) {
1505                 $sk =& $this->mOptions->getSkin();
1506                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1507                 $imagesexception = !empty($imagesfrom);
1508                 $text = false;
1509                 if ( $this->mOptions->getAllowExternalImages()
1510                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1511                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1512                                 # Image found
1513                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1514                         }
1515                 }
1516                 return $text;
1517         }
1518
1519         /**
1520          * Process [[ ]] wikilinks
1521          *
1522          * @private
1523          */
1524         function replaceInternalLinks( $s ) {
1525                 global $wgContLang;
1526                 static $fname = 'Parser::replaceInternalLinks' ;
1527
1528                 wfProfileIn( $fname );
1529
1530                 wfProfileIn( $fname.'-setup' );
1531                 static $tc = FALSE;
1532                 # the % is needed to support urlencoded titles as well
1533                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1534
1535                 $sk =& $this->mOptions->getSkin();
1536
1537                 #split the entire text string on occurences of [[
1538                 $a = explode( '[[', ' ' . $s );
1539                 #get the first element (all text up to first [[), and remove the space we added
1540                 $s = array_shift( $a );
1541                 $s = substr( $s, 1 );
1542
1543                 # Match a link having the form [[namespace:link|alternate]]trail
1544                 static $e1 = FALSE;
1545                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1546                 # Match cases where there is no "]]", which might still be images
1547                 static $e1_img = FALSE;
1548                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1549                 # Match the end of a line for a word that's not followed by whitespace,
1550                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1551                 $e2 = wfMsgForContent( 'linkprefix' );
1552
1553                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1554                 if( is_null( $this->mTitle ) ) {
1555                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1556                 }
1557                 $nottalk = !$this->mTitle->isTalkPage();
1558
1559                 if ( $useLinkPrefixExtension ) {
1560                         $m = array();
1561                         if ( preg_match( $e2, $s, $m ) ) {
1562                                 $first_prefix = $m[2];
1563                         } else {
1564                                 $first_prefix = false;
1565                         }
1566                 } else {
1567                         $prefix = '';
1568                 }
1569
1570                 if($wgContLang->hasVariants()) {
1571                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1572                 } else {
1573                         $selflink = array($this->mTitle->getPrefixedText());
1574                 }
1575                 $useSubpages = $this->areSubpagesAllowed();
1576                 wfProfileOut( $fname.'-setup' );
1577
1578                 # Loop for each link
1579                 for ($k = 0; isset( $a[$k] ); $k++) {
1580                         $line = $a[$k];
1581                         if ( $useLinkPrefixExtension ) {
1582                                 wfProfileIn( $fname.'-prefixhandling' );
1583                                 if ( preg_match( $e2, $s, $m ) ) {
1584                                         $prefix = $m[2];
1585                                         $s = $m[1];
1586                                 } else {
1587                                         $prefix='';
1588                                 }
1589                                 # first link
1590                                 if($first_prefix) {
1591                                         $prefix = $first_prefix;
1592                                         $first_prefix = false;
1593                                 }
1594                                 wfProfileOut( $fname.'-prefixhandling' );
1595                         }
1596
1597                         $might_be_img = false;
1598
1599                         wfProfileIn( "$fname-e1" );
1600                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1601                                 $text = $m[2];
1602                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1603                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1604                                 # the real problem is with the $e1 regex
1605                                 # See bug 1300.
1606                                 #
1607                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1608                                 # and no image is in sight. See bug 2095.
1609                                 #
1610                                 if( $text !== '' &&
1611                                         substr( $m[3], 0, 1 ) === ']' &&
1612                                         strpos($text, '[') !== false
1613                                 )
1614                                 {
1615                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1616                                         $m[3] = substr( $m[3], 1 );
1617                                 }
1618                                 # fix up urlencoded title texts
1619                                 if( strpos( $m[1], '%' ) !== false ) {
1620                                         # Should anchors '#' also be rejected?
1621                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1622                                 }
1623                                 $trail = $m[3];
1624                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1625                                 $might_be_img = true;
1626                                 $text = $m[2];
1627                                 if ( strpos( $m[1], '%' ) !== false ) {
1628                                        $m[1] = urldecode($m[1]);
1629                                 }
1630                                 $trail = "";
1631                         } else { # Invalid form; output directly
1632                                 $s .= $prefix . '[[' . $line ;
1633                                 wfProfileOut( "$fname-e1" );
1634                                 continue;
1635                         }
1636                         wfProfileOut( "$fname-e1" );
1637                         wfProfileIn( "$fname-misc" );
1638
1639                         # Don't allow internal links to pages containing
1640                         # PROTO: where PROTO is a valid URL protocol; these
1641                         # should be external links.
1642                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1643                                 $s .= $prefix . '[[' . $line ;
1644                                 continue;
1645                         }
1646
1647                         # Make subpage if necessary
1648                         if( $useSubpages ) {
1649                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1650                         } else {
1651                                 $link = $m[1];
1652                         }
1653
1654                         $noforce = (substr($m[1], 0, 1) != ':');
1655                         if (!$noforce) {
1656                                 # Strip off leading ':'
1657                                 $link = substr($link, 1);
1658                         }
1659
1660                         wfProfileOut( "$fname-misc" );
1661                         wfProfileIn( "$fname-title" );
1662                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1663                         if( !$nt ) {
1664                                 $s .= $prefix . '[[' . $line;
1665                                 wfProfileOut( "$fname-title" );
1666                                 continue;
1667                         }
1668
1669                         $ns = $nt->getNamespace();
1670                         $iw = $nt->getInterWiki();
1671                         wfProfileOut( "$fname-title" );
1672
1673                         if ($might_be_img) { # if this is actually an invalid link
1674                                 wfProfileIn( "$fname-might_be_img" );
1675                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1676                                         $found = false;
1677                                         while (isset ($a[$k+1]) ) {
1678                                                 #look at the next 'line' to see if we can close it there
1679                                                 $spliced = array_splice( $a, $k + 1, 1 );
1680                                                 $next_line = array_shift( $spliced );
1681                                                 $m = explode( ']]', $next_line, 3 );
1682                                                 if ( count( $m ) == 3 ) {
1683                                                         # the first ]] closes the inner link, the second the image
1684                                                         $found = true;
1685                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1686                                                         $trail = $m[2];
1687                                                         break;
1688                                                 } elseif ( count( $m ) == 2 ) {
1689                                                         #if there's exactly one ]] that's fine, we'll keep looking
1690                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1691                                                 } else {
1692                                                         #if $next_line is invalid too, we need look no further
1693                                                         $text .= '[[' . $next_line;
1694                                                         break;
1695                                                 }
1696                                         }
1697                                         if ( !$found ) {
1698                                                 # we couldn't find the end of this imageLink, so output it raw
1699                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1700                                                 $text = $this->replaceInternalLinks($text);
1701                                                 $s .= "{$prefix}[[$link|$text";
1702                                                 # note: no $trail, because without an end, there *is* no trail
1703                                                 wfProfileOut( "$fname-might_be_img" );
1704                                                 continue;
1705                                         }
1706                                 } else { #it's not an image, so output it raw
1707                                         $s .= "{$prefix}[[$link|$text";
1708                                         # note: no $trail, because without an end, there *is* no trail
1709                                         wfProfileOut( "$fname-might_be_img" );
1710                                         continue;
1711                                 }
1712                                 wfProfileOut( "$fname-might_be_img" );
1713                         }
1714
1715                         $wasblank = ( '' == $text );
1716                         if( $wasblank ) $text = $link;
1717
1718                         # Link not escaped by : , create the various objects
1719                         if( $noforce ) {
1720
1721                                 # Interwikis
1722                                 wfProfileIn( "$fname-interwiki" );
1723                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1724                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1725                                         $s = rtrim($s . "\n");
1726                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1727                                         wfProfileOut( "$fname-interwiki" );
1728                                         continue;
1729                                 }
1730                                 wfProfileOut( "$fname-interwiki" );
1731
1732                                 if ( $ns == NS_IMAGE ) {
1733                                         wfProfileIn( "$fname-image" );
1734                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1735                                                 # recursively parse links inside the image caption
1736                                                 # actually, this will parse them in any other parameters, too,
1737                                                 # but it might be hard to fix that, and it doesn't matter ATM
1738                                                 $text = $this->replaceExternalLinks($text);
1739                                                 $text = $this->replaceInternalLinks($text);
1740
1741                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1742                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1743                                                 $this->mOutput->addImage( $nt->getDBkey() );
1744
1745                                                 wfProfileOut( "$fname-image" );
1746                                                 continue;
1747                                         } else {
1748                                                 # We still need to record the image's presence on the page
1749                                                 $this->mOutput->addImage( $nt->getDBkey() );
1750                                         }
1751                                         wfProfileOut( "$fname-image" );
1752
1753                                 }
1754
1755                                 if ( $ns == NS_CATEGORY ) {
1756                                         wfProfileIn( "$fname-category" );
1757                                         $s = rtrim($s . "\n"); # bug 87
1758
1759                                         if ( $wasblank ) {
1760                                                 $sortkey = $this->getDefaultSort();
1761                                         } else {
1762                                                 $sortkey = $text;
1763                                         }
1764                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1765                                         $sortkey = str_replace( "\n", '', $sortkey );
1766                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1767                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1768
1769                                         /**
1770                                          * Strip the whitespace Category links produce, see bug 87
1771                                          * @todo We might want to use trim($tmp, "\n") here.
1772                                          */
1773                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1774
1775                                         wfProfileOut( "$fname-category" );
1776                                         continue;
1777                                 }
1778                         }
1779
1780                         # Self-link checking
1781                         if( $nt->getFragment() === '' ) {
1782                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1783                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1784                                         continue;
1785                                 }
1786                         }
1787
1788                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1789                         if( $ns == NS_MEDIA ) {
1790                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1791                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1792                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1793                                 $this->mOutput->addImage( $nt->getDBkey() );
1794                                 continue;
1795                         } elseif( $ns == NS_SPECIAL ) {
1796                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1797                                 continue;
1798                         } elseif( $ns == NS_IMAGE ) {
1799                                 $img = new Image( $nt );
1800                                 if( $img->exists() ) {
1801                                         // Force a blue link if the file exists; may be a remote
1802                                         // upload on the shared repository, and we want to see its
1803                                         // auto-generated page.
1804                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1805                                         $this->mOutput->addLink( $nt );
1806                                         continue;
1807                                 }
1808                         }
1809                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1810                 }
1811                 wfProfileOut( $fname );
1812                 return $s;
1813         }
1814
1815         /**
1816          * Make a link placeholder. The text returned can be later resolved to a real link with
1817          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1818          * parsing of interwiki links, and secondly to allow all existence checks and
1819          * article length checks (for stub links) to be bundled into a single query.
1820          *
1821          */
1822         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1823                 wfProfileIn( __METHOD__ );
1824                 if ( ! is_object($nt) ) {
1825                         # Fail gracefully
1826                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1827                 } else {
1828                         # Separate the link trail from the rest of the link
1829                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1830
1831                         if ( $nt->isExternal() ) {
1832                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1833                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1834                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1835                         } else {
1836                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1837                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1838                                 $this->mLinkHolders['queries'][] = $query;
1839                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1840                                 $this->mLinkHolders['titles'][] = $nt;
1841
1842                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1843                         }
1844                 }
1845                 wfProfileOut( __METHOD__ );
1846                 return $retVal;
1847         }
1848
1849         /**
1850          * Render a forced-blue link inline; protect against double expansion of
1851          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1852          * Since this little disaster has to split off the trail text to avoid
1853          * breaking URLs in the following text without breaking trails on the
1854          * wiki links, it's been made into a horrible function.
1855          *
1856          * @param Title $nt
1857          * @param string $text
1858          * @param string $query
1859          * @param string $trail
1860          * @param string $prefix
1861          * @return string HTML-wikitext mix oh yuck
1862          */
1863         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1864                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1865                 $sk =& $this->mOptions->getSkin();
1866                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1867                 return $this->armorLinks( $link ) . $trail;
1868         }
1869
1870         /**
1871          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1872          * going to go through further parsing steps before inline URL expansion.
1873          *
1874          * In particular this is important when using action=render, which causes
1875          * full URLs to be included.
1876          *
1877          * Oh man I hate our multi-layer parser!
1878          *
1879          * @param string more-or-less HTML
1880          * @return string less-or-more HTML with NOPARSE bits
1881          */
1882         function armorLinks( $text ) {
1883                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1884                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1885         }
1886
1887         /**
1888          * Return true if subpage links should be expanded on this page.
1889          * @return bool
1890          */
1891         function areSubpagesAllowed() {
1892                 # Some namespaces don't allow subpages
1893                 global $wgNamespacesWithSubpages;
1894                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1895         }
1896
1897         /**
1898          * Handle link to subpage if necessary
1899          * @param string $target the source of the link
1900          * @param string &$text the link text, modified as necessary
1901          * @return string the full name of the link
1902          * @private
1903          */
1904         function maybeDoSubpageLink($target, &$text) {
1905                 # Valid link forms:
1906                 # Foobar -- normal
1907                 # :Foobar -- override special treatment of prefix (images, language links)
1908                 # /Foobar -- convert to CurrentPage/Foobar
1909                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1910                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1911                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1912
1913                 $fname = 'Parser::maybeDoSubpageLink';
1914                 wfProfileIn( $fname );
1915                 $ret = $target; # default return value is no change
1916
1917                 # bug 7425
1918                 $target = trim( $target );
1919
1920                 # Some namespaces don't allow subpages,
1921                 # so only perform processing if subpages are allowed
1922                 if( $this->areSubpagesAllowed() ) {
1923                         # Look at the first character
1924                         if( $target != '' && $target{0} == '/' ) {
1925                                 # / at end means we don't want the slash to be shown
1926                                 if( substr( $target, -1, 1 ) == '/' ) {
1927                                         $target = substr( $target, 1, -1 );
1928                                         $noslash = $target;
1929                                 } else {
1930                                         $noslash = substr( $target, 1 );
1931                                 }
1932
1933                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1934                                 if( '' === $text ) {
1935                                         $text = $target;
1936                                 } # this might be changed for ugliness reasons
1937                         } else {
1938                                 # check for .. subpage backlinks
1939                                 $dotdotcount = 0;
1940                                 $nodotdot = $target;
1941                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1942                                         ++$dotdotcount;
1943                                         $nodotdot = substr( $nodotdot, 3 );
1944                                 }
1945                                 if($dotdotcount > 0) {
1946                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1947                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1948                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1949                                                 # / at the end means don't show full path
1950                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1951                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1952                                                         if( '' === $text ) {
1953                                                                 $text = $nodotdot;
1954                                                         }
1955                                                 }
1956                                                 $nodotdot = trim( $nodotdot );
1957                                                 if( $nodotdot != '' ) {
1958                                                         $ret .= '/' . $nodotdot;
1959                                                 }
1960                                         }
1961                                 }
1962                         }
1963                 }
1964
1965                 wfProfileOut( $fname );
1966                 return $ret;
1967         }
1968
1969         /**#@+
1970          * Used by doBlockLevels()
1971          * @private
1972          */
1973         /* private */ function closeParagraph() {
1974                 $result = '';
1975                 if ( '' != $this->mLastSection ) {
1976                         $result = '</' . $this->mLastSection  . ">\n";
1977                 }
1978                 $this->mInPre = false;
1979                 $this->mLastSection = '';
1980                 return $result;
1981         }
1982         # getCommon() returns the length of the longest common substring
1983         # of both arguments, starting at the beginning of both.
1984         #
1985         /* private */ function getCommon( $st1, $st2 ) {
1986                 $fl = strlen( $st1 );
1987                 $shorter = strlen( $st2 );
1988                 if ( $fl < $shorter ) { $shorter = $fl; }
1989
1990                 for ( $i = 0; $i < $shorter; ++$i ) {
1991                         if ( $st1{$i} != $st2{$i} ) { break; }
1992                 }
1993                 return $i;
1994         }
1995         # These next three functions open, continue, and close the list
1996         # element appropriate to the prefix character passed into them.
1997         #
1998         /* private */ function openList( $char ) {
1999                 $result = $this->closeParagraph();
2000
2001                 if ( '*' == $char ) { $result .= '<ul><li>'; }
2002                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
2003                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
2004                 else if ( ';' == $char ) {
2005                         $result .= '<dl><dt>';
2006                         $this->mDTopen = true;
2007                 }
2008                 else { $result = '<!-- ERR 1 -->'; }
2009
2010                 return $result;
2011         }
2012
2013         /* private */ function nextItem( $char ) {
2014                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
2015                 else if ( ':' == $char || ';' == $char ) {
2016                         $close = '</dd>';
2017                         if ( $this->mDTopen ) { $close = '</dt>'; }
2018                         if ( ';' == $char ) {
2019                                 $this->mDTopen = true;
2020                                 return $close . '<dt>';
2021                         } else {
2022                                 $this->mDTopen = false;
2023                                 return $close . '<dd>';
2024                         }
2025                 }
2026                 return '<!-- ERR 2 -->';
2027         }
2028
2029         /* private */ function closeList( $char ) {
2030                 if ( '*' == $char ) { $text = '</li></ul>'; }
2031                 else if ( '#' == $char ) { $text = '</li></ol>'; }
2032                 else if ( ':' == $char ) {
2033                         if ( $this->mDTopen ) {
2034                                 $this->mDTopen = false;
2035                                 $text = '</dt></dl>';
2036                         } else {
2037                                 $text = '</dd></dl>';
2038                         }
2039                 }
2040                 else {  return '<!-- ERR 3 -->'; }
2041                 return $text."\n";
2042         }
2043         /**#@-*/
2044
2045         /**
2046          * Make lists from lines starting with ':', '*', '#', etc.
2047          *
2048          * @private
2049          * @return string the lists rendered as HTML
2050          */
2051         function doBlockLevels( $text, $linestart ) {
2052                 $fname = 'Parser::doBlockLevels';
2053                 wfProfileIn( $fname );
2054
2055                 # Parsing through the text line by line.  The main thing
2056                 # happening here is handling of block-level elements p, pre,
2057                 # and making lists from lines starting with * # : etc.
2058                 #
2059                 $textLines = explode( "\n", $text );
2060
2061                 $lastPrefix = $output = '';
2062                 $this->mDTopen = $inBlockElem = false;
2063                 $prefixLength = 0;
2064                 $paragraphStack = false;
2065
2066                 if ( !$linestart ) {
2067                         $output .= array_shift( $textLines );
2068                 }
2069                 foreach ( $textLines as $oLine ) {
2070                         $lastPrefixLength = strlen( $lastPrefix );
2071                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2072                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2073                         if ( !$this->mInPre ) {
2074                                 # Multiple prefixes may abut each other for nested lists.
2075                                 $prefixLength = strspn( $oLine, '*#:;' );
2076                                 $pref = substr( $oLine, 0, $prefixLength );
2077
2078                                 # eh?
2079                                 $pref2 = str_replace( ';', ':', $pref );
2080                                 $t = substr( $oLine, $prefixLength );
2081                                 $this->mInPre = !empty($preOpenMatch);
2082                         } else {
2083                                 # Don't interpret any other prefixes in preformatted text
2084                                 $prefixLength = 0;
2085                                 $pref = $pref2 = '';
2086                                 $t = $oLine;
2087                         }
2088
2089                         # List generation
2090                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2091                                 # Same as the last item, so no need to deal with nesting or opening stuff
2092                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2093                                 $paragraphStack = false;
2094
2095                                 if ( substr( $pref, -1 ) == ';') {
2096                                         # The one nasty exception: definition lists work like this:
2097                                         # ; title : definition text
2098                                         # So we check for : in the remainder text to split up the
2099                                         # title and definition, without b0rking links.
2100                                         $term = $t2 = '';
2101                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2102                                                 $t = $t2;
2103                                                 $output .= $term . $this->nextItem( ':' );
2104                                         }
2105                                 }
2106                         } elseif( $prefixLength || $lastPrefixLength ) {
2107                                 # Either open or close a level...
2108                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2109                                 $paragraphStack = false;
2110
2111                                 while( $commonPrefixLength < $lastPrefixLength ) {
2112                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2113                                         --$lastPrefixLength;
2114                                 }
2115                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2116                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2117                                 }
2118                                 while ( $prefixLength > $commonPrefixLength ) {
2119                                         $char = substr( $pref, $commonPrefixLength, 1 );
2120                                         $output .= $this->openList( $char );
2121
2122                                         if ( ';' == $char ) {
2123                                                 # FIXME: This is dupe of code above
2124                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2125                                                         $t = $t2;
2126                                                         $output .= $term . $this->nextItem( ':' );
2127                                                 }
2128                                         }
2129                                         ++$commonPrefixLength;
2130                                 }
2131                                 $lastPrefix = $pref2;
2132                         }
2133                         if( 0 == $prefixLength ) {
2134                                 wfProfileIn( "$fname-paragraph" );
2135                                 # No prefix (not in list)--go to paragraph mode
2136                                 // XXX: use a stack for nestable elements like span, table and div
2137                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2138                                 $closematch = preg_match(
2139                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2140                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2141                                 if ( $openmatch or $closematch ) {
2142                                         $paragraphStack = false;
2143                                         # TODO bug 5718: paragraph closed
2144                                         $output .= $this->closeParagraph();
2145                                         if ( $preOpenMatch and !$preCloseMatch ) {
2146                                                 $this->mInPre = true;
2147                                         }
2148                                         if ( $closematch ) {
2149                                                 $inBlockElem = false;
2150                                         } else {
2151                                                 $inBlockElem = true;
2152                                         }
2153                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2154                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2155                                                 // pre
2156                                                 if ($this->mLastSection != 'pre') {
2157                                                         $paragraphStack = false;
2158                                                         $output .= $this->closeParagraph().'<pre>';
2159                                                         $this->mLastSection = 'pre';
2160                                                 }
2161                                                 $t = substr( $t, 1 );
2162                                         } else {
2163                                                 // paragraph
2164                                                 if ( '' == trim($t) ) {
2165                                                         if ( $paragraphStack ) {
2166                                                                 $output .= $paragraphStack.'<br />';
2167                                                                 $paragraphStack = false;
2168                                                                 $this->mLastSection = 'p';
2169                                                         } else {
2170                                                                 if ($this->mLastSection != 'p' ) {
2171                                                                         $output .= $this->closeParagraph();
2172                                                                         $this->mLastSection = '';
2173                                                                         $paragraphStack = '<p>';
2174                                                                 } else {
2175                                                                         $paragraphStack = '</p><p>';
2176                                                                 }
2177                                                         }
2178                                                 } else {
2179                                                         if ( $paragraphStack ) {
2180                                                                 $output .= $paragraphStack;
2181                                                                 $paragraphStack = false;
2182                                                                 $this->mLastSection = 'p';
2183                                                         } else if ($this->mLastSection != 'p') {
2184                                                                 $output .= $this->closeParagraph().'<p>';
2185                                                                 $this->mLastSection = 'p';
2186                                                         }
2187                                                 }
2188                                         }
2189                                 }
2190                                 wfProfileOut( "$fname-paragraph" );
2191                         }
2192                         // somewhere above we forget to get out of pre block (bug 785)
2193                         if($preCloseMatch && $this->mInPre) {
2194                                 $this->mInPre = false;
2195                         }
2196                         if ($paragraphStack === false) {
2197                                 $output .= $t."\n";
2198                         }
2199                 }
2200                 while ( $prefixLength ) {
2201                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2202                         --$prefixLength;
2203                 }
2204                 if ( '' != $this->mLastSection ) {
2205                         $output .= '</' . $this->mLastSection . '>';
2206                         $this->mLastSection = '';
2207                 }
2208
2209                 wfProfileOut( $fname );
2210                 return $output;
2211         }
2212
2213         /**
2214          * Split up a string on ':', ignoring any occurences inside tags
2215          * to prevent illegal overlapping.
2216          * @param string $str the string to split
2217          * @param string &$before set to everything before the ':'
2218          * @param string &$after set to everything after the ':'
2219          * return string the position of the ':', or false if none found
2220          */
2221         function findColonNoLinks($str, &$before, &$after) {
2222                 $fname = 'Parser::findColonNoLinks';
2223                 wfProfileIn( $fname );
2224
2225                 $pos = strpos( $str, ':' );
2226                 if( $pos === false ) {
2227                         // Nothing to find!
2228                         wfProfileOut( $fname );
2229                         return false;
2230                 }
2231
2232                 $lt = strpos( $str, '<' );
2233                 if( $lt === false || $lt > $pos ) {
2234                         // Easy; no tag nesting to worry about
2235                         $before = substr( $str, 0, $pos );
2236                         $after = substr( $str, $pos+1 );
2237                         wfProfileOut( $fname );
2238                         return $pos;
2239                 }
2240
2241                 // Ugly state machine to walk through avoiding tags.
2242                 $state = MW_COLON_STATE_TEXT;
2243                 $stack = 0;
2244                 $len = strlen( $str );
2245                 for( $i = 0; $i < $len; $i++ ) {
2246                         $c = $str{$i};
2247
2248                         switch( $state ) {
2249                         // (Using the number is a performance hack for common cases)
2250                         case 0: // MW_COLON_STATE_TEXT:
2251                                 switch( $c ) {
2252                                 case "<":
2253                                         // Could be either a <start> tag or an </end> tag
2254                                         $state = MW_COLON_STATE_TAGSTART;
2255                                         break;
2256                                 case ":":
2257                                         if( $stack == 0 ) {
2258                                                 // We found it!
2259                                                 $before = substr( $str, 0, $i );
2260                                                 $after = substr( $str, $i + 1 );
2261                                                 wfProfileOut( $fname );
2262                                                 return $i;
2263                                         }
2264                                         // Embedded in a tag; don't break it.
2265                                         break;
2266                                 default:
2267                                         // Skip ahead looking for something interesting
2268                                         $colon = strpos( $str, ':', $i );
2269                                         if( $colon === false ) {
2270                                                 // Nothing else interesting
2271                                                 wfProfileOut( $fname );
2272                                                 return false;
2273                                         }
2274                                         $lt = strpos( $str, '<', $i );
2275                                         if( $stack === 0 ) {
2276                                                 if( $lt === false || $colon < $lt ) {
2277                                                         // We found it!
2278                                                         $before = substr( $str, 0, $colon );
2279                                                         $after = substr( $str, $colon + 1 );
2280                                                         wfProfileOut( $fname );
2281                                                         return $i;
2282                                                 }
2283                                         }
2284                                         if( $lt === false ) {
2285                                                 // Nothing else interesting to find; abort!
2286                                                 // We're nested, but there's no close tags left. Abort!
2287                                                 break 2;
2288                                         }
2289                                         // Skip ahead to next tag start
2290                                         $i = $lt;
2291                                         $state = MW_COLON_STATE_TAGSTART;
2292                                 }
2293                                 break;
2294                         case 1: // MW_COLON_STATE_TAG:
2295                                 // In a <tag>
2296                                 switch( $c ) {
2297                                 case ">":
2298                                         $stack++;
2299                                         $state = MW_COLON_STATE_TEXT;
2300                                         break;
2301                                 case "/":
2302                                         // Slash may be followed by >?
2303                                         $state = MW_COLON_STATE_TAGSLASH;
2304                                         break;
2305                                 default:
2306                                         // ignore
2307                                 }
2308                                 break;
2309                         case 2: // MW_COLON_STATE_TAGSTART:
2310                                 switch( $c ) {
2311                                 case "/":
2312                                         $state = MW_COLON_STATE_CLOSETAG;
2313                                         break;
2314                                 case "!":
2315                                         $state = MW_COLON_STATE_COMMENT;
2316                                         break;
2317                                 case ">":
2318                                         // Illegal early close? This shouldn't happen D:
2319                                         $state = MW_COLON_STATE_TEXT;
2320                                         break;
2321                                 default:
2322                                         $state = MW_COLON_STATE_TAG;
2323                                 }
2324                                 break;
2325                         case 3: // MW_COLON_STATE_CLOSETAG:
2326                                 // In a </tag>
2327                                 if( $c == ">" ) {
2328                                         $stack--;
2329                                         if( $stack < 0 ) {
2330                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2331                                                 wfProfileOut( $fname );
2332                                                 return false;
2333                                         }
2334                                         $state = MW_COLON_STATE_TEXT;
2335                                 }
2336                                 break;
2337                         case MW_COLON_STATE_TAGSLASH:
2338                                 if( $c == ">" ) {
2339                                         // Yes, a self-closed tag <blah/>
2340                                         $state = MW_COLON_STATE_TEXT;
2341                                 } else {
2342                                         // Probably we're jumping the gun, and this is an attribute
2343                                         $state = MW_COLON_STATE_TAG;
2344                                 }
2345                                 break;
2346                         case 5: // MW_COLON_STATE_COMMENT:
2347                                 if( $c == "-" ) {
2348                                         $state = MW_COLON_STATE_COMMENTDASH;
2349                                 }
2350                                 break;
2351                         case MW_COLON_STATE_COMMENTDASH:
2352                                 if( $c == "-" ) {
2353                                         $state = MW_COLON_STATE_COMMENTDASHDASH;
2354                                 } else {
2355                                         $state = MW_COLON_STATE_COMMENT;
2356                                 }
2357                                 break;
2358                         case MW_COLON_STATE_COMMENTDASHDASH:
2359                                 if( $c == ">" ) {
2360                                         $state = MW_COLON_STATE_TEXT;
2361                                 } else {
2362                                         $state = MW_COLON_STATE_COMMENT;
2363                                 }
2364                                 break;
2365                         default:
2366                                 throw new MWException( "State machine error in $fname" );
2367                         }
2368                 }
2369                 if( $stack > 0 ) {
2370                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2371                         return false;
2372                 }
2373                 wfProfileOut( $fname );
2374                 return false;
2375         }
2376
2377         /**
2378          * Return value of a magic variable (like PAGENAME)
2379          *
2380          * @private
2381          */
2382         function getVariableValue( $index ) {
2383                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2384
2385                 /**
2386                  * Some of these require message or data lookups and can be
2387                  * expensive to check many times.
2388                  */
2389                 static $varCache = array();
2390                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) ) {
2391                         if ( isset( $varCache[$index] ) ) {
2392                                 return $varCache[$index];
2393                         }
2394                 }
2395
2396                 $ts = time();
2397                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2398
2399                 # Use the time zone
2400                 global $wgLocaltimezone;
2401                 if ( isset( $wgLocaltimezone ) ) {
2402                         $oldtz = getenv( 'TZ' );
2403                         putenv( 'TZ='.$wgLocaltimezone );
2404                 }
2405                 $localTimestamp = date( 'YmdHis', $ts );
2406                 $localMonth = date( 'm', $ts );
2407                 $localMonthName = date( 'n', $ts );
2408                 $localDay = date( 'j', $ts );
2409                 $localDay2 = date( 'd', $ts );
2410                 $localDayOfWeek = date( 'w', $ts );
2411                 $localWeek = date( 'W', $ts );
2412                 $localYear = date( 'Y', $ts );
2413                 $localHour = date( 'H', $ts );
2414                 if ( isset( $wgLocaltimezone ) ) {
2415                         putenv( 'TZ='.$oldtz );
2416                 }
2417
2418                 switch ( $index ) {
2419                         case 'currentmonth':
2420                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
2421                         case 'currentmonthname':
2422                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2423                         case 'currentmonthnamegen':
2424                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2425                         case 'currentmonthabbrev':
2426                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2427                         case 'currentday':
2428                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2429                         case 'currentday2':
2430                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2431                         case 'localmonth':
2432                                 return $varCache[$index] = $wgContLang->formatNum( $localMonth );
2433                         case 'localmonthname':
2434                                 return $varCache[$index] = $wgContLang->getMonthName( $localMonthName );
2435                         case 'localmonthnamegen':
2436                                 return $varCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2437                         case 'localmonthabbrev':
2438                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2439                         case 'localday':
2440                                 return $varCache[$index] = $wgContLang->formatNum( $localDay );
2441                         case 'localday2':
2442                                 return $varCache[$index] = $wgContLang->formatNum( $localDay2 );
2443                         case 'pagename':
2444                                 return $this->mTitle->getText();
2445                         case 'pagenamee':
2446                                 return $this->mTitle->getPartialURL();
2447                         case 'fullpagename':
2448                                 return $this->mTitle->getPrefixedText();
2449                         case 'fullpagenamee':
2450                                 return $this->mTitle->getPrefixedURL();
2451                         case 'subpagename':
2452                                 return $this->mTitle->getSubpageText();
2453                         case 'subpagenamee':
2454                                 return $this->mTitle->getSubpageUrlForm();
2455                         case 'basepagename':
2456                                 return $this->mTitle->getBaseText();
2457                         case 'basepagenamee':
2458                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2459                         case 'talkpagename':
2460                                 if( $this->mTitle->canTalk() ) {
2461                                         $talkPage = $this->mTitle->getTalkPage();
2462                                         return $talkPage->getPrefixedText();
2463                                 } else {
2464                                         return '';
2465                                 }
2466                         case 'talkpagenamee':
2467                                 if( $this->mTitle->canTalk() ) {
2468                                         $talkPage = $this->mTitle->getTalkPage();
2469                                         return $talkPage->getPrefixedUrl();
2470                                 } else {
2471                                         return '';
2472                                 }
2473                         case 'subjectpagename':
2474                                 $subjPage = $this->mTitle->getSubjectPage();
2475                                 return $subjPage->getPrefixedText();
2476                         case 'subjectpagenamee':
2477                                 $subjPage = $this->mTitle->getSubjectPage();
2478                                 return $subjPage->getPrefixedUrl();
2479                         case 'revisionid':
2480                                 return $this->mRevisionId;
2481                         case 'revisionday':
2482                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2483                         case 'revisionday2':
2484                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2485                         case 'revisionmonth':
2486                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2487                         case 'revisionyear':
2488                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2489                         case 'revisiontimestamp':
2490                                 return $this->getRevisionTimestamp();
2491                         case 'namespace':
2492                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2493                         case 'namespacee':
2494                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2495                         case 'talkspace':
2496                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2497                         case 'talkspacee':
2498                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2499                         case 'subjectspace':
2500                                 return $this->mTitle->getSubjectNsText();
2501                         case 'subjectspacee':
2502                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2503                         case 'currentdayname':
2504                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2505                         case 'currentyear':
2506                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2507                         case 'currenttime':
2508                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2509                         case 'currenthour':
2510                                 return $varCache[$index] = $wgContLang->formatNum( date( 'H', $ts ), true );
2511                         case 'currentweek':
2512                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2513                                 // int to remove the padding
2514                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2515                         case 'currentdow':
2516                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2517                         case 'localdayname':
2518                                 return $varCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2519                         case 'localyear':
2520                                 return $varCache[$index] = $wgContLang->formatNum( $localYear, true );
2521                         case 'localtime':
2522                                 return $varCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2523                         case 'localhour':
2524                                 return $varCache[$index] = $wgContLang->formatNum( $localHour, true );
2525                         case 'localweek':
2526                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2527                                 // int to remove the padding
2528                                 return $varCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2529                         case 'localdow':
2530                                 return $varCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2531                         case 'numberofarticles':
2532                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2533                         case 'numberoffiles':
2534                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2535                         case 'numberofusers':
2536                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2537                         case 'numberofpages':
2538                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2539                         case 'numberofadmins':
2540                                 return $varCache[$index]  = $wgContLang->formatNum( SiteStats::admins() );
2541                         case 'currenttimestamp':
2542                                 return $varCache[$index] = wfTimestampNow();
2543                         case 'localtimestamp':
2544                                 return $varCache[$index] = $localTimestamp;
2545                         case 'currentversion':
2546                                 return $varCache[$index] = SpecialVersion::getVersion();
2547                         case 'sitename':
2548                                 return $wgSitename;
2549                         case 'server':
2550                                 return $wgServer;
2551                         case 'servername':
2552                                 return $wgServerName;
2553                         case 'scriptpath':
2554                                 return $wgScriptPath;
2555                         case 'directionmark':
2556                                 return $wgContLang->getDirMark();
2557                         case 'contentlanguage':
2558                                 global $wgContLanguageCode;
2559                                 return $wgContLanguageCode;
2560                         default:
2561                                 $ret = null;
2562                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2563                                         return $ret;
2564                                 else
2565                                         return null;
2566                 }
2567         }
2568
2569         /**
2570          * initialise the magic variables (like CURRENTMONTHNAME)
2571          *
2572          * @private
2573          */
2574         function initialiseVariables() {
2575                 $fname = 'Parser::initialiseVariables';
2576                 wfProfileIn( $fname );
2577                 $variableIDs = MagicWord::getVariableIDs();
2578
2579                 $this->mVariables = array();
2580                 foreach ( $variableIDs as $id ) {
2581                         $mw =& MagicWord::get( $id );
2582                         $mw->addToArray( $this->mVariables, $id );
2583                 }
2584                 wfProfileOut( $fname );
2585         }
2586
2587         /**
2588          * parse any parentheses in format ((title|part|part))
2589          * and call callbacks to get a replacement text for any found piece
2590          *
2591          * @param string $text The text to parse
2592          * @param array $callbacks rules in form:
2593          *     '{' => array(                            # opening parentheses
2594          *                                      'end' => '}',   # closing parentheses
2595          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2596          *                                                                3 => callback         # replacement callback to call if {{{..}}} is found
2597          *                                                                )
2598          *                                      )
2599          *                                      'min' => 2,     # Minimum parenthesis count in cb
2600          *                                      'max' => 3,     # Maximum parenthesis count in cb
2601          * @private
2602          */
2603         function replace_callback ($text, $callbacks) {
2604                 wfProfileIn( __METHOD__ );
2605                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2606                 $lastOpeningBrace = -1;                 # last not closed parentheses
2607
2608                 $validOpeningBraces = implode( '', array_keys( $callbacks ) );
2609
2610                 $i = 0;
2611                 while ( $i < strlen( $text ) ) {
2612                         # Find next opening brace, closing brace or pipe
2613                         if ( $lastOpeningBrace == -1 ) {
2614                                 $currentClosing = '';
2615                                 $search = $validOpeningBraces;
2616                         } else {
2617                                 $currentClosing = $openingBraceStack[$lastOpeningBrace]['braceEnd'];
2618                                 $search = $validOpeningBraces . '|' . $currentClosing;
2619                         }
2620                         $rule = null;
2621                         $i += strcspn( $text, $search, $i );
2622                         if ( $i < strlen( $text ) ) {
2623                                 if ( $text[$i] == '|' ) {
2624                                         $found = 'pipe';
2625                                 } elseif ( $text[$i] == $currentClosing ) {
2626                                         $found = 'close';
2627                                 } elseif ( isset( $callbacks[$text[$i]] ) ) {
2628                                         $found = 'open';
2629                                         $rule = $callbacks[$text[$i]];
2630                                 } else {
2631                                         # Some versions of PHP have a strcspn which stops on null characters
2632                                         # Ignore and continue
2633                                         ++$i;
2634                                         continue;
2635                                 }
2636                         } else {
2637                                 # All done
2638                                 break;
2639                         }
2640
2641                         if ( $found == 'open' ) {
2642                                 # found opening brace, let's add it to parentheses stack
2643                                 $piece = array('brace' => $text[$i],
2644                                                            'braceEnd' => $rule['end'],
2645                                                            'title' => '',
2646                                                            'parts' => null);
2647
2648                                 # count opening brace characters
2649                                 $piece['count'] = strspn( $text, $piece['brace'], $i );
2650                                 $piece['startAt'] = $piece['partStart'] = $i + $piece['count'];
2651                                 $i += $piece['count'];
2652
2653                                 # we need to add to stack only if opening brace count is enough for one of the rules
2654                                 if ( $piece['count'] >= $rule['min'] ) {
2655                                         $lastOpeningBrace ++;
2656                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2657                                 }
2658                         } elseif ( $found == 'close' ) {
2659                                 # lets check if it is enough characters for closing brace
2660                                 $maxCount = $openingBraceStack[$lastOpeningBrace]['count'];
2661                                 $count = strspn( $text, $text[$i], $i, $maxCount );
2662
2663                                 # check for maximum matching characters (if there are 5 closing
2664                                 # characters, we will probably need only 3 - depending on the rules)
2665                                 $matchingCount = 0;
2666                                 $matchingCallback = null;
2667                                 $cbType = $callbacks[$openingBraceStack[$lastOpeningBrace]['brace']];
2668                                 if ( $count > $cbType['max'] ) {
2669                                         # The specified maximum exists in the callback array, unless the caller
2670                                         # has made an error
2671                                         $matchingCount = $cbType['max'];
2672                                 } else {
2673                                         # Count is less than the maximum
2674                                         # Skip any gaps in the callback array to find the true largest match
2675                                         # Need to use array_key_exists not isset because the callback can be null
2676                                         $matchingCount = $count;
2677                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $cbType['cb'] ) ) {
2678                                                 --$matchingCount;
2679                                         }
2680                                 }
2681
2682                                 if ($matchingCount <= 0) {
2683                                         $i += $count;
2684                                         continue;
2685                                 }
2686                                 $matchingCallback = $cbType['cb'][$matchingCount];
2687
2688                                 # let's set a title or last part (if '|' was found)
2689                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2690                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2691                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2692                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2693                                 } else {
2694                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2695                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2696                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2697                                 }
2698
2699                                 $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2700                                 $pieceEnd = $i + $matchingCount;
2701
2702                                 if( is_callable( $matchingCallback ) ) {
2703                                         $cbArgs = array (
2704                                                                          'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2705                                                                          'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2706                                                                          'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2707                                                                          'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == "\n")),
2708                                                                          );
2709                                         # finally we can call a user callback and replace piece of text
2710                                         $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2711                                         $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2712                                         $i = $pieceStart + strlen($replaceWith);
2713                                 } else {
2714                                         # null value for callback means that parentheses should be parsed, but not replaced
2715                                         $i += $matchingCount;
2716                                 }
2717
2718                                 # reset last opening parentheses, but keep it in case there are unused characters
2719                                 $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2720                                                            'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2721                                                            'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2722                                                            'title' => '',
2723                                                            'parts' => null,
2724                                                            'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2725                                 $openingBraceStack[$lastOpeningBrace--] = null;
2726
2727                                 if ($matchingCount < $piece['count']) {
2728                                         $piece['count'] -= $matchingCount;
2729                                         $piece['startAt'] -= $matchingCount;
2730                                         $piece['partStart'] = $piece['startAt'];
2731                                         # do we still qualify for any callback with remaining count?
2732                                         $currentCbList = $callbacks[$piece['brace']]['cb'];
2733                                         while ( $piece['count'] ) {
2734                                                 if ( array_key_exists( $piece['count'], $currentCbList ) ) {
2735                                                         $lastOpeningBrace++;
2736                                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2737                                                         break;
2738                                                 }
2739                                                 --$piece['count'];
2740                                         }
2741                                 }
2742                         } elseif ( $found == 'pipe' ) {
2743                                 # lets set a title if it is a first separator, or next part otherwise
2744                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2745                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2746                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2747                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2748                                         $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2749                                 } else {
2750                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2751                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2752                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2753                                 }
2754                                 $openingBraceStack[$lastOpeningBrace]['partStart'] = ++$i;
2755                         }
2756                 }
2757
2758                 wfProfileOut( __METHOD__ );
2759                 return $text;
2760         }
2761
2762         /**
2763          * Replace magic variables, templates, and template arguments
2764          * with the appropriate text. Templates are substituted recursively,
2765          * taking care to avoid infinite loops.
2766          *
2767          * Note that the substitution depends on value of $mOutputType:
2768          *  OT_WIKI: only {{subst:}} templates
2769          *  OT_MSG: only magic variables
2770          *  OT_HTML: all templates and magic variables
2771          *
2772          * @param string $tex The text to transform
2773          * @param array $args Key-value pairs representing template parameters to substitute
2774          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2775          * @private
2776          */
2777         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2778                 # Prevent too big inclusions
2779                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2780                         return $text;
2781                 }
2782
2783                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2784                 wfProfileIn( $fname );
2785
2786                 # This function is called recursively. To keep track of arguments we need a stack:
2787                 array_push( $this->mArgStack, $args );
2788
2789                 $braceCallbacks = array();
2790                 if ( !$argsOnly ) {
2791                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2792                 }
2793                 if ( $this->mOutputType != OT_MSG ) {
2794                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2795                 }
2796                 if ( $braceCallbacks ) {
2797                         $callbacks = array(
2798                                 '{' => array(
2799                                         'end' => '}',
2800                                         'cb' => $braceCallbacks,
2801                                         'min' => $argsOnly ? 3 : 2,
2802                                         'max' => isset( $braceCallbacks[3] ) ? 3 : 2,
2803                                 ),
2804                                 '[' => array(
2805                                         'end' => ']',
2806                                         'cb' => array(2=>null),
2807                                         'min' => 2,
2808                                         'max' => 2,
2809                                 )
2810                         );
2811                         $text = $this->replace_callback ($text, $callbacks);
2812
2813                         array_pop( $this->mArgStack );
2814                 }
2815                 wfProfileOut( $fname );
2816                 return $text;
2817         }
2818
2819         /**
2820          * Replace magic variables
2821          * @private
2822          */
2823         function variableSubstitution( $matches ) {
2824                 global $wgContLang;
2825                 $fname = 'Parser::variableSubstitution';
2826                 $varname = $wgContLang->lc($matches[1]);
2827                 wfProfileIn( $fname );
2828                 $skip = false;
2829                 if ( $this->mOutputType == OT_WIKI ) {
2830                         # Do only magic variables prefixed by SUBST
2831                         $mwSubst =& MagicWord::get( 'subst' );
2832                         if (!$mwSubst->matchStartAndRemove( $varname ))
2833                                 $skip = true;
2834                         # Note that if we don't substitute the variable below,
2835                         # we don't remove the {{subst:}} magic word, in case
2836                         # it is a template rather than a magic variable.
2837                 }
2838                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2839                         $id = $this->mVariables[$varname];
2840                         # Now check if we did really match, case sensitive or not
2841                         $mw =& MagicWord::get( $id );
2842                         if ($mw->match($matches[1])) {
2843                                 $text = $this->getVariableValue( $id );
2844                                 $this->mOutput->mContainsOldMagic = true;
2845                         } else {
2846                                 $text = $matches[0];
2847                         }
2848                 } else {
2849                         $text = $matches[0];
2850                 }
2851                 wfProfileOut( $fname );
2852                 return $text;
2853         }
2854
2855
2856         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2857         static function createAssocArgs( $args ) {
2858                 $assocArgs = array();
2859                 $index = 1;
2860                 foreach( $args as $arg ) {
2861                         $eqpos = strpos( $arg, '=' );
2862                         if ( $eqpos === false ) {
2863                                 $assocArgs[$index++] = $arg;
2864                         } else {
2865                                 $name = trim( substr( $arg, 0, $eqpos ) );
2866                                 $value = trim( substr( $arg, $eqpos+1 ) );
2867                                 if ( $value === false ) {
2868                                         $value = '';
2869                                 }
2870                                 if ( $name !== false ) {
2871                                         $assocArgs[$name] = $value;
2872                                 }
2873                         }
2874                 }
2875
2876                 return $assocArgs;
2877         }
2878
2879         /**
2880          * Return the text of a template, after recursively
2881          * replacing any variables or templates within the template.
2882          *
2883          * @param array $piece The parts of the template
2884          *  $piece['text']: matched text
2885          *  $piece['title']: the title, i.e. the part before the |
2886          *  $piece['parts']: the parameter array
2887          * @return string the text of the template
2888          * @private
2889          */
2890         function braceSubstitution( $piece ) {
2891                 global $wgContLang, $wgLang, $wgAllowDisplayTitle;
2892                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2893                 wfProfileIn( $fname );
2894                 wfProfileIn( __METHOD__.'-setup' );
2895
2896                 # Flags
2897                 $found = false;             # $text has been filled
2898                 $nowiki = false;            # wiki markup in $text should be escaped
2899                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2900                 $noargs = false;            # Don't replace triple-brace arguments in $text
2901                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2902                 $headingOffset = 0;         # Skip headings when number, to account for those that weren't transcluded.
2903                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2904                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2905
2906                 # Title object, where $text came from
2907                 $title = NULL;
2908
2909                 $linestart = '';
2910
2911
2912                 # $part1 is the bit before the first |, and must contain only title characters
2913                 # $args is a list of arguments, starting from index 0, not including $part1
2914
2915                 $titleText = $part1 = $piece['title'];
2916                 # If the third subpattern matched anything, it will start with |
2917
2918                 if (null == $piece['parts']) {
2919                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2920                         if ($replaceWith != $piece['text']) {
2921                                 $text = $replaceWith;
2922                                 $found = true;
2923                                 $noparse = true;
2924                                 $noargs = true;
2925                         }
2926                 }
2927
2928                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2929                 wfProfileOut( __METHOD__.'-setup' );
2930
2931                 # SUBST
2932                 wfProfileIn( __METHOD__.'-modifiers' );
2933                 if ( !$found ) {
2934                         $mwSubst =& MagicWord::get( 'subst' );
2935                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2936                                 # One of two possibilities is true:
2937                                 # 1) Found SUBST but not in the PST phase
2938                                 # 2) Didn't find SUBST and in the PST phase
2939                                 # In either case, return without further processing
2940                                 $text = $piece['text'];
2941                                 $found = true;
2942                                 $noparse = true;
2943                                 $noargs = true;
2944                         }
2945                 }
2946
2947                 # MSG, MSGNW and RAW
2948                 if ( !$found ) {
2949                         # Check for MSGNW:
2950                         $mwMsgnw =& MagicWord::get( 'msgnw' );
2951                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2952                                 $nowiki = true;
2953                         } else {
2954                                 # Remove obsolete MSG:
2955                                 $mwMsg =& MagicWord::get( 'msg' );
2956                                 $mwMsg->matchStartAndRemove( $part1 );
2957                         }
2958
2959                         # Check for RAW:
2960                         $mwRaw =& MagicWord::get( 'raw' );
2961                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2962                                 $forceRawInterwiki = true;
2963                         }
2964                 }
2965                 wfProfileOut( __METHOD__.'-modifiers' );
2966
2967                 //save path level before recursing into functions & templates.
2968                 $lastPathLevel = $this->mTemplatePath;
2969
2970                 # Parser functions
2971                 if ( !$found ) {
2972                         wfProfileIn( __METHOD__ . '-pfunc' );
2973
2974                         $colonPos = strpos( $part1, ':' );
2975                         if ( $colonPos !== false ) {
2976                                 # Case sensitive functions
2977                                 $function = substr( $part1, 0, $colonPos );
2978                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2979                                         $function = $this->mFunctionSynonyms[1][$function];
2980                                 } else {
2981                                         # Case insensitive functions
2982                                         $function = strtolower( $function );
2983                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2984                                                 $function = $this->mFunctionSynonyms[0][$function];
2985                                         } else {
2986                                                 $function = false;
2987                                         }
2988                                 }
2989                                 if ( $function ) {
2990                                         $funcArgs = array_map( 'trim', $args );
2991                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
2992                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
2993                                         $found = true;
2994
2995                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
2996                                         //$noargs = true;
2997                                         //$noparse = true;
2998
2999                                         if ( is_array( $result ) ) {
3000                                                 if ( isset( $result[0] ) ) {
3001                                                         $text = $linestart . $result[0];
3002                                                         unset( $result[0] );
3003                                                 }
3004
3005                                                 // Extract flags into the local scope
3006                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
3007                                                 extract( $result );
3008                                         } else {
3009                                                 $text = $linestart . $result;
3010                                         }
3011                                 }
3012                         }
3013                         wfProfileOut( __METHOD__ . '-pfunc' );
3014                 }
3015
3016                 # Template table test
3017
3018                 # Did we encounter this template already? If yes, it is in the cache
3019                 # and we need to check for loops.
3020                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
3021                         $found = true;
3022
3023                         # Infinite loop test
3024                         if ( isset( $this->mTemplatePath[$part1] ) ) {
3025                                 $noparse = true;
3026                                 $noargs = true;
3027                                 $found = true;
3028                                 $text = $linestart .
3029                                         "[[$part1]]<!-- WARNING: template loop detected -->";
3030                                 wfDebug( __METHOD__.": template loop broken at '$part1'\n" );
3031                         } else {
3032                                 # set $text to cached message.
3033                                 $text = $linestart . $this->mTemplates[$piece['title']];
3034                         }
3035                 }
3036
3037                 # Load from database
3038                 if ( !$found ) {
3039                         wfProfileIn( __METHOD__ . '-loadtpl' );
3040                         $ns = NS_TEMPLATE;
3041                         # declaring $subpage directly in the function call
3042                         # does not work correctly with references and breaks
3043                         # {{/subpage}}-style inclusions
3044                         $subpage = '';
3045                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3046                         if ($subpage !== '') {
3047                                 $ns = $this->mTitle->getNamespace();
3048                         }
3049                         $title = Title::newFromText( $part1, $ns );
3050
3051
3052                         if ( !is_null( $title ) ) {
3053                                 $titleText = $title->getPrefixedText();
3054                                 # Check for language variants if the template is not found
3055                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
3056                                         $wgContLang->findVariantLink($part1, $title);
3057                                 }
3058
3059                                 if ( !$title->isExternal() ) {
3060                                         if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
3061                                                 $text = SpecialPage::capturePath( $title );
3062                                                 if ( is_string( $text ) ) {
3063                                                         $found = true;
3064                                                         $noparse = true;
3065                                                         $noargs = true;
3066                                                         $isHTML = true;
3067                                                         $this->disableCache();
3068                                                 }
3069                                         } else {
3070                                                 $articleContent = $this->fetchTemplate( $title );
3071                                                 if ( $articleContent !== false ) {
3072                                                         $found = true;
3073                                                         $text = $articleContent;
3074                                                         $replaceHeadings = true;
3075                                                 }
3076                                         }
3077
3078                                         # If the title is valid but undisplayable, make a link to it
3079                                         if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3080                                                 $text = "[[:$titleText]]";
3081                                                 $found = true;
3082                                         }
3083                                 } elseif ( $title->isTrans() ) {
3084                                         // Interwiki transclusion
3085                                         if ( $this->ot['html'] && !$forceRawInterwiki ) {
3086                                                 $text = $this->interwikiTransclude( $title, 'render' );
3087                                                 $isHTML = true;
3088                                                 $noparse = true;
3089                                         } else {
3090                                                 $text = $this->interwikiTransclude( $title, 'raw' );
3091                                                 $replaceHeadings = true;
3092                                         }
3093                                         $found = true;
3094                                 }
3095
3096                                 # Template cache array insertion
3097                                 # Use the original $piece['title'] not the mangled $part1, so that
3098                                 # modifiers such as RAW: produce separate cache entries
3099                                 if( $found ) {
3100                                         if( $isHTML ) {
3101                                                 // A special page; don't store it in the template cache.
3102                                         } else {
3103                                                 $this->mTemplates[$piece['title']] = $text;
3104                                         }
3105                                         $text = $linestart . $text;
3106                                 }
3107                         }
3108                         wfProfileOut( __METHOD__ . '-loadtpl' );
3109                 }
3110
3111                 if ( $found && !$this->incrementIncludeSize( 'pre-expand', strlen( $text ) ) ) {
3112                         # Error, oversize inclusion
3113                         $text = $linestart .
3114                                 "[[$titleText]]<!-- WARNING: template omitted, pre-expand include size too large -->";
3115                         $noparse = true;
3116                         $noargs = true;
3117                 }
3118
3119                 # Recursive parsing, escaping and link table handling
3120                 # Only for HTML output
3121                 if ( $nowiki && $found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3122                         $text = wfEscapeWikiText( $text );
3123                 } elseif ( !$this->ot['msg'] && $found ) {
3124                         if ( $noargs ) {
3125                                 $assocArgs = array();
3126                         } else {
3127                                 # Clean up argument array
3128                                 $assocArgs = self::createAssocArgs($args);
3129                                 # Add a new element to the templace recursion path
3130                                 $this->mTemplatePath[$part1] = 1;
3131                         }
3132
3133                         if ( !$noparse ) {
3134                                 # If there are any <onlyinclude> tags, only include them
3135                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
3136                                         $replacer = new OnlyIncludeReplacer;
3137                                         StringUtils::delimiterReplaceCallback( '<onlyinclude>', '</onlyinclude>',
3138                                                 array( &$replacer, 'replace' ), $text );
3139                                         $text = $replacer->output;
3140                                 }
3141                                 # Remove <noinclude> sections and <includeonly> tags
3142                                 $text = StringUtils::delimiterReplace( '<noinclude>', '</noinclude>', '', $text );
3143                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
3144
3145                                 if( $this->ot['html'] || $this->ot['pre'] ) {
3146                                         # Strip <nowiki>, <pre>, etc.
3147                                         $text = $this->strip( $text, $this->mStripState );
3148                                         if ( $this->ot['html'] ) {
3149                                                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
3150                                         } elseif ( $this->ot['pre'] && $this->mOptions->getRemoveComments() ) {
3151                                                 $text = Sanitizer::removeHTMLcomments( $text );
3152                                         }
3153                                 }
3154                                 $text = $this->replaceVariables( $text, $assocArgs );
3155
3156                                 # If the template begins with a table or block-level
3157                                 # element, it should be treated as beginning a new line.
3158                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) /*}*/{
3159                                         $text = "\n" . $text;
3160                                 }
3161                         } elseif ( !$noargs ) {
3162                                 # $noparse and !$noargs
3163                                 # Just replace the arguments, not any double-brace items
3164                                 # This is used for rendered interwiki transclusion
3165                                 $text = $this->replaceVariables( $text, $assocArgs, true );
3166                         }
3167                 }
3168                 # Prune lower levels off the recursion check path
3169                 $this->mTemplatePath = $lastPathLevel;
3170
3171                 if ( $found && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3172                         # Error, oversize inclusion
3173                         $text = $linestart .
3174                                 "[[$titleText]]<!-- WARNING: template omitted, post-expand include size too large -->";
3175                         $noparse = true;
3176                         $noargs = true;
3177                 }
3178
3179                 if ( !$found ) {
3180                         wfProfileOut( $fname );
3181                         return $piece['text'];
3182                 } else {
3183                         wfProfileIn( __METHOD__ . '-placeholders' );
3184                         if ( $isHTML ) {
3185                                 # Replace raw HTML by a placeholder
3186                                 # Add a blank line preceding, to prevent it from mucking up
3187                                 # immediately preceding headings
3188                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
3189                         } else {
3190                                 # replace ==section headers==
3191                                 # XXX this needs to go away once we have a better parser.
3192                                 if ( !$this->ot['wiki'] && !$this->ot['pre'] && $replaceHeadings ) {
3193                                         if( !is_null( $title ) )
3194                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
3195                                         else
3196                                                 $encodedname = base64_encode("");
3197                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
3198                                                 PREG_SPLIT_DELIM_CAPTURE);
3199                                         $text = '';
3200                                         $nsec = $headingOffset;
3201                                         for( $i = 0; $i < count($m); $i += 2 ) {
3202                                                 $text .= $m[$i];
3203                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
3204                                                 $hl = $m[$i + 1];
3205                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
3206                                                         $text .= $hl;
3207                                                         continue;
3208                                                 }
3209                                                 $m2 = array();
3210                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
3211                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
3212                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
3213
3214                                                 $nsec++;
3215                                         }
3216                                 }
3217                         }
3218                         wfProfileOut( __METHOD__ . '-placeholders' );
3219                 }
3220
3221                 # Prune lower levels off the recursion check path
3222                 $this->mTemplatePath = $lastPathLevel;
3223
3224                 if ( !$found ) {
3225                         wfProfileOut( $fname );
3226                         return $piece['text'];
3227                 } else {
3228                         wfProfileOut( $fname );
3229                         return $text;
3230                 }
3231         }
3232
3233         /**
3234          * Fetch the unparsed text of a template and register a reference to it.
3235          */
3236         function fetchTemplate( $title ) {
3237                 $text = false;
3238                 // Loop to fetch the article, with up to 1 redirect
3239                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3240                         $rev = Revision::newFromTitle( $title );
3241                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
3242                         if ( $rev ) {
3243                                 $text = $rev->getText();
3244                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3245                                 global $wgLang;
3246                                 $message = $wgLang->lcfirst( $title->getText() );
3247                                 $text = wfMsgForContentNoTrans( $message );
3248                                 if( wfEmptyMsg( $message, $text ) ) {
3249                                         $text = false;
3250                                         break;
3251                                 }
3252                         } else {
3253                                 break;
3254                         }
3255                         if ( $text === false ) {
3256                                 break;
3257                         }
3258                         // Redirect?
3259                         $title = Title::newFromRedirect( $text );
3260                 }
3261                 return $text;
3262         }
3263
3264         /**
3265          * Transclude an interwiki link.
3266          */
3267         function interwikiTransclude( $title, $action ) {
3268                 global $wgEnableScaryTranscluding;
3269
3270                 if (!$wgEnableScaryTranscluding)
3271                         return wfMsg('scarytranscludedisabled');
3272
3273                 $url = $title->getFullUrl( "action=$action" );
3274
3275                 if (strlen($url) > 255)
3276                         return wfMsg('scarytranscludetoolong');
3277                 return $this->fetchScaryTemplateMaybeFromCache($url);
3278         }
3279
3280         function fetchScaryTemplateMaybeFromCache($url) {
3281                 global $wgTranscludeCacheExpiry;
3282                 $dbr =& wfGetDB(DB_SLAVE);
3283                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3284                                 array('tc_url' => $url));
3285                 if ($obj) {
3286                         $time = $obj->tc_time;
3287                         $text = $obj->tc_contents;
3288                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3289                                 return $text;
3290                         }
3291                 }
3292
3293                 $text = Http::get($url);
3294                 if (!$text)
3295                         return wfMsg('scarytranscludefailed', $url);
3296
3297                 $dbw =& wfGetDB(DB_MASTER);
3298                 $dbw->replace('transcache', array('tc_url'), array(
3299                         'tc_url' => $url,
3300                         'tc_time' => time(),
3301                         'tc_contents' => $text));
3302                 return $text;
3303         }
3304
3305
3306         /**
3307          * Triple brace replacement -- used for template arguments
3308          * @private
3309          */
3310         function argSubstitution( $matches ) {
3311                 $arg = trim( $matches['title'] );
3312                 $text = $matches['text'];
3313                 $inputArgs = end( $this->mArgStack );
3314
3315                 if ( array_key_exists( $arg, $inputArgs ) ) {
3316                         $text = $inputArgs[$arg];
3317                 } else if (($this->mOutputType == OT_HTML || $this->mOutputType == OT_PREPROCESS ) &&
3318                 null != $matches['parts'] && count($matches['parts']) > 0) {
3319                         $text = $matches['parts'][0];
3320                 }
3321                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3322                         $text = $matches['text'] .
3323                                 '<!-- WARNING: argument omitted, expansion size too large -->';
3324                 }
3325
3326                 return $text;
3327         }
3328
3329         /**
3330          * Increment an include size counter
3331          *
3332          * @param string $type The type of expansion
3333          * @param integer $size The size of the text
3334          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3335          */
3336         function incrementIncludeSize( $type, $size ) {
3337                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3338                         return false;
3339                 } else {
3340                         $this->mIncludeSizes[$type] += $size;
3341                         return true;
3342                 }
3343         }
3344
3345         /**
3346          * Detect __NOGALLERY__ magic word and set a placeholder
3347          */
3348         function stripNoGallery( &$text ) {
3349                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3350                 # do not add TOC
3351                 $mw = MagicWord::get( 'nogallery' );
3352                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3353         }
3354
3355         /**
3356          * Detect __TOC__ magic word and set a placeholder
3357          */
3358         function stripToc( $text ) {
3359                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3360                 # do not add TOC
3361                 $mw = MagicWord::get( 'notoc' );
3362                 if( $mw->matchAndRemove( $text ) ) {
3363                         $this->mShowToc = false;
3364                 }
3365
3366                 $mw = MagicWord::get( 'toc' );
3367                 if( $mw->match( $text ) ) {
3368                         $this->mShowToc = true;
3369                         $this->mForceTocPosition = true;
3370
3371                         // Set a placeholder. At the end we'll fill it in with the TOC.
3372                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3373
3374                         // Only keep the first one.
3375                         $text = $mw->replace( '', $text );
3376                 }
3377                 return $text;
3378         }
3379
3380         /**
3381          * This function accomplishes several tasks:
3382          * 1) Auto-number headings if that option is enabled
3383          * 2) Add an [edit] link to sections for logged in users who have enabled the option
3384          * 3) Add a Table of contents on the top for users who have enabled the option
3385          * 4) Auto-anchor headings
3386          *
3387          * It loops through all headlines, collects the necessary data, then splits up the
3388          * string and re-inserts the newly formatted headlines.
3389          *
3390          * @param string $text
3391          * @param boolean $isMain
3392          * @private
3393          */
3394         function formatHeadings( $text, $isMain=true ) {
3395                 global $wgMaxTocLevel, $wgContLang;
3396
3397                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3398                 if( !$this->mTitle->quickUserCan( 'edit' ) ) {
3399                         $showEditLink = 0;
3400                 } else {
3401                         $showEditLink = $this->mOptions->getEditSection();
3402                 }
3403
3404                 # Inhibit editsection links if requested in the page
3405                 $esw =& MagicWord::get( 'noeditsection' );
3406                 if( $esw->matchAndRemove( $text ) ) {
3407                         $showEditLink = 0;
3408                 }
3409
3410                 # Get all headlines for numbering them and adding funky stuff like [edit]
3411                 # links - this is for later, but we need the number of headlines right now
3412                 $matches = array();
3413                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3414
3415                 # if there are fewer than 4 headlines in the article, do not show TOC
3416                 # unless it's been explicitly enabled.
3417                 $enoughToc = $this->mShowToc &&
3418                         (($numMatches >= 4) || $this->mForceTocPosition);
3419
3420                 # Allow user to stipulate that a page should have a "new section"
3421                 # link added via __NEWSECTIONLINK__
3422                 $mw =& MagicWord::get( 'newsectionlink' );
3423                 if( $mw->matchAndRemove( $text ) )
3424                         $this->mOutput->setNewSection( true );
3425
3426                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3427                 # override above conditions and always show TOC above first header
3428                 $mw =& MagicWord::get( 'forcetoc' );
3429                 if ($mw->matchAndRemove( $text ) ) {
3430                         $this->mShowToc = true;
3431                         $enoughToc = true;
3432                 }
3433
3434                 # Never ever show TOC if no headers
3435                 if( $numMatches < 1 ) {
3436                         $enoughToc = false;
3437                 }
3438
3439                 # We need this to perform operations on the HTML
3440                 $sk =& $this->mOptions->getSkin();
3441
3442                 # headline counter
3443                 $headlineCount = 0;
3444                 $sectionCount = 0; # headlineCount excluding template sections
3445
3446                 # Ugh .. the TOC should have neat indentation levels which can be
3447                 # passed to the skin functions. These are determined here
3448                 $toc = '';
3449                 $full = '';
3450                 $head = array();
3451                 $sublevelCount = array();
3452                 $levelCount = array();
3453                 $toclevel = 0;
3454                 $level = 0;
3455                 $prevlevel = 0;
3456                 $toclevel = 0;
3457                 $prevtoclevel = 0;
3458
3459                 foreach( $matches[3] as $headline ) {
3460                         $istemplate = 0;
3461                         $templatetitle = '';
3462                         $templatesection = 0;
3463                         $numbering = '';
3464                         $mat = array();
3465                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3466                                 $istemplate = 1;
3467                                 $templatetitle = base64_decode($mat[1]);
3468                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3469                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3470                         }
3471
3472                         if( $toclevel ) {
3473                                 $prevlevel = $level;
3474                                 $prevtoclevel = $toclevel;
3475                         }
3476                         $level = $matches[1][$headlineCount];
3477
3478                         if( $doNumberHeadings || $enoughToc ) {
3479
3480                                 if ( $level > $prevlevel ) {
3481                                         # Increase TOC level
3482                                         $toclevel++;
3483                                         $sublevelCount[$toclevel] = 0;
3484                                         if( $toclevel<$wgMaxTocLevel ) {
3485                                                 $toc .= $sk->tocIndent();
3486                                         }
3487                                 }
3488                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3489                                         # Decrease TOC level, find level to jump to
3490
3491                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3492                                                 # Can only go down to level 1
3493                                                 $toclevel = 1;
3494                                         } else {
3495                                                 for ($i = $toclevel; $i > 0; $i--) {
3496                                                         if ( $levelCount[$i] == $level ) {
3497                                                                 # Found last matching level
3498                                                                 $toclevel = $i;
3499                                                                 break;
3500                                                         }
3501                                                         elseif ( $levelCount[$i] < $level ) {
3502                                                                 # Found first matching level below current level
3503                                                                 $toclevel = $i + 1;
3504                                                                 break;
3505                                                         }
3506                                                 }
3507                                         }
3508                                         if( $toclevel<$wgMaxTocLevel ) {
3509                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3510                                         }
3511                                 }
3512                                 else {
3513                                         # No change in level, end TOC line
3514                                         if( $toclevel<$wgMaxTocLevel ) {
3515                                                 $toc .= $sk->tocLineEnd();
3516                                         }
3517                                 }
3518
3519                                 $levelCount[$toclevel] = $level;
3520
3521                                 # count number of headlines for each level
3522                                 @$sublevelCount[$toclevel]++;
3523                                 $dot = 0;
3524                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3525                                         if( !empty( $sublevelCount[$i] ) ) {
3526                                                 if( $dot ) {
3527                                                         $numbering .= '.';
3528                                                 }
3529                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3530                                                 $dot = 1;
3531                                         }
3532                                 }
3533                         }
3534
3535                         # The canonized header is a version of the header text safe to use for links
3536                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3537                         $canonized_headline = $this->mStripState->unstripBoth( $headline );
3538
3539                         # Remove link placeholders by the link text.
3540                         #     <!--LINK number-->
3541                         # turns into
3542                         #     link text with suffix
3543                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3544                                                             "\$this->mLinkHolders['texts'][\$1]",
3545                                                             $canonized_headline );
3546                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3547                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3548                                                             $canonized_headline );
3549
3550                         # strip out HTML
3551                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3552                         $tocline = trim( $canonized_headline );
3553                         # Save headline for section edit hint before it's escaped
3554                         $headline_hint = trim( $canonized_headline );
3555                         $canonized_headline = Sanitizer::escapeId( $tocline );
3556                         $refers[$headlineCount] = $canonized_headline;
3557
3558                         # count how many in assoc. array so we can track dupes in anchors
3559                         isset( $refers[$canonized_headline] ) ? $refers[$canonized_headline]++ : $refers[$canonized_headline] = 1;
3560                         $refcount[$headlineCount]=$refers[$canonized_headline];
3561
3562                         # Don't number the heading if it is the only one (looks silly)
3563                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3564                                 # the two are different if the line contains a link
3565                                 $headline=$numbering . ' ' . $headline;
3566                         }
3567
3568                         # Create the anchor for linking from the TOC to the section
3569                         $anchor = $canonized_headline;
3570                         if($refcount[$headlineCount] > 1 ) {
3571                                 $anchor .= '_' . $refcount[$headlineCount];
3572                         }
3573                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3574                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3575                         }
3576                         # give headline the correct <h#> tag
3577                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3578                                 if( $istemplate )
3579                                         $editlink = $sk->editSectionLinkForOther($templatetitle, $templatesection);
3580                                 else
3581                                         $editlink = $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3582                         } else {
3583                                 $editlink = '';
3584                         }
3585                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
3586
3587                         $headlineCount++;
3588                         if( !$istemplate )
3589                                 $sectionCount++;
3590                 }
3591
3592                 if( $enoughToc ) {
3593                         if( $toclevel<$wgMaxTocLevel ) {
3594                                 $toc .= $sk->tocUnindent( $toclevel - 1 );
3595                         }
3596                         $toc = $sk->tocList( $toc );
3597                 }
3598
3599                 # split up and insert constructed headlines
3600
3601                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3602                 $i = 0;
3603
3604                 foreach( $blocks as $block ) {
3605                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3606                                 # This is the [edit] link that appears for the top block of text when
3607                                 # section editing is enabled
3608
3609                                 # Disabled because it broke block formatting
3610                                 # For example, a bullet point in the top line
3611                                 # $full .= $sk->editSectionLink(0);
3612                         }
3613                         $full .= $block;
3614                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3615                                 # Top anchor now in skin
3616                                 $full = $full.$toc;
3617                         }
3618
3619                         if( !empty( $head[$i] ) ) {
3620                                 $full .= $head[$i];
3621                         }
3622                         $i++;
3623                 }
3624                 if( $this->mForceTocPosition ) {
3625                         return str_replace( '<!--MWTOC-->', $toc, $full );
3626                 } else {
3627                         return $full;
3628                 }
3629         }
3630
3631         /**
3632          * Transform wiki markup when saving a page by doing \r\n -> \n
3633          * conversion, substitting signatures, {{subst:}} templates, etc.
3634          *
3635          * @param string $text the text to transform
3636          * @param Title &$title the Title object for the current article
3637          * @param User &$user the User object describing the current user
3638          * @param ParserOptions $options parsing options
3639          * @param bool $clearState whether to clear the parser state first
3640          * @return string the altered wiki markup
3641          * @public
3642          */
3643         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3644                 $this->mOptions = $options;
3645                 $this->mTitle =& $title;
3646                 $this->setOutputType( OT_WIKI );
3647
3648                 if ( $clearState ) {
3649                         $this->clearState();
3650                 }
3651
3652                 $stripState = new StripState;
3653                 $pairs = array(
3654                         "\r\n" => "\n",
3655                 );
3656                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3657                 $text = $this->strip( $text, $stripState, true, array( 'gallery' ) );
3658                 $text = $this->pstPass2( $text, $stripState, $user );
3659                 $text = $stripState->unstripBoth( $text );
3660                 return $text;
3661         }
3662
3663         /**
3664          * Pre-save transform helper function
3665          * @private
3666          */
3667         function pstPass2( $text, &$stripState, $user ) {
3668                 global $wgContLang, $wgLocaltimezone;
3669
3670                 /* Note: This is the timestamp saved as hardcoded wikitext to
3671                  * the database, we use $wgContLang here in order to give
3672                  * everyone the same signature and use the default one rather
3673                  * than the one selected in each user's preferences.
3674                  */
3675                 if ( isset( $wgLocaltimezone ) ) {
3676                         $oldtz = getenv( 'TZ' );
3677                         putenv( 'TZ='.$wgLocaltimezone );
3678                 }
3679                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3680                   ' (' . date( 'T' ) . ')';
3681                 if ( isset( $wgLocaltimezone ) ) {
3682                         putenv( 'TZ='.$oldtz );
3683                 }
3684
3685                 # Variable replacement
3686                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3687                 $text = $this->replaceVariables( $text );
3688
3689                 # Strip out <nowiki> etc. added via replaceVariables
3690                 $text = $this->strip( $text, $stripState, false, array( 'gallery' ) );
3691
3692                 # Signatures
3693                 $sigText = $this->getUserSig( $user );
3694                 $text = strtr( $text, array(
3695                         '~~~~~' => $d,
3696                         '~~~~' => "$sigText $d",
3697                         '~~~' => $sigText
3698                 ) );
3699
3700                 # Context links: [[|name]] and [[name (context)|]]
3701                 #
3702                 global $wgLegalTitleChars;
3703                 $tc = "[$wgLegalTitleChars]";
3704                 $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3705
3706                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3707                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3708                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3709
3710                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3711                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3712                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3713
3714                 $t = $this->mTitle->getText();
3715                 $m = array();
3716                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3717                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3718                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3719                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3720                 } else {
3721                         # if there's no context, don't bother duplicating the title
3722                         $text = preg_replace( $p2, '[[\\1]]', $text );
3723                 }
3724
3725                 # Trim trailing whitespace
3726                 $text = rtrim( $text );
3727
3728                 return $text;
3729         }
3730
3731         /**
3732          * Fetch the user's signature text, if any, and normalize to
3733          * validated, ready-to-insert wikitext.
3734          *
3735          * @param User $user
3736          * @return string
3737          * @private
3738          */
3739         function getUserSig( &$user ) {
3740                 $username = $user->getName();
3741                 $nickname = $user->getOption( 'nickname' );
3742                 $nickname = $nickname === '' ? $username : $nickname;
3743
3744                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3745                         # Sig. might contain markup; validate this
3746                         if( $this->validateSig( $nickname ) !== false ) {
3747                                 # Validated; clean up (if needed) and return it
3748                                 return $this->cleanSig( $nickname, true );
3749                         } else {
3750                                 # Failed to validate; fall back to the default
3751                                 $nickname = $username;
3752                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3753                         }
3754                 }
3755
3756                 // Make sure nickname doesnt get a sig in a sig
3757                 $nickname = $this->cleanSigInSig( $nickname );
3758
3759                 # If we're still here, make it a link to the user page
3760                 $userpage = $user->getUserPage();
3761                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3762         }
3763
3764         /**
3765          * Check that the user's signature contains no bad XML
3766          *
3767          * @param string $text
3768          * @return mixed An expanded string, or false if invalid.
3769          */
3770         function validateSig( $text ) {
3771                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3772         }
3773
3774         /**
3775          * Clean up signature text
3776          *
3777          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3778          * 2) Substitute all transclusions
3779          *
3780          * @param string $text
3781          * @param $parsing Whether we're cleaning (preferences save) or parsing
3782          * @return string Signature text
3783          */
3784         function cleanSig( $text, $parsing = false ) {
3785                 global $wgTitle;
3786                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
3787
3788                 $substWord = MagicWord::get( 'subst' );
3789                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3790                 $substText = '{{' . $substWord->getSynonym( 0 );
3791
3792                 $text = preg_replace( $substRegex, $substText, $text );
3793                 $text = $this->cleanSigInSig( $text );
3794                 $text = $this->replaceVariables( $text );
3795
3796                 $this->clearState();
3797                 return $text;
3798         }
3799
3800         /**
3801          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3802          * @param string $text
3803          * @return string Signature text with /~{3,5}/ removed
3804          */
3805         function cleanSigInSig( $text ) {
3806                 $text = preg_replace( '/~{3,5}/', '', $text );
3807                 return $text;
3808         }
3809
3810         /**
3811          * Set up some variables which are usually set up in parse()
3812          * so that an external function can call some class members with confidence
3813          * @public
3814          */
3815         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3816                 $this->mTitle =& $title;
3817                 $this->mOptions = $options;
3818                 $this->setOutputType( $outputType );
3819                 if ( $clearState ) {
3820                         $this->clearState();
3821                 }
3822         }
3823
3824         /**
3825          * Transform a MediaWiki message by replacing magic variables.
3826          *
3827          * @param string $text the text to transform
3828          * @param ParserOptions $options  options
3829          * @return string the text with variables substituted
3830          * @public
3831          */
3832         function transformMsg( $text, $options ) {
3833                 global $wgTitle;
3834                 static $executing = false;
3835
3836                 $fname = "Parser::transformMsg";
3837
3838                 # Guard against infinite recursion
3839                 if ( $executing ) {
3840                         return $text;
3841                 }
3842                 $executing = true;
3843
3844                 wfProfileIn($fname);
3845
3846                 if ( $wgTitle && !( $wgTitle instanceof FakeTitle ) ) {
3847                         $this->mTitle = $wgTitle;
3848                 } else {
3849                         $this->mTitle = Title::newFromText('msg');
3850                 }
3851                 $this->mOptions = $options;
3852                 $this->setOutputType( OT_MSG );
3853                 $this->clearState();
3854                 $text = $this->replaceVariables( $text );
3855
3856                 $executing = false;
3857                 wfProfileOut($fname);
3858                 return $text;
3859         }
3860
3861         /**
3862          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3863          * The callback should have the following form:
3864          *    function myParserHook( $text, $params, &$parser ) { ... }
3865          *
3866          * Transform and return $text. Use $parser for any required context, e.g. use
3867          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3868          *
3869          * @public
3870          *
3871          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3872          * @param mixed $callback The callback function (and object) to use for the tag
3873          *
3874          * @return The old value of the mTagHooks array associated with the hook
3875          */
3876         function setHook( $tag, $callback ) {
3877                 $tag = strtolower( $tag );
3878                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
3879                 $this->mTagHooks[$tag] = $callback;
3880
3881                 return $oldVal;
3882         }
3883
3884         /**
3885          * Create a function, e.g. {{sum:1|2|3}}
3886          * The callback function should have the form:
3887          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3888          *
3889          * The callback may either return the text result of the function, or an array with the text
3890          * in element 0, and a number of flags in the other elements. The names of the flags are
3891          * specified in the keys. Valid flags are:
3892          *   found                     The text returned is valid, stop processing the template. This
3893          *                             is on by default.
3894          *   nowiki                    Wiki markup in the return value should be escaped
3895          *   noparse                   Unsafe HTML tags should not be stripped, etc.
3896          *   noargs                    Don't replace triple-brace arguments in the return value
3897          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3898          *
3899          * @public
3900          *
3901          * @param string $id The magic word ID
3902          * @param mixed $callback The callback function (and object) to use
3903          * @param integer $flags a combination of the following flags:
3904          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
3905          *
3906          * @return The old callback function for this name, if any
3907          */
3908         function setFunctionHook( $id, $callback, $flags = 0 ) {
3909                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id] : null;
3910                 $this->mFunctionHooks[$id] = $callback;
3911
3912                 # Add to function cache
3913                 $mw = MagicWord::get( $id );
3914                 if( !$mw )
3915                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
3916
3917                 $synonyms = $mw->getSynonyms();
3918                 $sensitive = intval( $mw->isCaseSensitive() );
3919
3920                 foreach ( $synonyms as $syn ) {
3921                         # Case
3922                         if ( !$sensitive ) {
3923                                 $syn = strtolower( $syn );
3924                         }
3925                         # Add leading hash
3926                         if ( !( $flags & SFH_NO_HASH ) ) {
3927                                 $syn = '#' . $syn;
3928                         }
3929                         # Remove trailing colon
3930                         if ( substr( $syn, -1, 1 ) == ':' ) {
3931                                 $syn = substr( $syn, 0, -1 );
3932                         }
3933                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
3934                 }
3935                 return $oldVal;
3936         }
3937
3938         /**
3939          * Get all registered function hook identifiers
3940          *
3941          * @return array
3942          */
3943         function getFunctionHooks() {
3944                 return array_keys( $this->mFunctionHooks );
3945         }
3946
3947         /**
3948          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3949          * Placeholders created in Skin::makeLinkObj()
3950          * Returns an array of links found, indexed by PDBK:
3951          *  0 - broken
3952          *  1 - normal link
3953          *  2 - stub
3954          * $options is a bit field, RLH_FOR_UPDATE to select for update
3955          */
3956         function replaceLinkHolders( &$text, $options = 0 ) {
3957                 global $wgUser;
3958                 global $wgContLang;
3959
3960                 $fname = 'Parser::replaceLinkHolders';
3961                 wfProfileIn( $fname );
3962
3963                 $pdbks = array();
3964                 $colours = array();
3965                 $sk =& $this->mOptions->getSkin();
3966                 $linkCache =& LinkCache::singleton();
3967
3968                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3969                         wfProfileIn( $fname.'-check' );
3970                         $dbr =& wfGetDB( DB_SLAVE );
3971                         $page = $dbr->tableName( 'page' );
3972                         $threshold = $wgUser->getOption('stubthreshold');
3973
3974                         # Sort by namespace
3975                         asort( $this->mLinkHolders['namespaces'] );
3976
3977                         # Generate query
3978                         $query = false;
3979                         $current = null;
3980                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3981                                 # Make title object
3982                                 $title = $this->mLinkHolders['titles'][$key];
3983
3984                                 # Skip invalid entries.
3985                                 # Result will be ugly, but prevents crash.
3986                                 if ( is_null( $title ) ) {
3987                                         continue;
3988                                 }
3989                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3990
3991                                 # Check if it's a static known link, e.g. interwiki
3992                                 if ( $title->isAlwaysKnown() ) {
3993                                         $colours[$pdbk] = 1;
3994                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3995                                         $colours[$pdbk] = 1;
3996                                         $this->mOutput->addLink( $title, $id );
3997                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3998                                         $colours[$pdbk] = 0;
3999                                 } else {
4000                                         # Not in the link cache, add it to the query
4001                                         if ( !isset( $current ) ) {
4002                                                 $current = $ns;
4003                                                 $query =  "SELECT page_id, page_namespace, page_title";
4004                                                 if ( $threshold > 0 ) {
4005                                                         $query .= ', page_len, page_is_redirect';
4006                                                 }
4007                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
4008                                         } elseif ( $current != $ns ) {
4009                                                 $current = $ns;
4010                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
4011                                         } else {
4012                                                 $query .= ', ';
4013                                         }
4014
4015                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
4016                                 }
4017                         }
4018                         if ( $query ) {
4019                                 $query .= '))';
4020                                 if ( $options & RLH_FOR_UPDATE ) {
4021                                         $query .= ' FOR UPDATE';
4022                                 }
4023
4024                                 $res = $dbr->query( $query, $fname );
4025
4026                                 # Fetch data and form into an associative array
4027                                 # non-existent = broken
4028                                 # 1 = known
4029                                 # 2 = stub
4030                                 while ( $s = $dbr->fetchObject($res) ) {
4031                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4032                                         $pdbk = $title->getPrefixedDBkey();
4033                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
4034                                         $this->mOutput->addLink( $title, $s->page_id );
4035
4036                                         if ( $threshold >  0 ) {
4037                                                 $size = $s->page_len;
4038                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
4039                                                         $colours[$pdbk] = 1;
4040                                                 } else {
4041                                                         $colours[$pdbk] = 2;
4042                                                 }
4043                                         } else {
4044                                                 $colours[$pdbk] = 1;
4045                                         }
4046                                 }
4047                         }
4048                         wfProfileOut( $fname.'-check' );
4049
4050                         # Do a second query for different language variants of links and categories
4051                         if($wgContLang->hasVariants()){
4052                                 $linkBatch = new LinkBatch();
4053                                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
4054                                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
4055                                 $varCategories = array(); // category replacements oldDBkey => newDBkey
4056
4057                                 $categories = $this->mOutput->getCategoryLinks();
4058
4059                                 // Add variants of links to link batch
4060                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4061                                         $title = $this->mLinkHolders['titles'][$key];
4062                                         if ( is_null( $title ) )
4063                                                 continue;
4064
4065                                         $pdbk = $title->getPrefixedDBkey();
4066                                         $titleText = $title->getText();
4067
4068                                         // generate all variants of the link title text
4069                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
4070
4071                                         // if link was not found (in first query), add all variants to query
4072                                         if ( !isset($colours[$pdbk]) ){
4073                                                 foreach($allTextVariants as $textVariant){
4074                                                         if($textVariant != $titleText){
4075                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
4076                                                                 if(is_null($variantTitle)) continue;
4077                                                                 $linkBatch->addObj( $variantTitle );
4078                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4079                                                         }
4080                                                 }
4081                                         }
4082                                 }
4083
4084                                 // process categories, check if a category exists in some variant
4085                                 foreach( $categories as $category){
4086                                         $variants = $wgContLang->convertLinkToAllVariants($category);
4087                                         foreach($variants as $variant){
4088                                                 if($variant != $category){
4089                                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
4090                                                         if(is_null($variantTitle)) continue;
4091                                                         $linkBatch->addObj( $variantTitle );
4092                                                         $categoryMap[$variant] = $category;
4093                                                 }
4094                                         }
4095                                 }
4096
4097
4098                                 if(!$linkBatch->isEmpty()){
4099                                         // construct query
4100                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4101
4102                                         $variantQuery =  "SELECT page_id, page_namespace, page_title";
4103                                         if ( $threshold > 0 ) {
4104                                                 $variantQuery .= ', page_len, page_is_redirect';
4105                                         }
4106
4107                                         $variantQuery .= " FROM $page WHERE $titleClause";
4108                                         if ( $options & RLH_FOR_UPDATE ) {
4109                                                 $variantQuery .= ' FOR UPDATE';
4110                                         }
4111
4112                                         $varRes = $dbr->query( $variantQuery, $fname );
4113
4114                                         // for each found variants, figure out link holders and replace
4115                                         while ( $s = $dbr->fetchObject($varRes) ) {
4116
4117                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4118                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4119                                                 $vardbk = $variantTitle->getDBkey();
4120
4121                                                 $holderKeys = array();
4122                                                 if(isset($variantMap[$varPdbk])){
4123                                                         $holderKeys = $variantMap[$varPdbk];
4124                                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle );
4125                                                         $this->mOutput->addLink( $variantTitle, $s->page_id );
4126                                                 }
4127
4128                                                 // loop over link holders
4129                                                 foreach($holderKeys as $key){
4130                                                         $title = $this->mLinkHolders['titles'][$key];
4131                                                         if ( is_null( $title ) ) continue;
4132
4133                                                         $pdbk = $title->getPrefixedDBkey();
4134
4135                                                         if(!isset($colours[$pdbk])){
4136                                                                 // found link in some of the variants, replace the link holder data
4137                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4138                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4139
4140                                                                 // set pdbk and colour
4141                                                                 $pdbks[$key] = $varPdbk;
4142                                                                 if ( $threshold >  0 ) {
4143                                                                         $size = $s->page_len;
4144                                                                         if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
4145                                                                                 $colours[$varPdbk] = 1;
4146                                                                         } else {
4147                                                                                 $colours[$varPdbk] = 2;
4148                                                                         }
4149                                                                 }
4150                                                                 else {
4151                                                                         $colours[$varPdbk] = 1;
4152                                                                 }
4153                                                         }
4154                                                 }
4155
4156                                                 // check if the object is a variant of a category
4157                                                 if(isset($categoryMap[$vardbk])){
4158                                                         $oldkey = $categoryMap[$vardbk];
4159                                                         if($oldkey != $vardbk)
4160                                                                 $varCategories[$oldkey]=$vardbk;
4161                                                 }
4162                                         }
4163
4164                                         // rebuild the categories in original order (if there are replacements)
4165                                         if(count($varCategories)>0){
4166                                                 $newCats = array();
4167                                                 $originalCats = $this->mOutput->getCategories();
4168                                                 foreach($originalCats as $cat => $sortkey){
4169                                                         // make the replacement
4170                                                         if( array_key_exists($cat,$varCategories) )
4171                                                                 $newCats[$varCategories[$cat]] = $sortkey;
4172                                                         else $newCats[$cat] = $sortkey;
4173                                                 }
4174                                                 $this->mOutput->setCategoryLinks($newCats);
4175                                         }
4176                                 }
4177                         }
4178
4179                         # Construct search and replace arrays
4180                         wfProfileIn( $fname.'-construct' );
4181                         $replacePairs = array();
4182                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4183                                 $pdbk = $pdbks[$key];
4184                                 $searchkey = "<!--LINK $key-->";
4185                                 $title = $this->mLinkHolders['titles'][$key];
4186                                 if ( empty( $colours[$pdbk] ) ) {
4187                                         $linkCache->addBadLinkObj( $title );
4188                                         $colours[$pdbk] = 0;
4189                                         $this->mOutput->addLink( $title, 0 );
4190                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4191                                                                         $this->mLinkHolders['texts'][$key],
4192                                                                         $this->mLinkHolders['queries'][$key] );
4193                                 } elseif ( $colours[$pdbk] == 1 ) {
4194                                         $replacePairs[$searchkey] = $sk->makeKnownLinkObj( $title,
4195                                                                         $this->mLinkHolders['texts'][$key],
4196                                                                         $this->mLinkHolders['queries'][$key] );
4197                                 } elseif ( $colours[$pdbk] == 2 ) {
4198                                         $replacePairs[$searchkey] = $sk->makeStubLinkObj( $title,
4199                                                                         $this->mLinkHolders['texts'][$key],
4200                                                                         $this->mLinkHolders['queries'][$key] );
4201                                 }
4202                         }
4203                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4204                         wfProfileOut( $fname.'-construct' );
4205
4206                         # Do the thing
4207                         wfProfileIn( $fname.'-replace' );
4208                         $text = preg_replace_callback(
4209                                 '/(<!--LINK .*?-->)/',
4210                                 $replacer->cb(),
4211                                 $text);
4212
4213                         wfProfileOut( $fname.'-replace' );
4214                 }
4215
4216                 # Now process interwiki link holders
4217                 # This is quite a bit simpler than internal links
4218                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4219                         wfProfileIn( $fname.'-interwiki' );
4220                         # Make interwiki link HTML
4221                         $replacePairs = array();
4222                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4223                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4224                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4225                         }
4226                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4227
4228                         $text = preg_replace_callback(
4229                                 '/<!--IWLINK (.*?)-->/',
4230                                 $replacer->cb(),
4231                                 $text );
4232                         wfProfileOut( $fname.'-interwiki' );
4233                 }
4234
4235                 wfProfileOut( $fname );
4236                 return $colours;
4237         }
4238
4239         /**
4240          * Replace <!--LINK--> link placeholders with plain text of links
4241          * (not HTML-formatted).
4242          * @param string $text
4243          * @return string
4244          */
4245         function replaceLinkHoldersText( $text ) {
4246                 $fname = 'Parser::replaceLinkHoldersText';
4247                 wfProfileIn( $fname );
4248
4249                 $text = preg_replace_callback(
4250                         '/<!--(LINK|IWLINK) (.*?)-->/',
4251                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4252                         $text );
4253
4254                 wfProfileOut( $fname );
4255                 return $text;
4256         }
4257
4258         /**
4259          * @param array $matches
4260          * @return string
4261          * @private
4262          */
4263         function replaceLinkHoldersTextCallback( $matches ) {
4264                 $type = $matches[1];
4265                 $key  = $matches[2];
4266                 if( $type == 'LINK' ) {
4267                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4268                                 return $this->mLinkHolders['texts'][$key];
4269                         }
4270                 } elseif( $type == 'IWLINK' ) {
4271                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4272                                 return $this->mInterwikiLinkHolders['texts'][$key];
4273                         }
4274                 }
4275                 return $matches[0];
4276         }
4277
4278         /**
4279          * Tag hook handler for 'pre'.
4280          */
4281         function renderPreTag( $text, $attribs ) {
4282                 // Backwards-compatibility hack
4283                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4284
4285                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4286                 return wfOpenElement( 'pre', $attribs ) .
4287                         Xml::escapeTagsOnly( $content ) .
4288                         '</pre>';
4289         }
4290
4291         /**
4292          * Renders an image gallery from a text with one line per image.
4293          * text labels may be given by using |-style alternative text. E.g.
4294          *   Image:one.jpg|The number "1"
4295          *   Image:tree.jpg|A tree
4296          * given as text will return the HTML of a gallery with two images,
4297          * labeled 'The number "1"' and
4298          * 'A tree'.
4299          */
4300         function renderImageGallery( $text, $params ) {
4301                 $ig = new ImageGallery();
4302                 $ig->setShowBytes( false );
4303                 $ig->setShowFilename( false );
4304                 $ig->setParsing();
4305                 $ig->useSkin( $this->mOptions->getSkin() );
4306
4307                 if( isset( $params['caption'] ) ) {
4308                         $caption = $params['caption'];
4309                         $caption = htmlspecialchars( $caption );
4310                         $caption = $this->replaceInternalLinks( $caption );
4311                         $ig->setCaptionHtml( $caption );
4312                 }
4313
4314                 $lines = explode( "\n", $text );
4315                 foreach ( $lines as $line ) {
4316                         # match lines like these:
4317                         # Image:someimage.jpg|This is some image
4318                         $matches = array();
4319                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4320                         # Skip empty lines
4321                         if ( count( $matches ) == 0 ) {
4322                                 continue;
4323                         }
4324                         $tp = Title::newFromText( $matches[1] );
4325                         $nt =& $tp;
4326                         if( is_null( $nt ) ) {
4327                                 # Bogus title. Ignore these so we don't bomb out later.
4328                                 continue;
4329                         }
4330                         if ( isset( $matches[3] ) ) {
4331                                 $label = $matches[3];
4332                         } else {
4333                                 $label = '';
4334                         }
4335
4336                         $pout = $this->parse( $label,
4337                                 $this->mTitle,
4338                                 $this->mOptions,
4339                                 false, // Strip whitespace...?
4340                                 false  // Don't clear state!
4341                         );
4342                         $html = $pout->getText();
4343
4344                         $ig->add( new Image( $nt ), $html );
4345
4346                         # Only add real images (bug #5586)
4347                         if ( $nt->getNamespace() == NS_IMAGE ) {
4348                                 $this->mOutput->addImage( $nt->getDBkey() );
4349                         }
4350                 }
4351                 return $ig->toHTML();
4352         }
4353
4354         /**
4355          * Parse image options text and use it to make an image
4356          */
4357         function makeImage( $nt, $options ) {
4358                 global $wgUseImageResize, $wgDjvuRenderer;
4359
4360                 $align = '';
4361
4362                 # Check if the options text is of the form "options|alt text"
4363                 # Options are:
4364                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4365                 #  * left               no resizing, just left align. label is used for alt= only
4366                 #  * right              same, but right aligned
4367                 #  * none               same, but not aligned
4368                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4369                 #  * center             center the image
4370                 #  * framed             Keep original image size, no magnify-button.
4371
4372                 $part = explode( '|', $options);
4373
4374                 $mwThumb  =& MagicWord::get( 'img_thumbnail' );
4375                 $mwManualThumb =& MagicWord::get( 'img_manualthumb' );
4376                 $mwLeft   =& MagicWord::get( 'img_left' );
4377                 $mwRight  =& MagicWord::get( 'img_right' );
4378                 $mwNone   =& MagicWord::get( 'img_none' );
4379                 $mwWidth  =& MagicWord::get( 'img_width' );
4380                 $mwCenter =& MagicWord::get( 'img_center' );
4381                 $mwFramed =& MagicWord::get( 'img_framed' );
4382                 $mwPage   =& MagicWord::get( 'img_page' );
4383                 $caption = '';
4384
4385                 $width = $height = $framed = $thumb = false;
4386                 $page = null;
4387                 $manual_thumb = '' ;
4388
4389                 foreach( $part as $val ) {
4390                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
4391                                 $thumb=true;
4392                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
4393                                 # use manually specified thumbnail
4394                                 $thumb=true;
4395                                 $manual_thumb = $match;
4396                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
4397                                 # remember to set an alignment, don't render immediately
4398                                 $align = 'right';
4399                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
4400                                 # remember to set an alignment, don't render immediately
4401                                 $align = 'left';
4402                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
4403                                 # remember to set an alignment, don't render immediately
4404                                 $align = 'center';
4405                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
4406                                 # remember to set an alignment, don't render immediately
4407                                 $align = 'none';
4408                         } elseif ( isset( $wgDjvuRenderer ) && $wgDjvuRenderer
4409                                    && ! is_null( $match = $mwPage->matchVariableStartToEnd($val) ) ) {
4410                                 # Select a page in a multipage document
4411                                 $page = $match;
4412                         } elseif ( $wgUseImageResize && !$width && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
4413                                 wfDebug( "img_width match: $match\n" );
4414                                 # $match is the image width in pixels
4415                                 $m = array();
4416                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
4417                                         $width = intval( $m[1] );
4418                                         $height = intval( $m[2] );
4419                                 } else {
4420                                         $width = intval($match);
4421                                 }
4422                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
4423                                 $framed=true;
4424                         } else {
4425                                 $caption = $val;
4426                         }
4427                 }
4428                 # Strip bad stuff out of the alt text
4429                 $alt = $this->replaceLinkHoldersText( $caption );
4430
4431                 # make sure there are no placeholders in thumbnail attributes
4432                 # that are later expanded to html- so expand them now and
4433                 # remove the tags
4434                 $alt = $this->mStripState->unstripBoth( $alt );
4435                 $alt = Sanitizer::stripAllTags( $alt );
4436
4437                 # Linker does the rest
4438                 $sk =& $this->mOptions->getSkin();
4439                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb, $page );
4440         }
4441
4442         /**
4443          * Set a flag in the output object indicating that the content is dynamic and
4444          * shouldn't be cached.
4445          */
4446         function disableCache() {
4447                 wfDebug( "Parser output marked as uncacheable.\n" );
4448                 $this->mOutput->mCacheTime = -1;
4449         }
4450
4451         /**#@+
4452          * Callback from the Sanitizer for expanding items found in HTML attribute
4453          * values, so they can be safely tested and escaped.
4454          * @param string $text
4455          * @param array $args
4456          * @return string
4457          * @private
4458          */
4459         function attributeStripCallback( &$text, $args ) {
4460                 $text = $this->replaceVariables( $text, $args );
4461                 $text = $this->mStripState->unstripBoth( $text );
4462                 return $text;
4463         }
4464
4465         /**#@-*/
4466
4467         /**#@+
4468          * Accessor/mutator
4469          */
4470         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4471         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4472         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4473         /**#@-*/
4474
4475         /**#@+
4476          * Accessor
4477          */
4478         function getTags() { return array_keys( $this->mTagHooks ); }
4479         /**#@-*/
4480
4481
4482         /**
4483          * Break wikitext input into sections, and either pull or replace
4484          * some particular section's text.
4485          *
4486          * External callers should use the getSection and replaceSection methods.
4487          *
4488          * @param $text Page wikitext
4489          * @param $section Numbered section. 0 pulls the text before the first
4490          *                 heading; other numbers will pull the given section
4491          *                 along with its lower-level subsections.
4492          * @param $mode One of "get" or "replace"
4493          * @param $newtext Replacement text for section data.
4494          * @return string for "get", the extracted section text.
4495          *                for "replace", the whole page with the section replaced.
4496          */
4497         private function extractSections( $text, $section, $mode, $newtext='' ) {
4498                 # strip NOWIKI etc. to avoid confusion (true-parameter causes HTML
4499                 # comments to be stripped as well)
4500                 $stripState = new StripState;
4501
4502                 $oldOutputType = $this->mOutputType;
4503                 $oldOptions = $this->mOptions;
4504                 $this->mOptions = new ParserOptions();
4505                 $this->setOutputType( OT_WIKI );
4506
4507                 $striptext = $this->strip( $text, $stripState, true );
4508
4509                 $this->setOutputType( $oldOutputType );
4510                 $this->mOptions = $oldOptions;
4511
4512                 # now that we can be sure that no pseudo-sections are in the source,
4513                 # split it up by section
4514                 $uniq = preg_quote( $this->uniqPrefix(), '/' );
4515                 $comment = "(?:$uniq-!--.*?QINU)";
4516                 $secs = preg_split(
4517                 /*
4518                         "/
4519                         ^(
4520                         (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4521                         (?:
4522                                 (=+) # Should this be limited to 6?
4523                                 .+?  # Section title...
4524                                 \\2  # Ending = count must match start
4525                         |
4526                                 ^
4527                                 <h([1-6])\b.*?>
4528                                 .*?
4529                                 <\/h\\3\s*>
4530                         )
4531                         (?:$comment|<\/?noinclude>|\s+)* # Trailing whitespace ok
4532                         )$
4533                         /mix",
4534                 */
4535                         "/
4536                         (
4537                                 ^
4538                                 (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4539                                 (=+) # Should this be limited to 6?
4540                                 .+?  # Section title...
4541                                 \\2  # Ending = count must match start
4542                                 (?:$comment|<\/?noinclude>|[ \\t]+)* # Trailing whitespace ok
4543                                 $
4544                         |
4545                                 <h([1-6])\b.*?>
4546                                 .*?
4547                                 <\/h\\3\s*>
4548                         )
4549                         /mix",
4550                         $striptext, -1,
4551                         PREG_SPLIT_DELIM_CAPTURE);
4552
4553                 if( $mode == "get" ) {
4554                         if( $section == 0 ) {
4555                                 // "Section 0" returns the content before any other section.
4556                                 $rv = $secs[0];
4557                         } else {
4558                                 $rv = "";
4559                         }
4560                 } elseif( $mode == "replace" ) {
4561                         if( $section == 0 ) {
4562                                 $rv = $newtext . "\n\n";
4563                                 $remainder = true;
4564                         } else {
4565                                 $rv = $secs[0];
4566                                 $remainder = false;
4567                         }
4568                 }
4569                 $count = 0;
4570                 $sectionLevel = 0;
4571                 for( $index = 1; $index < count( $secs ); ) {
4572                         $headerLine = $secs[$index++];
4573                         if( $secs[$index] ) {
4574                                 // A wiki header
4575                                 $headerLevel = strlen( $secs[$index++] );
4576                         } else {
4577                                 // An HTML header
4578                                 $index++;
4579                                 $headerLevel = intval( $secs[$index++] );
4580                         }
4581                         $content = $secs[$index++];
4582
4583                         $count++;
4584                         if( $mode == "get" ) {
4585                                 if( $count == $section ) {
4586                                         $rv = $headerLine . $content;
4587                                         $sectionLevel = $headerLevel;
4588                                 } elseif( $count > $section ) {
4589                                         if( $sectionLevel && $headerLevel > $sectionLevel ) {
4590                                                 $rv .= $headerLine . $content;
4591                                         } else {
4592                                                 // Broke out to a higher-level section
4593                                                 break;
4594                                         }
4595                                 }
4596                         } elseif( $mode == "replace" ) {
4597                                 if( $count < $section ) {
4598                                         $rv .= $headerLine . $content;
4599                                 } elseif( $count == $section ) {
4600                                         $rv .= $newtext . "\n\n";
4601                                         $sectionLevel = $headerLevel;
4602                                 } elseif( $count > $section ) {
4603                                         if( $headerLevel <= $sectionLevel ) {
4604                                                 // Passed the section's sub-parts.
4605                                                 $remainder = true;
4606                                         }
4607                                         if( $remainder ) {
4608                                                 $rv .= $headerLine . $content;
4609                                         }
4610                                 }
4611                         }
4612                 }
4613                 # reinsert stripped tags
4614                 $rv = trim( $stripState->unstripBoth( $rv ) );
4615                 return $rv;
4616         }
4617
4618         /**
4619          * This function returns the text of a section, specified by a number ($section).
4620          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4621          * the first section before any such heading (section 0).
4622          *
4623          * If a section contains subsections, these are also returned.
4624          *
4625          * @param $text String: text to look in
4626          * @param $section Integer: section number
4627          * @return string text of the requested section
4628          */
4629         function getSection( $text, $section ) {
4630                 return $this->extractSections( $text, $section, "get" );
4631         }
4632
4633         function replaceSection( $oldtext, $section, $text ) {
4634                 return $this->extractSections( $oldtext, $section, "replace", $text );
4635         }
4636
4637         /**
4638          * Get the timestamp associated with the current revision, adjusted for
4639          * the default server-local timestamp
4640          */
4641         function getRevisionTimestamp() {
4642                 if ( is_null( $this->mRevisionTimestamp ) ) {
4643                         wfProfileIn( __METHOD__ );
4644                         global $wgContLang;
4645                         $dbr =& wfGetDB( DB_SLAVE );
4646                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4647                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4648
4649                         // Normalize timestamp to internal MW format for timezone processing.
4650                         // This has the added side-effect of replacing a null value with
4651                         // the current time, which gives us more sensible behavior for
4652                         // previews.
4653                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4654
4655                         // The cryptic '' timezone parameter tells to use the site-default
4656                         // timezone offset instead of the user settings.
4657                         //
4658                         // Since this value will be saved into the parser cache, served
4659                         // to other users, and potentially even used inside links and such,
4660                         // it needs to be consistent for all visitors.
4661                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4662
4663                         wfProfileOut( __METHOD__ );
4664                 }
4665                 return $this->mRevisionTimestamp;
4666         }
4667
4668         /**
4669          * Mutator for $mDefaultSort
4670          *
4671          * @param $sort New value
4672          */
4673         public function setDefaultSort( $sort ) {
4674                 $this->mDefaultSort = $sort;
4675         }
4676
4677         /**
4678          * Accessor for $mDefaultSort
4679          * Will use the title/prefixed title if none is set
4680          *
4681          * @return string
4682          */
4683         public function getDefaultSort() {
4684                 if( $this->mDefaultSort !== false ) {
4685                         return $this->mDefaultSort;
4686                 } else {
4687                         return $this->mTitle->getNamespace() == NS_CATEGORY
4688                                         ? $this->mTitle->getText()
4689                                         : $this->mTitle->getPrefixedText();
4690                 }
4691         }
4692
4693 }
4694
4695 class OnlyIncludeReplacer {
4696         var $output = '';
4697
4698         function replace( $matches ) {
4699                 if ( substr( $matches[1], -1 ) == "\n" ) {
4700                         $this->output .= substr( $matches[1], 0, -1 );
4701                 } else {
4702                         $this->output .= $matches[1];
4703                 }
4704         }
4705 }
4706
4707 class StripState {
4708         var $general, $nowiki;
4709
4710         function __construct() {
4711                 $this->general = new ReplacementArray;
4712                 $this->nowiki = new ReplacementArray;
4713         }
4714
4715         function unstripGeneral( $text ) {
4716                 wfProfileIn( __METHOD__ );
4717                 $text = $this->general->replace( $text );
4718                 wfProfileOut( __METHOD__ );
4719                 return $text;
4720         }
4721
4722         function unstripNoWiki( $text ) {
4723                 wfProfileIn( __METHOD__ );
4724                 $text = $this->nowiki->replace( $text );
4725                 wfProfileOut( __METHOD__ );
4726                 return $text;
4727         }
4728
4729         function unstripBoth( $text ) {
4730                 wfProfileIn( __METHOD__ );
4731                 $text = $this->general->replace( $text );
4732                 $text = $this->nowiki->replace( $text );
4733                 wfProfileOut( __METHOD__ );
4734                 return $text;
4735         }
4736 }
4737
4738 ?>