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          * @private
 266          * @param string $text Text we want to parse
 267          * @param Title &$title A title object
 268          * @param array $options
 269          * @param boolean $linestart
 270          * @param boolean $clearState
 271          * @param int $revid number to pass in {{REVISIONID}}
 272          * @return ParserOutput a ParserOutput
 273          */
 274         function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 275                 /**
 276                  * First pass--just handle <nowiki> sections, pass the rest off
 277                  * to internalParse() which does all the real work.
 278                  */
 279
 280                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 281                 $fname = 'Parser::parse-' . wfGetCaller();
 282                 wfProfileIn( __METHOD__ );
 283                 wfProfileIn( $fname );
 284
 285                 if ( $clearState ) {
 286                         $this->clearState();
 287                 }
 288
 289                 $this->mOptions = $options;
 290                 $this->mTitle =& $title;
 291                 $oldRevisionId = $this->mRevisionId;
 292                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 293                 if( $revid !== null ) {
 294                         $this->mRevisionId = $revid;
 295                         $this->mRevisionTimestamp = null;
 296                 }
 297                 $this->setOutputType( OT_HTML );
 298                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 299                 $text = $this->strip( $text, $this->mStripState );
 300                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 301                 $text = $this->internalParse( $text );
 302                 $text = $this->mStripState->unstripGeneral( $text );
 303
 304                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 305                 $fixtags = array(
 306                         # french spaces, last one Guillemet-left
 307                         # only if there is something before the space
 308                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 309                         # french spaces, Guillemet-right
 310                         '/(\\302\\253) /' => '\\1&nbsp;',
 311                 );
 312                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 313
 314                 # only once and last
 315                 $text = $this->doBlockLevels( $text, $linestart );
 316
 317                 $this->replaceLinkHolders( $text );
 318
 319                 # the position of the parserConvert() call should not be changed. it
 320                 # assumes that the links are all replaced and the only thing left
 321                 # is the <nowiki> mark.
 322                 # Side-effects: this calls $this->mOutput->setTitleText()
 323                 $text = $wgContLang->parserConvert( $text, $this );
 324
 325                 $text = $this->mStripState->unstripNoWiki( $text );
 326
 327                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 328
 329                 $text = Sanitizer::normalizeCharReferences( $text );
 330
 331                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 332                         $text = Parser::tidy($text);
 333                 } else {
 334                         # attempt to sanitize at least some nesting problems
 335                         # (bug #2702 and quite a few others)
 336                         $tidyregs = array(
 337                                 # ''Something [http://www.cool.com cool''] -->
 338                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 339                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 340                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 341                                 # fix up an anchor inside another anchor, only
 342                                 # at least for a single single nested link (bug 3695)
 343                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 344                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 345                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 346                                 # contains a div, so fix it up here; replace
 347                                 # div with escaped text
 348                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 349                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 350                                 # remove empty italic or bold tag pairs, some
 351                                 # introduced by rules above
 352                                 '/<([bi])><\/\\1>/' => '',
 353                         );
 354
 355                         $text = preg_replace(
 356                                 array_keys( $tidyregs ),
 357                                 array_values( $tidyregs ),
 358                                 $text );
 359                 }
 360
 361                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 362
 363                 # Information on include size limits, for the benefit of users who try to skirt them
 364                 if ( max( $this->mIncludeSizes ) > 1000 ) {
 365                         $max = $this->mOptions->getMaxIncludeSize();
 366                         $text .= "<!-- \n" .
 367                                 "Pre-expand include size: {$this->mIncludeSizes['pre-expand']} bytes\n" .
 368                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']} bytes\n" .
 369                                 "Template argument size: {$this->mIncludeSizes['arg']} bytes\n" .
 370                                 "Maximum: $max bytes\n" .
 371                                 "-->\n";
 372                 }
 373                 $this->mOutput->setText( $text );
 374                 $this->mRevisionId = $oldRevisionId;
 375                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 376                 wfProfileOut( $fname );
 377                 wfProfileOut( __METHOD__ );
 378
 379                 return $this->mOutput;
 380         }
 381
 382         /**
 383          * Recursive parser entry point that can be called from an extension tag
 384          * hook.
 385          */
 386         function recursiveTagParse( $text ) {
 387                 wfProfileIn( __METHOD__ );
 388                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 389                 $text = $this->strip( $text, $this->mStripState );
 390                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 391                 $text = $this->internalParse( $text );
 392                 wfProfileOut( __METHOD__ );
 393                 return $text;
 394         }
 395
 396         /**
 397          * Expand templates and variables in the text, producing valid, static wikitext.
 398          * Also removes comments.
 399          */
 400         function preprocess( $text, $title, $options ) {
 401                 wfProfileIn( __METHOD__ );
 402                 $this->clearState();
 403                 $this->setOutputType( OT_PREPROCESS );
 404                 $this->mOptions = $options;
 405                 $this->mTitle = $title;
 406                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 407                 $text = $this->strip( $text, $this->mStripState );
 408                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 409                 if ( $this->mOptions->getRemoveComments() ) {
 410                         $text = Sanitizer::removeHTMLcomments( $text );
 411                 }
 412                 $text = $this->replaceVariables( $text );
 413                 $text = $this->mStripState->unstripBoth( $text );
 414                 wfProfileOut( __METHOD__ );
 415                 return $text;
 416         }
 417
 418         /**
 419          * Get a random string
 420          *
 421          * @private
 422          * @static
 423          */
 424         function getRandomString() {
 425                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 426         }
 427
 428         function &getTitle() { return $this->mTitle; }
 429         function getOptions() { return $this->mOptions; }
 430
 431         function getFunctionLang() {
 432                 global $wgLang, $wgContLang;
 433                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 434         }
 435
 436         /**
 437          * Replaces all occurrences of HTML-style comments and the given tags
 438          * in the text with a random marker and returns teh next text. The output
 439          * parameter $matches will be an associative array filled with data in
 440          * the form:
 441          *   'UNIQ-xxxxx' => array(
 442          *     'element',
 443          *     'tag content',
 444          *     array( 'param' => 'x' ),
 445          *     '<element param="x">tag content</element>' ) )
 446          *
 447          * @param $elements list of element names. Comments are always extracted.
 448          * @param $text Source text string.
 449          * @param $uniq_prefix
 450          *
 451          * @private
 452          * @static
 453          */
 454         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 455                 static $n = 1;
 456                 $stripped = '';
 457                 $matches = array();
 458
 459                 $taglist = implode( '|', $elements );
 460                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 461
 462                 while ( '' != $text ) {
 463                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 464                         $stripped .= $p[0];
 465                         if( count( $p ) < 5 ) {
 466                                 break;
 467                         }
 468                         if( count( $p ) > 5 ) {
 469                                 // comment
 470                                 $element    = $p[4];
 471                                 $attributes = '';
 472                                 $close      = '';
 473                                 $inside     = $p[5];
 474                         } else {
 475                                 // tag
 476                                 $element    = $p[1];
 477                                 $attributes = $p[2];
 478                                 $close      = $p[3];
 479                                 $inside     = $p[4];
 480                         }
 481
 482                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . '-QINU';
 483                         $stripped .= $marker;
 484
 485                         if ( $close === '/>' ) {
 486                                 // Empty element tag, <tag />
 487                                 $content = null;
 488                                 $text = $inside;
 489                                 $tail = null;
 490                         } else {
 491                                 if( $element == '!--' ) {
 492                                         $end = '/(-->)/';
 493                                 } else {
 494                                         $end = "/(<\\/$element\\s*>)/i";
 495                                 }
 496                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 497                                 $content = $q[0];
 498                                 if( count( $q ) < 3 ) {
 499                                         # No end tag -- let it run out to the end of the text.
 500                                         $tail = '';
 501                                         $text = '';
 502                                 } else {
 503                                         $tail = $q[1];
 504                                         $text = $q[2];
 505                                 }
 506                         }
 507
 508                         $matches[$marker] = array( $element,
 509                                 $content,
 510                                 Sanitizer::decodeTagAttributes( $attributes ),
 511                                 "<$element$attributes$close$content$tail" );
 512                 }
 513                 return $stripped;
 514         }
 515
 516         /**
 517          * Strips and renders nowiki, pre, math, hiero
 518          * If $render is set, performs necessary rendering operations on plugins
 519          * Returns the text, and fills an array with data needed in unstrip()
 520          *
 521          * @param StripState $state
 522          *
 523          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 524          *  will be stripped in addition to other tags. This is important
 525          *  for section editing, where these comments cause confusion when
 526          *  counting the sections in the wikisource
 527          *
 528          * @param array dontstrip contains tags which should not be stripped;
 529          *  used to prevent stipping of <gallery> when saving (fixes bug 2700)
 530          *
 531          * @private
 532          */
 533         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 534                 global $wgContLang;
 535                 wfProfileIn( __METHOD__ );
 536                 $render = ($this->mOutputType == OT_HTML);
 537
 538                 $uniq_prefix = $this->mUniqPrefix;
 539                 $commentState = new ReplacementArray;
 540
 541                 $elements = array_merge(
 542                         array( 'nowiki', 'gallery' ),
 543                         array_keys( $this->mTagHooks ) );
 544                 global $wgRawHtml;
 545                 if( $wgRawHtml ) {
 546                         $elements[] = 'html';
 547                 }
 548                 if( $this->mOptions->getUseTeX() ) {
 549                         $elements[] = 'math';
 550                 }
 551
 552                 # Removing $dontstrip tags from $elements list (currently only 'gallery', fixing bug 2700)
 553                 foreach ( $elements AS $k => $v ) {
 554                         if ( !in_array ( $v , $dontstrip ) ) continue;
 555                         unset ( $elements[$k] );
 556                 }
 557
 558                 $matches = array();
 559                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 560
 561                 foreach( $matches as $marker => $data ) {
 562                         list( $element, $content, $params, $tag ) = $data;
 563                         if( $render ) {
 564                                 $tagName = strtolower( $element );
 565                                 wfProfileIn( __METHOD__."-render-$tagName" );
 566                                 switch( $tagName ) {
 567                                 case '!--':
 568                                         // Comment
 569                                         if( substr( $tag, -3 ) == '-->' ) {
 570                                                 $output = $tag;
 571                                         } else {
 572                                                 // Unclosed comment in input.
 573                                                 // Close it so later stripping can remove it
 574                                                 $output = "$tag-->";
 575                                         }
 576                                         break;
 577                                 case 'html':
 578                                         if( $wgRawHtml ) {
 579                                                 $output = $content;
 580                                                 break;
 581                                         }
 582                                         // Shouldn't happen otherwise. :)
 583                                 case 'nowiki':
 584                                         $output = Xml::escapeTagsOnly( $content );
 585                                         break;
 586                                 case 'math':
 587                                         $output = $wgContLang->armourMath( MathRenderer::renderMath( $content ) );
 588                                         break;
 589                                 case 'gallery':
 590                                         $output = $this->renderImageGallery( $content, $params );
 591                                         break;
 592                                 default:
 593                                         if( isset( $this->mTagHooks[$tagName] ) ) {
 594                                                 $output = call_user_func_array( $this->mTagHooks[$tagName],
 595                                                         array( $content, $params, $this ) );
 596                                         } else {
 597                                                 throw new MWException( "Invalid call hook $element" );
 598                                         }
 599                                 }
 600                                 wfProfileOut( __METHOD__."-render-$tagName" );
 601                         } else {
 602                                 // Just stripping tags; keep the source
 603                                 $output = $tag;
 604                         }
 605
 606                         // Unstrip the output, because unstrip() is no longer recursive so
 607                         // it won't do it itself
 608                         $output = $state->unstripBoth( $output );
 609
 610                         if( !$stripcomments && $element == '!--' ) {
 611                                 $commentState->setPair( $marker, $output );
 612                         } elseif ( $element == 'html' || $element == 'nowiki' ) {
 613                                 $state->nowiki->setPair( $marker, $output );
 614                         } else {
 615                                 $state->general->setPair( $marker, $output );
 616                         }
 617                 }
 618
 619                 # Unstrip comments unless explicitly told otherwise.
 620                 # (The comments are always stripped prior to this point, so as to
 621                 # not invoke any extension tags / parser hooks contained within
 622                 # a comment.)
 623                 if ( !$stripcomments ) {
 624                         // Put them all back and forget them
 625                         $text = $commentState->replace( $text );
 626                 }
 627
 628                 wfProfileOut( __METHOD__ );
 629                 return $text;
 630         }
 631
 632         /**
 633          * Restores pre, math, and other extensions removed by strip()
 634          *
 635          * always call unstripNoWiki() after this one
 636          * @private
 637          * @deprecated use $this->mStripState->unstrip()
 638          */
 639         function unstrip( $text, $state ) {
 640                 return $state->unstripGeneral( $text );
 641         }
 642
 643         /**
 644          * Always call this after unstrip() to preserve the order
 645          *
 646          * @private
 647          * @deprecated use $this->mStripState->unstrip()
 648          */
 649         function unstripNoWiki( $text, $state ) {
 650                 return $state->unstripNoWiki( $text );
 651         }
 652
 653         /**
 654          * @deprecated use $this->mStripState->unstripBoth()
 655          */
 656         function unstripForHTML( $text ) {
 657                 return $this->mStripState->unstripBoth( $text );
 658         }
 659
 660         /**
 661          * Add an item to the strip state
 662          * Returns the unique tag which must be inserted into the stripped text
 663          * The tag will be replaced with the original text in unstrip()
 664          *
 665          * @private
 666          */
 667         function insertStripItem( $text, &$state ) {
 668                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 669                 $state->general->setPair( $rnd, $text );
 670                 return $rnd;
 671         }
 672
 673         /**
 674          * Interface with html tidy, used if $wgUseTidy = true.
 675          * If tidy isn't able to correct the markup, the original will be
 676          * returned in all its glory with a warning comment appended.
 677          *
 678          * Either the external tidy program or the in-process tidy extension
 679          * will be used depending on availability. Override the default
 680          * $wgTidyInternal setting to disable the internal if it's not working.
 681          *
 682          * @param string $text Hideous HTML input
 683          * @return string Corrected HTML output
 684          * @public
 685          * @static
 686          */
 687         function tidy( $text ) {
 688                 global $wgTidyInternal;
 689                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 690 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 691 '<head><title>test</title></head><body>'.$text.'</body></html>';
 692                 if( $wgTidyInternal ) {
 693                         $correctedtext = Parser::internalTidy( $wrappedtext );
 694                 } else {
 695                         $correctedtext = Parser::externalTidy( $wrappedtext );
 696                 }
 697                 if( is_null( $correctedtext ) ) {
 698                         wfDebug( "Tidy error detected!\n" );
 699                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 700                 }
 701                 return $correctedtext;
 702         }
 703
 704         /**
 705          * Spawn an external HTML tidy process and get corrected markup back from it.
 706          *
 707          * @private
 708          * @static
 709          */
 710         function externalTidy( $text ) {
 711                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 712                 $fname = 'Parser::externalTidy';
 713                 wfProfileIn( $fname );
 714
 715                 $cleansource = '';
 716                 $opts = ' -utf8';
 717
 718                 $descriptorspec = array(
 719                         0 => array('pipe', 'r'),
 720                         1 => array('pipe', 'w'),
 721                         2 => array('file', '/dev/null', 'a')
 722                 );
 723                 $pipes = array();
 724                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 725                 if (is_resource($process)) {
 726                         // Theoretically, this style of communication could cause a deadlock
 727                         // here. If the stdout buffer fills up, then writes to stdin could
 728                         // block. This doesn't appear to happen with tidy, because tidy only
 729                         // writes to stdout after it's finished reading from stdin. Search
 730                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 731                         fwrite($pipes[0], $text);
 732                         fclose($pipes[0]);
 733                         while (!feof($pipes[1])) {
 734                                 $cleansource .= fgets($pipes[1], 1024);
 735                         }
 736                         fclose($pipes[1]);
 737                         proc_close($process);
 738                 }
 739
 740                 wfProfileOut( $fname );
 741
 742                 if( $cleansource == '' && $text != '') {
 743                         // Some kind of error happened, so we couldn't get the corrected text.
 744                         // Just give up; we'll use the source text and append a warning.
 745                         return null;
 746                 } else {
 747                         return $cleansource;
 748                 }
 749         }
 750
 751         /**
 752          * Use the HTML tidy PECL extension to use the tidy library in-process,
 753          * saving the overhead of spawning a new process. Currently written to
 754          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 755          *
 756          * 'pear install tidy' should be able to compile the extension module.
 757          *
 758          * @private
 759          * @static
 760          */
 761         function internalTidy( $text ) {
 762                 global $wgTidyConf;
 763                 $fname = 'Parser::internalTidy';
 764                 wfProfileIn( $fname );
 765
 766                 tidy_load_config( $wgTidyConf );
 767                 tidy_set_encoding( 'utf8' );
 768                 tidy_parse_string( $text );
 769                 tidy_clean_repair();
 770                 if( tidy_get_status() == 2 ) {
 771                         // 2 is magic number for fatal error
 772                         // http://www.php.net/manual/en/function.tidy-get-status.php
 773                         $cleansource = null;
 774                 } else {
 775                         $cleansource = tidy_get_output();
 776                 }
 777                 wfProfileOut( $fname );
 778                 return $cleansource;
 779         }
 780
 781         /**
 782          * parse the wiki syntax used to render tables
 783          *
 784          * @private
 785          */
 786         function doTableStuff ( $text ) {
 787                 $fname = 'Parser::doTableStuff';
 788                 wfProfileIn( $fname );
 789
 790                 $lines = explode ( "\n" , $text );
 791                 $td_history = array (); // Is currently a td tag open?
 792                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 793                 $tr_history = array (); // Is currently a tr tag open?
 794                 $tr_attributes = array (); // history of tr attributes
 795                 $has_opened_tr = array(); // Did this table open a <tr> element?
 796                 $indent_level = 0; // indent level of the table
 797                 foreach ( $lines as $key => $line )
 798                 {
 799                         $line = trim ( $line );
 800
 801                         if( $line == '' ) { // empty line, go to next line
 802                                 continue;
 803                         }
 804                         $first_character = $line{0};
 805                         $matches = array();
 806
 807                         if ( preg_match( '/^(:*)\{\|(.*)$/' , $line , $matches ) ) {
 808                                 // First check if we are starting a new table
 809                                 $indent_level = strlen( $matches[1] );
 810
 811                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 812                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 813
 814                                 $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 815                                 array_push ( $td_history , false );
 816                                 array_push ( $last_tag_history , '' );
 817                                 array_push ( $tr_history , false );
 818                                 array_push ( $tr_attributes , '' );
 819                                 array_push ( $has_opened_tr , false );
 820                         } else if ( count ( $td_history ) == 0 ) {
 821                                 // Don't do any of the following
 822                                 continue;
 823                         } else if ( substr ( $line , 0 , 2 ) == '|}' ) {
 824                                 // We are ending a table
 825                                 $line = '</table>' . substr ( $line , 2 );
 826                                 $last_tag = array_pop ( $last_tag_history );
 827
 828                                 if ( !array_pop ( $has_opened_tr ) ) {
 829                                         $line = "<tr><td></td></tr>{$line}";
 830                                 }
 831
 832                                 if ( array_pop ( $tr_history ) ) {
 833                                         $line = "</tr>{$line}";
 834                                 }
 835
 836                                 if ( array_pop ( $td_history ) ) {
 837                                         $line = "</{$last_tag}>{$line}";
 838                                 }
 839                                 array_pop ( $tr_attributes );
 840                                 $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );
 841                         } else if ( substr ( $line , 0 , 2 ) == '|-' ) {
 842                                 // Now we have a table row
 843                                 $line = preg_replace( '#^\|-+#', '', $line );
 844
 845                                 // Whats after the tag is now only attributes
 846                                 $attributes = $this->mStripState->unstripBoth( $line );
 847                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 848                                 array_pop ( $tr_attributes );
 849                                 array_push ( $tr_attributes , $attributes );
 850
 851                                 $line = '';
 852                                 $last_tag = array_pop ( $last_tag_history );
 853                                 array_pop ( $has_opened_tr );
 854                                 array_push ( $has_opened_tr , true );
 855
 856                                 if ( array_pop ( $tr_history ) ) {
 857                                         $line = '</tr>';
 858                                 }
 859
 860                                 if ( array_pop ( $td_history ) ) {
 861                                         $line = "</{$last_tag}>{$line}";
 862                                 }
 863
 864                                 $lines[$key] = $line;
 865                                 array_push ( $tr_history , false );
 866                                 array_push ( $td_history , false );
 867                                 array_push ( $last_tag_history , '' );
 868                         }
 869                         else if ( $first_character == '|' || $first_character == '!' || substr ( $line , 0 , 2 )  == '|+' ) {
 870                                 // This might be cell elements, td, th or captions
 871                                 if ( substr ( $line , 0 , 2 ) == '|+' ) {
 872                                         $first_character = '+';
 873                                         $line = substr ( $line , 1 );
 874                                 }
 875
 876                                 $line = substr ( $line , 1 );
 877
 878                                 if ( $first_character == '!' ) {
 879                                         $line = str_replace ( '!!' , '||' , $line );
 880                                 }
 881
 882                                 // Split up multiple cells on the same line.
 883                                 // FIXME : This can result in improper nesting of tags processed
 884                                 // by earlier parser steps, but should avoid splitting up eg
 885                                 // attribute values containing literal "||".
 886                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 887
 888                                 $lines[$key] = '';
 889
 890                                 // Loop through each table cell
 891                                 foreach ( $cells as $cell )
 892                                 {
 893                                         $previous = '';
 894                                         if ( $first_character != '+' )
 895                                         {
 896                                                 $tr_after = array_pop ( $tr_attributes );
 897                                                 if ( !array_pop ( $tr_history ) ) {
 898                                                         $previous = "<tr{$tr_after}>\n";
 899                                                 }
 900                                                 array_push ( $tr_history , true );
 901                                                 array_push ( $tr_attributes , '' );
 902                                                 array_pop ( $has_opened_tr );
 903                                                 array_push ( $has_opened_tr , true );
 904                                         }
 905
 906                                         $last_tag = array_pop ( $last_tag_history );
 907
 908                                         if ( array_pop ( $td_history ) ) {
 909                                                 $previous = "</{$last_tag}>{$previous}";
 910                                         }
 911
 912                                         if ( $first_character == '|' ) {
 913                                                 $last_tag = 'td';
 914                                         } else if ( $first_character == '!' ) {
 915                                                 $last_tag = 'th';
 916                                         } else if ( $first_character == '+' ) {
 917                                                 $last_tag = 'caption';
 918                                         } else {
 919                                                 $last_tag = '';
 920                                         }
 921
 922                                         array_push ( $last_tag_history , $last_tag );
 923
 924                                         // A cell could contain both parameters and data
 925                                         $cell_data = explode ( '|' , $cell , 2 );
 926
 927                                         // Bug 553: Note that a '|' inside an invalid link should not
 928                                         // be mistaken as delimiting cell parameters
 929                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 930                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 931                                         } else if ( count ( $cell_data ) == 1 )
 932                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 933                                         else {
 934                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 935                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 936                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 937                                         }
 938
 939                                         $lines[$key] .= $cell;
 940                                         array_push ( $td_history , true );
 941                                 }
 942                         }
 943                 }
 944
 945                 // Closing open td, tr && table
 946                 while ( count ( $td_history ) > 0 )
 947                 {
 948                         if ( array_pop ( $td_history ) ) {
 949                                 $lines[] = '</td>' ;
 950                         }
 951                         if ( array_pop ( $tr_history ) ) {
 952                                 $lines[] = '</tr>' ;
 953                         }
 954                         if ( !array_pop ( $has_opened_tr ) ) {
 955                                 $lines[] = "<tr><td></td></tr>" ;
 956                         }
 957
 958                         $lines[] = '</table>' ;
 959                 }
 960
 961                 $output = implode ( "\n" , $lines ) ;
 962
 963                 // special case: don't return empty table
 964                 if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {
 965                         $output = '';
 966                 }
 967
 968                 wfProfileOut( $fname );
 969
 970                 return $output;
 971         }
 972
 973         /**
 974          * Helper function for parse() that transforms wiki markup into
 975          * HTML. Only called for $mOutputType == OT_HTML.
 976          *
 977          * @private
 978          */
 979         function internalParse( $text ) {
 980                 $args = array();
 981                 $isMain = true;
 982                 $fname = 'Parser::internalParse';
 983                 wfProfileIn( $fname );
 984
 985                 # Hook to suspend the parser in this state
 986                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 987                         wfProfileOut( $fname );
 988                         return $text ;
 989                 }
 990
 991                 # Remove <noinclude> tags and <includeonly> sections
 992                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 993                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 994                 $text = StringUtils::delimiterReplace( '<includeonly>', '</includeonly>', '', $text );
 995
 996                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
 997
 998                 $text = $this->replaceVariables( $text, $args );
 999
1000                 // Tables need to come after variable replacement for things to work
1001                 // properly; putting them before other transformations should keep
1002                 // exciting things like link expansions from showing up in surprising
1003                 // places.
1004                 $text = $this->doTableStuff( $text );
1005
1006                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1007
1008                 $text = $this->stripToc( $text );
1009                 $this->stripNoGallery( $text );
1010                 $text = $this->doHeadings( $text );
1011                 if($this->mOptions->getUseDynamicDates()) {
1012                         $df =& DateFormatter::getInstance();
1013                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1014                 }
1015                 $text = $this->doAllQuotes( $text );
1016                 $text = $this->replaceInternalLinks( $text );
1017                 $text = $this->replaceExternalLinks( $text );
1018
1019                 # replaceInternalLinks may sometimes leave behind
1020                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1021                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
1022
1023                 $text = $this->doMagicLinks( $text );
1024                 $text = $this->formatHeadings( $text, $isMain );
1025
1026                 wfProfileOut( $fname );
1027                 return $text;
1028         }
1029
1030         /**
1031          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1032          * magic external links.
1033          *
1034          * @private
1035          */
1036         function &doMagicLinks( &$text ) {
1037                 wfProfileIn( __METHOD__ );
1038                 $text = preg_replace_callback(
1039                         '!(?:                           # Start cases
1040                             <a.*?</a> |                 # Skip link text
1041                             <.*?> |                     # Skip stuff inside HTML elements
1042                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
1043                             ISBN\s+(\b                  # ISBN, capture number as m[2]
1044                                      (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1045                                      (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1046                                      [0-9Xx]                 # check digit
1047                                    \b)
1048                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1049                 wfProfileOut( __METHOD__ );
1050                 return $text;
1051         }
1052
1053         function magicLinkCallback( $m ) {
1054                 if ( substr( $m[0], 0, 1 ) == '<' ) {
1055                         # Skip HTML element
1056                         return $m[0];
1057                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
1058                         $isbn = $m[2];
1059                         $num = strtr( $isbn, array(
1060                                 '-' => '',
1061                                 ' ' => '',
1062                                 'x' => 'X',
1063                         ));
1064                         $titleObj = SpecialPage::getTitleFor( 'Booksources' );
1065                         $text = '<a href="' .
1066                                 $titleObj->escapeLocalUrl( "isbn=$num" ) .
1067                                 "\" class=\"internal\">ISBN $isbn</a>";
1068                 } else {
1069                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
1070                                 $keyword = 'RFC';
1071                                 $urlmsg = 'rfcurl';
1072                                 $id = $m[1];
1073                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
1074                                 $keyword = 'PMID';
1075                                 $urlmsg = 'pubmedurl';
1076                                 $id = $m[1];
1077                         } else {
1078                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1079                                         substr($m[0], 0, 20 ) . '"' );
1080                         }
1081
1082                         $url = wfMsg( $urlmsg, $id);
1083                         $sk =& $this->mOptions->getSkin();
1084                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1085                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1086                 }
1087                 return $text;
1088         }
1089
1090         /**
1091          * Parse headers and return html
1092          *
1093          * @private
1094          */
1095         function doHeadings( $text ) {
1096                 $fname = 'Parser::doHeadings';
1097                 wfProfileIn( $fname );
1098                 for ( $i = 6; $i >= 1; --$i ) {
1099                         $h = str_repeat( '=', $i );
1100                         $text = preg_replace( "/^{$h}(.+){$h}\\s*$/m",
1101                           "<h{$i}>\\1</h{$i}>\\2", $text );
1102                 }
1103                 wfProfileOut( $fname );
1104                 return $text;
1105         }
1106
1107         /**
1108          * Replace single quotes with HTML markup
1109          * @private
1110          * @return string the altered text
1111          */
1112         function doAllQuotes( $text ) {
1113                 $fname = 'Parser::doAllQuotes';
1114                 wfProfileIn( $fname );
1115                 $outtext = '';
1116                 $lines = explode( "\n", $text );
1117                 foreach ( $lines as $line ) {
1118                         $outtext .= $this->doQuotes ( $line ) . "\n";
1119                 }
1120                 $outtext = substr($outtext, 0,-1);
1121                 wfProfileOut( $fname );
1122                 return $outtext;
1123         }
1124
1125         /**
1126          * Helper function for doAllQuotes()
1127          * @private
1128          */
1129         function doQuotes( $text ) {
1130                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1131                 if ( count( $arr ) == 1 )
1132                         return $text;
1133                 else
1134                 {
1135                         # First, do some preliminary work. This may shift some apostrophes from
1136                         # being mark-up to being text. It also counts the number of occurrences
1137                         # of bold and italics mark-ups.
1138                         $i = 0;
1139                         $numbold = 0;
1140                         $numitalics = 0;
1141                         foreach ( $arr as $r )
1142                         {
1143                                 if ( ( $i % 2 ) == 1 )
1144                                 {
1145                                         # If there are ever four apostrophes, assume the first is supposed to
1146                                         # be text, and the remaining three constitute mark-up for bold text.
1147                                         if ( strlen( $arr[$i] ) == 4 )
1148                                         {
1149                                                 $arr[$i-1] .= "'";
1150                                                 $arr[$i] = "'''";
1151                                         }
1152                                         # If there are more than 5 apostrophes in a row, assume they're all
1153                                         # text except for the last 5.
1154                                         else if ( strlen( $arr[$i] ) > 5 )
1155                                         {
1156                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1157                                                 $arr[$i] = "'''''";
1158                                         }
1159                                         # Count the number of occurrences of bold and italics mark-ups.
1160                                         # We are not counting sequences of five apostrophes.
1161                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1162                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1163                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1164                                 }
1165                                 $i++;
1166                         }
1167
1168                         # If there is an odd number of both bold and italics, it is likely
1169                         # that one of the bold ones was meant to be an apostrophe followed
1170                         # by italics. Which one we cannot know for certain, but it is more
1171                         # likely to be one that has a single-letter word before it.
1172                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1173                         {
1174                                 $i = 0;
1175                                 $firstsingleletterword = -1;
1176                                 $firstmultiletterword = -1;
1177                                 $firstspace = -1;
1178                                 foreach ( $arr as $r )
1179                                 {
1180                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1181                                         {
1182                                                 $x1 = substr ($arr[$i-1], -1);
1183                                                 $x2 = substr ($arr[$i-1], -2, 1);
1184                                                 if ($x1 == ' ') {
1185                                                         if ($firstspace == -1) $firstspace = $i;
1186                                                 } else if ($x2 == ' ') {
1187                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1188                                                 } else {
1189                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1190                                                 }
1191                                         }
1192                                         $i++;
1193                                 }
1194
1195                                 # If there is a single-letter word, use it!
1196                                 if ($firstsingleletterword > -1)
1197                                 {
1198                                         $arr [ $firstsingleletterword ] = "''";
1199                                         $arr [ $firstsingleletterword-1 ] .= "'";
1200                                 }
1201                                 # If not, but there's a multi-letter word, use that one.
1202                                 else if ($firstmultiletterword > -1)
1203                                 {
1204                                         $arr [ $firstmultiletterword ] = "''";
1205                                         $arr [ $firstmultiletterword-1 ] .= "'";
1206                                 }
1207                                 # ... otherwise use the first one that has neither.
1208                                 # (notice that it is possible for all three to be -1 if, for example,
1209                                 # there is only one pentuple-apostrophe in the line)
1210                                 else if ($firstspace > -1)
1211                                 {
1212                                         $arr [ $firstspace ] = "''";
1213                                         $arr [ $firstspace-1 ] .= "'";
1214                                 }
1215                         }
1216
1217                         # Now let's actually convert our apostrophic mush to HTML!
1218                         $output = '';
1219                         $buffer = '';
1220                         $state = '';
1221                         $i = 0;
1222                         foreach ($arr as $r)
1223                         {
1224                                 if (($i % 2) == 0)
1225                                 {
1226                                         if ($state == 'both')
1227                                                 $buffer .= $r;
1228                                         else
1229                                                 $output .= $r;
1230                                 }
1231                                 else
1232                                 {
1233                                         if (strlen ($r) == 2)
1234                                         {
1235                                                 if ($state == 'i')
1236                                                 { $output .= '</i>'; $state = ''; }
1237                                                 else if ($state == 'bi')
1238                                                 { $output .= '</i>'; $state = 'b'; }
1239                                                 else if ($state == 'ib')
1240                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1241                                                 else if ($state == 'both')
1242                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1243                                                 else # $state can be 'b' or ''
1244                                                 { $output .= '<i>'; $state .= 'i'; }
1245                                         }
1246                                         else if (strlen ($r) == 3)
1247                                         {
1248                                                 if ($state == 'b')
1249                                                 { $output .= '</b>'; $state = ''; }
1250                                                 else if ($state == 'bi')
1251                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1252                                                 else if ($state == 'ib')
1253                                                 { $output .= '</b>'; $state = 'i'; }
1254                                                 else if ($state == 'both')
1255                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1256                                                 else # $state can be 'i' or ''
1257                                                 { $output .= '<b>'; $state .= 'b'; }
1258                                         }
1259                                         else if (strlen ($r) == 5)
1260                                         {
1261                                                 if ($state == 'b')
1262                                                 { $output .= '</b><i>'; $state = 'i'; }
1263                                                 else if ($state == 'i')
1264                                                 { $output .= '</i><b>'; $state = 'b'; }
1265                                                 else if ($state == 'bi')
1266                                                 { $output .= '</i></b>'; $state = ''; }
1267                                                 else if ($state == 'ib')
1268                                                 { $output .= '</b></i>'; $state = ''; }
1269                                                 else if ($state == 'both')
1270                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1271                                                 else # ($state == '')
1272                                                 { $buffer = ''; $state = 'both'; }
1273                                         }
1274                                 }
1275                                 $i++;
1276                         }
1277                         # Now close all remaining tags.  Notice that the order is important.
1278                         if ($state == 'b' || $state == 'ib')
1279                                 $output .= '</b>';
1280                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1281                                 $output .= '</i>';
1282                         if ($state == 'bi')
1283                                 $output .= '</b>';
1284                         if ($state == 'both')
1285                                 $output .= '<b><i>'.$buffer.'</i></b>';
1286                         return $output;
1287                 }
1288         }
1289
1290         /**
1291          * Replace external links
1292          *
1293          * Note: this is all very hackish and the order of execution matters a lot.
1294          * Make sure to run maintenance/parserTests.php if you change this code.
1295          *
1296          * @private
1297          */
1298         function replaceExternalLinks( $text ) {
1299                 global $wgContLang;
1300                 $fname = 'Parser::replaceExternalLinks';
1301                 wfProfileIn( $fname );
1302
1303                 $sk =& $this->mOptions->getSkin();
1304
1305                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1306
1307                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1308
1309                 $i = 0;
1310                 while ( $i<count( $bits ) ) {
1311                         $url = $bits[$i++];
1312                         $protocol = $bits[$i++];
1313                         $text = $bits[$i++];
1314                         $trail = $bits[$i++];
1315
1316                         # The characters '<' and '>' (which were escaped by
1317                         # removeHTMLtags()) should not be included in
1318                         # URLs, per RFC 2396.
1319                         $m2 = array();
1320                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1321                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1322                                 $url = substr($url, 0, $m2[0][1]);
1323                         }
1324
1325                         # If the link text is an image URL, replace it with an <img> tag
1326                         # This happened by accident in the original parser, but some people used it extensively
1327                         $img = $this->maybeMakeExternalImage( $text );
1328                         if ( $img !== false ) {
1329                                 $text = $img;
1330                         }
1331
1332                         $dtrail = '';
1333
1334                         # Set linktype for CSS - if URL==text, link is essentially free
1335                         $linktype = ($text == $url) ? 'free' : 'text';
1336
1337                         # No link text, e.g. [http://domain.tld/some.link]
1338                         if ( $text == '' ) {
1339                                 # Autonumber if allowed. See bug #5918
1340                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1341                                         $text = '[' . ++$this->mAutonumber . ']';
1342                                         $linktype = 'autonumber';
1343                                 } else {
1344                                         # Otherwise just use the URL
1345                                         $text = htmlspecialchars( $url );
1346                                         $linktype = 'free';
1347                                 }
1348                         } else {
1349                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1350                                 # Check for trail
1351                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1352                         }
1353
1354                         $text = $wgContLang->markNoConversion($text);
1355
1356                         $url = Sanitizer::cleanUrl( $url );
1357
1358                         # Process the trail (i.e. everything after this link up until start of the next link),
1359                         # replacing any non-bracketed links
1360                         $trail = $this->replaceFreeExternalLinks( $trail );
1361
1362                         # Use the encoded URL
1363                         # This means that users can paste URLs directly into the text
1364                         # Funny characters like &ouml; aren't valid in URLs anyway
1365                         # This was changed in August 2004
1366                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1367
1368                         # Register link in the output object.
1369                         # Replace unnecessary URL escape codes with the referenced character
1370                         # This prevents spammers from hiding links from the filters
1371                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1372                         $this->mOutput->addExternalLink( $pasteurized );
1373                 }
1374
1375                 wfProfileOut( $fname );
1376                 return $s;
1377         }
1378
1379         /**
1380          * Replace anything that looks like a URL with a link
1381          * @private
1382          */
1383         function replaceFreeExternalLinks( $text ) {
1384                 global $wgContLang;
1385                 $fname = 'Parser::replaceFreeExternalLinks';
1386                 wfProfileIn( $fname );
1387
1388                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1389                 $s = array_shift( $bits );
1390                 $i = 0;
1391
1392                 $sk =& $this->mOptions->getSkin();
1393
1394                 while ( $i < count( $bits ) ){
1395                         $protocol = $bits[$i++];
1396                         $remainder = $bits[$i++];
1397
1398                         $m = array();
1399                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1400                                 # Found some characters after the protocol that look promising
1401                                 $url = $protocol . $m[1];
1402                                 $trail = $m[2];
1403
1404                                 # special case: handle urls as url args:
1405                                 # http://www.example.com/foo?=http://www.example.com/bar
1406                                 if(strlen($trail) == 0 &&
1407                                         isset($bits[$i]) &&
1408                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1409                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1410                                 {
1411                                         # add protocol, arg
1412                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1413                                         $i += 2;
1414                                         $trail = $m[2];
1415                                 }
1416
1417                                 # The characters '<' and '>' (which were escaped by
1418                                 # removeHTMLtags()) should not be included in
1419                                 # URLs, per RFC 2396.
1420                                 $m2 = array();
1421                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1422                                         $trail = substr($url, $m2[0][1]) . $trail;
1423                                         $url = substr($url, 0, $m2[0][1]);
1424                                 }
1425
1426                                 # Move trailing punctuation to $trail
1427                                 $sep = ',;\.:!?';
1428                                 # If there is no left bracket, then consider right brackets fair game too
1429                                 if ( strpos( $url, '(' ) === false ) {
1430                                         $sep .= ')';
1431                                 }
1432
1433                                 $numSepChars = strspn( strrev( $url ), $sep );
1434                                 if ( $numSepChars ) {
1435                                         $trail = substr( $url, -$numSepChars ) . $trail;
1436                                         $url = substr( $url, 0, -$numSepChars );
1437                                 }
1438
1439                                 $url = Sanitizer::cleanUrl( $url );
1440
1441                                 # Is this an external image?
1442                                 $text = $this->maybeMakeExternalImage( $url );
1443                                 if ( $text === false ) {
1444                                         # Not an image, make a link
1445                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1446                                         # Register it in the output object...
1447                                         # Replace unnecessary URL escape codes with their equivalent characters
1448                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1449                                         $this->mOutput->addExternalLink( $pasteurized );
1450                                 }
1451                                 $s .= $text . $trail;
1452                         } else {
1453                                 $s .= $protocol . $remainder;
1454                         }
1455                 }
1456                 wfProfileOut( $fname );
1457                 return $s;
1458         }
1459
1460         /**
1461          * Replace unusual URL escape codes with their equivalent characters
1462          * @param string
1463          * @return string
1464          * @static
1465          * @fixme This can merge genuinely required bits in the path or query string,
1466          *        breaking legit URLs. A proper fix would treat the various parts of
1467          *        the URL differently; as a workaround, just use the output for
1468          *        statistical records, not for actual linking/output.
1469          */
1470         static function replaceUnusualEscapes( $url ) {
1471                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1472                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1473         }
1474
1475         /**
1476          * Callback function used in replaceUnusualEscapes().
1477          * Replaces unusual URL escape codes with their equivalent character
1478          * @static
1479          * @private
1480          */
1481         private static function replaceUnusualEscapesCallback( $matches ) {
1482                 $char = urldecode( $matches[0] );
1483                 $ord = ord( $char );
1484                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1485                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1486                         // No, shouldn't be escaped
1487                         return $char;
1488                 } else {
1489                         // Yes, leave it escaped
1490                         return $matches[0];
1491                 }
1492         }
1493
1494         /**
1495          * make an image if it's allowed, either through the global
1496          * option or through the exception
1497          * @private
1498          */
1499         function maybeMakeExternalImage( $url ) {
1500                 $sk =& $this->mOptions->getSkin();
1501                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1502                 $imagesexception = !empty($imagesfrom);
1503                 $text = false;
1504                 if ( $this->mOptions->getAllowExternalImages()
1505                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1506                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1507                                 # Image found
1508                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1509                         }
1510                 }
1511                 return $text;
1512         }
1513
1514         /**
1515          * Process [[ ]] wikilinks
1516          *
1517          * @private
1518          */
1519         function replaceInternalLinks( $s ) {
1520                 global $wgContLang;
1521                 static $fname = 'Parser::replaceInternalLinks' ;
1522
1523                 wfProfileIn( $fname );
1524
1525                 wfProfileIn( $fname.'-setup' );
1526                 static $tc = FALSE;
1527                 # the % is needed to support urlencoded titles as well
1528                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1529
1530                 $sk =& $this->mOptions->getSkin();
1531
1532                 #split the entire text string on occurences of [[
1533                 $a = explode( '[[', ' ' . $s );
1534                 #get the first element (all text up to first [[), and remove the space we added
1535                 $s = array_shift( $a );
1536                 $s = substr( $s, 1 );
1537
1538                 # Match a link having the form [[namespace:link|alternate]]trail
1539                 static $e1 = FALSE;
1540                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1541                 # Match cases where there is no "]]", which might still be images
1542                 static $e1_img = FALSE;
1543                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1544                 # Match the end of a line for a word that's not followed by whitespace,
1545                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1546                 $e2 = wfMsgForContent( 'linkprefix' );
1547
1548                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1549
1550                 if( is_null( $this->mTitle ) ) {
1551                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1552                 }
1553                 $nottalk = !$this->mTitle->isTalkPage();
1554
1555                 if ( $useLinkPrefixExtension ) {
1556                         $m = array();
1557                         if ( preg_match( $e2, $s, $m ) ) {
1558                                 $first_prefix = $m[2];
1559                         } else {
1560                                 $first_prefix = false;
1561                         }
1562                 } else {
1563                         $prefix = '';
1564                 }
1565
1566                 if($wgContLang->hasVariants())
1567                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1568                 else
1569                         $selflink = array($this->mTitle->getPrefixedText());
1570                 $useSubpages = $this->areSubpagesAllowed();
1571                 wfProfileOut( $fname.'-setup' );
1572
1573                 # Loop for each link
1574                 for ($k = 0; isset( $a[$k] ); $k++) {
1575                         $line = $a[$k];
1576                         if ( $useLinkPrefixExtension ) {
1577                                 wfProfileIn( $fname.'-prefixhandling' );
1578                                 if ( preg_match( $e2, $s, $m ) ) {
1579                                         $prefix = $m[2];
1580                                         $s = $m[1];
1581                                 } else {
1582                                         $prefix='';
1583                                 }
1584                                 # first link
1585                                 if($first_prefix) {
1586                                         $prefix = $first_prefix;
1587                                         $first_prefix = false;
1588                                 }
1589                                 wfProfileOut( $fname.'-prefixhandling' );
1590                         }
1591
1592                         $might_be_img = false;
1593
1594                         wfProfileIn( "$fname-e1" );
1595                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1596                                 $text = $m[2];
1597                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1598                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1599                                 # the real problem is with the $e1 regex
1600                                 # See bug 1300.
1601                                 #
1602                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1603                                 # and no image is in sight. See bug 2095.
1604                                 #
1605                                 if( $text !== '' &&
1606                                         substr( $m[3], 0, 1 ) === ']' &&
1607                                         strpos($text, '[') !== false
1608                                 )
1609                                 {
1610                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1611                                         $m[3] = substr( $m[3], 1 );
1612                                 }
1613                                 # fix up urlencoded title texts
1614                                 if( strpos( $m[1], '%' ) !== false ) {
1615                                         # Should anchors '#' also be rejected?
1616                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1617                                 }
1618                                 $trail = $m[3];
1619                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1620                                 $might_be_img = true;
1621                                 $text = $m[2];
1622                                 if ( strpos( $m[1], '%' ) !== false ) {
1623                                        $m[1] = urldecode($m[1]);
1624                                 }
1625                                 $trail = "";
1626                         } else { # Invalid form; output directly
1627                                 $s .= $prefix . '[[' . $line ;
1628                                 wfProfileOut( "$fname-e1" );
1629                                 continue;
1630                         }
1631                         wfProfileOut( "$fname-e1" );
1632                         wfProfileIn( "$fname-misc" );
1633
1634                         # Don't allow internal links to pages containing
1635                         # PROTO: where PROTO is a valid URL protocol; these
1636                         # should be external links.
1637                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1638                                 $s .= $prefix . '[[' . $line ;
1639                                 continue;
1640                         }
1641
1642                         # Make subpage if necessary
1643                         if( $useSubpages ) {
1644                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1645                         } else {
1646                                 $link = $m[1];
1647                         }
1648
1649                         $noforce = (substr($m[1], 0, 1) != ':');
1650                         if (!$noforce) {
1651                                 # Strip off leading ':'
1652                                 $link = substr($link, 1);
1653                         }
1654
1655                         wfProfileOut( "$fname-misc" );
1656                         wfProfileIn( "$fname-title" );
1657                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1658                         if( !$nt ) {
1659                                 $s .= $prefix . '[[' . $line;
1660                                 wfProfileOut( "$fname-title" );
1661                                 continue;
1662                         }
1663
1664                         $ns = $nt->getNamespace();
1665                         $iw = $nt->getInterWiki();
1666                         wfProfileOut( "$fname-title" );
1667
1668                         if ($might_be_img) { # if this is actually an invalid link
1669                                 wfProfileIn( "$fname-might_be_img" );
1670                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1671                                         $found = false;
1672                                         while (isset ($a[$k+1]) ) {
1673                                                 #look at the next 'line' to see if we can close it there
1674                                                 $spliced = array_splice( $a, $k + 1, 1 );
1675                                                 $next_line = array_shift( $spliced );
1676                                                 $m = explode( ']]', $next_line, 3 );
1677                                                 if ( count( $m ) == 3 ) {
1678                                                         # the first ]] closes the inner link, the second the image
1679                                                         $found = true;
1680                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1681                                                         $trail = $m[2];
1682                                                         break;
1683                                                 } elseif ( count( $m ) == 2 ) {
1684                                                         #if there's exactly one ]] that's fine, we'll keep looking
1685                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1686                                                 } else {
1687                                                         #if $next_line is invalid too, we need look no further
1688                                                         $text .= '[[' . $next_line;
1689                                                         break;
1690                                                 }
1691                                         }
1692                                         if ( !$found ) {
1693                                                 # we couldn't find the end of this imageLink, so output it raw
1694                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1695                                                 $text = $this->replaceInternalLinks($text);
1696                                                 $s .= "{$prefix}[[$link|$text";
1697                                                 # note: no $trail, because without an end, there *is* no trail
1698                                                 wfProfileOut( "$fname-might_be_img" );
1699                                                 continue;
1700                                         }
1701                                 } else { #it's not an image, so output it raw
1702                                         $s .= "{$prefix}[[$link|$text";
1703                                         # note: no $trail, because without an end, there *is* no trail
1704                                         wfProfileOut( "$fname-might_be_img" );
1705                                         continue;
1706                                 }
1707                                 wfProfileOut( "$fname-might_be_img" );
1708                         }
1709
1710                         $wasblank = ( '' == $text );
1711                         if( $wasblank ) $text = $link;
1712
1713                         # Link not escaped by : , create the various objects
1714                         if( $noforce ) {
1715
1716                                 # Interwikis
1717                                 wfProfileIn( "$fname-interwiki" );
1718                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1719                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1720                                         $s = rtrim($s . "\n");
1721                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1722                                         wfProfileOut( "$fname-interwiki" );
1723                                         continue;
1724                                 }
1725                                 wfProfileOut( "$fname-interwiki" );
1726
1727                                 if ( $ns == NS_IMAGE ) {
1728                                         wfProfileIn( "$fname-image" );
1729                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1730                                                 # recursively parse links inside the image caption
1731                                                 # actually, this will parse them in any other parameters, too,
1732                                                 # but it might be hard to fix that, and it doesn't matter ATM
1733                                                 $text = $this->replaceExternalLinks($text);
1734                                                 $text = $this->replaceInternalLinks($text);
1735
1736                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1737                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1738                                                 $this->mOutput->addImage( $nt->getDBkey() );
1739
1740                                                 wfProfileOut( "$fname-image" );
1741                                                 continue;
1742                                         } else {
1743                                                 # We still need to record the image's presence on the page
1744                                                 $this->mOutput->addImage( $nt->getDBkey() );
1745                                         }
1746                                         wfProfileOut( "$fname-image" );
1747
1748                                 }
1749
1750                                 if ( $ns == NS_CATEGORY ) {
1751                                         wfProfileIn( "$fname-category" );
1752                                         $s = rtrim($s . "\n"); # bug 87
1753
1754                                         if ( $wasblank ) {
1755                                                 $sortkey = $this->getDefaultSort();
1756                                         } else {
1757                                                 $sortkey = $text;
1758                                         }
1759                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1760                                         $sortkey = str_replace( "\n", '', $sortkey );
1761                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1762                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1763
1764                                         /**
1765                                          * Strip the whitespace Category links produce, see bug 87
1766                                          * @todo We might want to use trim($tmp, "\n") here.
1767                                          */
1768                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1769
1770                                         wfProfileOut( "$fname-category" );
1771                                         continue;
1772                                 }
1773                         }
1774
1775                         if( ( in_array( $nt->getPrefixedText(), $selflink ) ) &&
1776                             ( $nt->getFragment() === '' ) ) {
1777                                 # Self-links are handled specially; generally de-link and change to bold.
1778                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1779                                 continue;
1780                         }
1781
1782                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1783                         if( $ns == NS_MEDIA ) {
1784                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1785                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1786                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1787                                 $this->mOutput->addImage( $nt->getDBkey() );
1788                                 continue;
1789                         } elseif( $ns == NS_SPECIAL ) {
1790                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1791                                 continue;
1792                         } elseif( $ns == NS_IMAGE ) {
1793                                 $img = new Image( $nt );
1794                                 if( $img->exists() ) {
1795                                         // Force a blue link if the file exists; may be a remote
1796                                         // upload on the shared repository, and we want to see its
1797                                         // auto-generated page.
1798                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1799                                         $this->mOutput->addLink( $nt );
1800                                         continue;
1801                                 }
1802                         }
1803                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1804                 }
1805                 wfProfileOut( $fname );
1806                 return $s;
1807         }
1808
1809         /**
1810          * Make a link placeholder. The text returned can be later resolved to a real link with
1811          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1812          * parsing of interwiki links, and secondly to allow all existence checks and
1813          * article length checks (for stub links) to be bundled into a single query.
1814          *
1815          */
1816         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1817                 wfProfileIn( __METHOD__ );
1818                 if ( ! is_object($nt) ) {
1819                         # Fail gracefully
1820                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1821                 } else {
1822                         # Separate the link trail from the rest of the link
1823                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1824
1825                         if ( $nt->isExternal() ) {
1826                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1827                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1828                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1829                         } else {
1830                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1831                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1832                                 $this->mLinkHolders['queries'][] = $query;
1833                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1834                                 $this->mLinkHolders['titles'][] = $nt;
1835
1836                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1837                         }
1838                 }
1839                 wfProfileOut( __METHOD__ );
1840                 return $retVal;
1841         }
1842
1843         /**
1844          * Render a forced-blue link inline; protect against double expansion of
1845          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1846          * Since this little disaster has to split off the trail text to avoid
1847          * breaking URLs in the following text without breaking trails on the
1848          * wiki links, it's been made into a horrible function.
1849          *
1850          * @param Title $nt
1851          * @param string $text
1852          * @param string $query
1853          * @param string $trail
1854          * @param string $prefix
1855          * @return string HTML-wikitext mix oh yuck
1856          */
1857         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1858                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1859                 $sk =& $this->mOptions->getSkin();
1860                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1861                 return $this->armorLinks( $link ) . $trail;
1862         }
1863
1864         /**
1865          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1866          * going to go through further parsing steps before inline URL expansion.
1867          *
1868          * In particular this is important when using action=render, which causes
1869          * full URLs to be included.
1870          *
1871          * Oh man I hate our multi-layer parser!
1872          *
1873          * @param string more-or-less HTML
1874          * @return string less-or-more HTML with NOPARSE bits
1875          */
1876         function armorLinks( $text ) {
1877                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1878                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1879         }
1880
1881         /**
1882          * Return true if subpage links should be expanded on this page.
1883          * @return bool
1884          */
1885         function areSubpagesAllowed() {
1886                 # Some namespaces don't allow subpages
1887                 global $wgNamespacesWithSubpages;
1888                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1889         }
1890
1891         /**
1892          * Handle link to subpage if necessary
1893          * @param string $target the source of the link
1894          * @param string &$text the link text, modified as necessary
1895          * @return string the full name of the link
1896          * @private
1897          */
1898         function maybeDoSubpageLink($target, &$text) {
1899                 # Valid link forms:
1900                 # Foobar -- normal
1901                 # :Foobar -- override special treatment of prefix (images, language links)
1902                 # /Foobar -- convert to CurrentPage/Foobar
1903                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1904                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1905                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1906
1907                 $fname = 'Parser::maybeDoSubpageLink';
1908                 wfProfileIn( $fname );
1909                 $ret = $target; # default return value is no change
1910
1911                 # bug 7425
1912                 $target = trim( $target );
1913
1914                 # Some namespaces don't allow subpages,
1915                 # so only perform processing if subpages are allowed
1916                 if( $this->areSubpagesAllowed() ) {
1917                         # Look at the first character
1918                         if( $target != '' && $target{0} == '/' ) {
1919                                 # / at end means we don't want the slash to be shown
1920                                 if( substr( $target, -1, 1 ) == '/' ) {
1921                                         $target = substr( $target, 1, -1 );
1922                                         $noslash = $target;
1923                                 } else {
1924                                         $noslash = substr( $target, 1 );
1925                                 }
1926
1927                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1928                                 if( '' === $text ) {
1929                                         $text = $target;
1930                                 } # this might be changed for ugliness reasons
1931                         } else {
1932                                 # check for .. subpage backlinks
1933                                 $dotdotcount = 0;
1934                                 $nodotdot = $target;
1935                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1936                                         ++$dotdotcount;
1937                                         $nodotdot = substr( $nodotdot, 3 );
1938                                 }
1939                                 if($dotdotcount > 0) {
1940                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1941                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1942                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1943                                                 # / at the end means don't show full path
1944                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1945                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1946                                                         if( '' === $text ) {
1947                                                                 $text = $nodotdot;
1948                                                         }
1949                                                 }
1950                                                 $nodotdot = trim( $nodotdot );
1951                                                 if( $nodotdot != '' ) {
1952                                                         $ret .= '/' . $nodotdot;
1953                                                 }
1954                                         }
1955                                 }
1956                         }
1957                 }
1958
1959                 wfProfileOut( $fname );
1960                 return $ret;
1961         }
1962
1963         /**#@+
1964          * Used by doBlockLevels()
1965          * @private
1966          */
1967         /* private */ function closeParagraph() {
1968                 $result = '';
1969                 if ( '' != $this->mLastSection ) {
1970                         $result = '</' . $this->mLastSection  . ">\n";
1971                 }
1972                 $this->mInPre = false;
1973                 $this->mLastSection = '';
1974                 return $result;
1975         }
1976         # getCommon() returns the length of the longest common substring
1977         # of both arguments, starting at the beginning of both.
1978         #
1979         /* private */ function getCommon( $st1, $st2 ) {
1980                 $fl = strlen( $st1 );
1981                 $shorter = strlen( $st2 );
1982                 if ( $fl < $shorter ) { $shorter = $fl; }
1983
1984                 for ( $i = 0; $i < $shorter; ++$i ) {
1985                         if ( $st1{$i} != $st2{$i} ) { break; }
1986                 }
1987                 return $i;
1988         }
1989         # These next three functions open, continue, and close the list
1990         # element appropriate to the prefix character passed into them.
1991         #
1992         /* private */ function openList( $char ) {
1993                 $result = $this->closeParagraph();
1994
1995                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1996                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1997                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1998                 else if ( ';' == $char ) {
1999                         $result .= '<dl><dt>';
2000                         $this->mDTopen = true;
2001                 }
2002                 else { $result = '<!-- ERR 1 -->'; }
2003
2004                 return $result;
2005         }
2006
2007         /* private */ function nextItem( $char ) {
2008                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
2009                 else if ( ':' == $char || ';' == $char ) {
2010                         $close = '</dd>';
2011                         if ( $this->mDTopen ) { $close = '</dt>'; }
2012                         if ( ';' == $char ) {
2013                                 $this->mDTopen = true;
2014                                 return $close . '<dt>';
2015                         } else {
2016                                 $this->mDTopen = false;
2017                                 return $close . '<dd>';
2018                         }
2019                 }
2020                 return '<!-- ERR 2 -->';
2021         }
2022
2023         /* private */ function closeList( $char ) {
2024                 if ( '*' == $char ) { $text = '</li></ul>'; }
2025                 else if ( '#' == $char ) { $text = '</li></ol>'; }
2026                 else if ( ':' == $char ) {
2027                         if ( $this->mDTopen ) {
2028                                 $this->mDTopen = false;
2029                                 $text = '</dt></dl>';
2030                         } else {
2031                                 $text = '</dd></dl>';
2032                         }
2033                 }
2034                 else {  return '<!-- ERR 3 -->'; }
2035                 return $text."\n";
2036         }
2037         /**#@-*/
2038
2039         /**
2040          * Make lists from lines starting with ':', '*', '#', etc.
2041          *
2042          * @private
2043          * @return string the lists rendered as HTML
2044          */
2045         function doBlockLevels( $text, $linestart ) {
2046                 $fname = 'Parser::doBlockLevels';
2047                 wfProfileIn( $fname );
2048
2049                 # Parsing through the text line by line.  The main thing
2050                 # happening here is handling of block-level elements p, pre,
2051                 # and making lists from lines starting with * # : etc.
2052                 #
2053                 $textLines = explode( "\n", $text );
2054
2055                 $lastPrefix = $output = '';
2056                 $this->mDTopen = $inBlockElem = false;
2057                 $prefixLength = 0;
2058                 $paragraphStack = false;
2059
2060                 if ( !$linestart ) {
2061                         $output .= array_shift( $textLines );
2062                 }
2063                 foreach ( $textLines as $oLine ) {
2064                         $lastPrefixLength = strlen( $lastPrefix );
2065                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2066                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2067                         if ( !$this->mInPre ) {
2068                                 # Multiple prefixes may abut each other for nested lists.
2069                                 $prefixLength = strspn( $oLine, '*#:;' );
2070                                 $pref = substr( $oLine, 0, $prefixLength );
2071
2072                                 # eh?
2073                                 $pref2 = str_replace( ';', ':', $pref );
2074                                 $t = substr( $oLine, $prefixLength );
2075                                 $this->mInPre = !empty($preOpenMatch);
2076                         } else {
2077                                 # Don't interpret any other prefixes in preformatted text
2078                                 $prefixLength = 0;
2079                                 $pref = $pref2 = '';
2080                                 $t = $oLine;
2081                         }
2082
2083                         # List generation
2084                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2085                                 # Same as the last item, so no need to deal with nesting or opening stuff
2086                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2087                                 $paragraphStack = false;
2088
2089                                 if ( substr( $pref, -1 ) == ';') {
2090                                         # The one nasty exception: definition lists work like this:
2091                                         # ; title : definition text
2092                                         # So we check for : in the remainder text to split up the
2093                                         # title and definition, without b0rking links.
2094                                         $term = $t2 = '';
2095                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2096                                                 $t = $t2;
2097                                                 $output .= $term . $this->nextItem( ':' );
2098                                         }
2099                                 }
2100                         } elseif( $prefixLength || $lastPrefixLength ) {
2101                                 # Either open or close a level...
2102                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2103                                 $paragraphStack = false;
2104
2105                                 while( $commonPrefixLength < $lastPrefixLength ) {
2106                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2107                                         --$lastPrefixLength;
2108                                 }
2109                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2110                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2111                                 }
2112                                 while ( $prefixLength > $commonPrefixLength ) {
2113                                         $char = substr( $pref, $commonPrefixLength, 1 );
2114                                         $output .= $this->openList( $char );
2115
2116                                         if ( ';' == $char ) {
2117                                                 # FIXME: This is dupe of code above
2118                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2119                                                         $t = $t2;
2120                                                         $output .= $term . $this->nextItem( ':' );
2121                                                 }
2122                                         }
2123                                         ++$commonPrefixLength;
2124                                 }
2125                                 $lastPrefix = $pref2;
2126                         }
2127                         if( 0 == $prefixLength ) {
2128                                 wfProfileIn( "$fname-paragraph" );
2129                                 # No prefix (not in list)--go to paragraph mode
2130                                 // XXX: use a stack for nestable elements like span, table and div
2131                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2132                                 $closematch = preg_match(
2133                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2134                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2135                                 if ( $openmatch or $closematch ) {
2136                                         $paragraphStack = false;
2137                                         # TODO bug 5718: paragraph closed
2138                                         $output .= $this->closeParagraph();
2139                                         if ( $preOpenMatch and !$preCloseMatch ) {
2140                                                 $this->mInPre = true;
2141                                         }
2142                                         if ( $closematch ) {
2143                                                 $inBlockElem = false;
2144                                         } else {
2145                                                 $inBlockElem = true;
2146                                         }
2147                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2148                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2149                                                 // pre
2150                                                 if ($this->mLastSection != 'pre') {
2151                                                         $paragraphStack = false;
2152                                                         $output .= $this->closeParagraph().'<pre>';
2153                                                         $this->mLastSection = 'pre';
2154                                                 }
2155                                                 $t = substr( $t, 1 );
2156                                         } else {
2157                                                 // paragraph
2158                                                 if ( '' == trim($t) ) {
2159                                                         if ( $paragraphStack ) {
2160                                                                 $output .= $paragraphStack.'<br />';
2161                                                                 $paragraphStack = false;
2162                                                                 $this->mLastSection = 'p';
2163                                                         } else {
2164                                                                 if ($this->mLastSection != 'p' ) {
2165                                                                         $output .= $this->closeParagraph();
2166                                                                         $this->mLastSection = '';
2167                                                                         $paragraphStack = '<p>';
2168                                                                 } else {
2169                                                                         $paragraphStack = '</p><p>';
2170                                                                 }
2171                                                         }
2172                                                 } else {
2173                                                         if ( $paragraphStack ) {
2174                                                                 $output .= $paragraphStack;
2175                                                                 $paragraphStack = false;
2176                                                                 $this->mLastSection = 'p';
2177                                                         } else if ($this->mLastSection != 'p') {
2178                                                                 $output .= $this->closeParagraph().'<p>';
2179                                                                 $this->mLastSection = 'p';
2180                                                         }
2181                                                 }
2182                                         }
2183                                 }
2184                                 wfProfileOut( "$fname-paragraph" );
2185                         }
2186                         // somewhere above we forget to get out of pre block (bug 785)
2187                         if($preCloseMatch && $this->mInPre) {
2188                                 $this->mInPre = false;
2189                         }
2190                         if ($paragraphStack === false) {
2191                                 $output .= $t."\n";
2192                         }
2193                 }
2194                 while ( $prefixLength ) {
2195                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2196                         --$prefixLength;
2197                 }
2198                 if ( '' != $this->mLastSection ) {
2199                         $output .= '</' . $this->mLastSection . '>';
2200                         $this->mLastSection = '';
2201                 }
2202
2203                 wfProfileOut( $fname );
2204                 return $output;
2205         }
2206
2207         /**
2208          * Split up a string on ':', ignoring any occurences inside tags
2209          * to prevent illegal overlapping.
2210          * @param string $str the string to split
2211          * @param string &$before set to everything before the ':'
2212          * @param string &$after set to everything after the ':'
2213          * return string the position of the ':', or false if none found
2214          */
2215         function findColonNoLinks($str, &$before, &$after) {
2216                 $fname = 'Parser::findColonNoLinks';
2217                 wfProfileIn( $fname );
2218
2219                 $pos = strpos( $str, ':' );
2220                 if( $pos === false ) {
2221                         // Nothing to find!
2222                         wfProfileOut( $fname );
2223                         return false;
2224                 }
2225
2226                 $lt = strpos( $str, '<' );
2227                 if( $lt === false || $lt > $pos ) {
2228                         // Easy; no tag nesting to worry about
2229                         $before = substr( $str, 0, $pos );
2230                         $after = substr( $str, $pos+1 );
2231                         wfProfileOut( $fname );
2232                         return $pos;
2233                 }
2234
2235                 // Ugly state machine to walk through avoiding tags.
2236                 $state = MW_COLON_STATE_TEXT;
2237                 $stack = 0;
2238                 $len = strlen( $str );
2239                 for( $i = 0; $i < $len; $i++ ) {
2240                         $c = $str{$i};
2241
2242                         switch( $state ) {
2243                         // (Using the number is a performance hack for common cases)
2244                         case 0: // MW_COLON_STATE_TEXT:
2245                                 switch( $c ) {
2246                                 case "<":
2247                                         // Could be either a <start> tag or an </end> tag
2248                                         $state = MW_COLON_STATE_TAGSTART;
2249                                         break;
2250                                 case ":":
2251                                         if( $stack == 0 ) {
2252                                                 // We found it!
2253                                                 $before = substr( $str, 0, $i );
2254                                                 $after = substr( $str, $i + 1 );
2255                                                 wfProfileOut( $fname );
2256                                                 return $i;
2257                                         }
2258                                         // Embedded in a tag; don't break it.
2259                                         break;
2260                                 default:
2261                                         // Skip ahead looking for something interesting
2262                                         $colon = strpos( $str, ':', $i );
2263                                         if( $colon === false ) {
2264                                                 // Nothing else interesting
2265                                                 wfProfileOut( $fname );
2266                                                 return false;
2267                                         }
2268                                         $lt = strpos( $str, '<', $i );
2269                                         if( $stack === 0 ) {
2270                                                 if( $lt === false || $colon < $lt ) {
2271                                                         // We found it!
2272                                                         $before = substr( $str, 0, $colon );
2273                                                         $after = substr( $str, $colon + 1 );
2274                                                         wfProfileOut( $fname );
2275                                                         return $i;
2276                                                 }
2277                                         }
2278                                         if( $lt === false ) {
2279                                                 // Nothing else interesting to find; abort!
2280                                                 // We're nested, but there's no close tags left. Abort!
2281                                                 break 2;
2282                                         }
2283                                         // Skip ahead to next tag start
2284                                         $i = $lt;
2285                                         $state = MW_COLON_STATE_TAGSTART;
2286                                 }
2287                                 break;
2288                         case 1: // MW_COLON_STATE_TAG:
2289                                 // In a <tag>
2290                                 switch( $c ) {
2291                                 case ">":
2292                                         $stack++;
2293                                         $state = MW_COLON_STATE_TEXT;
2294                                         break;
2295                                 case "/":
2296                                         // Slash may be followed by >?
2297                                         $state = MW_COLON_STATE_TAGSLASH;
2298                                         break;
2299                                 default:
2300                                         // ignore
2301                                 }
2302                                 break;
2303                         case 2: // MW_COLON_STATE_TAGSTART:
2304                                 switch( $c ) {
2305                                 case "/":
2306                                         $state = MW_COLON_STATE_CLOSETAG;
2307                                         break;
2308                                 case "!":
2309                                         $state = MW_COLON_STATE_COMMENT;
2310                                         break;
2311                                 case ">":
2312                                         // Illegal early close? This shouldn't happen D:
2313                                         $state = MW_COLON_STATE_TEXT;
2314                                         break;
2315                                 default:
2316                                         $state = MW_COLON_STATE_TAG;
2317                                 }
2318                                 break;
2319                         case 3: // MW_COLON_STATE_CLOSETAG:
2320                                 // In a </tag>
2321                                 if( $c == ">" ) {
2322                                         $stack--;
2323                                         if( $stack < 0 ) {
2324                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2325                                                 wfProfileOut( $fname );
2326                                                 return false;
2327                                         }
2328                                         $state = MW_COLON_STATE_TEXT;
2329                                 }
2330                                 break;
2331                         case MW_COLON_STATE_TAGSLASH:
2332                                 if( $c == ">" ) {
2333                                         // Yes, a self-closed tag <blah/>
2334                                         $state = MW_COLON_STATE_TEXT;
2335                                 } else {
2336                                         // Probably we're jumping the gun, and this is an attribute
2337                                         $state = MW_COLON_STATE_TAG;
2338                                 }
2339                                 break;
2340                         case 5: // MW_COLON_STATE_COMMENT:
2341                                 if( $c == "-" ) {
2342                                         $state = MW_COLON_STATE_COMMENTDASH;
2343                                 }
2344                                 break;
2345                         case MW_COLON_STATE_COMMENTDASH:
2346                                 if( $c == "-" ) {
2347                                         $state = MW_COLON_STATE_COMMENTDASHDASH;
2348                                 } else {
2349                                         $state = MW_COLON_STATE_COMMENT;
2350                                 }
2351                                 break;
2352                         case MW_COLON_STATE_COMMENTDASHDASH:
2353                                 if( $c == ">" ) {
2354                                         $state = MW_COLON_STATE_TEXT;
2355                                 } else {
2356                                         $state = MW_COLON_STATE_COMMENT;
2357                                 }
2358                                 break;
2359                         default:
2360                                 throw new MWException( "State machine error in $fname" );
2361                         }
2362                 }
2363                 if( $stack > 0 ) {
2364                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2365                         return false;
2366                 }
2367                 wfProfileOut( $fname );
2368                 return false;
2369         }
2370
2371         /**
2372          * Return value of a magic variable (like PAGENAME)
2373          *
2374          * @private
2375          */
2376         function getVariableValue( $index ) {
2377                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2378
2379                 /**
2380                  * Some of these require message or data lookups and can be
2381                  * expensive to check many times.
2382                  */
2383                 static $varCache = array();
2384                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) ) {
2385                         if ( isset( $varCache[$index] ) ) {
2386                                 return $varCache[$index];
2387                         }
2388                 }
2389
2390                 $ts = time();
2391                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2392
2393                 # Use the time zone
2394                 global $wgLocaltimezone;
2395                 if ( isset( $wgLocaltimezone ) ) {
2396                         $oldtz = getenv( 'TZ' );
2397                         putenv( 'TZ='.$wgLocaltimezone );
2398                 }
2399                 $localTimestamp = date( 'YmdHis', $ts );
2400                 $localMonth = date( 'm', $ts );
2401                 $localMonthName = date( 'n', $ts );
2402                 $localDay = date( 'j', $ts );
2403                 $localDay2 = date( 'd', $ts );
2404                 $localDayOfWeek = date( 'w', $ts );
2405                 $localWeek = date( 'W', $ts );
2406                 $localYear = date( 'Y', $ts );
2407                 $localHour = date( 'H', $ts );
2408                 if ( isset( $wgLocaltimezone ) ) {
2409                         putenv( 'TZ='.$oldtz );
2410                 }
2411
2412                 switch ( $index ) {
2413                         case 'currentmonth':
2414                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
2415                         case 'currentmonthname':
2416                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2417                         case 'currentmonthnamegen':
2418                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2419                         case 'currentmonthabbrev':
2420                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2421                         case 'currentday':
2422                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2423                         case 'currentday2':
2424                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2425                         case 'localmonth':
2426                                 return $varCache[$index] = $wgContLang->formatNum( $localMonth );
2427                         case 'localmonthname':
2428                                 return $varCache[$index] = $wgContLang->getMonthName( $localMonthName );
2429                         case 'localmonthnamegen':
2430                                 return $varCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2431                         case 'localmonthabbrev':
2432                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2433                         case 'localday':
2434                                 return $varCache[$index] = $wgContLang->formatNum( $localDay );
2435                         case 'localday2':
2436                                 return $varCache[$index] = $wgContLang->formatNum( $localDay2 );
2437                         case 'pagename':
2438                                 return $this->mTitle->getText();
2439                         case 'pagenamee':
2440                                 return $this->mTitle->getPartialURL();
2441                         case 'fullpagename':
2442                                 return $this->mTitle->getPrefixedText();
2443                         case 'fullpagenamee':
2444                                 return $this->mTitle->getPrefixedURL();
2445                         case 'subpagename':
2446                                 return $this->mTitle->getSubpageText();
2447                         case 'subpagenamee':
2448                                 return $this->mTitle->getSubpageUrlForm();
2449                         case 'basepagename':
2450                                 return $this->mTitle->getBaseText();
2451                         case 'basepagenamee':
2452                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2453                         case 'talkpagename':
2454                                 if( $this->mTitle->canTalk() ) {
2455                                         $talkPage = $this->mTitle->getTalkPage();
2456                                         return $talkPage->getPrefixedText();
2457                                 } else {
2458                                         return '';
2459                                 }
2460                         case 'talkpagenamee':
2461                                 if( $this->mTitle->canTalk() ) {
2462                                         $talkPage = $this->mTitle->getTalkPage();
2463                                         return $talkPage->getPrefixedUrl();
2464                                 } else {
2465                                         return '';
2466                                 }
2467                         case 'subjectpagename':
2468                                 $subjPage = $this->mTitle->getSubjectPage();
2469                                 return $subjPage->getPrefixedText();
2470                         case 'subjectpagenamee':
2471                                 $subjPage = $this->mTitle->getSubjectPage();
2472                                 return $subjPage->getPrefixedUrl();
2473                         case 'revisionid':
2474                                 return $this->mRevisionId;
2475                         case 'revisionday':
2476                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2477                         case 'revisionday2':
2478                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2479                         case 'revisionmonth':
2480                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2481                         case 'revisionyear':
2482                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2483                         case 'revisiontimestamp':
2484                                 return $this->getRevisionTimestamp();
2485                         case 'namespace':
2486                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2487                         case 'namespacee':
2488                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2489                         case 'talkspace':
2490                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2491                         case 'talkspacee':
2492                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2493                         case 'subjectspace':
2494                                 return $this->mTitle->getSubjectNsText();
2495                         case 'subjectspacee':
2496                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2497                         case 'currentdayname':
2498                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2499                         case 'currentyear':
2500                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2501                         case 'currenttime':
2502                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2503                         case 'currenthour':
2504                                 return $varCache[$index] = $wgContLang->formatNum( date( 'H', $ts ), true );
2505                         case 'currentweek':
2506                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2507                                 // int to remove the padding
2508                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2509                         case 'currentdow':
2510                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2511                         case 'localdayname':
2512                                 return $varCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2513                         case 'localyear':
2514                                 return $varCache[$index] = $wgContLang->formatNum( $localYear, true );
2515                         case 'localtime':
2516                                 return $varCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2517                         case 'localhour':
2518                                 return $varCache[$index] = $wgContLang->formatNum( $localHour, true );
2519                         case 'localweek':
2520                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2521                                 // int to remove the padding
2522                                 return $varCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2523                         case 'localdow':
2524                                 return $varCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2525                         case 'numberofarticles':
2526                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2527                         case 'numberoffiles':
2528                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2529                         case 'numberofusers':
2530                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2531                         case 'numberofpages':
2532                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2533                         case 'numberofadmins':
2534                                 return $varCache[$index]  = $wgContLang->formatNum( SiteStats::admins() );
2535                         case 'currenttimestamp':
2536                                 return $varCache[$index] = wfTimestampNow();
2537                         case 'localtimestamp':
2538                                 return $varCache[$index] = $localTimestamp;
2539                         case 'currentversion':
2540                                 return $varCache[$index] = SpecialVersion::getVersion();
2541                         case 'sitename':
2542                                 return $wgSitename;
2543                         case 'server':
2544                                 return $wgServer;
2545                         case 'servername':
2546                                 return $wgServerName;
2547                         case 'scriptpath':
2548                                 return $wgScriptPath;
2549                         case 'directionmark':
2550                                 return $wgContLang->getDirMark();
2551                         case 'contentlanguage':
2552                                 global $wgContLanguageCode;
2553                                 return $wgContLanguageCode;
2554                         default:
2555                                 $ret = null;
2556                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2557                                         return $ret;
2558                                 else
2559                                         return null;
2560                 }
2561         }
2562
2563         /**
2564          * initialise the magic variables (like CURRENTMONTHNAME)
2565          *
2566          * @private
2567          */
2568         function initialiseVariables() {
2569                 $fname = 'Parser::initialiseVariables';
2570                 wfProfileIn( $fname );
2571                 $variableIDs = MagicWord::getVariableIDs();
2572
2573                 $this->mVariables = array();
2574                 foreach ( $variableIDs as $id ) {
2575                         $mw =& MagicWord::get( $id );
2576                         $mw->addToArray( $this->mVariables, $id );
2577                 }
2578                 wfProfileOut( $fname );
2579         }
2580
2581         /**
2582          * parse any parentheses in format ((title|part|part))
2583          * and call callbacks to get a replacement text for any found piece
2584          *
2585          * @param string $text The text to parse
2586          * @param array $callbacks rules in form:
2587          *     '{' => array(                            # opening parentheses
2588          *                                      'end' => '}',   # closing parentheses
2589          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2590          *                                                                3 => callback         # replacement callback to call if {{{..}}} is found
2591          *                                                                )
2592          *                                      )
2593          *                                      'min' => 2,     # Minimum parenthesis count in cb
2594          *                                      'max' => 3,     # Maximum parenthesis count in cb
2595          * @private
2596          */
2597         function replace_callback ($text, $callbacks) {
2598                 wfProfileIn( __METHOD__ );
2599                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2600                 $lastOpeningBrace = -1;                 # last not closed parentheses
2601
2602                 $validOpeningBraces = implode( '', array_keys( $callbacks ) );
2603
2604                 $i = 0;
2605                 while ( $i < strlen( $text ) ) {
2606                         # Find next opening brace, closing brace or pipe
2607                         if ( $lastOpeningBrace == -1 ) {
2608                                 $currentClosing = '';
2609                                 $search = $validOpeningBraces;
2610                         } else {
2611                                 $currentClosing = $openingBraceStack[$lastOpeningBrace]['braceEnd'];
2612                                 $search = $validOpeningBraces . '|' . $currentClosing;
2613                         }
2614                         $rule = null;
2615                         $i += strcspn( $text, $search, $i );
2616                         if ( $i < strlen( $text ) ) {
2617                                 if ( $text[$i] == '|' ) {
2618                                         $found = 'pipe';
2619                                 } elseif ( $text[$i] == $currentClosing ) {
2620                                         $found = 'close';
2621                                 } elseif ( isset( $callbacks[$text[$i]] ) ) {
2622                                         $found = 'open';
2623                                         $rule = $callbacks[$text[$i]];
2624                                 } else {
2625                                         # Some versions of PHP have a strcspn which stops on null characters
2626                                         # Ignore and continue
2627                                         ++$i;
2628                                         continue;
2629                                 }
2630                         } else {
2631                                 # All done
2632                                 break;
2633                         }
2634
2635                         if ( $found == 'open' ) {
2636                                 # found opening brace, let's add it to parentheses stack
2637                                 $piece = array('brace' => $text[$i],
2638                                                            'braceEnd' => $rule['end'],
2639                                                            'title' => '',
2640                                                            'parts' => null);
2641
2642                                 # count opening brace characters
2643                                 $piece['count'] = strspn( $text, $piece['brace'], $i );
2644                                 $piece['startAt'] = $piece['partStart'] = $i + $piece['count'];
2645                                 $i += $piece['count'];
2646
2647                                 # we need to add to stack only if opening brace count is enough for one of the rules
2648                                 if ( $piece['count'] >= $rule['min'] ) {
2649                                         $lastOpeningBrace ++;
2650                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2651                                 }
2652                         } elseif ( $found == 'close' ) {
2653                                 # lets check if it is enough characters for closing brace
2654                                 $maxCount = $openingBraceStack[$lastOpeningBrace]['count'];
2655                                 $count = strspn( $text, $text[$i], $i, $maxCount );
2656
2657                                 # check for maximum matching characters (if there are 5 closing
2658                                 # characters, we will probably need only 3 - depending on the rules)
2659                                 $matchingCount = 0;
2660                                 $matchingCallback = null;
2661                                 $cbType = $callbacks[$openingBraceStack[$lastOpeningBrace]['brace']];
2662                                 if ( $count > $cbType['max'] ) {
2663                                         # The specified maximum exists in the callback array, unless the caller
2664                                         # has made an error
2665                                         $matchingCount = $cbType['max'];
2666                                 } else {
2667                                         # Count is less than the maximum
2668                                         # Skip any gaps in the callback array to find the true largest match
2669                                         # Need to use array_key_exists not isset because the callback can be null
2670                                         $matchingCount = $count;
2671                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $cbType['cb'] ) ) {
2672                                                 --$matchingCount;
2673                                         }
2674                                 }
2675
2676                                 if ($matchingCount <= 0) {
2677                                         $i += $count;
2678                                         continue;
2679                                 }
2680                                 $matchingCallback = $cbType['cb'][$matchingCount];
2681
2682                                 # let's set a title or last part (if '|' was found)
2683                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2684                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2685                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2686                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2687                                 } else {
2688                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2689                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2690                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2691                                 }
2692
2693                                 $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2694                                 $pieceEnd = $i + $matchingCount;
2695
2696                                 if( is_callable( $matchingCallback ) ) {
2697                                         $cbArgs = array (
2698                                                                          'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2699                                                                          'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2700                                                                          'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2701                                                                          'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == "\n")),
2702                                                                          );
2703                                         # finally we can call a user callback and replace piece of text
2704                                         $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2705                                         $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2706                                         $i = $pieceStart + strlen($replaceWith);
2707                                 } else {
2708                                         # null value for callback means that parentheses should be parsed, but not replaced
2709                                         $i += $matchingCount;
2710                                 }
2711
2712                                 # reset last opening parentheses, but keep it in case there are unused characters
2713                                 $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2714                                                            'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2715                                                            'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2716                                                            'title' => '',
2717                                                            'parts' => null,
2718                                                            'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2719                                 $openingBraceStack[$lastOpeningBrace--] = null;
2720
2721                                 if ($matchingCount < $piece['count']) {
2722                                         $piece['count'] -= $matchingCount;
2723                                         $piece['startAt'] -= $matchingCount;
2724                                         $piece['partStart'] = $piece['startAt'];
2725                                         # do we still qualify for any callback with remaining count?
2726                                         $currentCbList = $callbacks[$piece['brace']]['cb'];
2727                                         while ( $piece['count'] ) {
2728                                                 if ( array_key_exists( $piece['count'], $currentCbList ) ) {
2729                                                         $lastOpeningBrace++;
2730                                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2731                                                         break;
2732                                                 }
2733                                                 --$piece['count'];
2734                                         }
2735                                 }
2736                         } elseif ( $found == 'pipe' ) {
2737                                 # lets set a title if it is a first separator, or next part otherwise
2738                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2739                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2740                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2741                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2742                                         $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2743                                 } else {
2744                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2745                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2746                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2747                                 }
2748                                 $openingBraceStack[$lastOpeningBrace]['partStart'] = ++$i;
2749                         }
2750                 }
2751
2752                 wfProfileOut( __METHOD__ );
2753                 return $text;
2754         }
2755
2756         /**
2757          * Replace magic variables, templates, and template arguments
2758          * with the appropriate text. Templates are substituted recursively,
2759          * taking care to avoid infinite loops.
2760          *
2761          * Note that the substitution depends on value of $mOutputType:
2762          *  OT_WIKI: only {{subst:}} templates
2763          *  OT_MSG: only magic variables
2764          *  OT_HTML: all templates and magic variables
2765          *
2766          * @param string $tex The text to transform
2767          * @param array $args Key-value pairs representing template parameters to substitute
2768          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2769          * @private
2770          */
2771         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2772                 # Prevent too big inclusions
2773                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2774                         return $text;
2775                 }
2776
2777                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2778                 wfProfileIn( $fname );
2779
2780                 # This function is called recursively. To keep track of arguments we need a stack:
2781                 array_push( $this->mArgStack, $args );
2782
2783                 $braceCallbacks = array();
2784                 if ( !$argsOnly ) {
2785                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2786                 }
2787                 if ( $this->mOutputType != OT_MSG ) {
2788                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2789                 }
2790                 if ( $braceCallbacks ) {
2791                         $callbacks = array(
2792                                 '{' => array(
2793                                         'end' => '}',
2794                                         'cb' => $braceCallbacks,
2795                                         'min' => $argsOnly ? 3 : 2,
2796                                         'max' => isset( $braceCallbacks[3] ) ? 3 : 2,
2797                                 ),
2798                                 '[' => array(
2799                                         'end' => ']',
2800                                         'cb' => array(2=>null),
2801                                         'min' => 2,
2802                                         'max' => 2,
2803                                 )
2804                         );
2805                         $text = $this->replace_callback ($text, $callbacks);
2806
2807                         array_pop( $this->mArgStack );
2808                 }
2809                 wfProfileOut( $fname );
2810                 return $text;
2811         }
2812
2813         /**
2814          * Replace magic variables
2815          * @private
2816          */
2817         function variableSubstitution( $matches ) {
2818                 global $wgContLang;
2819                 $fname = 'Parser::variableSubstitution';
2820                 $varname = $wgContLang->lc($matches[1]);
2821                 wfProfileIn( $fname );
2822                 $skip = false;
2823                 if ( $this->mOutputType == OT_WIKI ) {
2824                         # Do only magic variables prefixed by SUBST
2825                         $mwSubst =& MagicWord::get( 'subst' );
2826                         if (!$mwSubst->matchStartAndRemove( $varname ))
2827                                 $skip = true;
2828                         # Note that if we don't substitute the variable below,
2829                         # we don't remove the {{subst:}} magic word, in case
2830                         # it is a template rather than a magic variable.
2831                 }
2832                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2833                         $id = $this->mVariables[$varname];
2834                         # Now check if we did really match, case sensitive or not
2835                         $mw =& MagicWord::get( $id );
2836                         if ($mw->match($matches[1])) {
2837                                 $text = $this->getVariableValue( $id );
2838                                 $this->mOutput->mContainsOldMagic = true;
2839                         } else {
2840                                 $text = $matches[0];
2841                         }
2842                 } else {
2843                         $text = $matches[0];
2844                 }
2845                 wfProfileOut( $fname );
2846                 return $text;
2847         }
2848
2849
2850         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2851         static function createAssocArgs( $args ) {
2852                 $assocArgs = array();
2853                 $index = 1;
2854                 foreach( $args as $arg ) {
2855                         $eqpos = strpos( $arg, '=' );
2856                         if ( $eqpos === false ) {
2857                                 $assocArgs[$index++] = $arg;
2858                         } else {
2859                                 $name = trim( substr( $arg, 0, $eqpos ) );
2860                                 $value = trim( substr( $arg, $eqpos+1 ) );
2861                                 if ( $value === false ) {
2862                                         $value = '';
2863                                 }
2864                                 if ( $name !== false ) {
2865                                         $assocArgs[$name] = $value;
2866                                 }
2867                         }
2868                 }
2869
2870                 return $assocArgs;
2871         }
2872
2873         /**
2874          * Return the text of a template, after recursively
2875          * replacing any variables or templates within the template.
2876          *
2877          * @param array $piece The parts of the template
2878          *  $piece['text']: matched text
2879          *  $piece['title']: the title, i.e. the part before the |
2880          *  $piece['parts']: the parameter array
2881          * @return string the text of the template
2882          * @private
2883          */
2884         function braceSubstitution( $piece ) {
2885                 global $wgContLang, $wgLang, $wgAllowDisplayTitle;
2886                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2887                 wfProfileIn( $fname );
2888                 wfProfileIn( __METHOD__.'-setup' );
2889
2890                 # Flags
2891                 $found = false;             # $text has been filled
2892                 $nowiki = false;            # wiki markup in $text should be escaped
2893                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2894                 $noargs = false;            # Don't replace triple-brace arguments in $text
2895                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2896                 $headingOffset = 0;         # Skip headings when number, to account for those that weren't transcluded.
2897                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2898                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2899
2900                 # Title object, where $text came from
2901                 $title = NULL;
2902
2903                 $linestart = '';
2904
2905
2906                 # $part1 is the bit before the first |, and must contain only title characters
2907                 # $args is a list of arguments, starting from index 0, not including $part1
2908
2909                 $titleText = $part1 = $piece['title'];
2910                 # If the third subpattern matched anything, it will start with |
2911
2912                 if (null == $piece['parts']) {
2913                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2914                         if ($replaceWith != $piece['text']) {
2915                                 $text = $replaceWith;
2916                                 $found = true;
2917                                 $noparse = true;
2918                                 $noargs = true;
2919                         }
2920                 }
2921
2922                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2923                 wfProfileOut( __METHOD__.'-setup' );
2924
2925                 # SUBST
2926                 wfProfileIn( __METHOD__.'-modifiers' );
2927                 if ( !$found ) {
2928                         $mwSubst =& MagicWord::get( 'subst' );
2929                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2930                                 # One of two possibilities is true:
2931                                 # 1) Found SUBST but not in the PST phase
2932                                 # 2) Didn't find SUBST and in the PST phase
2933                                 # In either case, return without further processing
2934                                 $text = $piece['text'];
2935                                 $found = true;
2936                                 $noparse = true;
2937                                 $noargs = true;
2938                         }
2939                 }
2940
2941                 # MSG, MSGNW and RAW
2942                 if ( !$found ) {
2943                         # Check for MSGNW:
2944                         $mwMsgnw =& MagicWord::get( 'msgnw' );
2945                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2946                                 $nowiki = true;
2947                         } else {
2948                                 # Remove obsolete MSG:
2949                                 $mwMsg =& MagicWord::get( 'msg' );
2950                                 $mwMsg->matchStartAndRemove( $part1 );
2951                         }
2952
2953                         # Check for RAW:
2954                         $mwRaw =& MagicWord::get( 'raw' );
2955                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2956                                 $forceRawInterwiki = true;
2957                         }
2958                 }
2959                 wfProfileOut( __METHOD__.'-modifiers' );
2960
2961                 //save path level before recursing into functions & templates.
2962                 $lastPathLevel = $this->mTemplatePath;
2963
2964                 # Parser functions
2965                 if ( !$found ) {
2966                         wfProfileIn( __METHOD__ . '-pfunc' );
2967
2968                         $colonPos = strpos( $part1, ':' );
2969                         if ( $colonPos !== false ) {
2970                                 # Case sensitive functions
2971                                 $function = substr( $part1, 0, $colonPos );
2972                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2973                                         $function = $this->mFunctionSynonyms[1][$function];
2974                                 } else {
2975                                         # Case insensitive functions
2976                                         $function = strtolower( $function );
2977                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2978                                                 $function = $this->mFunctionSynonyms[0][$function];
2979                                         } else {
2980                                                 $function = false;
2981                                         }
2982                                 }
2983                                 if ( $function ) {
2984                                         $funcArgs = array_map( 'trim', $args );
2985                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
2986                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
2987                                         $found = true;
2988
2989                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
2990                                         //$noargs = true;
2991                                         //$noparse = true;
2992
2993                                         if ( is_array( $result ) ) {
2994                                                 if ( isset( $result[0] ) ) {
2995                                                         $text = $linestart . $result[0];
2996                                                         unset( $result[0] );
2997                                                 }
2998
2999                                                 // Extract flags into the local scope
3000                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
3001                                                 extract( $result );
3002                                         } else {
3003                                                 $text = $linestart . $result;
3004                                         }
3005                                 }
3006                         }
3007                         wfProfileOut( __METHOD__ . '-pfunc' );
3008                 }
3009
3010                 # Template table test
3011
3012                 # Did we encounter this template already? If yes, it is in the cache
3013                 # and we need to check for loops.
3014                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
3015                         $found = true;
3016
3017                         # Infinite loop test
3018                         if ( isset( $this->mTemplatePath[$part1] ) ) {
3019                                 $noparse = true;
3020                                 $noargs = true;
3021                                 $found = true;
3022                                 $text = $linestart .
3023                                         "[[$part1]]<!-- WARNING: template loop detected -->";
3024                                 wfDebug( __METHOD__.": template loop broken at '$part1'\n" );
3025                         } else {
3026                                 # set $text to cached message.
3027                                 $text = $linestart . $this->mTemplates[$piece['title']];
3028                         }
3029                 }
3030
3031                 # Load from database
3032                 if ( !$found ) {
3033                         wfProfileIn( __METHOD__ . '-loadtpl' );
3034                         $ns = NS_TEMPLATE;
3035                         # declaring $subpage directly in the function call
3036                         # does not work correctly with references and breaks
3037                         # {{/subpage}}-style inclusions
3038                         $subpage = '';
3039                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3040                         if ($subpage !== '') {
3041                                 $ns = $this->mTitle->getNamespace();
3042                         }
3043                         $title = Title::newFromText( $part1, $ns );
3044
3045
3046                         if ( !is_null( $title ) ) {
3047                                 $titleText = $title->getPrefixedText();
3048                                 # Check for language variants if the template is not found
3049                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
3050                                         $wgContLang->findVariantLink($part1, $title);
3051                                 }
3052
3053                                 if ( !$title->isExternal() ) {
3054                                         if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
3055                                                 $text = SpecialPage::capturePath( $title );
3056                                                 if ( is_string( $text ) ) {
3057                                                         $found = true;
3058                                                         $noparse = true;
3059                                                         $noargs = true;
3060                                                         $isHTML = true;
3061                                                         $this->disableCache();
3062                                                 }
3063                                         } else {
3064                                                 $articleContent = $this->fetchTemplate( $title );
3065                                                 if ( $articleContent !== false ) {
3066                                                         $found = true;
3067                                                         $text = $articleContent;
3068                                                         $replaceHeadings = true;
3069                                                 }
3070                                         }
3071
3072                                         # If the title is valid but undisplayable, make a link to it
3073                                         if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3074                                                 $text = "[[:$titleText]]";
3075                                                 $found = true;
3076                                         }
3077                                 } elseif ( $title->isTrans() ) {
3078                                         // Interwiki transclusion
3079                                         if ( $this->ot['html'] && !$forceRawInterwiki ) {
3080                                                 $text = $this->interwikiTransclude( $title, 'render' );
3081                                                 $isHTML = true;
3082                                                 $noparse = true;
3083                                         } else {
3084                                                 $text = $this->interwikiTransclude( $title, 'raw' );
3085                                                 $replaceHeadings = true;
3086                                         }
3087                                         $found = true;
3088                                 }
3089
3090                                 # Template cache array insertion
3091                                 # Use the original $piece['title'] not the mangled $part1, so that
3092                                 # modifiers such as RAW: produce separate cache entries
3093                                 if( $found ) {
3094                                         if( $isHTML ) {
3095                                                 // A special page; don't store it in the template cache.
3096                                         } else {
3097                                                 $this->mTemplates[$piece['title']] = $text;
3098                                         }
3099                                         $text = $linestart . $text;
3100                                 }
3101                         }
3102                         wfProfileOut( __METHOD__ . '-loadtpl' );
3103                 }
3104
3105                 if ( $found && !$this->incrementIncludeSize( 'pre-expand', strlen( $text ) ) ) {
3106                         # Error, oversize inclusion
3107                         $text = $linestart .
3108                                 "[[$titleText]]<!-- WARNING: template omitted, pre-expand include size too large -->";
3109                         $noparse = true;
3110                         $noargs = true;
3111                 }
3112
3113                 # Recursive parsing, escaping and link table handling
3114                 # Only for HTML output
3115                 if ( $nowiki && $found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3116                         $text = wfEscapeWikiText( $text );
3117                 } elseif ( !$this->ot['msg'] && $found ) {
3118                         if ( $noargs ) {
3119                                 $assocArgs = array();
3120                         } else {
3121                                 # Clean up argument array
3122                                 $assocArgs = self::createAssocArgs($args);
3123                                 # Add a new element to the templace recursion path
3124                                 $this->mTemplatePath[$part1] = 1;
3125                         }
3126
3127                         if ( !$noparse ) {
3128                                 # If there are any <onlyinclude> tags, only include them
3129                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
3130                                         $replacer = new OnlyIncludeReplacer;
3131                                         StringUtils::delimiterReplaceCallback( '<onlyinclude>', '</onlyinclude>',
3132                                                 array( &$replacer, 'replace' ), $text );
3133                                         $text = $replacer->output;
3134                                 }
3135                                 # Remove <noinclude> sections and <includeonly> tags
3136                                 $text = StringUtils::delimiterReplace( '<noinclude>', '</noinclude>', '', $text );
3137                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
3138
3139                                 if( $this->ot['html'] || $this->ot['pre'] ) {
3140                                         # Strip <nowiki>, <pre>, etc.
3141                                         $text = $this->strip( $text, $this->mStripState );
3142                                         if ( $this->ot['html'] ) {
3143                                                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
3144                                         } elseif ( $this->ot['pre'] && $this->mOptions->getRemoveComments() ) {
3145                                                 $text = Sanitizer::removeHTMLcomments( $text );
3146                                         }
3147                                 }
3148                                 $text = $this->replaceVariables( $text, $assocArgs );
3149
3150                                 # If the template begins with a table or block-level
3151                                 # element, it should be treated as beginning a new line.
3152                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) /*}*/{
3153                                         $text = "\n" . $text;
3154                                 }
3155                         } elseif ( !$noargs ) {
3156                                 # $noparse and !$noargs
3157                                 # Just replace the arguments, not any double-brace items
3158                                 # This is used for rendered interwiki transclusion
3159                                 $text = $this->replaceVariables( $text, $assocArgs, true );
3160                         }
3161                 }
3162                 # Prune lower levels off the recursion check path
3163                 $this->mTemplatePath = $lastPathLevel;
3164
3165                 if ( $found && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3166                         # Error, oversize inclusion
3167                         $text = $linestart .
3168                                 "[[$titleText]]<!-- WARNING: template omitted, post-expand include size too large -->";
3169                         $noparse = true;
3170                         $noargs = true;
3171                 }
3172
3173                 if ( !$found ) {
3174                         wfProfileOut( $fname );
3175                         return $piece['text'];
3176                 } else {
3177                         wfProfileIn( __METHOD__ . '-placeholders' );
3178                         if ( $isHTML ) {
3179                                 # Replace raw HTML by a placeholder
3180                                 # Add a blank line preceding, to prevent it from mucking up
3181                                 # immediately preceding headings
3182                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
3183                         } else {
3184                                 # replace ==section headers==
3185                                 # XXX this needs to go away once we have a better parser.
3186                                 if ( !$this->ot['wiki'] && !$this->ot['pre'] && $replaceHeadings ) {
3187                                         if( !is_null( $title ) )
3188                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
3189                                         else
3190                                                 $encodedname = base64_encode("");
3191                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
3192                                                 PREG_SPLIT_DELIM_CAPTURE);
3193                                         $text = '';
3194                                         $nsec = $headingOffset;
3195                                         for( $i = 0; $i < count($m); $i += 2 ) {
3196                                                 $text .= $m[$i];
3197                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
3198                                                 $hl = $m[$i + 1];
3199                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
3200                                                         $text .= $hl;
3201                                                         continue;
3202                                                 }
3203                                                 $m2 = array();
3204                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
3205                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
3206                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
3207
3208                                                 $nsec++;
3209                                         }
3210                                 }
3211                         }
3212                         wfProfileOut( __METHOD__ . '-placeholders' );
3213                 }
3214
3215                 # Prune lower levels off the recursion check path
3216                 $this->mTemplatePath = $lastPathLevel;
3217
3218                 if ( !$found ) {
3219                         wfProfileOut( $fname );
3220                         return $piece['text'];
3221                 } else {
3222                         wfProfileOut( $fname );
3223                         return $text;
3224                 }
3225         }
3226
3227         /**
3228          * Fetch the unparsed text of a template and register a reference to it.
3229          */
3230         function fetchTemplate( $title ) {
3231                 $text = false;
3232                 // Loop to fetch the article, with up to 1 redirect
3233                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3234                         $rev = Revision::newFromTitle( $title );
3235                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
3236                         if ( !$rev ) {
3237                                 break;
3238                         }
3239                         $text = $rev->getText();
3240                         if ( $text === false ) {
3241                                 break;
3242                         }
3243                         // Redirect?
3244                         $title = Title::newFromRedirect( $text );
3245                 }
3246                 return $text;
3247         }
3248
3249         /**
3250          * Transclude an interwiki link.
3251          */
3252         function interwikiTransclude( $title, $action ) {
3253                 global $wgEnableScaryTranscluding;
3254
3255                 if (!$wgEnableScaryTranscluding)
3256                         return wfMsg('scarytranscludedisabled');
3257
3258                 $url = $title->getFullUrl( "action=$action" );
3259
3260                 if (strlen($url) > 255)
3261                         return wfMsg('scarytranscludetoolong');
3262                 return $this->fetchScaryTemplateMaybeFromCache($url);
3263         }
3264
3265         function fetchScaryTemplateMaybeFromCache($url) {
3266                 global $wgTranscludeCacheExpiry;
3267                 $dbr =& wfGetDB(DB_SLAVE);
3268                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3269                                 array('tc_url' => $url));
3270                 if ($obj) {
3271                         $time = $obj->tc_time;
3272                         $text = $obj->tc_contents;
3273                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3274                                 return $text;
3275                         }
3276                 }
3277
3278                 $text = Http::get($url);
3279                 if (!$text)
3280                         return wfMsg('scarytranscludefailed', $url);
3281
3282                 $dbw =& wfGetDB(DB_MASTER);
3283                 $dbw->replace('transcache', array('tc_url'), array(
3284                         'tc_url' => $url,
3285                         'tc_time' => time(),
3286                         'tc_contents' => $text));
3287                 return $text;
3288         }
3289
3290
3291         /**
3292          * Triple brace replacement -- used for template arguments
3293          * @private
3294          */
3295         function argSubstitution( $matches ) {
3296                 $arg = trim( $matches['title'] );
3297                 $text = $matches['text'];
3298                 $inputArgs = end( $this->mArgStack );
3299
3300                 if ( array_key_exists( $arg, $inputArgs ) ) {
3301                         $text = $inputArgs[$arg];
3302                 } else if (($this->mOutputType == OT_HTML || $this->mOutputType == OT_PREPROCESS ) &&
3303                 null != $matches['parts'] && count($matches['parts']) > 0) {
3304                         $text = $matches['parts'][0];
3305                 }
3306                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3307                         $text = $matches['text'] .
3308                                 '<!-- WARNING: argument omitted, expansion size too large -->';
3309                 }
3310
3311                 return $text;
3312         }
3313
3314         /**
3315          * Increment an include size counter
3316          *
3317          * @param string $type The type of expansion
3318          * @param integer $size The size of the text
3319          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3320          */
3321         function incrementIncludeSize( $type, $size ) {
3322                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3323                         return false;
3324                 } else {
3325                         $this->mIncludeSizes[$type] += $size;
3326                         return true;
3327                 }
3328         }
3329
3330         /**
3331          * Detect __NOGALLERY__ magic word and set a placeholder
3332          */
3333         function stripNoGallery( &$text ) {
3334                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3335                 # do not add TOC
3336                 $mw = MagicWord::get( 'nogallery' );
3337                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3338         }
3339
3340         /**
3341          * Detect __TOC__ magic word and set a placeholder
3342          */
3343         function stripToc( $text ) {
3344                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3345                 # do not add TOC
3346                 $mw = MagicWord::get( 'notoc' );
3347                 if( $mw->matchAndRemove( $text ) ) {
3348                         $this->mShowToc = false;
3349                 }
3350
3351                 $mw = MagicWord::get( 'toc' );
3352                 if( $mw->match( $text ) ) {
3353                         $this->mShowToc = true;
3354                         $this->mForceTocPosition = true;
3355
3356                         // Set a placeholder. At the end we'll fill it in with the TOC.
3357                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3358
3359                         // Only keep the first one.
3360                         $text = $mw->replace( '', $text );
3361                 }
3362                 return $text;
3363         }
3364
3365         /**
3366          * This function accomplishes several tasks:
3367          * 1) Auto-number headings if that option is enabled
3368          * 2) Add an [edit] link to sections for logged in users who have enabled the option
3369          * 3) Add a Table of contents on the top for users who have enabled the option
3370          * 4) Auto-anchor headings
3371          *
3372          * It loops through all headlines, collects the necessary data, then splits up the
3373          * string and re-inserts the newly formatted headlines.
3374          *
3375          * @param string $text
3376          * @param boolean $isMain
3377          * @private
3378          */
3379         function formatHeadings( $text, $isMain=true ) {
3380                 global $wgMaxTocLevel, $wgContLang;
3381
3382                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3383                 if( !$this->mTitle->userCanEdit() ) {
3384                         $showEditLink = 0;
3385                 } else {
3386                         $showEditLink = $this->mOptions->getEditSection();
3387                 }
3388
3389                 # Inhibit editsection links if requested in the page
3390                 $esw =& MagicWord::get( 'noeditsection' );
3391                 if( $esw->matchAndRemove( $text ) ) {
3392                         $showEditLink = 0;
3393                 }
3394
3395                 # Get all headlines for numbering them and adding funky stuff like [edit]
3396                 # links - this is for later, but we need the number of headlines right now
3397                 $matches = array();
3398                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
3399
3400                 # if there are fewer than 4 headlines in the article, do not show TOC
3401                 # unless it's been explicitly enabled.
3402                 $enoughToc = $this->mShowToc &&
3403                         (($numMatches >= 4) || $this->mForceTocPosition);
3404
3405                 # Allow user to stipulate that a page should have a "new section"
3406                 # link added via __NEWSECTIONLINK__
3407                 $mw =& MagicWord::get( 'newsectionlink' );
3408                 if( $mw->matchAndRemove( $text ) )
3409                         $this->mOutput->setNewSection( true );
3410
3411                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3412                 # override above conditions and always show TOC above first header
3413                 $mw =& MagicWord::get( 'forcetoc' );
3414                 if ($mw->matchAndRemove( $text ) ) {
3415                         $this->mShowToc = true;
3416                         $enoughToc = true;
3417                 }
3418
3419                 # Never ever show TOC if no headers
3420                 if( $numMatches < 1 ) {
3421                         $enoughToc = false;
3422                 }
3423
3424                 # We need this to perform operations on the HTML
3425                 $sk =& $this->mOptions->getSkin();
3426
3427                 # headline counter
3428                 $headlineCount = 0;
3429                 $sectionCount = 0; # headlineCount excluding template sections
3430
3431                 # Ugh .. the TOC should have neat indentation levels which can be
3432                 # passed to the skin functions. These are determined here
3433                 $toc = '';
3434                 $full = '';
3435                 $head = array();
3436                 $sublevelCount = array();
3437                 $levelCount = array();
3438                 $toclevel = 0;
3439                 $level = 0;
3440                 $prevlevel = 0;
3441                 $toclevel = 0;
3442                 $prevtoclevel = 0;
3443
3444                 foreach( $matches[3] as $headline ) {
3445                         $istemplate = 0;
3446                         $templatetitle = '';
3447                         $templatesection = 0;
3448                         $numbering = '';
3449                         $mat = array();
3450                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3451                                 $istemplate = 1;
3452                                 $templatetitle = base64_decode($mat[1]);
3453                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3454                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3455                         }
3456
3457                         if( $toclevel ) {
3458                                 $prevlevel = $level;
3459                                 $prevtoclevel = $toclevel;
3460                         }
3461                         $level = $matches[1][$headlineCount];
3462
3463                         if( $doNumberHeadings || $enoughToc ) {
3464
3465                                 if ( $level > $prevlevel ) {
3466                                         # Increase TOC level
3467                                         $toclevel++;
3468                                         $sublevelCount[$toclevel] = 0;
3469                                         if( $toclevel<$wgMaxTocLevel ) {
3470                                                 $toc .= $sk->tocIndent();
3471                                         }
3472                                 }
3473                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3474                                         # Decrease TOC level, find level to jump to
3475
3476                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3477                                                 # Can only go down to level 1
3478                                                 $toclevel = 1;
3479                                         } else {
3480                                                 for ($i = $toclevel; $i > 0; $i--) {
3481                                                         if ( $levelCount[$i] == $level ) {
3482                                                                 # Found last matching level
3483                                                                 $toclevel = $i;
3484                                                                 break;
3485                                                         }
3486                                                         elseif ( $levelCount[$i] < $level ) {
3487                                                                 # Found first matching level below current level
3488                                                                 $toclevel = $i + 1;
3489                                                                 break;
3490                                                         }
3491                                                 }
3492                                         }
3493                                         if( $toclevel<$wgMaxTocLevel ) {
3494                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3495                                         }
3496                                 }
3497                                 else {
3498                                         # No change in level, end TOC line
3499                                         if( $toclevel<$wgMaxTocLevel ) {
3500                                                 $toc .= $sk->tocLineEnd();
3501                                         }
3502                                 }
3503
3504                                 $levelCount[$toclevel] = $level;
3505
3506                                 # count number of headlines for each level
3507                                 @$sublevelCount[$toclevel]++;
3508                                 $dot = 0;
3509                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3510                                         if( !empty( $sublevelCount[$i] ) ) {
3511                                                 if( $dot ) {
3512                                                         $numbering .= '.';
3513                                                 }
3514                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3515                                                 $dot = 1;
3516                                         }
3517                                 }
3518                         }
3519
3520                         # The canonized header is a version of the header text safe to use for links
3521                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3522                         $canonized_headline = $this->mStripState->unstripBoth( $headline );
3523
3524                         # Remove link placeholders by the link text.
3525                         #     <!--LINK number-->
3526                         # turns into
3527                         #     link text with suffix
3528                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3529                                                             "\$this->mLinkHolders['texts'][\$1]",
3530                                                             $canonized_headline );
3531                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3532                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3533                                                             $canonized_headline );
3534
3535                         # strip out HTML
3536                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3537                         $tocline = trim( $canonized_headline );
3538                         # Save headline for section edit hint before it's escaped
3539                         $headline_hint = trim( $canonized_headline );
3540                         $canonized_headline = Sanitizer::escapeId( $tocline );
3541                         $refers[$headlineCount] = $canonized_headline;
3542
3543                         # count how many in assoc. array so we can track dupes in anchors
3544                         isset( $refers[$canonized_headline] ) ? $refers[$canonized_headline]++ : $refers[$canonized_headline] = 1;
3545                         $refcount[$headlineCount]=$refers[$canonized_headline];
3546
3547                         # Don't number the heading if it is the only one (looks silly)
3548                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3549                                 # the two are different if the line contains a link
3550                                 $headline=$numbering . ' ' . $headline;
3551                         }
3552
3553                         # Create the anchor for linking from the TOC to the section
3554                         $anchor = $canonized_headline;
3555                         if($refcount[$headlineCount] > 1 ) {
3556                                 $anchor .= '_' . $refcount[$headlineCount];
3557                         }
3558                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3559                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3560                         }
3561                         # give headline the correct <h#> tag
3562                         $head[$headlineCount] = "<a name=\"$anchor\"></a><h".$level.$matches[2][$headlineCount];
3563
3564                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3565                                 if ( empty( $head[$headlineCount] ) ) {
3566                                         $head[$headlineCount] = '';
3567                                 }
3568                                 if( $istemplate )
3569                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
3570                                 else
3571                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3572                         }
3573                         // Yes, the headline logically goes before the edit section.  Why isn't it there
3574                         // in source?  Ask the CSS people.  The float gets screwed up if you do that.
3575                         // This might be moved to before the editsection at some point so that it will
3576                         // display a bit more prettily without CSS, so please don't rely on the order.
3577                         $head[$headlineCount] .= ' <span class="mw-headline">'.$headline.'</span></h'.$level.'>';
3578
3579                         $headlineCount++;
3580                         if( !$istemplate )
3581                                 $sectionCount++;
3582                 }
3583
3584                 if( $enoughToc ) {
3585                         if( $toclevel<$wgMaxTocLevel ) {
3586                                 $toc .= $sk->tocUnindent( $toclevel - 1 );
3587                         }
3588                         $toc = $sk->tocList( $toc );
3589                 }
3590
3591                 # split up and insert constructed headlines
3592
3593                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3594                 $i = 0;
3595
3596                 foreach( $blocks as $block ) {
3597                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3598                                 # This is the [edit] link that appears for the top block of text when
3599                                 # section editing is enabled
3600
3601                                 # Disabled because it broke block formatting
3602                                 # For example, a bullet point in the top line
3603                                 # $full .= $sk->editSectionLink(0);
3604                         }
3605                         $full .= $block;
3606                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3607                                 # Top anchor now in skin
3608                                 $full = $full.$toc;
3609                         }
3610
3611                         if( !empty( $head[$i] ) ) {
3612                                 $full .= $head[$i];
3613                         }
3614                         $i++;
3615                 }
3616                 if( $this->mForceTocPosition ) {
3617                         return str_replace( '<!--MWTOC-->', $toc, $full );
3618                 } else {
3619                         return $full;
3620                 }
3621         }
3622
3623         /**
3624          * Transform wiki markup when saving a page by doing \r\n -> \n
3625          * conversion, substitting signatures, {{subst:}} templates, etc.
3626          *
3627          * @param string $text the text to transform
3628          * @param Title &$title the Title object for the current article
3629          * @param User &$user the User object describing the current user
3630          * @param ParserOptions $options parsing options
3631          * @param bool $clearState whether to clear the parser state first
3632          * @return string the altered wiki markup
3633          * @public
3634          */
3635         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3636                 $this->mOptions = $options;
3637                 $this->mTitle =& $title;
3638                 $this->setOutputType( OT_WIKI );
3639
3640                 if ( $clearState ) {
3641                         $this->clearState();
3642                 }
3643
3644                 $stripState = new StripState;
3645                 $pairs = array(
3646                         "\r\n" => "\n",
3647                 );
3648                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3649                 $text = $this->strip( $text, $stripState, true, array( 'gallery' ) );
3650                 $text = $this->pstPass2( $text, $stripState, $user );
3651                 $text = $stripState->unstripBoth( $text );
3652                 return $text;
3653         }
3654
3655         /**
3656          * Pre-save transform helper function
3657          * @private
3658          */
3659         function pstPass2( $text, &$stripState, $user ) {
3660                 global $wgContLang, $wgLocaltimezone;
3661
3662                 /* Note: This is the timestamp saved as hardcoded wikitext to
3663                  * the database, we use $wgContLang here in order to give
3664                  * everyone the same signature and use the default one rather
3665                  * than the one selected in each user's preferences.
3666                  */
3667                 if ( isset( $wgLocaltimezone ) ) {
3668                         $oldtz = getenv( 'TZ' );
3669                         putenv( 'TZ='.$wgLocaltimezone );
3670                 }
3671                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3672                   ' (' . date( 'T' ) . ')';
3673                 if ( isset( $wgLocaltimezone ) ) {
3674                         putenv( 'TZ='.$oldtz );
3675                 }
3676
3677                 # Variable replacement
3678                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3679                 $text = $this->replaceVariables( $text );
3680
3681                 # Strip out <nowiki> etc. added via replaceVariables
3682                 $text = $this->strip( $text, $stripState, false, array( 'gallery' ) );
3683
3684                 # Signatures
3685                 $sigText = $this->getUserSig( $user );
3686                 $text = strtr( $text, array(
3687                         '~~~~~' => $d,
3688                         '~~~~' => "$sigText $d",
3689                         '~~~' => $sigText
3690                 ) );
3691
3692                 # Context links: [[|name]] and [[name (context)|]]
3693                 #
3694                 global $wgLegalTitleChars;
3695                 $tc = "[$wgLegalTitleChars]";
3696                 $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3697
3698                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3699                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3700                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3701
3702                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3703                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3704                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3705
3706                 $t = $this->mTitle->getText();
3707                 $m = array();
3708                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3709                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3710                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3711                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3712                 } else {
3713                         # if there's no context, don't bother duplicating the title
3714                         $text = preg_replace( $p2, '[[\\1]]', $text );
3715                 }
3716
3717                 # Trim trailing whitespace
3718                 # __END__ tag allows for trailing
3719                 # whitespace to be deliberately included
3720                 $text = rtrim( $text );
3721                 $mw =& MagicWord::get( 'end' );
3722                 $mw->matchAndRemove( $text );
3723
3724                 return $text;
3725         }
3726
3727         /**
3728          * Fetch the user's signature text, if any, and normalize to
3729          * validated, ready-to-insert wikitext.
3730          *
3731          * @param User $user
3732          * @return string
3733          * @private
3734          */
3735         function getUserSig( &$user ) {
3736                 $username = $user->getName();
3737                 $nickname = $user->getOption( 'nickname' );
3738                 $nickname = $nickname === '' ? $username : $nickname;
3739
3740                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3741                         # Sig. might contain markup; validate this
3742                         if( $this->validateSig( $nickname ) !== false ) {
3743                                 # Validated; clean up (if needed) and return it
3744                                 return $this->cleanSig( $nickname, true );
3745                         } else {
3746                                 # Failed to validate; fall back to the default
3747                                 $nickname = $username;
3748                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3749                         }
3750                 }
3751
3752                 // Make sure nickname doesnt get a sig in a sig
3753                 $nickname = $this->cleanSigInSig( $nickname );
3754
3755                 # If we're still here, make it a link to the user page
3756                 $userpage = $user->getUserPage();
3757                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3758         }
3759
3760         /**
3761          * Check that the user's signature contains no bad XML
3762          *
3763          * @param string $text
3764          * @return mixed An expanded string, or false if invalid.
3765          */
3766         function validateSig( $text ) {
3767                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3768         }
3769
3770         /**
3771          * Clean up signature text
3772          *
3773          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3774          * 2) Substitute all transclusions
3775          *
3776          * @param string $text
3777          * @param $parsing Whether we're cleaning (preferences save) or parsing
3778          * @return string Signature text
3779          */
3780         function cleanSig( $text, $parsing = false ) {
3781                 global $wgTitle;
3782                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
3783
3784                 $substWord = MagicWord::get( 'subst' );
3785                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3786                 $substText = '{{' . $substWord->getSynonym( 0 );
3787
3788                 $text = preg_replace( $substRegex, $substText, $text );
3789                 $text = $this->cleanSigInSig( $text );
3790                 $text = $this->replaceVariables( $text );
3791
3792                 $this->clearState();
3793                 return $text;
3794         }
3795
3796         /**
3797          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3798          * @param string $text
3799          * @return string Signature text with /~{3,5}/ removed
3800          */
3801         function cleanSigInSig( $text ) {
3802                 $text = preg_replace( '/~{3,5}/', '', $text );
3803                 return $text;
3804         }
3805
3806         /**
3807          * Set up some variables which are usually set up in parse()
3808          * so that an external function can call some class members with confidence
3809          * @public
3810          */
3811         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3812                 $this->mTitle =& $title;
3813                 $this->mOptions = $options;
3814                 $this->setOutputType( $outputType );
3815                 if ( $clearState ) {
3816                         $this->clearState();
3817                 }
3818         }
3819
3820         /**
3821          * Transform a MediaWiki message by replacing magic variables.
3822          *
3823          * @param string $text the text to transform
3824          * @param ParserOptions $options  options
3825          * @return string the text with variables substituted
3826          * @public
3827          */
3828         function transformMsg( $text, $options ) {
3829                 global $wgTitle;
3830                 static $executing = false;
3831
3832                 $fname = "Parser::transformMsg";
3833
3834                 # Guard against infinite recursion
3835                 if ( $executing ) {
3836                         return $text;
3837                 }
3838                 $executing = true;
3839
3840                 wfProfileIn($fname);
3841
3842                 if ( $wgTitle ) {
3843                         $this->mTitle = $wgTitle;
3844                 } else {
3845                         $this->mTitle = Title::newFromText('msg');
3846                 }
3847                 $this->mOptions = $options;
3848                 $this->setOutputType( OT_MSG );
3849                 $this->clearState();
3850                 $text = $this->replaceVariables( $text );
3851
3852                 $executing = false;
3853                 wfProfileOut($fname);
3854                 return $text;
3855         }
3856
3857         /**
3858          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3859          * The callback should have the following form:
3860          *    function myParserHook( $text, $params, &$parser ) { ... }
3861          *
3862          * Transform and return $text. Use $parser for any required context, e.g. use
3863          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3864          *
3865          * @public
3866          *
3867          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3868          * @param mixed $callback The callback function (and object) to use for the tag
3869          *
3870          * @return The old value of the mTagHooks array associated with the hook
3871          */
3872         function setHook( $tag, $callback ) {
3873                 $tag = strtolower( $tag );
3874                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
3875                 $this->mTagHooks[$tag] = $callback;
3876
3877                 return $oldVal;
3878         }
3879
3880         /**
3881          * Create a function, e.g. {{sum:1|2|3}}
3882          * The callback function should have the form:
3883          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3884          *
3885          * The callback may either return the text result of the function, or an array with the text
3886          * in element 0, and a number of flags in the other elements. The names of the flags are
3887          * specified in the keys. Valid flags are:
3888          *   found                     The text returned is valid, stop processing the template. This
3889          *                             is on by default.
3890          *   nowiki                    Wiki markup in the return value should be escaped
3891          *   noparse                   Unsafe HTML tags should not be stripped, etc.
3892          *   noargs                    Don't replace triple-brace arguments in the return value
3893          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3894          *
3895          * @public
3896          *
3897          * @param string $id The magic word ID
3898          * @param mixed $callback The callback function (and object) to use
3899          * @param integer $flags a combination of the following flags:
3900          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
3901          *
3902          * @return The old callback function for this name, if any
3903          */
3904         function setFunctionHook( $id, $callback, $flags = 0 ) {
3905                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id] : null;
3906                 $this->mFunctionHooks[$id] = $callback;
3907
3908                 # Add to function cache
3909                 $mw = MagicWord::get( $id );
3910                 if( !$mw )
3911                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
3912
3913                 $synonyms = $mw->getSynonyms();
3914                 $sensitive = intval( $mw->isCaseSensitive() );
3915
3916                 foreach ( $synonyms as $syn ) {
3917                         # Case
3918                         if ( !$sensitive ) {
3919                                 $syn = strtolower( $syn );
3920                         }
3921                         # Add leading hash
3922                         if ( !( $flags & SFH_NO_HASH ) ) {
3923                                 $syn = '#' . $syn;
3924                         }
3925                         # Remove trailing colon
3926                         if ( substr( $syn, -1, 1 ) == ':' ) {
3927                                 $syn = substr( $syn, 0, -1 );
3928                         }
3929                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
3930                 }
3931                 return $oldVal;
3932         }
3933
3934         /**
3935          * Get all registered function hook identifiers
3936          *
3937          * @return array
3938          */
3939         function getFunctionHooks() {
3940                 return array_keys( $this->mFunctionHooks );
3941         }
3942
3943         /**
3944          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3945          * Placeholders created in Skin::makeLinkObj()
3946          * Returns an array of links found, indexed by PDBK:
3947          *  0 - broken
3948          *  1 - normal link
3949          *  2 - stub
3950          * $options is a bit field, RLH_FOR_UPDATE to select for update
3951          */
3952         function replaceLinkHolders( &$text, $options = 0 ) {
3953                 global $wgUser;
3954                 global $wgContLang;
3955
3956                 $fname = 'Parser::replaceLinkHolders';
3957                 wfProfileIn( $fname );
3958
3959                 $pdbks = array();
3960                 $colours = array();
3961                 $sk =& $this->mOptions->getSkin();
3962                 $linkCache =& LinkCache::singleton();
3963
3964                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3965                         wfProfileIn( $fname.'-check' );
3966                         $dbr =& wfGetDB( DB_SLAVE );
3967                         $page = $dbr->tableName( 'page' );
3968                         $threshold = $wgUser->getOption('stubthreshold');
3969
3970                         # Sort by namespace
3971                         asort( $this->mLinkHolders['namespaces'] );
3972
3973                         # Generate query
3974                         $query = false;
3975                         $current = null;
3976                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3977                                 # Make title object
3978                                 $title = $this->mLinkHolders['titles'][$key];
3979
3980                                 # Skip invalid entries.
3981                                 # Result will be ugly, but prevents crash.
3982                                 if ( is_null( $title ) ) {
3983                                         continue;
3984                                 }
3985                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3986
3987                                 # Check if it's a static known link, e.g. interwiki
3988                                 if ( $title->isAlwaysKnown() ) {
3989                                         $colours[$pdbk] = 1;
3990                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3991                                         $colours[$pdbk] = 1;
3992                                         $this->mOutput->addLink( $title, $id );
3993                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3994                                         $colours[$pdbk] = 0;
3995                                 } else {
3996                                         # Not in the link cache, add it to the query
3997                                         if ( !isset( $current ) ) {
3998                                                 $current = $ns;
3999                                                 $query =  "SELECT page_id, page_namespace, page_title";
4000                                                 if ( $threshold > 0 ) {
4001                                                         $query .= ', page_len, page_is_redirect';
4002                                                 }
4003                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
4004                                         } elseif ( $current != $ns ) {
4005                                                 $current = $ns;
4006                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
4007                                         } else {
4008                                                 $query .= ', ';
4009                                         }
4010
4011                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
4012                                 }
4013                         }
4014                         if ( $query ) {
4015                                 $query .= '))';
4016                                 if ( $options & RLH_FOR_UPDATE ) {
4017                                         $query .= ' FOR UPDATE';
4018                                 }
4019
4020                                 $res = $dbr->query( $query, $fname );
4021
4022                                 # Fetch data and form into an associative array
4023                                 # non-existent = broken
4024                                 # 1 = known
4025                                 # 2 = stub
4026                                 while ( $s = $dbr->fetchObject($res) ) {
4027                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4028                                         $pdbk = $title->getPrefixedDBkey();
4029                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
4030                                         $this->mOutput->addLink( $title, $s->page_id );
4031
4032                                         if ( $threshold >  0 ) {
4033                                                 $size = $s->page_len;
4034                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
4035                                                         $colours[$pdbk] = 1;
4036                                                 } else {
4037                                                         $colours[$pdbk] = 2;
4038                                                 }
4039                                         } else {
4040                                                 $colours[$pdbk] = 1;
4041                                         }
4042                                 }
4043                         }
4044                         wfProfileOut( $fname.'-check' );
4045
4046                         # Do a second query for different language variants of links and categories
4047                         if($wgContLang->hasVariants()){
4048                                 $linkBatch = new LinkBatch();
4049                                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
4050                                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
4051                                 $varCategories = array(); // category replacements oldDBkey => newDBkey
4052
4053                                 $categories = $this->mOutput->getCategoryLinks();
4054
4055                                 // Add variants of links to link batch
4056                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4057                                         $title = $this->mLinkHolders['titles'][$key];
4058                                         if ( is_null( $title ) )
4059                                                 continue;
4060
4061                                         $pdbk = $title->getPrefixedDBkey();
4062                                         $titleText = $title->getText();
4063
4064                                         // generate all variants of the link title text
4065                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
4066
4067                                         // if link was not found (in first query), add all variants to query
4068                                         if ( !isset($colours[$pdbk]) ){
4069                                                 foreach($allTextVariants as $textVariant){
4070                                                         if($textVariant != $titleText){
4071                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
4072                                                                 if(is_null($variantTitle)) continue;
4073                                                                 $linkBatch->addObj( $variantTitle );
4074                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4075                                                         }
4076                                                 }
4077                                         }
4078                                 }
4079
4080                                 // process categories, check if a category exists in some variant
4081                                 foreach( $categories as $category){
4082                                         $variants = $wgContLang->convertLinkToAllVariants($category);
4083                                         foreach($variants as $variant){
4084                                                 if($variant != $category){
4085                                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
4086                                                         if(is_null($variantTitle)) continue;
4087                                                         $linkBatch->addObj( $variantTitle );
4088                                                         $categoryMap[$variant] = $category;
4089                                                 }
4090                                         }
4091                                 }
4092
4093
4094                                 if(!$linkBatch->isEmpty()){
4095                                         // construct query
4096                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4097
4098                                         $variantQuery =  "SELECT page_id, page_namespace, page_title";
4099                                         if ( $threshold > 0 ) {
4100                                                 $variantQuery .= ', page_len, page_is_redirect';
4101                                         }
4102
4103                                         $variantQuery .= " FROM $page WHERE $titleClause";
4104                                         if ( $options & RLH_FOR_UPDATE ) {
4105                                                 $variantQuery .= ' FOR UPDATE';
4106                                         }
4107
4108                                         $varRes = $dbr->query( $variantQuery, $fname );
4109
4110                                         // for each found variants, figure out link holders and replace
4111                                         while ( $s = $dbr->fetchObject($varRes) ) {
4112
4113                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4114                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4115                                                 $vardbk = $variantTitle->getDBkey();
4116
4117                                                 $holderKeys = array();
4118                                                 if(isset($variantMap[$varPdbk])){
4119                                                         $holderKeys = $variantMap[$varPdbk];
4120                                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle );
4121                                                         $this->mOutput->addLink( $variantTitle, $s->page_id );
4122                                                 }
4123
4124                                                 // loop over link holders
4125                                                 foreach($holderKeys as $key){
4126                                                         $title = $this->mLinkHolders['titles'][$key];
4127                                                         if ( is_null( $title ) ) continue;
4128
4129                                                         $pdbk = $title->getPrefixedDBkey();
4130
4131                                                         if(!isset($colours[$pdbk])){
4132                                                                 // found link in some of the variants, replace the link holder data
4133                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4134                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4135
4136                                                                 // set pdbk and colour
4137                                                                 $pdbks[$key] = $varPdbk;
4138                                                                 if ( $threshold >  0 ) {
4139                                                                         $size = $s->page_len;
4140                                                                         if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
4141                                                                                 $colours[$varPdbk] = 1;
4142                                                                         } else {
4143                                                                                 $colours[$varPdbk] = 2;
4144                                                                         }
4145                                                                 }
4146                                                                 else {
4147                                                                         $colours[$varPdbk] = 1;
4148                                                                 }
4149                                                         }
4150                                                 }
4151
4152                                                 // check if the object is a variant of a category
4153                                                 if(isset($categoryMap[$vardbk])){
4154                                                         $oldkey = $categoryMap[$vardbk];
4155                                                         if($oldkey != $vardbk)
4156                                                                 $varCategories[$oldkey]=$vardbk;
4157                                                 }
4158                                         }
4159
4160                                         // rebuild the categories in original order (if there are replacements)
4161                                         if(count($varCategories)>0){
4162                                                 $newCats = array();
4163                                                 $originalCats = $this->mOutput->getCategories();
4164                                                 foreach($originalCats as $cat => $sortkey){
4165                                                         // make the replacement
4166                                                         if( array_key_exists($cat,$varCategories) )
4167                                                                 $newCats[$varCategories[$cat]] = $sortkey;
4168                                                         else $newCats[$cat] = $sortkey;
4169                                                 }
4170                                                 $this->mOutput->setCategoryLinks($newCats);
4171                                         }
4172                                 }
4173                         }
4174
4175                         # Construct search and replace arrays
4176                         wfProfileIn( $fname.'-construct' );
4177                         $replacePairs = array();
4178                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4179                                 $pdbk = $pdbks[$key];
4180                                 $searchkey = "<!--LINK $key-->";
4181                                 $title = $this->mLinkHolders['titles'][$key];
4182                                 if ( empty( $colours[$pdbk] ) ) {
4183                                         $linkCache->addBadLinkObj( $title );
4184                                         $colours[$pdbk] = 0;
4185                                         $this->mOutput->addLink( $title, 0 );
4186                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4187                                                                         $this->mLinkHolders['texts'][$key],
4188                                                                         $this->mLinkHolders['queries'][$key] );
4189                                 } elseif ( $colours[$pdbk] == 1 ) {
4190                                         $replacePairs[$searchkey] = $sk->makeKnownLinkObj( $title,
4191                                                                         $this->mLinkHolders['texts'][$key],
4192                                                                         $this->mLinkHolders['queries'][$key] );
4193                                 } elseif ( $colours[$pdbk] == 2 ) {
4194                                         $replacePairs[$searchkey] = $sk->makeStubLinkObj( $title,
4195                                                                         $this->mLinkHolders['texts'][$key],
4196                                                                         $this->mLinkHolders['queries'][$key] );
4197                                 }
4198                         }
4199                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4200                         wfProfileOut( $fname.'-construct' );
4201
4202                         # Do the thing
4203                         wfProfileIn( $fname.'-replace' );
4204                         $text = preg_replace_callback(
4205                                 '/(<!--LINK .*?-->)/',
4206                                 $replacer->cb(),
4207                                 $text);
4208
4209                         wfProfileOut( $fname.'-replace' );
4210                 }
4211
4212                 # Now process interwiki link holders
4213                 # This is quite a bit simpler than internal links
4214                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4215                         wfProfileIn( $fname.'-interwiki' );
4216                         # Make interwiki link HTML
4217                         $replacePairs = array();
4218                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4219                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4220                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4221                         }
4222                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4223
4224                         $text = preg_replace_callback(
4225                                 '/<!--IWLINK (.*?)-->/',
4226                                 $replacer->cb(),
4227                                 $text );
4228                         wfProfileOut( $fname.'-interwiki' );
4229                 }
4230
4231                 wfProfileOut( $fname );
4232                 return $colours;
4233         }
4234
4235         /**
4236          * Replace <!--LINK--> link placeholders with plain text of links
4237          * (not HTML-formatted).
4238          * @param string $text
4239          * @return string
4240          */
4241         function replaceLinkHoldersText( $text ) {
4242                 $fname = 'Parser::replaceLinkHoldersText';
4243                 wfProfileIn( $fname );
4244
4245                 $text = preg_replace_callback(
4246                         '/<!--(LINK|IWLINK) (.*?)-->/',
4247                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4248                         $text );
4249
4250                 wfProfileOut( $fname );
4251                 return $text;
4252         }
4253
4254         /**
4255          * @param array $matches
4256          * @return string
4257          * @private
4258          */
4259         function replaceLinkHoldersTextCallback( $matches ) {
4260                 $type = $matches[1];
4261                 $key  = $matches[2];
4262                 if( $type == 'LINK' ) {
4263                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4264                                 return $this->mLinkHolders['texts'][$key];
4265                         }
4266                 } elseif( $type == 'IWLINK' ) {
4267                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4268                                 return $this->mInterwikiLinkHolders['texts'][$key];
4269                         }
4270                 }
4271                 return $matches[0];
4272         }
4273
4274         /**
4275          * Tag hook handler for 'pre'.
4276          */
4277         function renderPreTag( $text, $attribs ) {
4278                 // Backwards-compatibility hack
4279                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4280
4281                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4282                 return wfOpenElement( 'pre', $attribs ) .
4283                         Xml::escapeTagsOnly( $content ) .
4284                         '</pre>';
4285         }
4286
4287         /**
4288          * Renders an image gallery from a text with one line per image.
4289          * text labels may be given by using |-style alternative text. E.g.
4290          *   Image:one.jpg|The number "1"
4291          *   Image:tree.jpg|A tree
4292          * given as text will return the HTML of a gallery with two images,
4293          * labeled 'The number "1"' and
4294          * 'A tree'.
4295          */
4296         function renderImageGallery( $text, $params ) {
4297                 $ig = new ImageGallery();
4298                 $ig->setShowBytes( false );
4299                 $ig->setShowFilename( false );
4300                 $ig->setParsing();
4301                 $ig->useSkin( $this->mOptions->getSkin() );
4302
4303                 if( isset( $params['caption'] ) ) {
4304                         $caption = $params['caption'];
4305                         $caption = htmlspecialchars( $caption );
4306                         $caption = $this->replaceInternalLinks( $caption );
4307                         $ig->setCaptionSafe( $caption );
4308                 }
4309
4310                 $lines = explode( "\n", $text );
4311                 foreach ( $lines as $line ) {
4312                         # match lines like these:
4313                         # Image:someimage.jpg|This is some image
4314                         $matches = array();
4315                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4316                         # Skip empty lines
4317                         if ( count( $matches ) == 0 ) {
4318                                 continue;
4319                         }
4320                         $tp = Title::newFromText( $matches[1] );
4321                         $nt =& $tp;
4322                         if( is_null( $nt ) ) {
4323                                 # Bogus title. Ignore these so we don't bomb out later.
4324                                 continue;
4325                         }
4326                         if ( isset( $matches[3] ) ) {
4327                                 $label = $matches[3];
4328                         } else {
4329                                 $label = '';
4330                         }
4331
4332                         $pout = $this->parse( $label,
4333                                 $this->mTitle,
4334                                 $this->mOptions,
4335                                 false, // Strip whitespace...?
4336                                 false  // Don't clear state!
4337                         );
4338                         $html = $pout->getText();
4339
4340                         $ig->add( new Image( $nt ), $html );
4341
4342                         # Only add real images (bug #5586)
4343                         if ( $nt->getNamespace() == NS_IMAGE ) {
4344                                 $this->mOutput->addImage( $nt->getDBkey() );
4345                         }
4346                 }
4347                 return $ig->toHTML();
4348         }
4349
4350         /**
4351          * Parse image options text and use it to make an image
4352          */
4353         function makeImage( $nt, $options ) {
4354                 global $wgUseImageResize, $wgDjvuRenderer;
4355
4356                 $align = '';
4357
4358                 # Check if the options text is of the form "options|alt text"
4359                 # Options are:
4360                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4361                 #  * left               no resizing, just left align. label is used for alt= only
4362                 #  * right              same, but right aligned
4363                 #  * none               same, but not aligned
4364                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4365                 #  * center             center the image
4366                 #  * framed             Keep original image size, no magnify-button.
4367
4368                 $part = explode( '|', $options);
4369
4370                 $mwThumb  =& MagicWord::get( 'img_thumbnail' );
4371                 $mwManualThumb =& MagicWord::get( 'img_manualthumb' );
4372                 $mwLeft   =& MagicWord::get( 'img_left' );
4373                 $mwRight  =& MagicWord::get( 'img_right' );
4374                 $mwNone   =& MagicWord::get( 'img_none' );
4375                 $mwWidth  =& MagicWord::get( 'img_width' );
4376                 $mwCenter =& MagicWord::get( 'img_center' );
4377                 $mwFramed =& MagicWord::get( 'img_framed' );
4378                 $mwPage   =& MagicWord::get( 'img_page' );
4379                 $caption = '';
4380
4381                 $width = $height = $framed = $thumb = false;
4382                 $page = null;
4383                 $manual_thumb = '' ;
4384
4385                 foreach( $part as $val ) {
4386                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
4387                                 $thumb=true;
4388                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
4389                                 # use manually specified thumbnail
4390                                 $thumb=true;
4391                                 $manual_thumb = $match;
4392                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
4393                                 # remember to set an alignment, don't render immediately
4394                                 $align = 'right';
4395                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
4396                                 # remember to set an alignment, don't render immediately
4397                                 $align = 'left';
4398                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
4399                                 # remember to set an alignment, don't render immediately
4400                                 $align = 'center';
4401                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
4402                                 # remember to set an alignment, don't render immediately
4403                                 $align = 'none';
4404                         } elseif ( isset( $wgDjvuRenderer ) && $wgDjvuRenderer
4405                                    && ! is_null( $match = $mwPage->matchVariableStartToEnd($val) ) ) {
4406                                 # Select a page in a multipage document
4407                                 $page = $match;
4408                         } elseif ( $wgUseImageResize && !$width && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
4409                                 wfDebug( "img_width match: $match\n" );
4410                                 # $match is the image width in pixels
4411                                 $m = array();
4412                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
4413                                         $width = intval( $m[1] );
4414                                         $height = intval( $m[2] );
4415                                 } else {
4416                                         $width = intval($match);
4417                                 }
4418                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
4419                                 $framed=true;
4420                         } else {
4421                                 $caption = $val;
4422                         }
4423                 }
4424                 # Strip bad stuff out of the alt text
4425                 $alt = $this->replaceLinkHoldersText( $caption );
4426
4427                 # make sure there are no placeholders in thumbnail attributes
4428                 # that are later expanded to html- so expand them now and
4429                 # remove the tags
4430                 $alt = $this->mStripState->unstripBoth( $alt );
4431                 $alt = Sanitizer::stripAllTags( $alt );
4432
4433                 # Linker does the rest
4434                 $sk =& $this->mOptions->getSkin();
4435                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb, $page );
4436         }
4437
4438         /**
4439          * Set a flag in the output object indicating that the content is dynamic and
4440          * shouldn't be cached.
4441          */
4442         function disableCache() {
4443                 wfDebug( "Parser output marked as uncacheable.\n" );
4444                 $this->mOutput->mCacheTime = -1;
4445         }
4446
4447         /**#@+
4448          * Callback from the Sanitizer for expanding items found in HTML attribute
4449          * values, so they can be safely tested and escaped.
4450          * @param string $text
4451          * @param array $args
4452          * @return string
4453          * @private
4454          */
4455         function attributeStripCallback( &$text, $args ) {
4456                 $text = $this->replaceVariables( $text, $args );
4457                 $text = $this->mStripState->unstripBoth( $text );
4458                 return $text;
4459         }
4460
4461         /**#@-*/
4462
4463         /**#@+
4464          * Accessor/mutator
4465          */
4466         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4467         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4468         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4469         /**#@-*/
4470
4471         /**#@+
4472          * Accessor
4473          */
4474         function getTags() { return array_keys( $this->mTagHooks ); }
4475         /**#@-*/
4476
4477
4478         /**
4479          * Break wikitext input into sections, and either pull or replace
4480          * some particular section's text.
4481          *
4482          * External callers should use the getSection and replaceSection methods.
4483          *
4484          * @param $text Page wikitext
4485          * @param $section Numbered section. 0 pulls the text before the first
4486          *                 heading; other numbers will pull the given section
4487          *                 along with its lower-level subsections.
4488          * @param $mode One of "get" or "replace"
4489          * @param $newtext Replacement text for section data.
4490          * @return string for "get", the extracted section text.
4491          *                for "replace", the whole page with the section replaced.
4492          */
4493         private function extractSections( $text, $section, $mode, $newtext='' ) {
4494                 # strip NOWIKI etc. to avoid confusion (true-parameter causes HTML
4495                 # comments to be stripped as well)
4496                 $stripState = new StripState;
4497
4498                 $oldOutputType = $this->mOutputType;
4499                 $oldOptions = $this->mOptions;
4500                 $this->mOptions = new ParserOptions();
4501                 $this->setOutputType( OT_WIKI );
4502
4503                 $striptext = $this->strip( $text, $stripState, true );
4504
4505                 $this->setOutputType( $oldOutputType );
4506                 $this->mOptions = $oldOptions;
4507
4508                 # now that we can be sure that no pseudo-sections are in the source,
4509                 # split it up by section
4510                 $uniq = preg_quote( $this->uniqPrefix(), '/' );
4511                 $comment = "(?:$uniq-!--.*?QINU)";
4512                 $secs = preg_split(
4513                 /*
4514                         "/
4515                         ^(
4516                         (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4517                         (?:
4518                                 (=+) # Should this be limited to 6?
4519                                 .+?  # Section title...
4520                                 \\2  # Ending = count must match start
4521                         |
4522                                 ^
4523                                 <h([1-6])\b.*?>
4524                                 .*?
4525                                 <\/h\\3\s*>
4526                         )
4527                         (?:$comment|<\/?noinclude>|\s+)* # Trailing whitespace ok
4528                         )$
4529                         /mix",
4530                 */
4531                         "/
4532                         (
4533                                 ^
4534                                 (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4535                                 (=+) # Should this be limited to 6?
4536                                 .+?  # Section title...
4537                                 \\2  # Ending = count must match start
4538                                 (?:$comment|<\/?noinclude>|[ \\t]+)* # Trailing whitespace ok
4539                                 $
4540                         |
4541                                 <h([1-6])\b.*?>
4542                                 .*?
4543                                 <\/h\\3\s*>
4544                         )
4545                         /mix",
4546                         $striptext, -1,
4547                         PREG_SPLIT_DELIM_CAPTURE);
4548
4549                 if( $mode == "get" ) {
4550                         if( $section == 0 ) {
4551                                 // "Section 0" returns the content before any other section.
4552                                 $rv = $secs[0];
4553                         } else {
4554                                 $rv = "";
4555                         }
4556                 } elseif( $mode == "replace" ) {
4557                         if( $section == 0 ) {
4558                                 $rv = $newtext . "\n\n";
4559                                 $remainder = true;
4560                         } else {
4561                                 $rv = $secs[0];
4562                                 $remainder = false;
4563                         }
4564                 }
4565                 $count = 0;
4566                 $sectionLevel = 0;
4567                 for( $index = 1; $index < count( $secs ); ) {
4568                         $headerLine = $secs[$index++];
4569                         if( $secs[$index] ) {
4570                                 // A wiki header
4571                                 $headerLevel = strlen( $secs[$index++] );
4572                         } else {
4573                                 // An HTML header
4574                                 $index++;
4575                                 $headerLevel = intval( $secs[$index++] );
4576                         }
4577                         $content = $secs[$index++];
4578
4579                         $count++;
4580                         if( $mode == "get" ) {
4581                                 if( $count == $section ) {
4582                                         $rv = $headerLine . $content;
4583                                         $sectionLevel = $headerLevel;
4584                                 } elseif( $count > $section ) {
4585                                         if( $sectionLevel && $headerLevel > $sectionLevel ) {
4586                                                 $rv .= $headerLine . $content;
4587                                         } else {
4588                                                 // Broke out to a higher-level section
4589                                                 break;
4590                                         }
4591                                 }
4592                         } elseif( $mode == "replace" ) {
4593                                 if( $count < $section ) {
4594                                         $rv .= $headerLine . $content;
4595                                 } elseif( $count == $section ) {
4596                                         $rv .= $newtext . "\n\n";
4597                                         $sectionLevel = $headerLevel;
4598                                 } elseif( $count > $section ) {
4599                                         if( $headerLevel <= $sectionLevel ) {
4600                                                 // Passed the section's sub-parts.
4601                                                 $remainder = true;
4602                                         }
4603                                         if( $remainder ) {
4604                                                 $rv .= $headerLine . $content;
4605                                         }
4606                                 }
4607                         }
4608                 }
4609                 # reinsert stripped tags
4610                 $rv = trim( $stripState->unstripBoth( $rv ) );
4611                 return $rv;
4612         }
4613
4614         /**
4615          * This function returns the text of a section, specified by a number ($section).
4616          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4617          * the first section before any such heading (section 0).
4618          *
4619          * If a section contains subsections, these are also returned.
4620          *
4621          * @param $text String: text to look in
4622          * @param $section Integer: section number
4623          * @return string text of the requested section
4624          */
4625         function getSection( $text, $section ) {
4626                 return $this->extractSections( $text, $section, "get" );
4627         }
4628
4629         function replaceSection( $oldtext, $section, $text ) {
4630                 return $this->extractSections( $oldtext, $section, "replace", $text );
4631         }
4632
4633         /**
4634          * Get the timestamp associated with the current revision, adjusted for
4635          * the default server-local timestamp
4636          */
4637         function getRevisionTimestamp() {
4638                 if ( is_null( $this->mRevisionTimestamp ) ) {
4639                         wfProfileIn( __METHOD__ );
4640                         global $wgContLang;
4641                         $dbr =& wfGetDB( DB_SLAVE );
4642                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4643                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4644
4645                         // Normalize timestamp to internal MW format for timezone processing.
4646                         // This has the added side-effect of replacing a null value with
4647                         // the current time, which gives us more sensible behavior for
4648                         // previews.
4649                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4650
4651                         // The cryptic '' timezone parameter tells to use the site-default
4652                         // timezone offset instead of the user settings.
4653                         //
4654                         // Since this value will be saved into the parser cache, served
4655                         // to other users, and potentially even used inside links and such,
4656                         // it needs to be consistent for all visitors.
4657                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4658
4659                         wfProfileOut( __METHOD__ );
4660                 }
4661                 return $this->mRevisionTimestamp;
4662         }
4663
4664         /**
4665          * Mutator for $mDefaultSort
4666          *
4667          * @param $sort New value
4668          */
4669         public function setDefaultSort( $sort ) {
4670                 $this->mDefaultSort = $sort;
4671         }
4672
4673         /**
4674          * Accessor for $mDefaultSort
4675          * Will use the title/prefixed title if none is set
4676          *
4677          * @return string
4678          */
4679         public function getDefaultSort() {
4680                 if( $this->mDefaultSort !== false ) {
4681                         return $this->mDefaultSort;
4682                 } else {
4683                         return $this->mTitle->getNamespace() == NS_CATEGORY
4684                                         ? $this->mTitle->getText()
4685                                         : $this->mTitle->getPrefixedText();
4686                 }
4687         }
4688
4689 }
4690
4691 /**
4692  * @todo document
4693  * @package MediaWiki
4694  */
4695 class ParserOutput
4696 {
4697         var $mText,             # The output text
4698                 $mLanguageLinks,    # List of the full text of language links, in the order they appear
4699                 $mCategories,       # Map of category names to sort keys
4700                 $mContainsOldMagic, # Boolean variable indicating if the input contained variables like {{CURRENTDAY}}
4701                 $mCacheTime,        # Time when this object was generated, or -1 for uncacheable. Used in ParserCache.
4702                 $mVersion,          # Compatibility check
4703                 $mTitleText,        # title text of the chosen language variant
4704                 $mLinks,            # 2-D map of NS/DBK to ID for the links in the document. ID=zero for broken.
4705                 $mTemplates,        # 2-D map of NS/DBK to ID for the template references. ID=zero for broken.
4706                 $mImages,           # DB keys of the images used, in the array key only
4707                 $mExternalLinks,    # External link URLs, in the key only
4708                 $mHTMLtitle,            # Display HTML title
4709                 $mSubtitle,                     # Additional subtitle
4710                 $mNewSection,           # Show a new section link?
4711                 $mNoGallery;            # No gallery on category page? (__NOGALLERY__)
4712
4713         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
4714                 $containsOldMagic = false, $titletext = '' )
4715         {
4716                 $this->mText = $text;
4717                 $this->mLanguageLinks = $languageLinks;
4718                 $this->mCategories = $categoryLinks;
4719                 $this->mContainsOldMagic = $containsOldMagic;
4720                 $this->mCacheTime = '';
4721                 $this->mVersion = MW_PARSER_VERSION;
4722                 $this->mTitleText = $titletext;
4723                 $this->mLinks = array();
4724                 $this->mTemplates = array();
4725                 $this->mImages = array();
4726                 $this->mExternalLinks = array();
4727                 $this->mHTMLtitle = "" ;
4728                 $this->mSubtitle = "" ;
4729                 $this->mNewSection = false;
4730                 $this->mNoGallery = false;
4731         }
4732
4733         function getText()                   { return $this->mText; }
4734         function &getLanguageLinks()          { return $this->mLanguageLinks; }
4735         function getCategoryLinks()          { return array_keys( $this->mCategories ); }
4736         function &getCategories()            { return $this->mCategories; }
4737         function getCacheTime()              { return $this->mCacheTime; }
4738         function getTitleText()              { return $this->mTitleText; }
4739         function &getLinks()                 { return $this->mLinks; }
4740         function &getTemplates()             { return $this->mTemplates; }
4741         function &getImages()                { return $this->mImages; }
4742         function &getExternalLinks()         { return $this->mExternalLinks; }
4743         function getNoGallery()              { return $this->mNoGallery; }
4744         function getSubtitle()               { return $this->mSubtitle; }
4745
4746         function containsOldMagic()          { return $this->mContainsOldMagic; }
4747         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
4748         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
4749         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategories, $cl ); }
4750         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
4751         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
4752         function setTitleText( $t )          { return wfSetVar($this->mTitleText, $t); }
4753         function setSubtitle( $st )          { return wfSetVar( $this->mSubtitle, $st ); }
4754
4755         function addCategory( $c, $sort )    { $this->mCategories[$c] = $sort; }
4756         function addImage( $name )           { $this->mImages[$name] = 1; }
4757         function addLanguageLink( $t )       { $this->mLanguageLinks[] = $t; }
4758         function addExternalLink( $url )     { $this->mExternalLinks[$url] = 1; }
4759
4760         function setNewSection( $value ) {
4761                 $this->mNewSection = (bool)$value;
4762         }
4763         function getNewSection() {
4764                 return (bool)$this->mNewSection;
4765         }
4766
4767         function addLink( $title, $id = null ) {
4768                 $ns = $title->getNamespace();
4769                 $dbk = $title->getDBkey();
4770                 if ( !isset( $this->mLinks[$ns] ) ) {
4771                         $this->mLinks[$ns] = array();
4772                 }
4773                 if ( is_null( $id ) ) {
4774                         $id = $title->getArticleID();
4775                 }
4776                 $this->mLinks[$ns][$dbk] = $id;
4777         }
4778
4779         function addTemplate( $title, $id ) {
4780                 $ns = $title->getNamespace();
4781                 $dbk = $title->getDBkey();
4782                 if ( !isset( $this->mTemplates[$ns] ) ) {
4783                         $this->mTemplates[$ns] = array();
4784                 }
4785                 $this->mTemplates[$ns][$dbk] = $id;
4786         }
4787
4788         /**
4789          * Return true if this cached output object predates the global or
4790          * per-article cache invalidation timestamps, or if it comes from
4791          * an incompatible older version.
4792          *
4793          * @param string $touched the affected article's last touched timestamp
4794          * @return bool
4795          * @public
4796          */
4797         function expired( $touched ) {
4798                 global $wgCacheEpoch;
4799                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
4800                        $this->getCacheTime() < $touched ||
4801                        $this->getCacheTime() <= $wgCacheEpoch ||
4802                        !isset( $this->mVersion ) ||
4803                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
4804         }
4805 }
4806
4807 /**
4808  * Set options of the Parser
4809  * @todo document
4810  * @package MediaWiki
4811  */
4812 class ParserOptions
4813 {
4814         # All variables are supposed to be private in theory, although in practise this is not the case.
4815         var $mUseTeX;                    # Use texvc to expand <math> tags
4816         var $mUseDynamicDates;           # Use DateFormatter to format dates
4817         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
4818         var $mAllowExternalImages;       # Allow external images inline
4819         var $mAllowExternalImagesFrom;   # If not, any exception?
4820         var $mSkin;                      # Reference to the preferred skin
4821         var $mDateFormat;                # Date format index
4822         var $mEditSection;               # Create "edit section" links
4823         var $mNumberHeadings;            # Automatically number headings
4824         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
4825         var $mTidy;                      # Ask for tidy cleanup
4826         var $mInterfaceMessage;          # Which lang to call for PLURAL and GRAMMAR
4827         var $mMaxIncludeSize;            # Maximum size of template expansions, in bytes
4828         var $mRemoveComments;            # Remove HTML comments. ONLY APPLIES TO PREPROCESS OPERATIONS
4829
4830         var $mUser;                      # Stored user object, just used to initialise the skin
4831
4832         function getUseTeX()                        { return $this->mUseTeX; }
4833         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
4834         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
4835         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
4836         function getAllowExternalImagesFrom()       { return $this->mAllowExternalImagesFrom; }
4837         function getEditSection()                   { return $this->mEditSection; }
4838         function getNumberHeadings()                { return $this->mNumberHeadings; }
4839         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
4840         function getTidy()                          { return $this->mTidy; }
4841         function getInterfaceMessage()              { return $this->mInterfaceMessage; }
4842         function getMaxIncludeSize()                { return $this->mMaxIncludeSize; }
4843         function getRemoveComments()                { return $this->mRemoveComments; }
4844
4845         function &getSkin() {
4846                 if ( !isset( $this->mSkin ) ) {
4847                         $this->mSkin = $this->mUser->getSkin();
4848                 }
4849                 return $this->mSkin;
4850         }
4851
4852         function getDateFormat() {
4853                 if ( !isset( $this->mDateFormat ) ) {
4854                         $this->mDateFormat = $this->mUser->getDatePreference();
4855                 }
4856                 return $this->mDateFormat;
4857         }
4858
4859         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
4860         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
4861         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
4862         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
4863         function setAllowExternalImagesFrom( $x )   { return wfSetVar( $this->mAllowExternalImagesFrom, $x ); }
4864         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
4865         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
4866         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
4867         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
4868         function setTidy( $x )                      { return wfSetVar( $this->mTidy, $x); }
4869         function setSkin( $x )                      { $this->mSkin = $x; }
4870         function setInterfaceMessage( $x )          { return wfSetVar( $this->mInterfaceMessage, $x); }
4871         function setMaxIncludeSize( $x )            { return wfSetVar( $this->mMaxIncludeSize, $x ); }
4872         function setRemoveComments( $x )            { return wfSetVar( $this->mRemoveComments, $x ); }
4873
4874         function ParserOptions( $user = null ) {
4875                 $this->initialiseFromUser( $user );
4876         }
4877
4878         /**
4879          * Get parser options
4880          * @static
4881          */
4882         static function newFromUser( $user ) {
4883                 return new ParserOptions( $user );
4884         }
4885
4886         /** Get user options */
4887         function initialiseFromUser( $userInput ) {
4888                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages;
4889                 global $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion, $wgMaxArticleSize;
4890                 $fname = 'ParserOptions::initialiseFromUser';
4891                 wfProfileIn( $fname );
4892                 if ( !$userInput ) {
4893                         global $wgUser;
4894                         if ( isset( $wgUser ) ) {
4895                                 $user = $wgUser;
4896                         } else {
4897                                 $user = new User;
4898                         }
4899                 } else {
4900                         $user =& $userInput;
4901                 }
4902
4903                 $this->mUser = $user;
4904
4905                 $this->mUseTeX = $wgUseTeX;
4906                 $this->mUseDynamicDates = $wgUseDynamicDates;
4907                 $this->mInterwikiMagic = $wgInterwikiMagic;
4908                 $this->mAllowExternalImages = $wgAllowExternalImages;
4909                 $this->mAllowExternalImagesFrom = $wgAllowExternalImagesFrom;
4910                 $this->mSkin = null; # Deferred
4911                 $this->mDateFormat = null; # Deferred
4912                 $this->mEditSection = true;
4913                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
4914                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
4915                 $this->mTidy = false;
4916                 $this->mInterfaceMessage = false;
4917                 $this->mMaxIncludeSize = $wgMaxArticleSize * 1024;
4918                 $this->mRemoveComments = true;
4919                 wfProfileOut( $fname );
4920         }
4921 }
4922
4923 class OnlyIncludeReplacer {
4924         var $output = '';
4925
4926         function replace( $matches ) {
4927                 if ( substr( $matches[1], -1 ) == "\n" ) {
4928                         $this->output .= substr( $matches[1], 0, -1 );
4929                 } else {
4930                         $this->output .= $matches[1];
4931                 }
4932         }
4933 }
4934
4935 class StripState {
4936         var $general, $nowiki;
4937
4938         function __construct() {
4939                 $this->general = new ReplacementArray;
4940                 $this->nowiki = new ReplacementArray;
4941         }
4942
4943         function unstripGeneral( $text ) {
4944                 wfProfileIn( __METHOD__ );
4945                 $text = $this->general->replace( $text );
4946                 wfProfileOut( __METHOD__ );
4947                 return $text;
4948         }
4949
4950         function unstripNoWiki( $text ) {
4951                 wfProfileIn( __METHOD__ );
4952                 $text = $this->nowiki->replace( $text );
4953                 wfProfileOut( __METHOD__ );
4954                 return $text;
4955         }
4956
4957         function unstripBoth( $text ) {
4958                 wfProfileIn( __METHOD__ );
4959                 $text = $this->general->replace( $text );
4960                 $text = $this->nowiki->replace( $text );
4961                 wfProfileOut( __METHOD__ );
4962                 return $text;
4963         }
4964 }
4965
4966 ?>