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 = array();
 101         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 102         var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
 103         var $mIncludeSizes;
 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
 117         /**#@-*/
 118
 119         /**
 120          * Constructor
 121          *
 122          * @public
 123          */
 124         function Parser() {
 125                 $this->mTagHooks = array();
 126                 $this->mFunctionHooks = array();
 127                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 128                 $this->mFirstCall = true;
 129         }
 130
 131         /**
 132          * Do various kinds of initialisation on the first call of the parser
 133          */
 134         function firstCallInit() {
 135                 if ( !$this->mFirstCall ) {
 136                         return;
 137                 }
 138
 139                 wfProfileIn( __METHOD__ );
 140                 global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
 141
 142                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 143
 144                 $this->setFunctionHook( 'int', array( 'CoreParserFunctions', 'intFunction' ), SFH_NO_HASH );
 145                 $this->setFunctionHook( 'ns', array( 'CoreParserFunctions', 'ns' ), SFH_NO_HASH );
 146                 $this->setFunctionHook( 'urlencode', array( 'CoreParserFunctions', 'urlencode' ), SFH_NO_HASH );
 147                 $this->setFunctionHook( 'lcfirst', array( 'CoreParserFunctions', 'lcfirst' ), SFH_NO_HASH );
 148                 $this->setFunctionHook( 'ucfirst', array( 'CoreParserFunctions', 'ucfirst' ), SFH_NO_HASH );
 149                 $this->setFunctionHook( 'lc', array( 'CoreParserFunctions', 'lc' ), SFH_NO_HASH );
 150                 $this->setFunctionHook( 'uc', array( 'CoreParserFunctions', 'uc' ), SFH_NO_HASH );
 151                 $this->setFunctionHook( 'localurl', array( 'CoreParserFunctions', 'localurl' ), SFH_NO_HASH );
 152                 $this->setFunctionHook( 'localurle', array( 'CoreParserFunctions', 'localurle' ), SFH_NO_HASH );
 153                 $this->setFunctionHook( 'fullurl', array( 'CoreParserFunctions', 'fullurl' ), SFH_NO_HASH );
 154                 $this->setFunctionHook( 'fullurle', array( 'CoreParserFunctions', 'fullurle' ), SFH_NO_HASH );
 155                 $this->setFunctionHook( 'formatnum', array( 'CoreParserFunctions', 'formatnum' ), SFH_NO_HASH );
 156                 $this->setFunctionHook( 'grammar', array( 'CoreParserFunctions', 'grammar' ), SFH_NO_HASH );
 157                 $this->setFunctionHook( 'plural', array( 'CoreParserFunctions', 'plural' ), SFH_NO_HASH );
 158                 $this->setFunctionHook( 'numberofpages', array( 'CoreParserFunctions', 'numberofpages' ), SFH_NO_HASH );
 159                 $this->setFunctionHook( 'numberofusers', array( 'CoreParserFunctions', 'numberofusers' ), SFH_NO_HASH );
 160                 $this->setFunctionHook( 'numberofarticles', array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 161                 $this->setFunctionHook( 'numberoffiles', array( 'CoreParserFunctions', 'numberoffiles' ), SFH_NO_HASH );
 162                 $this->setFunctionHook( 'numberofadmins', array( 'CoreParserFunctions', 'numberofadmins' ), SFH_NO_HASH );
 163                 $this->setFunctionHook( 'language', array( 'CoreParserFunctions', 'language' ), SFH_NO_HASH );
 164                 $this->setFunctionHook( 'padleft', array( 'CoreParserFunctions', 'padleft' ), SFH_NO_HASH );
 165                 $this->setFunctionHook( 'padright', array( 'CoreParserFunctions', 'padright' ), SFH_NO_HASH );
 166                 $this->setFunctionHook( 'anchorencode', array( 'CoreParserFunctions', 'anchorencode' ), SFH_NO_HASH );
 167                 $this->setFunctionHook( 'special', array( 'CoreParserFunctions', 'special' ) );
 168
 169                 if ( $wgAllowDisplayTitle ) {
 170                         $this->setFunctionHook( 'displaytitle', array( 'CoreParserFunctions', 'displaytitle' ), SFH_NO_HASH );
 171                 }
 172                 if ( $wgAllowSlowParserFunctions ) {
 173                         $this->setFunctionHook( 'pagesinnamespace', array( 'CoreParserFunctions', 'pagesinnamespace' ), SFH_NO_HASH );
 174                 }
 175
 176                 $this->initialiseVariables();
 177
 178                 $this->mFirstCall = false;
 179                 wfProfileOut( __METHOD__ );
 180         }
 181
 182         /**
 183          * Clear Parser state
 184          *
 185          * @private
 186          */
 187         function clearState() {
 188                 wfProfileIn( __METHOD__ );
 189                 if ( $this->mFirstCall ) {
 190                         $this->firstCallInit();
 191                 }
 192                 $this->mOutput = new ParserOutput;
 193                 $this->mAutonumber = 0;
 194                 $this->mLastSection = '';
 195                 $this->mDTopen = false;
 196                 $this->mIncludeCount = array();
 197                 $this->mStripState = array();
 198                 $this->mArgStack = array();
 199                 $this->mInPre = false;
 200                 $this->mInterwikiLinkHolders = array(
 201                         'texts' => array(),
 202                         'titles' => array()
 203                 );
 204                 $this->mLinkHolders = array(
 205                         'namespaces' => array(),
 206                         'dbkeys' => array(),
 207                         'queries' => array(),
 208                         'texts' => array(),
 209                         'titles' => array()
 210                 );
 211                 $this->mRevisionId = null;
 212
 213                 /**
 214                  * Prefix for temporary replacement strings for the multipass parser.
 215                  * \x07 should never appear in input as it's disallowed in XML.
 216                  * Using it at the front also gives us a little extra robustness
 217                  * since it shouldn't match when butted up against identifier-like
 218                  * string constructs.
 219                  */
 220                 $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 221
 222                 # Clear these on every parse, bug 4549
 223                 $this->mTemplates = array();
 224                 $this->mTemplatePath = array();
 225
 226                 $this->mShowToc = true;
 227                 $this->mForceTocPosition = false;
 228                 $this->mIncludeSizes = array(
 229                         'pre-expand' => 0,
 230                         'post-expand' => 0,
 231                         'arg' => 0
 232                 );
 233
 234                 wfRunHooks( 'ParserClearState', array( &$this ) );
 235                 wfProfileOut( __METHOD__ );
 236         }
 237
 238         function setOutputType( $ot ) {
 239                 $this->mOutputType = $ot;
 240                 // Shortcut alias
 241                 $this->ot = array(
 242                         'html' => $ot == OT_HTML,
 243                         'wiki' => $ot == OT_WIKI,
 244                         'msg' => $ot == OT_MSG,
 245                         'pre' => $ot == OT_PREPROCESS,
 246                 );
 247         }
 248
 249         /**
 250          * Accessor for mUniqPrefix.
 251          *
 252          * @public
 253          */
 254         function uniqPrefix() {
 255                 return $this->mUniqPrefix;
 256         }
 257
 258         /**
 259          * Convert wikitext to HTML
 260          * Do not call this function recursively.
 261          *
 262          * @private
 263          * @param string $text Text we want to parse
 264          * @param Title &$title A title object
 265          * @param array $options
 266          * @param boolean $linestart
 267          * @param boolean $clearState
 268          * @param int $revid number to pass in {{REVISIONID}}
 269          * @return ParserOutput a ParserOutput
 270          */
 271         function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 272                 /**
 273                  * First pass--just handle <nowiki> sections, pass the rest off
 274                  * to internalParse() which does all the real work.
 275                  */
 276
 277                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 278                 $fname = 'Parser::parse-' . wfGetCaller();
 279                 wfProfileIn( $fname );
 280
 281                 if ( $clearState ) {
 282                         $this->clearState();
 283                 }
 284
 285                 $this->mOptions = $options;
 286                 $this->mTitle =& $title;
 287                 $oldRevisionId = $this->mRevisionId;
 288                 if( $revid !== null ) {
 289                         $this->mRevisionId = $revid;
 290                 }
 291                 $this->setOutputType( OT_HTML );
 292
 293                 //$text = $this->strip( $text, $this->mStripState );
 294                 // VOODOO MAGIC FIX! Sometimes the above segfaults in PHP5.
 295                 $x =& $this->mStripState;
 296
 297                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 298                 $text = $this->strip( $text, $x );
 299                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 300
 301                 $text = $this->internalParse( $text );
 302
 303                 $text = $this->unstrip( $text, $this->mStripState );
 304
 305                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 306                 $fixtags = array(
 307                         # french spaces, last one Guillemet-left
 308                         # only if there is something before the space
 309                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 310                         # french spaces, Guillemet-right
 311                         '/(\\302\\253) /' => '\\1&nbsp;',
 312                 );
 313                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 314
 315                 # only once and last
 316                 $text = $this->doBlockLevels( $text, $linestart );
 317
 318                 $this->replaceLinkHolders( $text );
 319
 320                 # the position of the parserConvert() call should not be changed. it
 321                 # assumes that the links are all replaced and the only thing left
 322                 # is the <nowiki> mark.
 323                 # Side-effects: this calls $this->mOutput->setTitleText()
 324                 $text = $wgContLang->parserConvert( $text, $this );
 325
 326                 $text = $this->unstripNoWiki( $text, $this->mStripState );
 327
 328                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 329
 330                 $text = Sanitizer::normalizeCharReferences( $text );
 331
 332                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 333                         $text = Parser::tidy($text);
 334                 } else {
 335                         # attempt to sanitize at least some nesting problems
 336                         # (bug #2702 and quite a few others)
 337                         $tidyregs = array(
 338                                 # ''Something [http://www.cool.com cool''] -->
 339                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 340                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 341                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 342                                 # fix up an anchor inside another anchor, only
 343                                 # at least for a single single nested link (bug 3695)
 344                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 345                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 346                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 347                                 # contains a div, so fix it up here; replace
 348                                 # div with escaped text
 349                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 350                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 351                                 # remove empty italic or bold tag pairs, some
 352                                 # introduced by rules above
 353                                 '/<([bi])><\/\\1>/' => '',
 354                         );
 355
 356                         $text = preg_replace(
 357                                 array_keys( $tidyregs ),
 358                                 array_values( $tidyregs ),
 359                                 $text );
 360                 }
 361
 362                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 363
 364                 # Information on include size limits, for the benefit of users who try to skirt them
 365                 if ( max( $this->mIncludeSizes ) > 1000 ) {
 366                         $max = $this->mOptions->getMaxIncludeSize();
 367                         $text .= "<!-- \n" .
 368                                 "Pre-expand include size: {$this->mIncludeSizes['pre-expand']} bytes\n" .
 369                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']} bytes\n" .
 370                                 "Template argument size: {$this->mIncludeSizes['arg']} bytes\n" .
 371                                 "Maximum: $max bytes\n" .
 372                                 "-->\n";
 373                 }
 374                 $this->mOutput->setText( $text );
 375                 $this->mRevisionId = $oldRevisionId;
 376                 wfProfileOut( $fname );
 377
 378                 return $this->mOutput;
 379         }
 380
 381         /**
 382          * Recursive parser entry point that can be called from an extension tag
 383          * hook.
 384          */
 385         function recursiveTagParse( $text ) {
 386                 wfProfileIn( __METHOD__ );
 387                 $x =& $this->mStripState;
 388                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 389                 $text = $this->strip( $text, $x );
 390                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 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                 $x =& $this->mStripState;
 407                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$x ) );
 408                 $text = $this->strip( $text, $x );
 409                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$x ) );
 410                 if ( $this->mOptions->getRemoveComments() ) {
 411                         $text = Sanitizer::removeHTMLcomments( $text );
 412                 }
 413                 $text = $this->replaceVariables( $text );
 414                 $text = $this->unstrip( $text, $x );
 415                 $text = $this->unstripNowiki( $text, $x );
 416                 wfProfileOut( __METHOD__ );
 417                 return $text;
 418         }
 419
 420         /**
 421          * Get a random string
 422          *
 423          * @private
 424          * @static
 425          */
 426         function getRandomString() {
 427                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 428         }
 429
 430         function &getTitle() { return $this->mTitle; }
 431         function getOptions() { return $this->mOptions; }
 432
 433         function getFunctionLang() {
 434                 global $wgLang, $wgContLang;
 435                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 436         }
 437
 438         /**
 439          * Replaces all occurrences of HTML-style comments and the given tags
 440          * in the text with a random marker and returns teh next text. The output
 441          * parameter $matches will be an associative array filled with data in
 442          * the form:
 443          *   'UNIQ-xxxxx' => array(
 444          *     'element',
 445          *     'tag content',
 446          *     array( 'param' => 'x' ),
 447          *     '<element param="x">tag content</element>' ) )
 448          *
 449          * @param $elements list of element names. Comments are always extracted.
 450          * @param $text Source text string.
 451          * @param $uniq_prefix
 452          *
 453          * @private
 454          * @static
 455          */
 456         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 457                 static $n = 1;
 458                 $stripped = '';
 459                 $matches = array();
 460
 461                 $taglist = implode( '|', $elements );
 462                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 463
 464                 while ( '' != $text ) {
 465                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 466                         $stripped .= $p[0];
 467                         if( count( $p ) < 5 ) {
 468                                 break;
 469                         }
 470                         if( count( $p ) > 5 ) {
 471                                 // comment
 472                                 $element    = $p[4];
 473                                 $attributes = '';
 474                                 $close      = '';
 475                                 $inside     = $p[5];
 476                         } else {
 477                                 // tag
 478                                 $element    = $p[1];
 479                                 $attributes = $p[2];
 480                                 $close      = $p[3];
 481                                 $inside     = $p[4];
 482                         }
 483
 484                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . '-QINU';
 485                         $stripped .= $marker;
 486
 487                         if ( $close === '/>' ) {
 488                                 // Empty element tag, <tag />
 489                                 $content = null;
 490                                 $text = $inside;
 491                                 $tail = null;
 492                         } else {
 493                                 if( $element == '!--' ) {
 494                                         $end = '/(-->)/';
 495                                 } else {
 496                                         $end = "/(<\\/$element\\s*>)/i";
 497                                 }
 498                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 499                                 $content = $q[0];
 500                                 if( count( $q ) < 3 ) {
 501                                         # No end tag -- let it run out to the end of the text.
 502                                         $tail = '';
 503                                         $text = '';
 504                                 } else {
 505                                         $tail = $q[1];
 506                                         $text = $q[2];
 507                                 }
 508                         }
 509
 510                         $matches[$marker] = array( $element,
 511                                 $content,
 512                                 Sanitizer::decodeTagAttributes( $attributes ),
 513                                 "<$element$attributes$close$content$tail" );
 514                 }
 515                 return $stripped;
 516         }
 517
 518         /**
 519          * Strips and renders nowiki, pre, math, hiero
 520          * If $render is set, performs necessary rendering operations on plugins
 521          * Returns the text, and fills an array with data needed in unstrip()
 522          * If the $state is already a valid strip state, it adds to the state
 523          *
 524          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 525          *  will be stripped in addition to other tags. This is important
 526          *  for section editing, where these comments cause confusion when
 527          *  counting the sections in the wikisource
 528          *
 529          * @param array dontstrip contains tags which should not be stripped;
 530          *  used to prevent stipping of <gallery> when saving (fixes bug 2700)
 531          *
 532          * @private
 533          */
 534         function strip( $text, &$state, $stripcomments = false , $dontstrip = array () ) {
 535                 wfProfileIn( __METHOD__ );
 536                 $render = ($this->mOutputType == OT_HTML);
 537
 538                 $uniq_prefix = $this->mUniqPrefix;
 539                 $commentState = array();
 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 = wfEscapeHTMLTagsOnly( $content );
 585                                         break;
 586                                 case 'math':
 587                                         $output = 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 = $this->unstrip( $output, $state );
 609
 610                         if( !$stripcomments && $element == '!--' ) {
 611                                 $commentState[$marker] = $output;
 612                         } elseif ( $element == 'html' || $element == 'nowiki' ) {
 613                                 $state['nowiki'][$marker] = $output;
 614                         } else {
 615                                 $state['general'][$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 = strtr( $text, $commentState );
 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          */
 638         function unstrip( $text, $state ) {
 639                 if ( !isset( $state['general'] ) ) {
 640                         return $text;
 641                 }
 642
 643                 wfProfileIn( __METHOD__ );
 644                 # TODO: good candidate for FSS
 645                 $text = strtr( $text, $state['general'] );
 646                 wfProfileOut( __METHOD__ );
 647                 return $text;
 648         }
 649
 650         /**
 651          * Always call this after unstrip() to preserve the order
 652          *
 653          * @private
 654          */
 655         function unstripNoWiki( $text, $state ) {
 656                 if ( !isset( $state['nowiki'] ) ) {
 657                         return $text;
 658                 }
 659
 660                 wfProfileIn( __METHOD__ );
 661                 # TODO: good candidate for FSS
 662                 $text = strtr( $text, $state['nowiki'] );
 663                 wfProfileOut( __METHOD__ );
 664
 665                 return $text;
 666         }
 667
 668         /**
 669          * Add an item to the strip state
 670          * Returns the unique tag which must be inserted into the stripped text
 671          * The tag will be replaced with the original text in unstrip()
 672          *
 673          * @private
 674          */
 675         function insertStripItem( $text, &$state ) {
 676                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 677                 if ( !$state ) {
 678                         $state = array();
 679                 }
 680                 $state['general'][$rnd] = $text;
 681                 return $rnd;
 682         }
 683
 684         /**
 685          * Interface with html tidy, used if $wgUseTidy = true.
 686          * If tidy isn't able to correct the markup, the original will be
 687          * returned in all its glory with a warning comment appended.
 688          *
 689          * Either the external tidy program or the in-process tidy extension
 690          * will be used depending on availability. Override the default
 691          * $wgTidyInternal setting to disable the internal if it's not working.
 692          *
 693          * @param string $text Hideous HTML input
 694          * @return string Corrected HTML output
 695          * @public
 696          * @static
 697          */
 698         function tidy( $text ) {
 699                 global $wgTidyInternal;
 700                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 701 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 702 '<head><title>test</title></head><body>'.$text.'</body></html>';
 703                 if( $wgTidyInternal ) {
 704                         $correctedtext = Parser::internalTidy( $wrappedtext );
 705                 } else {
 706                         $correctedtext = Parser::externalTidy( $wrappedtext );
 707                 }
 708                 if( is_null( $correctedtext ) ) {
 709                         wfDebug( "Tidy error detected!\n" );
 710                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 711                 }
 712                 return $correctedtext;
 713         }
 714
 715         /**
 716          * Spawn an external HTML tidy process and get corrected markup back from it.
 717          *
 718          * @private
 719          * @static
 720          */
 721         function externalTidy( $text ) {
 722                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 723                 $fname = 'Parser::externalTidy';
 724                 wfProfileIn( $fname );
 725
 726                 $cleansource = '';
 727                 $opts = ' -utf8';
 728
 729                 $descriptorspec = array(
 730                         0 => array('pipe', 'r'),
 731                         1 => array('pipe', 'w'),
 732                         2 => array('file', '/dev/null', 'a')
 733                 );
 734                 $pipes = array();
 735                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 736                 if (is_resource($process)) {
 737                         // Theoretically, this style of communication could cause a deadlock
 738                         // here. If the stdout buffer fills up, then writes to stdin could
 739                         // block. This doesn't appear to happen with tidy, because tidy only
 740                         // writes to stdout after it's finished reading from stdin. Search
 741                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 742                         fwrite($pipes[0], $text);
 743                         fclose($pipes[0]);
 744                         while (!feof($pipes[1])) {
 745                                 $cleansource .= fgets($pipes[1], 1024);
 746                         }
 747                         fclose($pipes[1]);
 748                         proc_close($process);
 749                 }
 750
 751                 wfProfileOut( $fname );
 752
 753                 if( $cleansource == '' && $text != '') {
 754                         // Some kind of error happened, so we couldn't get the corrected text.
 755                         // Just give up; we'll use the source text and append a warning.
 756                         return null;
 757                 } else {
 758                         return $cleansource;
 759                 }
 760         }
 761
 762         /**
 763          * Use the HTML tidy PECL extension to use the tidy library in-process,
 764          * saving the overhead of spawning a new process. Currently written to
 765          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 766          *
 767          * 'pear install tidy' should be able to compile the extension module.
 768          *
 769          * @private
 770          * @static
 771          */
 772         function internalTidy( $text ) {
 773                 global $wgTidyConf;
 774                 $fname = 'Parser::internalTidy';
 775                 wfProfileIn( $fname );
 776
 777                 tidy_load_config( $wgTidyConf );
 778                 tidy_set_encoding( 'utf8' );
 779                 tidy_parse_string( $text );
 780                 tidy_clean_repair();
 781                 if( tidy_get_status() == 2 ) {
 782                         // 2 is magic number for fatal error
 783                         // http://www.php.net/manual/en/function.tidy-get-status.php
 784                         $cleansource = null;
 785                 } else {
 786                         $cleansource = tidy_get_output();
 787                 }
 788                 wfProfileOut( $fname );
 789                 return $cleansource;
 790         }
 791
 792         /**
 793          * parse the wiki syntax used to render tables
 794          *
 795          * @private
 796          */
 797         function doTableStuff ( $t ) {
 798                 $fname = 'Parser::doTableStuff';
 799                 wfProfileIn( $fname );
 800
 801                 $t = explode ( "\n" , $t ) ;
 802                 $td = array () ; # Is currently a td tag open?
 803                 $ltd = array () ; # Was it TD or TH?
 804                 $tr = array () ; # Is currently a tr tag open?
 805                 $ltr = array () ; # tr attributes
 806                 $has_opened_tr = array(); # Did this table open a <tr> element?
 807                 $indent_level = 0; # indent level of the table
 808                 foreach ( $t AS $k => $x )
 809                 {
 810                         $x = trim ( $x ) ;
 811                         $fc = substr ( $x , 0 , 1 ) ;
 812                         $matches = array();
 813                         if ( preg_match( '/^(:*)\{\|(.*)$/', $x, $matches ) ) {
 814                                 $indent_level = strlen( $matches[1] );
 815
 816                                 $attributes = $this->unstripForHTML( $matches[2] );
 817
 818                                 $t[$k] = str_repeat( '<dl><dd>', $indent_level ) .
 819                                         '<table' . Sanitizer::fixTagAttributes ( $attributes, 'table' ) . '>' ;
 820                                 array_push ( $td , false ) ;
 821                                 array_push ( $ltd , '' ) ;
 822                                 array_push ( $tr , false ) ;
 823                                 array_push ( $ltr , '' ) ;
 824                                 array_push ( $has_opened_tr, false );
 825                         }
 826                         else if ( count ( $td ) == 0 ) { } # Don't do any of the following
 827                         else if ( '|}' == substr ( $x , 0 , 2 ) ) {
 828                                 $z = "</table>" . substr ( $x , 2);
 829                                 $l = array_pop ( $ltd ) ;
 830                                 if ( !array_pop ( $has_opened_tr ) ) $z = "<tr><td></td></tr>" . $z ;
 831                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 832                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 833                                 array_pop ( $ltr ) ;
 834                                 $t[$k] = $z . str_repeat( '</dd></dl>', $indent_level );
 835                         }
 836                         else if ( '|-' == substr ( $x , 0 , 2 ) ) { # Allows for |---------------
 837                                 $x = substr ( $x , 1 ) ;
 838                                 while ( $x != '' && substr ( $x , 0 , 1 ) == '-' ) $x = substr ( $x , 1 ) ;
 839                                 $z = '' ;
 840                                 $l = array_pop ( $ltd ) ;
 841                                 array_pop ( $has_opened_tr );
 842                                 array_push ( $has_opened_tr , true ) ;
 843                                 if ( array_pop ( $tr ) ) $z = '</tr>' . $z ;
 844                                 if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 845                                 array_pop ( $ltr ) ;
 846                                 $t[$k] = $z ;
 847                                 array_push ( $tr , false ) ;
 848                                 array_push ( $td , false ) ;
 849                                 array_push ( $ltd , '' ) ;
 850                                 $attributes = $this->unstripForHTML( $x );
 851                                 array_push ( $ltr , Sanitizer::fixTagAttributes ( $attributes, 'tr' ) ) ;
 852                         }
 853                         else if ( '|' == $fc || '!' == $fc || '|+' == substr ( $x , 0 , 2 ) ) { # Caption
 854                                 # $x is a table row
 855                                 if ( '|+' == substr ( $x , 0 , 2 ) ) {
 856                                         $fc = '+' ;
 857                                         $x = substr ( $x , 1 ) ;
 858                                 }
 859                                 $after = substr ( $x , 1 ) ;
 860                                 if ( $fc == '!' ) $after = str_replace ( '!!' , '||' , $after ) ;
 861
 862                                 // Split up multiple cells on the same line.
 863                                 // FIXME: This can result in improper nesting of tags processed
 864                                 // by earlier parser steps, but should avoid splitting up eg
 865                                 // attribute values containing literal "||".
 866                                 $after = wfExplodeMarkup( '||', $after );
 867
 868                                 $t[$k] = '' ;
 869
 870                                 # Loop through each table cell
 871                                 foreach ( $after AS $theline )
 872                                 {
 873                                         $z = '' ;
 874                                         if ( $fc != '+' )
 875                                         {
 876                                                 $tra = array_pop ( $ltr ) ;
 877                                                 if ( !array_pop ( $tr ) ) $z = '<tr'.$tra.">\n" ;
 878                                                 array_push ( $tr , true ) ;
 879                                                 array_push ( $ltr , '' ) ;
 880                                                 array_pop ( $has_opened_tr );
 881                                                 array_push ( $has_opened_tr , true ) ;
 882                                         }
 883
 884                                         $l = array_pop ( $ltd ) ;
 885                                         if ( array_pop ( $td ) ) $z = '</'.$l.'>' . $z ;
 886                                         if ( $fc == '|' ) {
 887                                             $l = 'td' ;
 888                                         } else if ( $fc == '!' ) {
 889                                             $l = 'th' ;
 890                                         } else if ( $fc == '+' ) {
 891                                             $l = 'caption' ;
 892                                         } else {
 893                                             $l = '' ;
 894                                         }
 895                                         array_push ( $ltd , $l ) ;
 896
 897                                         # Cell parameters
 898                                         $y = explode ( '|' , $theline , 2 ) ;
 899                                         # Note that a '|' inside an invalid link should not
 900                                         # be mistaken as delimiting cell parameters
 901                                         if ( strpos( $y[0], '[[' ) !== false ) {
 902                                                 $y = array ($theline);
 903                                         }
 904                                         if ( count ( $y ) == 1 )
 905                                                 $y = "{$z}<{$l}>{$y[0]}" ;
 906                                         else {
 907                                                 $attributes = $this->unstripForHTML( $y[0] );
 908                                                 $y = "{$z}<{$l}".Sanitizer::fixTagAttributes($attributes, $l).">{$y[1]}" ;
 909                                         }
 910                                         $t[$k] .= $y ;
 911                                         array_push ( $td , true ) ;
 912                                 }
 913                         }
 914                 }
 915
 916                 # Closing open td, tr && table
 917                 while ( count ( $td ) > 0 )
 918                 {
 919                         $l = array_pop ( $ltd ) ;
 920                         if ( array_pop ( $td ) ) $t[] = '</td>' ;
 921                         if ( array_pop ( $tr ) ) $t[] = '</tr>' ;
 922                         if ( !array_pop ( $has_opened_tr ) ) $t[] = "<tr><td></td></tr>" ;
 923                         $t[] = '</table>' ;
 924                 }
 925
 926                 $t = implode ( "\n" , $t ) ;
 927                 # special case: don't return empty table
 928                 if($t == "<table>\n<tr><td></td></tr>\n</table>")
 929                         $t = '';
 930                 wfProfileOut( $fname );
 931                 return $t ;
 932         }
 933
 934         /**
 935          * Helper function for parse() that transforms wiki markup into
 936          * HTML. Only called for $mOutputType == OT_HTML.
 937          *
 938          * @private
 939          */
 940         function internalParse( $text ) {
 941                 $args = array();
 942                 $isMain = true;
 943                 $fname = 'Parser::internalParse';
 944                 wfProfileIn( $fname );
 945
 946                 # Hook to suspend the parser in this state
 947                 $x =& $this->mStripState; // FIXME: Please check that this initialization is correct.
 948                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$x ) ) ) {
 949                         wfProfileOut( $fname );
 950                         return $text ;
 951                 }
 952
 953                 # Remove <noinclude> tags and <includeonly> sections
 954                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 955                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 956                 $text = preg_replace( '/<includeonly>.*?<\/includeonly>/s', '', $text );
 957
 958                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
 959
 960                 $text = $this->replaceVariables( $text, $args );
 961
 962                 // Tables need to come after variable replacement for things to work
 963                 // properly; putting them before other transformations should keep
 964                 // exciting things like link expansions from showing up in surprising
 965                 // places.
 966                 $text = $this->doTableStuff( $text );
 967
 968                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
 969
 970                 $text = $this->stripToc( $text );
 971                 $this->stripNoGallery( $text );
 972                 $text = $this->doHeadings( $text );
 973                 if($this->mOptions->getUseDynamicDates()) {
 974                         $df =& DateFormatter::getInstance();
 975                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
 976                 }
 977                 $text = $this->doAllQuotes( $text );
 978                 $text = $this->replaceInternalLinks( $text );
 979                 $text = $this->replaceExternalLinks( $text );
 980
 981                 # replaceInternalLinks may sometimes leave behind
 982                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
 983                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
 984
 985                 $text = $this->doMagicLinks( $text );
 986                 $text = $this->formatHeadings( $text, $isMain );
 987
 988                 wfProfileOut( $fname );
 989                 return $text;
 990         }
 991
 992         /**
 993          * Replace special strings like "ISBN xxx" and "RFC xxx" with
 994          * magic external links.
 995          *
 996          * @private
 997          */
 998         function &doMagicLinks( &$text ) {
 999                 wfProfileIn( __METHOD__ );
1000                 $text = preg_replace_callback(
1001                         '!(?:                           # Start cases
1002                             <a.*?</a> |                 # Skip link text
1003                             <.*?> |                     # Skip stuff inside HTML elements
1004                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
1005                             ISBN\s+([0-9Xx-]+)          # ISBN, capture number as m[2]
1006                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1007                 wfProfileOut( __METHOD__ );
1008                 return $text;
1009         }
1010
1011         function magicLinkCallback( $m ) {
1012                 if ( substr( $m[0], 0, 1 ) == '<' ) {
1013                         # Skip HTML element
1014                         return $m[0];
1015                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
1016                         $isbn = $m[2];
1017                         $num = strtr( $isbn, array(
1018                                 '-' => '',
1019                                 ' ' => '',
1020                                 'x' => 'X',
1021                         ));
1022                         $titleObj = SpecialPage::getTitleFor( 'Booksources' );
1023                         $text = '<a href="' .
1024                                 $titleObj->escapeLocalUrl( "isbn=$num" ) .
1025                                 "\" class=\"internal\">ISBN $isbn</a>";
1026                 } else {
1027                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
1028                                 $keyword = 'RFC';
1029                                 $urlmsg = 'rfcurl';
1030                                 $id = $m[1];
1031                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
1032                                 $keyword = 'PMID';
1033                                 $urlmsg = 'pubmedurl';
1034                                 $id = $m[1];
1035                         } else {
1036                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1037                                         substr($m[0], 0, 20 ) . '"' );
1038                         }
1039
1040                         $url = wfMsg( $urlmsg, $id);
1041                         $sk =& $this->mOptions->getSkin();
1042                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1043                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1044                 }
1045                 return $text;
1046         }
1047
1048         /**
1049          * Parse headers and return html
1050          *
1051          * @private
1052          */
1053         function doHeadings( $text ) {
1054                 $fname = 'Parser::doHeadings';
1055                 wfProfileIn( $fname );
1056                 for ( $i = 6; $i >= 1; --$i ) {
1057                         $h = str_repeat( '=', $i );
1058                         $text = preg_replace( "/^{$h}(.+){$h}\\s*$/m",
1059                           "<h{$i}>\\1</h{$i}>\\2", $text );
1060                 }
1061                 wfProfileOut( $fname );
1062                 return $text;
1063         }
1064
1065         /**
1066          * Replace single quotes with HTML markup
1067          * @private
1068          * @return string the altered text
1069          */
1070         function doAllQuotes( $text ) {
1071                 $fname = 'Parser::doAllQuotes';
1072                 wfProfileIn( $fname );
1073                 $outtext = '';
1074                 $lines = explode( "\n", $text );
1075                 foreach ( $lines as $line ) {
1076                         $outtext .= $this->doQuotes ( $line ) . "\n";
1077                 }
1078                 $outtext = substr($outtext, 0,-1);
1079                 wfProfileOut( $fname );
1080                 return $outtext;
1081         }
1082
1083         /**
1084          * Helper function for doAllQuotes()
1085          * @private
1086          */
1087         function doQuotes( $text ) {
1088                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1089                 if ( count( $arr ) == 1 )
1090                         return $text;
1091                 else
1092                 {
1093                         # First, do some preliminary work. This may shift some apostrophes from
1094                         # being mark-up to being text. It also counts the number of occurrences
1095                         # of bold and italics mark-ups.
1096                         $i = 0;
1097                         $numbold = 0;
1098                         $numitalics = 0;
1099                         foreach ( $arr as $r )
1100                         {
1101                                 if ( ( $i % 2 ) == 1 )
1102                                 {
1103                                         # If there are ever four apostrophes, assume the first is supposed to
1104                                         # be text, and the remaining three constitute mark-up for bold text.
1105                                         if ( strlen( $arr[$i] ) == 4 )
1106                                         {
1107                                                 $arr[$i-1] .= "'";
1108                                                 $arr[$i] = "'''";
1109                                         }
1110                                         # If there are more than 5 apostrophes in a row, assume they're all
1111                                         # text except for the last 5.
1112                                         else if ( strlen( $arr[$i] ) > 5 )
1113                                         {
1114                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1115                                                 $arr[$i] = "'''''";
1116                                         }
1117                                         # Count the number of occurrences of bold and italics mark-ups.
1118                                         # We are not counting sequences of five apostrophes.
1119                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1120                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1121                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1122                                 }
1123                                 $i++;
1124                         }
1125
1126                         # If there is an odd number of both bold and italics, it is likely
1127                         # that one of the bold ones was meant to be an apostrophe followed
1128                         # by italics. Which one we cannot know for certain, but it is more
1129                         # likely to be one that has a single-letter word before it.
1130                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1131                         {
1132                                 $i = 0;
1133                                 $firstsingleletterword = -1;
1134                                 $firstmultiletterword = -1;
1135                                 $firstspace = -1;
1136                                 foreach ( $arr as $r )
1137                                 {
1138                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1139                                         {
1140                                                 $x1 = substr ($arr[$i-1], -1);
1141                                                 $x2 = substr ($arr[$i-1], -2, 1);
1142                                                 if ($x1 == ' ') {
1143                                                         if ($firstspace == -1) $firstspace = $i;
1144                                                 } else if ($x2 == ' ') {
1145                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1146                                                 } else {
1147                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1148                                                 }
1149                                         }
1150                                         $i++;
1151                                 }
1152
1153                                 # If there is a single-letter word, use it!
1154                                 if ($firstsingleletterword > -1)
1155                                 {
1156                                         $arr [ $firstsingleletterword ] = "''";
1157                                         $arr [ $firstsingleletterword-1 ] .= "'";
1158                                 }
1159                                 # If not, but there's a multi-letter word, use that one.
1160                                 else if ($firstmultiletterword > -1)
1161                                 {
1162                                         $arr [ $firstmultiletterword ] = "''";
1163                                         $arr [ $firstmultiletterword-1 ] .= "'";
1164                                 }
1165                                 # ... otherwise use the first one that has neither.
1166                                 # (notice that it is possible for all three to be -1 if, for example,
1167                                 # there is only one pentuple-apostrophe in the line)
1168                                 else if ($firstspace > -1)
1169                                 {
1170                                         $arr [ $firstspace ] = "''";
1171                                         $arr [ $firstspace-1 ] .= "'";
1172                                 }
1173                         }
1174
1175                         # Now let's actually convert our apostrophic mush to HTML!
1176                         $output = '';
1177                         $buffer = '';
1178                         $state = '';
1179                         $i = 0;
1180                         foreach ($arr as $r)
1181                         {
1182                                 if (($i % 2) == 0)
1183                                 {
1184                                         if ($state == 'both')
1185                                                 $buffer .= $r;
1186                                         else
1187                                                 $output .= $r;
1188                                 }
1189                                 else
1190                                 {
1191                                         if (strlen ($r) == 2)
1192                                         {
1193                                                 if ($state == 'i')
1194                                                 { $output .= '</i>'; $state = ''; }
1195                                                 else if ($state == 'bi')
1196                                                 { $output .= '</i>'; $state = 'b'; }
1197                                                 else if ($state == 'ib')
1198                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1199                                                 else if ($state == 'both')
1200                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1201                                                 else # $state can be 'b' or ''
1202                                                 { $output .= '<i>'; $state .= 'i'; }
1203                                         }
1204                                         else if (strlen ($r) == 3)
1205                                         {
1206                                                 if ($state == 'b')
1207                                                 { $output .= '</b>'; $state = ''; }
1208                                                 else if ($state == 'bi')
1209                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1210                                                 else if ($state == 'ib')
1211                                                 { $output .= '</b>'; $state = 'i'; }
1212                                                 else if ($state == 'both')
1213                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1214                                                 else # $state can be 'i' or ''
1215                                                 { $output .= '<b>'; $state .= 'b'; }
1216                                         }
1217                                         else if (strlen ($r) == 5)
1218                                         {
1219                                                 if ($state == 'b')
1220                                                 { $output .= '</b><i>'; $state = 'i'; }
1221                                                 else if ($state == 'i')
1222                                                 { $output .= '</i><b>'; $state = 'b'; }
1223                                                 else if ($state == 'bi')
1224                                                 { $output .= '</i></b>'; $state = ''; }
1225                                                 else if ($state == 'ib')
1226                                                 { $output .= '</b></i>'; $state = ''; }
1227                                                 else if ($state == 'both')
1228                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1229                                                 else # ($state == '')
1230                                                 { $buffer = ''; $state = 'both'; }
1231                                         }
1232                                 }
1233                                 $i++;
1234                         }
1235                         # Now close all remaining tags.  Notice that the order is important.
1236                         if ($state == 'b' || $state == 'ib')
1237                                 $output .= '</b>';
1238                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1239                                 $output .= '</i>';
1240                         if ($state == 'bi')
1241                                 $output .= '</b>';
1242                         if ($state == 'both')
1243                                 $output .= '<b><i>'.$buffer.'</i></b>';
1244                         return $output;
1245                 }
1246         }
1247
1248         /**
1249          * Replace external links
1250          *
1251          * Note: this is all very hackish and the order of execution matters a lot.
1252          * Make sure to run maintenance/parserTests.php if you change this code.
1253          *
1254          * @private
1255          */
1256         function replaceExternalLinks( $text ) {
1257                 global $wgContLang;
1258                 $fname = 'Parser::replaceExternalLinks';
1259                 wfProfileIn( $fname );
1260
1261                 $sk =& $this->mOptions->getSkin();
1262
1263                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1264
1265                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1266
1267                 $i = 0;
1268                 while ( $i<count( $bits ) ) {
1269                         $url = $bits[$i++];
1270                         $protocol = $bits[$i++];
1271                         $text = $bits[$i++];
1272                         $trail = $bits[$i++];
1273
1274                         # The characters '<' and '>' (which were escaped by
1275                         # removeHTMLtags()) should not be included in
1276                         # URLs, per RFC 2396.
1277                         $m2 = array();
1278                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1279                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1280                                 $url = substr($url, 0, $m2[0][1]);
1281                         }
1282
1283                         # If the link text is an image URL, replace it with an <img> tag
1284                         # This happened by accident in the original parser, but some people used it extensively
1285                         $img = $this->maybeMakeExternalImage( $text );
1286                         if ( $img !== false ) {
1287                                 $text = $img;
1288                         }
1289
1290                         $dtrail = '';
1291
1292                         # Set linktype for CSS - if URL==text, link is essentially free
1293                         $linktype = ($text == $url) ? 'free' : 'text';
1294
1295                         # No link text, e.g. [http://domain.tld/some.link]
1296                         if ( $text == '' ) {
1297                                 # Autonumber if allowed. See bug #5918
1298                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1299                                         $text = '[' . ++$this->mAutonumber . ']';
1300                                         $linktype = 'autonumber';
1301                                 } else {
1302                                         # Otherwise just use the URL
1303                                         $text = htmlspecialchars( $url );
1304                                         $linktype = 'free';
1305                                 }
1306                         } else {
1307                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1308                                 # Check for trail
1309                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1310                         }
1311
1312                         $text = $wgContLang->markNoConversion($text);
1313
1314                         $url = Sanitizer::cleanUrl( $url );
1315
1316                         # Process the trail (i.e. everything after this link up until start of the next link),
1317                         # replacing any non-bracketed links
1318                         $trail = $this->replaceFreeExternalLinks( $trail );
1319
1320                         # Use the encoded URL
1321                         # This means that users can paste URLs directly into the text
1322                         # Funny characters like &ouml; aren't valid in URLs anyway
1323                         # This was changed in August 2004
1324                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1325
1326                         # Register link in the output object.
1327                         # Replace unnecessary URL escape codes with the referenced character
1328                         # This prevents spammers from hiding links from the filters
1329                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1330                         $this->mOutput->addExternalLink( $pasteurized );
1331                 }
1332
1333                 wfProfileOut( $fname );
1334                 return $s;
1335         }
1336
1337         /**
1338          * Replace anything that looks like a URL with a link
1339          * @private
1340          */
1341         function replaceFreeExternalLinks( $text ) {
1342                 global $wgContLang;
1343                 $fname = 'Parser::replaceFreeExternalLinks';
1344                 wfProfileIn( $fname );
1345
1346                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1347                 $s = array_shift( $bits );
1348                 $i = 0;
1349
1350                 $sk =& $this->mOptions->getSkin();
1351
1352                 while ( $i < count( $bits ) ){
1353                         $protocol = $bits[$i++];
1354                         $remainder = $bits[$i++];
1355
1356                         $m = array();
1357                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1358                                 # Found some characters after the protocol that look promising
1359                                 $url = $protocol . $m[1];
1360                                 $trail = $m[2];
1361
1362                                 # special case: handle urls as url args:
1363                                 # http://www.example.com/foo?=http://www.example.com/bar
1364                                 if(strlen($trail) == 0 &&
1365                                         isset($bits[$i]) &&
1366                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1367                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1368                                 {
1369                                         # add protocol, arg
1370                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1371                                         $i += 2;
1372                                         $trail = $m[2];
1373                                 }
1374
1375                                 # The characters '<' and '>' (which were escaped by
1376                                 # removeHTMLtags()) should not be included in
1377                                 # URLs, per RFC 2396.
1378                                 $m2 = array();
1379                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1380                                         $trail = substr($url, $m2[0][1]) . $trail;
1381                                         $url = substr($url, 0, $m2[0][1]);
1382                                 }
1383
1384                                 # Move trailing punctuation to $trail
1385                                 $sep = ',;\.:!?';
1386                                 # If there is no left bracket, then consider right brackets fair game too
1387                                 if ( strpos( $url, '(' ) === false ) {
1388                                         $sep .= ')';
1389                                 }
1390
1391                                 $numSepChars = strspn( strrev( $url ), $sep );
1392                                 if ( $numSepChars ) {
1393                                         $trail = substr( $url, -$numSepChars ) . $trail;
1394                                         $url = substr( $url, 0, -$numSepChars );
1395                                 }
1396
1397                                 $url = Sanitizer::cleanUrl( $url );
1398
1399                                 # Is this an external image?
1400                                 $text = $this->maybeMakeExternalImage( $url );
1401                                 if ( $text === false ) {
1402                                         # Not an image, make a link
1403                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1404                                         # Register it in the output object...
1405                                         # Replace unnecessary URL escape codes with their equivalent characters
1406                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1407                                         $this->mOutput->addExternalLink( $pasteurized );
1408                                 }
1409                                 $s .= $text . $trail;
1410                         } else {
1411                                 $s .= $protocol . $remainder;
1412                         }
1413                 }
1414                 wfProfileOut( $fname );
1415                 return $s;
1416         }
1417
1418         /**
1419          * Replace unusual URL escape codes with their equivalent characters
1420          * @param string
1421          * @return string
1422          * @static
1423          * @fixme This can merge genuinely required bits in the path or query string,
1424          *        breaking legit URLs. A proper fix would treat the various parts of
1425          *        the URL differently; as a workaround, just use the output for
1426          *        statistical records, not for actual linking/output.
1427          */
1428         static function replaceUnusualEscapes( $url ) {
1429                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1430                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1431         }
1432
1433         /**
1434          * Callback function used in replaceUnusualEscapes().
1435          * Replaces unusual URL escape codes with their equivalent character
1436          * @static
1437          * @private
1438          */
1439         private static function replaceUnusualEscapesCallback( $matches ) {
1440                 $char = urldecode( $matches[0] );
1441                 $ord = ord( $char );
1442                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1443                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1444                         // No, shouldn't be escaped
1445                         return $char;
1446                 } else {
1447                         // Yes, leave it escaped
1448                         return $matches[0];
1449                 }
1450         }
1451
1452         /**
1453          * make an image if it's allowed, either through the global
1454          * option or through the exception
1455          * @private
1456          */
1457         function maybeMakeExternalImage( $url ) {
1458                 $sk =& $this->mOptions->getSkin();
1459                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1460                 $imagesexception = !empty($imagesfrom);
1461                 $text = false;
1462                 if ( $this->mOptions->getAllowExternalImages()
1463                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1464                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1465                                 # Image found
1466                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1467                         }
1468                 }
1469                 return $text;
1470         }
1471
1472         /**
1473          * Process [[ ]] wikilinks
1474          *
1475          * @private
1476          */
1477         function replaceInternalLinks( $s ) {
1478                 global $wgContLang;
1479                 static $fname = 'Parser::replaceInternalLinks' ;
1480
1481                 wfProfileIn( $fname );
1482
1483                 wfProfileIn( $fname.'-setup' );
1484                 static $tc = FALSE;
1485                 # the % is needed to support urlencoded titles as well
1486                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1487
1488                 $sk =& $this->mOptions->getSkin();
1489
1490                 #split the entire text string on occurences of [[
1491                 $a = explode( '[[', ' ' . $s );
1492                 #get the first element (all text up to first [[), and remove the space we added
1493                 $s = array_shift( $a );
1494                 $s = substr( $s, 1 );
1495
1496                 # Match a link having the form [[namespace:link|alternate]]trail
1497                 static $e1 = FALSE;
1498                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1499                 # Match cases where there is no "]]", which might still be images
1500                 static $e1_img = FALSE;
1501                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1502                 # Match the end of a line for a word that's not followed by whitespace,
1503                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1504                 $e2 = wfMsgForContent( 'linkprefix' );
1505
1506                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1507
1508                 if( is_null( $this->mTitle ) ) {
1509                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1510                 }
1511                 $nottalk = !$this->mTitle->isTalkPage();
1512
1513                 if ( $useLinkPrefixExtension ) {
1514                         $m = array();
1515                         if ( preg_match( $e2, $s, $m ) ) {
1516                                 $first_prefix = $m[2];
1517                         } else {
1518                                 $first_prefix = false;
1519                         }
1520                 } else {
1521                         $prefix = '';
1522                 }
1523
1524                 $selflink = $this->mTitle->getPrefixedText();
1525                 $useSubpages = $this->areSubpagesAllowed();
1526                 wfProfileOut( $fname.'-setup' );
1527
1528                 # Loop for each link
1529                 for ($k = 0; isset( $a[$k] ); $k++) {
1530                         $line = $a[$k];
1531                         if ( $useLinkPrefixExtension ) {
1532                                 wfProfileIn( $fname.'-prefixhandling' );
1533                                 if ( preg_match( $e2, $s, $m ) ) {
1534                                         $prefix = $m[2];
1535                                         $s = $m[1];
1536                                 } else {
1537                                         $prefix='';
1538                                 }
1539                                 # first link
1540                                 if($first_prefix) {
1541                                         $prefix = $first_prefix;
1542                                         $first_prefix = false;
1543                                 }
1544                                 wfProfileOut( $fname.'-prefixhandling' );
1545                         }
1546
1547                         $might_be_img = false;
1548
1549                         wfProfileIn( "$fname-e1" );
1550                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1551                                 $text = $m[2];
1552                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1553                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1554                                 # the real problem is with the $e1 regex
1555                                 # See bug 1300.
1556                                 #
1557                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1558                                 # and no image is in sight. See bug 2095.
1559                                 #
1560                                 if( $text !== '' &&
1561                                         substr( $m[3], 0, 1 ) === ']' &&
1562                                         strpos($text, '[') !== false
1563                                 )
1564                                 {
1565                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1566                                         $m[3] = substr( $m[3], 1 );
1567                                 }
1568                                 # fix up urlencoded title texts
1569                                 if( strpos( $m[1], '%' ) !== false ) {
1570                                         # Should anchors '#' also be rejected?
1571                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1572                                 }
1573                                 $trail = $m[3];
1574                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1575                                 $might_be_img = true;
1576                                 $text = $m[2];
1577                                 if ( strpos( $m[1], '%' ) !== false ) {
1578                                        $m[1] = urldecode($m[1]);
1579                                 }
1580                                 $trail = "";
1581                         } else { # Invalid form; output directly
1582                                 $s .= $prefix . '[[' . $line ;
1583                                 wfProfileOut( "$fname-e1" );
1584                                 continue;
1585                         }
1586                         wfProfileOut( "$fname-e1" );
1587                         wfProfileIn( "$fname-misc" );
1588
1589                         # Don't allow internal links to pages containing
1590                         # PROTO: where PROTO is a valid URL protocol; these
1591                         # should be external links.
1592                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1593                                 $s .= $prefix . '[[' . $line ;
1594                                 continue;
1595                         }
1596
1597                         # Make subpage if necessary
1598                         if( $useSubpages ) {
1599                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1600                         } else {
1601                                 $link = $m[1];
1602                         }
1603
1604                         $noforce = (substr($m[1], 0, 1) != ':');
1605                         if (!$noforce) {
1606                                 # Strip off leading ':'
1607                                 $link = substr($link, 1);
1608                         }
1609
1610                         wfProfileOut( "$fname-misc" );
1611                         wfProfileIn( "$fname-title" );
1612                         $nt = Title::newFromText( $this->unstripNoWiki($link, $this->mStripState) );
1613                         if( !$nt ) {
1614                                 $s .= $prefix . '[[' . $line;
1615                                 wfProfileOut( "$fname-title" );
1616                                 continue;
1617                         }
1618
1619                         $ns = $nt->getNamespace();
1620                         $iw = $nt->getInterWiki();
1621                         wfProfileOut( "$fname-title" );
1622
1623                         if ($might_be_img) { # if this is actually an invalid link
1624                                 wfProfileIn( "$fname-might_be_img" );
1625                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1626                                         $found = false;
1627                                         while (isset ($a[$k+1]) ) {
1628                                                 #look at the next 'line' to see if we can close it there
1629                                                 $spliced = array_splice( $a, $k + 1, 1 );
1630                                                 $next_line = array_shift( $spliced );
1631                                                 $m = explode( ']]', $next_line, 3 );
1632                                                 if ( count( $m ) == 3 ) {
1633                                                         # the first ]] closes the inner link, the second the image
1634                                                         $found = true;
1635                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1636                                                         $trail = $m[2];
1637                                                         break;
1638                                                 } elseif ( count( $m ) == 2 ) {
1639                                                         #if there's exactly one ]] that's fine, we'll keep looking
1640                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1641                                                 } else {
1642                                                         #if $next_line is invalid too, we need look no further
1643                                                         $text .= '[[' . $next_line;
1644                                                         break;
1645                                                 }
1646                                         }
1647                                         if ( !$found ) {
1648                                                 # we couldn't find the end of this imageLink, so output it raw
1649                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1650                                                 $text = $this->replaceInternalLinks($text);
1651                                                 $s .= "{$prefix}[[$link|$text";
1652                                                 # note: no $trail, because without an end, there *is* no trail
1653                                                 wfProfileOut( "$fname-might_be_img" );
1654                                                 continue;
1655                                         }
1656                                 } else { #it's not an image, so output it raw
1657                                         $s .= "{$prefix}[[$link|$text";
1658                                         # note: no $trail, because without an end, there *is* no trail
1659                                         wfProfileOut( "$fname-might_be_img" );
1660                                         continue;
1661                                 }
1662                                 wfProfileOut( "$fname-might_be_img" );
1663                         }
1664
1665                         $wasblank = ( '' == $text );
1666                         if( $wasblank ) $text = $link;
1667
1668                         # Link not escaped by : , create the various objects
1669                         if( $noforce ) {
1670
1671                                 # Interwikis
1672                                 wfProfileIn( "$fname-interwiki" );
1673                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1674                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1675                                         $s = rtrim($s . "\n");
1676                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1677                                         wfProfileOut( "$fname-interwiki" );
1678                                         continue;
1679                                 }
1680                                 wfProfileOut( "$fname-interwiki" );
1681
1682                                 if ( $ns == NS_IMAGE ) {
1683                                         wfProfileIn( "$fname-image" );
1684                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1685                                                 # recursively parse links inside the image caption
1686                                                 # actually, this will parse them in any other parameters, too,
1687                                                 # but it might be hard to fix that, and it doesn't matter ATM
1688                                                 $text = $this->replaceExternalLinks($text);
1689                                                 $text = $this->replaceInternalLinks($text);
1690
1691                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1692                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1693                                                 $this->mOutput->addImage( $nt->getDBkey() );
1694
1695                                                 wfProfileOut( "$fname-image" );
1696                                                 continue;
1697                                         } else {
1698                                                 # We still need to record the image's presence on the page
1699                                                 $this->mOutput->addImage( $nt->getDBkey() );
1700                                         }
1701                                         wfProfileOut( "$fname-image" );
1702
1703                                 }
1704
1705                                 if ( $ns == NS_CATEGORY ) {
1706                                         wfProfileIn( "$fname-category" );
1707                                         $s = rtrim($s . "\n"); # bug 87
1708
1709                                         if ( $wasblank ) {
1710                                                 if ( $this->mTitle->getNamespace() == NS_CATEGORY ) {
1711                                                         $sortkey = $this->mTitle->getText();
1712                                                 } else {
1713                                                         $sortkey = $this->mTitle->getPrefixedText();
1714                                                 }
1715                                         } else {
1716                                                 $sortkey = $text;
1717                                         }
1718                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1719                                         $sortkey = str_replace( "\n", '', $sortkey );
1720                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1721                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1722
1723                                         /**
1724                                          * Strip the whitespace Category links produce, see bug 87
1725                                          * @todo We might want to use trim($tmp, "\n") here.
1726                                          */
1727                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1728
1729                                         wfProfileOut( "$fname-category" );
1730                                         continue;
1731                                 }
1732                         }
1733
1734                         if( ( $nt->getPrefixedText() === $selflink ) &&
1735                             ( $nt->getFragment() === '' ) ) {
1736                                 # Self-links are handled specially; generally de-link and change to bold.
1737                                 $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1738                                 continue;
1739                         }
1740
1741                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1742                         if( $ns == NS_MEDIA ) {
1743                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1744                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1745                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1746                                 $this->mOutput->addImage( $nt->getDBkey() );
1747                                 continue;
1748                         } elseif( $ns == NS_SPECIAL ) {
1749                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1750                                 continue;
1751                         } elseif( $ns == NS_IMAGE ) {
1752                                 $img = new Image( $nt );
1753                                 if( $img->exists() ) {
1754                                         // Force a blue link if the file exists; may be a remote
1755                                         // upload on the shared repository, and we want to see its
1756                                         // auto-generated page.
1757                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1758                                         $this->mOutput->addLink( $nt );
1759                                         continue;
1760                                 }
1761                         }
1762                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1763                 }
1764                 wfProfileOut( $fname );
1765                 return $s;
1766         }
1767
1768         /**
1769          * Make a link placeholder. The text returned can be later resolved to a real link with
1770          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1771          * parsing of interwiki links, and secondly to allow all existence checks and
1772          * article length checks (for stub links) to be bundled into a single query.
1773          *
1774          */
1775         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1776                 wfProfileIn( __METHOD__ );
1777                 if ( ! is_object($nt) ) {
1778                         # Fail gracefully
1779                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1780                 } else {
1781                         # Separate the link trail from the rest of the link
1782                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1783
1784                         if ( $nt->isExternal() ) {
1785                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1786                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1787                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1788                         } else {
1789                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1790                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1791                                 $this->mLinkHolders['queries'][] = $query;
1792                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1793                                 $this->mLinkHolders['titles'][] = $nt;
1794
1795                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1796                         }
1797                 }
1798                 wfProfileOut( __METHOD__ );
1799                 return $retVal;
1800         }
1801
1802         /**
1803          * Render a forced-blue link inline; protect against double expansion of
1804          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1805          * Since this little disaster has to split off the trail text to avoid
1806          * breaking URLs in the following text without breaking trails on the
1807          * wiki links, it's been made into a horrible function.
1808          *
1809          * @param Title $nt
1810          * @param string $text
1811          * @param string $query
1812          * @param string $trail
1813          * @param string $prefix
1814          * @return string HTML-wikitext mix oh yuck
1815          */
1816         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1817                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1818                 $sk =& $this->mOptions->getSkin();
1819                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1820                 return $this->armorLinks( $link ) . $trail;
1821         }
1822
1823         /**
1824          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1825          * going to go through further parsing steps before inline URL expansion.
1826          *
1827          * In particular this is important when using action=render, which causes
1828          * full URLs to be included.
1829          *
1830          * Oh man I hate our multi-layer parser!
1831          *
1832          * @param string more-or-less HTML
1833          * @return string less-or-more HTML with NOPARSE bits
1834          */
1835         function armorLinks( $text ) {
1836                 return preg_replace( "/\b(" . wfUrlProtocols() . ')/',
1837                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1838         }
1839
1840         /**
1841          * Return true if subpage links should be expanded on this page.
1842          * @return bool
1843          */
1844         function areSubpagesAllowed() {
1845                 # Some namespaces don't allow subpages
1846                 global $wgNamespacesWithSubpages;
1847                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1848         }
1849
1850         /**
1851          * Handle link to subpage if necessary
1852          * @param string $target the source of the link
1853          * @param string &$text the link text, modified as necessary
1854          * @return string the full name of the link
1855          * @private
1856          */
1857         function maybeDoSubpageLink($target, &$text) {
1858                 # Valid link forms:
1859                 # Foobar -- normal
1860                 # :Foobar -- override special treatment of prefix (images, language links)
1861                 # /Foobar -- convert to CurrentPage/Foobar
1862                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1863                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1864                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1865
1866                 $fname = 'Parser::maybeDoSubpageLink';
1867                 wfProfileIn( $fname );
1868                 $ret = $target; # default return value is no change
1869
1870                 # bug 7425
1871                 $target = trim( $target );
1872
1873                 # Some namespaces don't allow subpages,
1874                 # so only perform processing if subpages are allowed
1875                 if( $this->areSubpagesAllowed() ) {
1876                         # Look at the first character
1877                         if( $target != '' && $target{0} == '/' ) {
1878                                 # / at end means we don't want the slash to be shown
1879                                 if( substr( $target, -1, 1 ) == '/' ) {
1880                                         $target = substr( $target, 1, -1 );
1881                                         $noslash = $target;
1882                                 } else {
1883                                         $noslash = substr( $target, 1 );
1884                                 }
1885
1886                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1887                                 if( '' === $text ) {
1888                                         $text = $target;
1889                                 } # this might be changed for ugliness reasons
1890                         } else {
1891                                 # check for .. subpage backlinks
1892                                 $dotdotcount = 0;
1893                                 $nodotdot = $target;
1894                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1895                                         ++$dotdotcount;
1896                                         $nodotdot = substr( $nodotdot, 3 );
1897                                 }
1898                                 if($dotdotcount > 0) {
1899                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1900                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1901                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1902                                                 # / at the end means don't show full path
1903                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1904                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1905                                                         if( '' === $text ) {
1906                                                                 $text = $nodotdot;
1907                                                         }
1908                                                 }
1909                                                 $nodotdot = trim( $nodotdot );
1910                                                 if( $nodotdot != '' ) {
1911                                                         $ret .= '/' . $nodotdot;
1912                                                 }
1913                                         }
1914                                 }
1915                         }
1916                 }
1917
1918                 wfProfileOut( $fname );
1919                 return $ret;
1920         }
1921
1922         /**#@+
1923          * Used by doBlockLevels()
1924          * @private
1925          */
1926         /* private */ function closeParagraph() {
1927                 $result = '';
1928                 if ( '' != $this->mLastSection ) {
1929                         $result = '</' . $this->mLastSection  . ">\n";
1930                 }
1931                 $this->mInPre = false;
1932                 $this->mLastSection = '';
1933                 return $result;
1934         }
1935         # getCommon() returns the length of the longest common substring
1936         # of both arguments, starting at the beginning of both.
1937         #
1938         /* private */ function getCommon( $st1, $st2 ) {
1939                 $fl = strlen( $st1 );
1940                 $shorter = strlen( $st2 );
1941                 if ( $fl < $shorter ) { $shorter = $fl; }
1942
1943                 for ( $i = 0; $i < $shorter; ++$i ) {
1944                         if ( $st1{$i} != $st2{$i} ) { break; }
1945                 }
1946                 return $i;
1947         }
1948         # These next three functions open, continue, and close the list
1949         # element appropriate to the prefix character passed into them.
1950         #
1951         /* private */ function openList( $char ) {
1952                 $result = $this->closeParagraph();
1953
1954                 if ( '*' == $char ) { $result .= '<ul><li>'; }
1955                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
1956                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
1957                 else if ( ';' == $char ) {
1958                         $result .= '<dl><dt>';
1959                         $this->mDTopen = true;
1960                 }
1961                 else { $result = '<!-- ERR 1 -->'; }
1962
1963                 return $result;
1964         }
1965
1966         /* private */ function nextItem( $char ) {
1967                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
1968                 else if ( ':' == $char || ';' == $char ) {
1969                         $close = '</dd>';
1970                         if ( $this->mDTopen ) { $close = '</dt>'; }
1971                         if ( ';' == $char ) {
1972                                 $this->mDTopen = true;
1973                                 return $close . '<dt>';
1974                         } else {
1975                                 $this->mDTopen = false;
1976                                 return $close . '<dd>';
1977                         }
1978                 }
1979                 return '<!-- ERR 2 -->';
1980         }
1981
1982         /* private */ function closeList( $char ) {
1983                 if ( '*' == $char ) { $text = '</li></ul>'; }
1984                 else if ( '#' == $char ) { $text = '</li></ol>'; }
1985                 else if ( ':' == $char ) {
1986                         if ( $this->mDTopen ) {
1987                                 $this->mDTopen = false;
1988                                 $text = '</dt></dl>';
1989                         } else {
1990                                 $text = '</dd></dl>';
1991                         }
1992                 }
1993                 else {  return '<!-- ERR 3 -->'; }
1994                 return $text."\n";
1995         }
1996         /**#@-*/
1997
1998         /**
1999          * Make lists from lines starting with ':', '*', '#', etc.
2000          *
2001          * @private
2002          * @return string the lists rendered as HTML
2003          */
2004         function doBlockLevels( $text, $linestart ) {
2005                 $fname = 'Parser::doBlockLevels';
2006                 wfProfileIn( $fname );
2007
2008                 # Parsing through the text line by line.  The main thing
2009                 # happening here is handling of block-level elements p, pre,
2010                 # and making lists from lines starting with * # : etc.
2011                 #
2012                 $textLines = explode( "\n", $text );
2013
2014                 $lastPrefix = $output = '';
2015                 $this->mDTopen = $inBlockElem = false;
2016                 $prefixLength = 0;
2017                 $paragraphStack = false;
2018
2019                 if ( !$linestart ) {
2020                         $output .= array_shift( $textLines );
2021                 }
2022                 foreach ( $textLines as $oLine ) {
2023                         $lastPrefixLength = strlen( $lastPrefix );
2024                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2025                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2026                         if ( !$this->mInPre ) {
2027                                 # Multiple prefixes may abut each other for nested lists.
2028                                 $prefixLength = strspn( $oLine, '*#:;' );
2029                                 $pref = substr( $oLine, 0, $prefixLength );
2030
2031                                 # eh?
2032                                 $pref2 = str_replace( ';', ':', $pref );
2033                                 $t = substr( $oLine, $prefixLength );
2034                                 $this->mInPre = !empty($preOpenMatch);
2035                         } else {
2036                                 # Don't interpret any other prefixes in preformatted text
2037                                 $prefixLength = 0;
2038                                 $pref = $pref2 = '';
2039                                 $t = $oLine;
2040                         }
2041
2042                         # List generation
2043                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2044                                 # Same as the last item, so no need to deal with nesting or opening stuff
2045                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2046                                 $paragraphStack = false;
2047
2048                                 if ( substr( $pref, -1 ) == ';') {
2049                                         # The one nasty exception: definition lists work like this:
2050                                         # ; title : definition text
2051                                         # So we check for : in the remainder text to split up the
2052                                         # title and definition, without b0rking links.
2053                                         $term = $t2 = '';
2054                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2055                                                 $t = $t2;
2056                                                 $output .= $term . $this->nextItem( ':' );
2057                                         }
2058                                 }
2059                         } elseif( $prefixLength || $lastPrefixLength ) {
2060                                 # Either open or close a level...
2061                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2062                                 $paragraphStack = false;
2063
2064                                 while( $commonPrefixLength < $lastPrefixLength ) {
2065                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2066                                         --$lastPrefixLength;
2067                                 }
2068                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2069                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2070                                 }
2071                                 while ( $prefixLength > $commonPrefixLength ) {
2072                                         $char = substr( $pref, $commonPrefixLength, 1 );
2073                                         $output .= $this->openList( $char );
2074
2075                                         if ( ';' == $char ) {
2076                                                 # FIXME: This is dupe of code above
2077                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2078                                                         $t = $t2;
2079                                                         $output .= $term . $this->nextItem( ':' );
2080                                                 }
2081                                         }
2082                                         ++$commonPrefixLength;
2083                                 }
2084                                 $lastPrefix = $pref2;
2085                         }
2086                         if( 0 == $prefixLength ) {
2087                                 wfProfileIn( "$fname-paragraph" );
2088                                 # No prefix (not in list)--go to paragraph mode
2089                                 // XXX: use a stack for nestable elements like span, table and div
2090                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/center|<\\/tr|<\\/td|<\\/th)/iS', $t );
2091                                 $closematch = preg_match(
2092                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2093                                         '<td|<th|<div|<\\/div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<center)/iS', $t );
2094                                 if ( $openmatch or $closematch ) {
2095                                         $paragraphStack = false;
2096                                         # TODO bug 5718: paragraph closed
2097                                         $output .= $this->closeParagraph();
2098                                         if ( $preOpenMatch and !$preCloseMatch ) {
2099                                                 $this->mInPre = true;
2100                                         }
2101                                         if ( $closematch ) {
2102                                                 $inBlockElem = false;
2103                                         } else {
2104                                                 $inBlockElem = true;
2105                                         }
2106                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2107                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2108                                                 // pre
2109                                                 if ($this->mLastSection != 'pre') {
2110                                                         $paragraphStack = false;
2111                                                         $output .= $this->closeParagraph().'<pre>';
2112                                                         $this->mLastSection = 'pre';
2113                                                 }
2114                                                 $t = substr( $t, 1 );
2115                                         } else {
2116                                                 // paragraph
2117                                                 if ( '' == trim($t) ) {
2118                                                         if ( $paragraphStack ) {
2119                                                                 $output .= $paragraphStack.'<br />';
2120                                                                 $paragraphStack = false;
2121                                                                 $this->mLastSection = 'p';
2122                                                         } else {
2123                                                                 if ($this->mLastSection != 'p' ) {
2124                                                                         $output .= $this->closeParagraph();
2125                                                                         $this->mLastSection = '';
2126                                                                         $paragraphStack = '<p>';
2127                                                                 } else {
2128                                                                         $paragraphStack = '</p><p>';
2129                                                                 }
2130                                                         }
2131                                                 } else {
2132                                                         if ( $paragraphStack ) {
2133                                                                 $output .= $paragraphStack;
2134                                                                 $paragraphStack = false;
2135                                                                 $this->mLastSection = 'p';
2136                                                         } else if ($this->mLastSection != 'p') {
2137                                                                 $output .= $this->closeParagraph().'<p>';
2138                                                                 $this->mLastSection = 'p';
2139                                                         }
2140                                                 }
2141                                         }
2142                                 }
2143                                 wfProfileOut( "$fname-paragraph" );
2144                         }
2145                         // somewhere above we forget to get out of pre block (bug 785)
2146                         if($preCloseMatch && $this->mInPre) {
2147                                 $this->mInPre = false;
2148                         }
2149                         if ($paragraphStack === false) {
2150                                 $output .= $t."\n";
2151                         }
2152                 }
2153                 while ( $prefixLength ) {
2154                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2155                         --$prefixLength;
2156                 }
2157                 if ( '' != $this->mLastSection ) {
2158                         $output .= '</' . $this->mLastSection . '>';
2159                         $this->mLastSection = '';
2160                 }
2161
2162                 wfProfileOut( $fname );
2163                 return $output;
2164         }
2165
2166         /**
2167          * Split up a string on ':', ignoring any occurences inside tags
2168          * to prevent illegal overlapping.
2169          * @param string $str the string to split
2170          * @param string &$before set to everything before the ':'
2171          * @param string &$after set to everything after the ':'
2172          * return string the position of the ':', or false if none found
2173          */
2174         function findColonNoLinks($str, &$before, &$after) {
2175                 $fname = 'Parser::findColonNoLinks';
2176                 wfProfileIn( $fname );
2177
2178                 $pos = strpos( $str, ':' );
2179                 if( $pos === false ) {
2180                         // Nothing to find!
2181                         wfProfileOut( $fname );
2182                         return false;
2183                 }
2184
2185                 $lt = strpos( $str, '<' );
2186                 if( $lt === false || $lt > $pos ) {
2187                         // Easy; no tag nesting to worry about
2188                         $before = substr( $str, 0, $pos );
2189                         $after = substr( $str, $pos+1 );
2190                         wfProfileOut( $fname );
2191                         return $pos;
2192                 }
2193
2194                 // Ugly state machine to walk through avoiding tags.
2195                 $state = MW_COLON_STATE_TEXT;
2196                 $stack = 0;
2197                 $len = strlen( $str );
2198                 for( $i = 0; $i < $len; $i++ ) {
2199                         $c = $str{$i};
2200
2201                         switch( $state ) {
2202                         // (Using the number is a performance hack for common cases)
2203                         case 0: // MW_COLON_STATE_TEXT:
2204                                 switch( $c ) {
2205                                 case "<":
2206                                         // Could be either a <start> tag or an </end> tag
2207                                         $state = MW_COLON_STATE_TAGSTART;
2208                                         break;
2209                                 case ":":
2210                                         if( $stack == 0 ) {
2211                                                 // We found it!
2212                                                 $before = substr( $str, 0, $i );
2213                                                 $after = substr( $str, $i + 1 );
2214                                                 wfProfileOut( $fname );
2215                                                 return $i;
2216                                         }
2217                                         // Embedded in a tag; don't break it.
2218                                         break;
2219                                 default:
2220                                         // Skip ahead looking for something interesting
2221                                         $colon = strpos( $str, ':', $i );
2222                                         if( $colon === false ) {
2223                                                 // Nothing else interesting
2224                                                 wfProfileOut( $fname );
2225                                                 return false;
2226                                         }
2227                                         $lt = strpos( $str, '<', $i );
2228                                         if( $stack === 0 ) {
2229                                                 if( $lt === false || $colon < $lt ) {
2230                                                         // We found it!
2231                                                         $before = substr( $str, 0, $colon );
2232                                                         $after = substr( $str, $colon + 1 );
2233                                                         wfProfileOut( $fname );
2234                                                         return $i;
2235                                                 }
2236                                         }
2237                                         if( $lt === false ) {
2238                                                 // Nothing else interesting to find; abort!
2239                                                 // We're nested, but there's no close tags left. Abort!
2240                                                 break 2;
2241                                         }
2242                                         // Skip ahead to next tag start
2243                                         $i = $lt;
2244                                         $state = MW_COLON_STATE_TAGSTART;
2245                                 }
2246                                 break;
2247                         case 1: // MW_COLON_STATE_TAG:
2248                                 // In a <tag>
2249                                 switch( $c ) {
2250                                 case ">":
2251                                         $stack++;
2252                                         $state = MW_COLON_STATE_TEXT;
2253                                         break;
2254                                 case "/":
2255                                         // Slash may be followed by >?
2256                                         $state = MW_COLON_STATE_TAGSLASH;
2257                                         break;
2258                                 default:
2259                                         // ignore
2260                                 }
2261                                 break;
2262                         case 2: // MW_COLON_STATE_TAGSTART:
2263                                 switch( $c ) {
2264                                 case "/":
2265                                         $state = MW_COLON_STATE_CLOSETAG;
2266                                         break;
2267                                 case "!":
2268                                         $state = MW_COLON_STATE_COMMENT;
2269                                         break;
2270                                 case ">":
2271                                         // Illegal early close? This shouldn't happen D:
2272                                         $state = MW_COLON_STATE_TEXT;
2273                                         break;
2274                                 default:
2275                                         $state = MW_COLON_STATE_TAG;
2276                                 }
2277                                 break;
2278                         case 3: // MW_COLON_STATE_CLOSETAG:
2279                                 // In a </tag>
2280                                 if( $c == ">" ) {
2281                                         $stack--;
2282                                         if( $stack < 0 ) {
2283                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2284                                                 wfProfileOut( $fname );
2285                                                 return false;
2286                                         }
2287                                         $state = MW_COLON_STATE_TEXT;
2288                                 }
2289                                 break;
2290                         case MW_COLON_STATE_TAGSLASH:
2291                                 if( $c == ">" ) {
2292                                         // Yes, a self-closed tag <blah/>
2293                                         $state = MW_COLON_STATE_TEXT;
2294                                 } else {
2295                                         // Probably we're jumping the gun, and this is an attribute
2296                                         $state = MW_COLON_STATE_TAG;
2297                                 }
2298                                 break;
2299                         case 5: // MW_COLON_STATE_COMMENT:
2300                                 if( $c == "-" ) {
2301                                         $state = MW_COLON_STATE_COMMENTDASH;
2302                                 }
2303                                 break;
2304                         case MW_COLON_STATE_COMMENTDASH:
2305                                 if( $c == "-" ) {
2306                                         $state = MW_COLON_STATE_COMMENTDASHDASH;
2307                                 } else {
2308                                         $state = MW_COLON_STATE_COMMENT;
2309                                 }
2310                                 break;
2311                         case MW_COLON_STATE_COMMENTDASHDASH:
2312                                 if( $c == ">" ) {
2313                                         $state = MW_COLON_STATE_TEXT;
2314                                 } else {
2315                                         $state = MW_COLON_STATE_COMMENT;
2316                                 }
2317                                 break;
2318                         default:
2319                                 throw new MWException( "State machine error in $fname" );
2320                         }
2321                 }
2322                 if( $stack > 0 ) {
2323                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2324                         return false;
2325                 }
2326                 wfProfileOut( $fname );
2327                 return false;
2328         }
2329
2330         /**
2331          * Return value of a magic variable (like PAGENAME)
2332          *
2333          * @private
2334          */
2335         function getVariableValue( $index ) {
2336                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2337
2338                 /**
2339                  * Some of these require message or data lookups and can be
2340                  * expensive to check many times.
2341                  */
2342                 static $varCache = array();
2343                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) ) {
2344                         if ( isset( $varCache[$index] ) ) {
2345                                 return $varCache[$index];
2346                         }
2347                 }
2348
2349                 $ts = time();
2350                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2351
2352                 # Use the time zone
2353                 global $wgLocaltimezone;
2354                 if ( isset( $wgLocaltimezone ) ) {
2355                         $oldtz = getenv( 'TZ' );
2356                         putenv( 'TZ='.$wgLocaltimezone );
2357                 }
2358                 $localTimestamp = date( 'YmdHis', $ts );
2359                 $localMonth = date( 'm', $ts );
2360                 $localMonthName = date( 'n', $ts );
2361                 $localDay = date( 'j', $ts );
2362                 $localDay2 = date( 'd', $ts );
2363                 $localDayOfWeek = date( 'w', $ts );
2364                 $localWeek = date( 'W', $ts );
2365                 $localYear = date( 'Y', $ts );
2366                 $localHour = date( 'H', $ts );
2367                 if ( isset( $wgLocaltimezone ) ) {
2368                         putenv( 'TZ='.$oldtz );
2369                 }
2370
2371                 switch ( $index ) {
2372                         case 'currentmonth':
2373                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
2374                         case 'currentmonthname':
2375                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2376                         case 'currentmonthnamegen':
2377                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2378                         case 'currentmonthabbrev':
2379                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2380                         case 'currentday':
2381                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2382                         case 'currentday2':
2383                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2384                         case 'localmonth':
2385                                 return $varCache[$index] = $wgContLang->formatNum( $localMonth );
2386                         case 'localmonthname':
2387                                 return $varCache[$index] = $wgContLang->getMonthName( $localMonthName );
2388                         case 'localmonthnamegen':
2389                                 return $varCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2390                         case 'localmonthabbrev':
2391                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2392                         case 'localday':
2393                                 return $varCache[$index] = $wgContLang->formatNum( $localDay );
2394                         case 'localday2':
2395                                 return $varCache[$index] = $wgContLang->formatNum( $localDay2 );
2396                         case 'pagename':
2397                                 return $this->mTitle->getText();
2398                         case 'pagenamee':
2399                                 return $this->mTitle->getPartialURL();
2400                         case 'fullpagename':
2401                                 return $this->mTitle->getPrefixedText();
2402                         case 'fullpagenamee':
2403                                 return $this->mTitle->getPrefixedURL();
2404                         case 'subpagename':
2405                                 return $this->mTitle->getSubpageText();
2406                         case 'subpagenamee':
2407                                 return $this->mTitle->getSubpageUrlForm();
2408                         case 'basepagename':
2409                                 return $this->mTitle->getBaseText();
2410                         case 'basepagenamee':
2411                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2412                         case 'talkpagename':
2413                                 if( $this->mTitle->canTalk() ) {
2414                                         $talkPage = $this->mTitle->getTalkPage();
2415                                         return $talkPage->getPrefixedText();
2416                                 } else {
2417                                         return '';
2418                                 }
2419                         case 'talkpagenamee':
2420                                 if( $this->mTitle->canTalk() ) {
2421                                         $talkPage = $this->mTitle->getTalkPage();
2422                                         return $talkPage->getPrefixedUrl();
2423                                 } else {
2424                                         return '';
2425                                 }
2426                         case 'subjectpagename':
2427                                 $subjPage = $this->mTitle->getSubjectPage();
2428                                 return $subjPage->getPrefixedText();
2429                         case 'subjectpagenamee':
2430                                 $subjPage = $this->mTitle->getSubjectPage();
2431                                 return $subjPage->getPrefixedUrl();
2432                         case 'revisionid':
2433                                 return $this->mRevisionId;
2434                         case 'revisionday':
2435                                 return intval( substr( wfRevisionTimestamp( $this->mRevisionId ), 6, 2 ) );
2436                         case 'revisionday2':
2437                                 return substr( wfRevisionTimestamp( $this->mRevisionId ), 6, 2 );
2438                         case 'revisionmonth':
2439                                 return intval( substr( wfRevisionTimestamp( $this->mRevisionId ), 4, 2 ) );
2440                         case 'revisionyear':
2441                                 return substr( wfRevisionTimestamp( $this->mRevisionId ), 0, 4 );
2442                         case 'revisiontimestamp':
2443                                 return wfRevisionTimestamp( $this->mRevisionId );
2444                         case 'namespace':
2445                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2446                         case 'namespacee':
2447                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2448                         case 'talkspace':
2449                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2450                         case 'talkspacee':
2451                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2452                         case 'subjectspace':
2453                                 return $this->mTitle->getSubjectNsText();
2454                         case 'subjectspacee':
2455                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2456                         case 'currentdayname':
2457                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2458                         case 'currentyear':
2459                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2460                         case 'currenttime':
2461                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2462                         case 'currenthour':
2463                                 return $varCache[$index] = $wgContLang->formatNum( date( 'H', $ts ), true );
2464                         case 'currentweek':
2465                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2466                                 // int to remove the padding
2467                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2468                         case 'currentdow':
2469                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2470                         case 'localdayname':
2471                                 return $varCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2472                         case 'localyear':
2473                                 return $varCache[$index] = $wgContLang->formatNum( $localYear, true );
2474                         case 'localtime':
2475                                 return $varCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2476                         case 'localhour':
2477                                 return $varCache[$index] = $wgContLang->formatNum( $localHour, true );
2478                         case 'localweek':
2479                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2480                                 // int to remove the padding
2481                                 return $varCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2482                         case 'localdow':
2483                                 return $varCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2484                         case 'numberofarticles':
2485                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfArticles() );
2486                         case 'numberoffiles':
2487                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfFiles() );
2488                         case 'numberofusers':
2489                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfUsers() );
2490                         case 'numberofpages':
2491                                 return $varCache[$index] = $wgContLang->formatNum( wfNumberOfPages() );
2492                         case 'numberofadmins':
2493                                 return $varCache[$index]  = $wgContLang->formatNum( wfNumberOfAdmins() );
2494                         case 'currenttimestamp':
2495                                 return $varCache[$index] = wfTimestampNow();
2496                         case 'localtimestamp':
2497                                 return $varCache[$index] = $localTimestamp;
2498                         case 'currentversion':
2499                                 return $varCache[$index] = SpecialVersion::getVersion();
2500                         case 'sitename':
2501                                 return $wgSitename;
2502                         case 'server':
2503                                 return $wgServer;
2504                         case 'servername':
2505                                 return $wgServerName;
2506                         case 'scriptpath':
2507                                 return $wgScriptPath;
2508                         case 'directionmark':
2509                                 return $wgContLang->getDirMark();
2510                         case 'contentlanguage':
2511                                 global $wgContLanguageCode;
2512                                 return $wgContLanguageCode;
2513                         default:
2514                                 $ret = null;
2515                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2516                                         return $ret;
2517                                 else
2518                                         return null;
2519                 }
2520         }
2521
2522         /**
2523          * initialise the magic variables (like CURRENTMONTHNAME)
2524          *
2525          * @private
2526          */
2527         function initialiseVariables() {
2528                 $fname = 'Parser::initialiseVariables';
2529                 wfProfileIn( $fname );
2530                 $variableIDs = MagicWord::getVariableIDs();
2531
2532                 $this->mVariables = array();
2533                 foreach ( $variableIDs as $id ) {
2534                         $mw =& MagicWord::get( $id );
2535                         $mw->addToArray( $this->mVariables, $id );
2536                 }
2537                 wfProfileOut( $fname );
2538         }
2539
2540         /**
2541          * parse any parentheses in format ((title|part|part))
2542          * and call callbacks to get a replacement text for any found piece
2543          *
2544          * @param string $text The text to parse
2545          * @param array $callbacks rules in form:
2546          *     '{' => array(                            # opening parentheses
2547          *                                      'end' => '}',   # closing parentheses
2548          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2549          *                                                                3 => callback         # replacement callback to call if {{{..}}} is found
2550          *                                                                )
2551          *                                      )
2552          *                                      'min' => 2,     # Minimum parenthesis count in cb
2553          *                                      'max' => 3,     # Maximum parenthesis count in cb
2554          * @private
2555          */
2556         function replace_callback ($text, $callbacks) {
2557                 wfProfileIn( __METHOD__ );
2558                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2559                 $lastOpeningBrace = -1;                 # last not closed parentheses
2560
2561                 $validOpeningBraces = implode( '', array_keys( $callbacks ) );
2562
2563                 $i = 0;
2564                 while ( $i < strlen( $text ) ) {
2565                         # Find next opening brace, closing brace or pipe
2566                         if ( $lastOpeningBrace == -1 ) {
2567                                 $currentClosing = '';
2568                                 $search = $validOpeningBraces;
2569                         } else {
2570                                 $currentClosing = $openingBraceStack[$lastOpeningBrace]['braceEnd'];
2571                                 $search = $validOpeningBraces . '|' . $currentClosing;
2572                         }
2573                         $rule = null;
2574                         $i += strcspn( $text, $search, $i );
2575                         if ( $i < strlen( $text ) ) {
2576                                 if ( $text[$i] == '|' ) {
2577                                         $found = 'pipe';
2578                                 } elseif ( $text[$i] == $currentClosing ) {
2579                                         $found = 'close';
2580                                 } elseif ( isset( $callbacks[$text[$i]] ) ) {
2581                                         $found = 'open';
2582                                         $rule = $callbacks[$text[$i]];
2583                                 } else {
2584                                         # Some versions of PHP have a strcspn which stops on null characters
2585                                         # Ignore and continue
2586                                         ++$i;
2587                                         continue;
2588                                 }
2589                         } else {
2590                                 # All done
2591                                 break;
2592                         }
2593
2594                         if ( $found == 'open' ) {
2595                                 # found opening brace, let's add it to parentheses stack
2596                                 $piece = array('brace' => $text[$i],
2597                                                            'braceEnd' => $rule['end'],
2598                                                            'title' => '',
2599                                                            'parts' => null);
2600
2601                                 # count opening brace characters
2602                                 $piece['count'] = strspn( $text, $piece['brace'], $i );
2603                                 $piece['startAt'] = $piece['partStart'] = $i + $piece['count'];
2604                                 $i += $piece['count'];
2605
2606                                 # we need to add to stack only if opening brace count is enough for one of the rules
2607                                 if ( $piece['count'] >= $rule['min'] ) {
2608                                         $lastOpeningBrace ++;
2609                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2610                                 }
2611                         } elseif ( $found == 'close' ) {
2612                                 # lets check if it is enough characters for closing brace
2613                                 $maxCount = $openingBraceStack[$lastOpeningBrace]['count'];
2614                                 $count = strspn( $text, $text[$i], $i, $maxCount );
2615
2616                                 # check for maximum matching characters (if there are 5 closing
2617                                 # characters, we will probably need only 3 - depending on the rules)
2618                                 $matchingCount = 0;
2619                                 $matchingCallback = null;
2620                                 $cbType = $callbacks[$openingBraceStack[$lastOpeningBrace]['brace']];
2621                                 if ( $count > $cbType['max'] ) {
2622                                         # The specified maximum exists in the callback array, unless the caller
2623                                         # has made an error
2624                                         $matchingCount = $cbType['max'];
2625                                 } else {
2626                                         # Count is less than the maximum
2627                                         # Skip any gaps in the callback array to find the true largest match
2628                                         # Need to use array_key_exists not isset because the callback can be null
2629                                         $matchingCount = $count;
2630                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $cbType['cb'] ) ) {
2631                                                 --$matchingCount;
2632                                         }
2633                                 }
2634
2635                                 if ($matchingCount <= 0) {
2636                                         $i += $count;
2637                                         continue;
2638                                 }
2639                                 $matchingCallback = $cbType['cb'][$matchingCount];
2640
2641                                 # let's set a title or last part (if '|' was found)
2642                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2643                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2644                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2645                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2646                                 } else {
2647                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2648                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2649                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2650                                 }
2651
2652                                 $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2653                                 $pieceEnd = $i + $matchingCount;
2654
2655                                 if( is_callable( $matchingCallback ) ) {
2656                                         $cbArgs = array (
2657                                                                          'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2658                                                                          'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2659                                                                          'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2660                                                                          'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == "\n")),
2661                                                                          );
2662                                         # finally we can call a user callback and replace piece of text
2663                                         $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2664                                         $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2665                                         $i = $pieceStart + strlen($replaceWith);
2666                                 } else {
2667                                         # null value for callback means that parentheses should be parsed, but not replaced
2668                                         $i += $matchingCount;
2669                                 }
2670
2671                                 # reset last opening parentheses, but keep it in case there are unused characters
2672                                 $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2673                                                            'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2674                                                            'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2675                                                            'title' => '',
2676                                                            'parts' => null,
2677                                                            'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2678                                 $openingBraceStack[$lastOpeningBrace--] = null;
2679
2680                                 if ($matchingCount < $piece['count']) {
2681                                         $piece['count'] -= $matchingCount;
2682                                         $piece['startAt'] -= $matchingCount;
2683                                         $piece['partStart'] = $piece['startAt'];
2684                                         # do we still qualify for any callback with remaining count?
2685                                         $currentCbList = $callbacks[$piece['brace']]['cb'];
2686                                         while ( $piece['count'] ) {
2687                                                 if ( array_key_exists( $piece['count'], $currentCbList ) ) {
2688                                                         $lastOpeningBrace++;
2689                                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2690                                                         break;
2691                                                 }
2692                                                 --$piece['count'];
2693                                         }
2694                                 }
2695                         } elseif ( $found == 'pipe' ) {
2696                                 # lets set a title if it is a first separator, or next part otherwise
2697                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2698                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2699                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2700                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2701                                         $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2702                                 } else {
2703                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2704                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2705                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2706                                 }
2707                                 $openingBraceStack[$lastOpeningBrace]['partStart'] = ++$i;
2708                         }
2709                 }
2710
2711                 wfProfileOut( __METHOD__ );
2712                 return $text;
2713         }
2714
2715         /**
2716          * Replace magic variables, templates, and template arguments
2717          * with the appropriate text. Templates are substituted recursively,
2718          * taking care to avoid infinite loops.
2719          *
2720          * Note that the substitution depends on value of $mOutputType:
2721          *  OT_WIKI: only {{subst:}} templates
2722          *  OT_MSG: only magic variables
2723          *  OT_HTML: all templates and magic variables
2724          *
2725          * @param string $tex The text to transform
2726          * @param array $args Key-value pairs representing template parameters to substitute
2727          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2728          * @private
2729          */
2730         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2731                 # Prevent too big inclusions
2732                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2733                         return $text;
2734                 }
2735
2736                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2737                 wfProfileIn( $fname );
2738
2739                 # This function is called recursively. To keep track of arguments we need a stack:
2740                 array_push( $this->mArgStack, $args );
2741
2742                 $braceCallbacks = array();
2743                 if ( !$argsOnly ) {
2744                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2745                 }
2746                 if ( $this->mOutputType != OT_MSG ) {
2747                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2748                 }
2749                 if ( $braceCallbacks ) {
2750                         $callbacks = array(
2751                                 '{' => array(
2752                                         'end' => '}',
2753                                         'cb' => $braceCallbacks,
2754                                         'min' => $argsOnly ? 3 : 2,
2755                                         'max' => isset( $braceCallbacks[3] ) ? 3 : 2,
2756                                 ),
2757                                 '[' => array(
2758                                         'end' => ']',
2759                                         'cb' => array(2=>null),
2760                                         'min' => 2,
2761                                         'max' => 2,
2762                                 )
2763                         );
2764                         $text = $this->replace_callback ($text, $callbacks);
2765
2766                         array_pop( $this->mArgStack );
2767                 }
2768                 wfProfileOut( $fname );
2769                 return $text;
2770         }
2771
2772         /**
2773          * Replace magic variables
2774          * @private
2775          */
2776         function variableSubstitution( $matches ) {
2777                 global $wgContLang;
2778                 $fname = 'Parser::variableSubstitution';
2779                 $varname = $wgContLang->lc($matches[1]);
2780                 wfProfileIn( $fname );
2781                 $skip = false;
2782                 if ( $this->mOutputType == OT_WIKI ) {
2783                         # Do only magic variables prefixed by SUBST
2784                         $mwSubst =& MagicWord::get( 'subst' );
2785                         if (!$mwSubst->matchStartAndRemove( $varname ))
2786                                 $skip = true;
2787                         # Note that if we don't substitute the variable below,
2788                         # we don't remove the {{subst:}} magic word, in case
2789                         # it is a template rather than a magic variable.
2790                 }
2791                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2792                         $id = $this->mVariables[$varname];
2793                         # Now check if we did really match, case sensitive or not
2794                         $mw =& MagicWord::get( $id );
2795                         if ($mw->match($matches[1])) {
2796                                 $text = $this->getVariableValue( $id );
2797                                 $this->mOutput->mContainsOldMagic = true;
2798                         } else {
2799                                 $text = $matches[0];
2800                         }
2801                 } else {
2802                         $text = $matches[0];
2803                 }
2804                 wfProfileOut( $fname );
2805                 return $text;
2806         }
2807
2808         /**
2809          * Return the text of a template, after recursively
2810          * replacing any variables or templates within the template.
2811          *
2812          * @param array $piece The parts of the template
2813          *  $piece['text']: matched text
2814          *  $piece['title']: the title, i.e. the part before the |
2815          *  $piece['parts']: the parameter array
2816          * @return string the text of the template
2817          * @private
2818          */
2819         function braceSubstitution( $piece ) {
2820                 global $wgContLang, $wgLang, $wgAllowDisplayTitle;
2821                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2822                 wfProfileIn( $fname );
2823                 wfProfileIn( __METHOD__.'-setup' );
2824
2825                 # Flags
2826                 $found = false;             # $text has been filled
2827                 $nowiki = false;            # wiki markup in $text should be escaped
2828                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2829                 $noargs = false;            # Don't replace triple-brace arguments in $text
2830                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2831                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2832                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2833
2834                 # Title object, where $text came from
2835                 $title = NULL;
2836
2837                 $linestart = '';
2838
2839
2840                 # $part1 is the bit before the first |, and must contain only title characters
2841                 # $args is a list of arguments, starting from index 0, not including $part1
2842
2843                 $titleText = $part1 = $piece['title'];
2844                 # If the third subpattern matched anything, it will start with |
2845
2846                 if (null == $piece['parts']) {
2847                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2848                         if ($replaceWith != $piece['text']) {
2849                                 $text = $replaceWith;
2850                                 $found = true;
2851                                 $noparse = true;
2852                                 $noargs = true;
2853                         }
2854                 }
2855
2856                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2857                 wfProfileOut( __METHOD__.'-setup' );
2858
2859                 # SUBST
2860                 wfProfileIn( __METHOD__.'-modifiers' );
2861                 if ( !$found ) {
2862                         $mwSubst =& MagicWord::get( 'subst' );
2863                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2864                                 # One of two possibilities is true:
2865                                 # 1) Found SUBST but not in the PST phase
2866                                 # 2) Didn't find SUBST and in the PST phase
2867                                 # In either case, return without further processing
2868                                 $text = $piece['text'];
2869                                 $found = true;
2870                                 $noparse = true;
2871                                 $noargs = true;
2872                         }
2873                 }
2874
2875                 # MSG, MSGNW and RAW
2876                 if ( !$found ) {
2877                         # Check for MSGNW:
2878                         $mwMsgnw =& MagicWord::get( 'msgnw' );
2879                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2880                                 $nowiki = true;
2881                         } else {
2882                                 # Remove obsolete MSG:
2883                                 $mwMsg =& MagicWord::get( 'msg' );
2884                                 $mwMsg->matchStartAndRemove( $part1 );
2885                         }
2886
2887                         # Check for RAW:
2888                         $mwRaw =& MagicWord::get( 'raw' );
2889                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2890                                 $forceRawInterwiki = true;
2891                         }
2892                 }
2893                 wfProfileOut( __METHOD__.'-modifiers' );
2894
2895                 # Parser functions
2896                 if ( !$found ) {
2897                         wfProfileIn( __METHOD__ . '-pfunc' );
2898
2899                         $colonPos = strpos( $part1, ':' );
2900                         if ( $colonPos !== false ) {
2901                                 # Case sensitive functions
2902                                 $function = substr( $part1, 0, $colonPos );
2903                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2904                                         $function = $this->mFunctionSynonyms[1][$function];
2905                                 } else {
2906                                         # Case insensitive functions
2907                                         $function = strtolower( $function );
2908                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2909                                                 $function = $this->mFunctionSynonyms[0][$function];
2910                                         } else {
2911                                                 $function = false;
2912                                         }
2913                                 }
2914                                 if ( $function ) {
2915                                         $funcArgs = array_map( 'trim', $args );
2916                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
2917                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
2918                                         $found = true;
2919
2920                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
2921                                         //$noargs = true;
2922                                         //$noparse = true;
2923
2924                                         if ( is_array( $result ) ) {
2925                                                 if ( isset( $result[0] ) ) {
2926                                                         $text = $linestart . $result[0];
2927                                                         unset( $result[0] );
2928                                                 }
2929
2930                                                 // Extract flags into the local scope
2931                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
2932                                                 extract( $result );
2933                                         } else {
2934                                                 $text = $linestart . $result;
2935                                         }
2936                                 }
2937                         }
2938                         wfProfileOut( __METHOD__ . '-pfunc' );
2939                 }
2940
2941                 # Template table test
2942
2943                 # Did we encounter this template already? If yes, it is in the cache
2944                 # and we need to check for loops.
2945                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
2946                         $found = true;
2947
2948                         # Infinite loop test
2949                         if ( isset( $this->mTemplatePath[$part1] ) ) {
2950                                 $noparse = true;
2951                                 $noargs = true;
2952                                 $found = true;
2953                                 $text = $linestart .
2954                                         "[[$part1]]<!-- WARNING: template loop detected -->";
2955                                 wfDebug( __METHOD__.": template loop broken at '$part1'\n" );
2956                         } else {
2957                                 # set $text to cached message.
2958                                 $text = $linestart . $this->mTemplates[$piece['title']];
2959                         }
2960                 }
2961
2962                 # Load from database
2963                 $lastPathLevel = $this->mTemplatePath;
2964                 if ( !$found ) {
2965                         wfProfileIn( __METHOD__ . '-loadtpl' );
2966                         $ns = NS_TEMPLATE;
2967                         # declaring $subpage directly in the function call
2968                         # does not work correctly with references and breaks
2969                         # {{/subpage}}-style inclusions
2970                         $subpage = '';
2971                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
2972                         if ($subpage !== '') {
2973                                 $ns = $this->mTitle->getNamespace();
2974                         }
2975                         $title = Title::newFromText( $part1, $ns );
2976
2977
2978                         if ( !is_null( $title ) ) {
2979                                 $titleText = $title->getPrefixedText();
2980                                 $checkVariantLink = sizeof($wgContLang->getVariants())>1;
2981                                 # Check for language variants if the template is not found
2982                                 if($checkVariantLink && $title->getArticleID() == 0){
2983                                         $wgContLang->findVariantLink($part1, $title);
2984                                 }
2985
2986                                 if ( !$title->isExternal() ) {
2987                                         if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
2988                                                 $text = SpecialPage::capturePath( $title );
2989                                                 if ( is_string( $text ) ) {
2990                                                         $found = true;
2991                                                         $noparse = true;
2992                                                         $noargs = true;
2993                                                         $isHTML = true;
2994                                                         $this->disableCache();
2995                                                 }
2996                                         } else {
2997                                                 $articleContent = $this->fetchTemplate( $title );
2998                                                 if ( $articleContent !== false ) {
2999                                                         $found = true;
3000                                                         $text = $articleContent;
3001                                                         $replaceHeadings = true;
3002                                                 }
3003                                         }
3004
3005                                         # If the title is valid but undisplayable, make a link to it
3006                                         if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3007                                                 $text = "[[:$titleText]]";
3008                                                 $found = true;
3009                                         }
3010                                 } elseif ( $title->isTrans() ) {
3011                                         // Interwiki transclusion
3012                                         if ( $this->ot['html'] && !$forceRawInterwiki ) {
3013                                                 $text = $this->interwikiTransclude( $title, 'render' );
3014                                                 $isHTML = true;
3015                                                 $noparse = true;
3016                                         } else {
3017                                                 $text = $this->interwikiTransclude( $title, 'raw' );
3018                                                 $replaceHeadings = true;
3019                                         }
3020                                         $found = true;
3021                                 }
3022
3023                                 # Template cache array insertion
3024                                 # Use the original $piece['title'] not the mangled $part1, so that
3025                                 # modifiers such as RAW: produce separate cache entries
3026                                 if( $found ) {
3027                                         if( $isHTML ) {
3028                                                 // A special page; don't store it in the template cache.
3029                                         } else {
3030                                                 $this->mTemplates[$piece['title']] = $text;
3031                                         }
3032                                         $text = $linestart . $text;
3033                                 }
3034                         }
3035                         wfProfileOut( __METHOD__ . '-loadtpl' );
3036                 }
3037
3038                 if ( $found && !$this->incrementIncludeSize( 'pre-expand', strlen( $text ) ) ) {
3039                         # Error, oversize inclusion
3040                         $text = $linestart .
3041                                 "[[$titleText]]<!-- WARNING: template omitted, pre-expand include size too large -->";
3042                         $noparse = true;
3043                         $noargs = true;
3044                 }
3045
3046                 # Recursive parsing, escaping and link table handling
3047                 # Only for HTML output
3048                 if ( $nowiki && $found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3049                         $text = wfEscapeWikiText( $text );
3050                 } elseif ( !$this->ot['msg'] && $found ) {
3051                         if ( $noargs ) {
3052                                 $assocArgs = array();
3053                         } else {
3054                                 # Clean up argument array
3055                                 $assocArgs = array();
3056                                 $index = 1;
3057                                 foreach( $args as $arg ) {
3058                                         $eqpos = strpos( $arg, '=' );
3059                                         if ( $eqpos === false ) {
3060                                                 $assocArgs[$index++] = $arg;
3061                                         } else {
3062                                                 $name = trim( substr( $arg, 0, $eqpos ) );
3063                                                 $value = trim( substr( $arg, $eqpos+1 ) );
3064                                                 if ( $value === false ) {
3065                                                         $value = '';
3066                                                 }
3067                                                 if ( $name !== false ) {
3068                                                         $assocArgs[$name] = $value;
3069                                                 }
3070                                         }
3071                                 }
3072
3073                                 # Add a new element to the templace recursion path
3074                                 $this->mTemplatePath[$part1] = 1;
3075                         }
3076
3077                         if ( !$noparse ) {
3078                                 # If there are any <onlyinclude> tags, only include them
3079                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
3080                                         $m = array();
3081                                         preg_match_all( '/<onlyinclude>(.*?)\n?<\/onlyinclude>/s', $text, $m );
3082                                         $text = '';
3083                                         foreach ($m[1] as $piece)
3084                                                 $text .= $piece;
3085                                 }
3086                                 # Remove <noinclude> sections and <includeonly> tags
3087                                 $text = preg_replace( '/<noinclude>.*?<\/noinclude>/s', '', $text );
3088                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
3089
3090                                 if( $this->ot['html'] || $this->ot['pre'] ) {
3091                                         # Strip <nowiki>, <pre>, etc.
3092                                         $text = $this->strip( $text, $this->mStripState );
3093                                         if ( $this->ot['html'] ) {
3094                                                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
3095                                         } elseif ( $this->ot['pre'] && $this->mOptions->getRemoveComments() ) {
3096                                                 $text = Sanitizer::removeHTMLcomments( $text );
3097                                         }
3098                                 }
3099                                 $text = $this->replaceVariables( $text, $assocArgs );
3100
3101                                 # If the template begins with a table or block-level
3102                                 # element, it should be treated as beginning a new line.
3103                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) /*}*/{
3104                                         $text = "\n" . $text;
3105                                 }
3106                         } elseif ( !$noargs ) {
3107                                 # $noparse and !$noargs
3108                                 # Just replace the arguments, not any double-brace items
3109                                 # This is used for rendered interwiki transclusion
3110                                 $text = $this->replaceVariables( $text, $assocArgs, true );
3111                         }
3112                 }
3113                 # Prune lower levels off the recursion check path
3114                 $this->mTemplatePath = $lastPathLevel;
3115
3116                 if ( $found && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3117                         # Error, oversize inclusion
3118                         $text = $linestart .
3119                                 "[[$titleText]]<!-- WARNING: template omitted, post-expand include size too large -->";
3120                         $noparse = true;
3121                         $noargs = true;
3122                 }
3123
3124                 if ( !$found ) {
3125                         wfProfileOut( $fname );
3126                         return $piece['text'];
3127                 } else {
3128                         wfProfileIn( __METHOD__ . '-placeholders' );
3129                         if ( $isHTML ) {
3130                                 # Replace raw HTML by a placeholder
3131                                 # Add a blank line preceding, to prevent it from mucking up
3132                                 # immediately preceding headings
3133                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
3134                         } else {
3135                                 # replace ==section headers==
3136                                 # XXX this needs to go away once we have a better parser.
3137                                 if ( !$this->ot['wiki'] && !$this->ot['pre'] && $replaceHeadings ) {
3138                                         if( !is_null( $title ) )
3139                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
3140                                         else
3141                                                 $encodedname = base64_encode("");
3142                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
3143                                                 PREG_SPLIT_DELIM_CAPTURE);
3144                                         $text = '';
3145                                         $nsec = 0;
3146                                         for( $i = 0; $i < count($m); $i += 2 ) {
3147                                                 $text .= $m[$i];
3148                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
3149                                                 $hl = $m[$i + 1];
3150                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
3151                                                         $text .= $hl;
3152                                                         continue;
3153                                                 }
3154                                                 $m2 = array();
3155                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
3156                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
3157                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
3158
3159                                                 $nsec++;
3160                                         }
3161                                 }
3162                         }
3163                         wfProfileOut( __METHOD__ . '-placeholders' );
3164                 }
3165
3166                 # Prune lower levels off the recursion check path
3167                 $this->mTemplatePath = $lastPathLevel;
3168
3169                 if ( !$found ) {
3170                         wfProfileOut( $fname );
3171                         return $piece['text'];
3172                 } else {
3173                         wfProfileOut( $fname );
3174                         return $text;
3175                 }
3176         }
3177
3178         /**
3179          * Fetch the unparsed text of a template and register a reference to it.
3180          */
3181         function fetchTemplate( $title ) {
3182                 $text = false;
3183                 // Loop to fetch the article, with up to 1 redirect
3184                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3185                         $rev = Revision::newFromTitle( $title );
3186                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
3187                         if ( !$rev ) {
3188                                 break;
3189                         }
3190                         $text = $rev->getText();
3191                         if ( $text === false ) {
3192                                 break;
3193                         }
3194                         // Redirect?
3195                         $title = Title::newFromRedirect( $text );
3196                 }
3197                 return $text;
3198         }
3199
3200         /**
3201          * Transclude an interwiki link.
3202          */
3203         function interwikiTransclude( $title, $action ) {
3204                 global $wgEnableScaryTranscluding, $wgCanonicalNamespaceNames;
3205
3206                 if (!$wgEnableScaryTranscluding)
3207                         return wfMsg('scarytranscludedisabled');
3208
3209                 // The namespace will actually only be 0 or 10, depending on whether there was a leading :
3210                 // But we'll handle it generally anyway
3211                 if ( $title->getNamespace() ) {
3212                         // Use the canonical namespace, which should work anywhere
3213                         $articleName = $wgCanonicalNamespaceNames[$title->getNamespace()] . ':' . $title->getDBkey();
3214                 } else {
3215                         $articleName = $title->getDBkey();
3216                 }
3217
3218                 $url = str_replace('$1', urlencode($articleName), Title::getInterwikiLink($title->getInterwiki()));
3219                 $url .= "?action=$action";
3220                 if (strlen($url) > 255)
3221                         return wfMsg('scarytranscludetoolong');
3222                 return $this->fetchScaryTemplateMaybeFromCache($url);
3223         }
3224
3225         function fetchScaryTemplateMaybeFromCache($url) {
3226                 global $wgTranscludeCacheExpiry;
3227                 $dbr =& wfGetDB(DB_SLAVE);
3228                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3229                                 array('tc_url' => $url));
3230                 if ($obj) {
3231                         $time = $obj->tc_time;
3232                         $text = $obj->tc_contents;
3233                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3234                                 return $text;
3235                         }
3236                 }
3237
3238                 $text = Http::get($url);
3239                 if (!$text)
3240                         return wfMsg('scarytranscludefailed', $url);
3241
3242                 $dbw =& wfGetDB(DB_MASTER);
3243                 $dbw->replace('transcache', array('tc_url'), array(
3244                         'tc_url' => $url,
3245                         'tc_time' => time(),
3246                         'tc_contents' => $text));
3247                 return $text;
3248         }
3249
3250
3251         /**
3252          * Triple brace replacement -- used for template arguments
3253          * @private
3254          */
3255         function argSubstitution( $matches ) {
3256                 $arg = trim( $matches['title'] );
3257                 $text = $matches['text'];
3258                 $inputArgs = end( $this->mArgStack );
3259
3260                 if ( array_key_exists( $arg, $inputArgs ) ) {
3261                         $text = $inputArgs[$arg];
3262                 } else if (($this->mOutputType == OT_HTML || $this->mOutputType == OT_PREPROCESS ) &&
3263                 null != $matches['parts'] && count($matches['parts']) > 0) {
3264                         $text = $matches['parts'][0];
3265                 }
3266                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3267                         $text = $matches['text'] .
3268                                 '<!-- WARNING: argument omitted, expansion size too large -->';
3269                 }
3270
3271                 return $text;
3272         }
3273
3274         /**
3275          * Increment an include size counter
3276          *
3277          * @param string $type The type of expansion
3278          * @param integer $size The size of the text
3279          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3280          */
3281         function incrementIncludeSize( $type, $size ) {
3282                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3283                         return false;
3284                 } else {
3285                         $this->mIncludeSizes[$type] += $size;
3286                         return true;
3287                 }
3288         }
3289
3290         /**
3291          * Detect __NOGALLERY__ magic word and set a placeholder
3292          */
3293         function stripNoGallery( &$text ) {
3294                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3295                 # do not add TOC
3296                 $mw = MagicWord::get( 'nogallery' );
3297                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3298         }
3299
3300         /**
3301          * Detect __TOC__ magic word and set a placeholder
3302          */
3303         function stripToc( $text ) {
3304                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3305                 # do not add TOC
3306                 $mw = MagicWord::get( 'notoc' );
3307                 if( $mw->matchAndRemove( $text ) ) {
3308                         $this->mShowToc = false;
3309                 }
3310
3311                 $mw = MagicWord::get( 'toc' );
3312                 if( $mw->match( $text ) ) {
3313                         $this->mShowToc = true;
3314                         $this->mForceTocPosition = true;
3315
3316                         // Set a placeholder. At the end we'll fill it in with the TOC.
3317                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3318
3319                         // Only keep the first one.
3320                         $text = $mw->replace( '', $text );
3321                 }
3322                 return $text;
3323         }
3324
3325         /**
3326          * This function accomplishes several tasks:
3327          * 1) Auto-number headings if that option is enabled
3328          * 2) Add an [edit] link to sections for logged in users who have enabled the option
3329          * 3) Add a Table of contents on the top for users who have enabled the option
3330          * 4) Auto-anchor headings
3331          *
3332          * It loops through all headlines, collects the necessary data, then splits up the
3333          * string and re-inserts the newly formatted headlines.
3334          *
3335          * @param string $text
3336          * @param boolean $isMain
3337          * @private
3338          */
3339         function formatHeadings( $text, $isMain=true ) {
3340                 global $wgMaxTocLevel, $wgContLang;
3341
3342                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3343                 if( !$this->mTitle->userCanEdit() ) {
3344                         $showEditLink = 0;
3345                 } else {
3346                         $showEditLink = $this->mOptions->getEditSection();
3347                 }
3348
3349                 # Inhibit editsection links if requested in the page
3350                 $esw =& MagicWord::get( 'noeditsection' );
3351                 if( $esw->matchAndRemove( $text ) ) {
3352                         $showEditLink = 0;
3353                 }
3354
3355                 # Get all headlines for numbering them and adding funky stuff like [edit]
3356                 # links - this is for later, but we need the number of headlines right now
3357                 $matches = array();
3358                 $numMatches = preg_match_all( '/<H([1-6])(.*?'.'>)(.*?)<\/H[1-6] *>/i', $text, $matches );
3359
3360                 # if there are fewer than 4 headlines in the article, do not show TOC
3361                 # unless it's been explicitly enabled.
3362                 $enoughToc = $this->mShowToc &&
3363                         (($numMatches >= 4) || $this->mForceTocPosition);
3364
3365                 # Allow user to stipulate that a page should have a "new section"
3366                 # link added via __NEWSECTIONLINK__
3367                 $mw =& MagicWord::get( 'newsectionlink' );
3368                 if( $mw->matchAndRemove( $text ) )
3369                         $this->mOutput->setNewSection( true );
3370
3371                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3372                 # override above conditions and always show TOC above first header
3373                 $mw =& MagicWord::get( 'forcetoc' );
3374                 if ($mw->matchAndRemove( $text ) ) {
3375                         $this->mShowToc = true;
3376                         $enoughToc = true;
3377                 }
3378
3379                 # Never ever show TOC if no headers
3380                 if( $numMatches < 1 ) {
3381                         $enoughToc = false;
3382                 }
3383
3384                 # We need this to perform operations on the HTML
3385                 $sk =& $this->mOptions->getSkin();
3386
3387                 # headline counter
3388                 $headlineCount = 0;
3389                 $sectionCount = 0; # headlineCount excluding template sections
3390
3391                 # Ugh .. the TOC should have neat indentation levels which can be
3392                 # passed to the skin functions. These are determined here
3393                 $toc = '';
3394                 $full = '';
3395                 $head = array();
3396                 $sublevelCount = array();
3397                 $levelCount = array();
3398                 $toclevel = 0;
3399                 $level = 0;
3400                 $prevlevel = 0;
3401                 $toclevel = 0;
3402                 $prevtoclevel = 0;
3403
3404                 foreach( $matches[3] as $headline ) {
3405                         $istemplate = 0;
3406                         $templatetitle = '';
3407                         $templatesection = 0;
3408                         $numbering = '';
3409                         $mat = array();
3410                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3411                                 $istemplate = 1;
3412                                 $templatetitle = base64_decode($mat[1]);
3413                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3414                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3415                         }
3416
3417                         if( $toclevel ) {
3418                                 $prevlevel = $level;
3419                                 $prevtoclevel = $toclevel;
3420                         }
3421                         $level = $matches[1][$headlineCount];
3422
3423                         if( $doNumberHeadings || $enoughToc ) {
3424
3425                                 if ( $level > $prevlevel ) {
3426                                         # Increase TOC level
3427                                         $toclevel++;
3428                                         $sublevelCount[$toclevel] = 0;
3429                                         if( $toclevel<$wgMaxTocLevel ) {
3430                                                 $toc .= $sk->tocIndent();
3431                                         }
3432                                 }
3433                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3434                                         # Decrease TOC level, find level to jump to
3435
3436                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3437                                                 # Can only go down to level 1
3438                                                 $toclevel = 1;
3439                                         } else {
3440                                                 for ($i = $toclevel; $i > 0; $i--) {
3441                                                         if ( $levelCount[$i] == $level ) {
3442                                                                 # Found last matching level
3443                                                                 $toclevel = $i;
3444                                                                 break;
3445                                                         }
3446                                                         elseif ( $levelCount[$i] < $level ) {
3447                                                                 # Found first matching level below current level
3448                                                                 $toclevel = $i + 1;
3449                                                                 break;
3450                                                         }
3451                                                 }
3452                                         }
3453                                         if( $toclevel<$wgMaxTocLevel ) {
3454                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3455                                         }
3456                                 }
3457                                 else {
3458                                         # No change in level, end TOC line
3459                                         if( $toclevel<$wgMaxTocLevel ) {
3460                                                 $toc .= $sk->tocLineEnd();
3461                                         }
3462                                 }
3463
3464                                 $levelCount[$toclevel] = $level;
3465
3466                                 # count number of headlines for each level
3467                                 @$sublevelCount[$toclevel]++;
3468                                 $dot = 0;
3469                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3470                                         if( !empty( $sublevelCount[$i] ) ) {
3471                                                 if( $dot ) {
3472                                                         $numbering .= '.';
3473                                                 }
3474                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3475                                                 $dot = 1;
3476                                         }
3477                                 }
3478                         }
3479
3480                         # The canonized header is a version of the header text safe to use for links
3481                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3482                         $canonized_headline = $this->unstrip( $headline, $this->mStripState );
3483                         $canonized_headline = $this->unstripNoWiki( $canonized_headline, $this->mStripState );
3484
3485                         # Remove link placeholders by the link text.
3486                         #     <!--LINK number-->
3487                         # turns into
3488                         #     link text with suffix
3489                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3490                                                             "\$this->mLinkHolders['texts'][\$1]",
3491                                                             $canonized_headline );
3492                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3493                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3494                                                             $canonized_headline );
3495
3496                         # strip out HTML
3497                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3498                         $tocline = trim( $canonized_headline );
3499                         # Save headline for section edit hint before it's escaped
3500                         $headline_hint = trim( $canonized_headline );
3501                         $canonized_headline = Sanitizer::escapeId( $tocline );
3502                         $refers[$headlineCount] = $canonized_headline;
3503
3504                         # count how many in assoc. array so we can track dupes in anchors
3505                         @$refers[$canonized_headline]++;
3506                         $refcount[$headlineCount]=$refers[$canonized_headline];
3507
3508                         # Don't number the heading if it is the only one (looks silly)
3509                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3510                                 # the two are different if the line contains a link
3511                                 $headline=$numbering . ' ' . $headline;
3512                         }
3513
3514                         # Create the anchor for linking from the TOC to the section
3515                         $anchor = $canonized_headline;
3516                         if($refcount[$headlineCount] > 1 ) {
3517                                 $anchor .= '_' . $refcount[$headlineCount];
3518                         }
3519                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3520                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3521                         }
3522                         # give headline the correct <h#> tag
3523                         @$head[$headlineCount] .= "<a name=\"$anchor\"></a><div class=\"mw-h$level\">";
3524
3525                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3526                                 if ( empty( $head[$headlineCount] ) ) {
3527                                         $head[$headlineCount] = '';
3528                                 }
3529                                 if( $istemplate )
3530                                         $head[$headlineCount] .= $sk->editSectionLinkForOther($templatetitle, $templatesection);
3531                                 else
3532                                         $head[$headlineCount] .= $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3533                         }
3534                         // Yes, the headline logically goes before the edit section.  It can't
3535                         // be put there, however, because of bugs in IE's and Firefox's float
3536                         // rendering. See <https://bugzilla.mozilla.org/show_bug.cgi?id=50630>.
3537                         $head[$headlineCount] .= " <h$level{$matches[2][$headlineCount]}$headline</h$level></div>";
3538
3539                         $headlineCount++;
3540                         if( !$istemplate )
3541                                 $sectionCount++;
3542                 }
3543
3544                 if( $enoughToc ) {
3545                         if( $toclevel<$wgMaxTocLevel ) {
3546                                 $toc .= $sk->tocUnindent( $toclevel - 1 );
3547                         }
3548                         $toc = $sk->tocList( $toc );
3549                 }
3550
3551                 # split up and insert constructed headlines
3552
3553                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3554                 $i = 0;
3555
3556                 foreach( $blocks as $block ) {
3557                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3558                                 # This is the [edit] link that appears for the top block of text when
3559                                 # section editing is enabled
3560
3561                                 # Disabled because it broke block formatting
3562                                 # For example, a bullet point in the top line
3563                                 # $full .= $sk->editSectionLink(0);
3564                         }
3565                         $full .= $block;
3566                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3567                                 # Top anchor now in skin
3568                                 $full = $full.$toc;
3569                         }
3570
3571                         if( !empty( $head[$i] ) ) {
3572                                 $full .= $head[$i];
3573                         }
3574                         $i++;
3575                 }
3576                 if( $this->mForceTocPosition ) {
3577                         return str_replace( '<!--MWTOC-->', $toc, $full );
3578                 } else {
3579                         return $full;
3580                 }
3581         }
3582
3583         /**
3584          * Transform wiki markup when saving a page by doing \r\n -> \n
3585          * conversion, substitting signatures, {{subst:}} templates, etc.
3586          *
3587          * @param string $text the text to transform
3588          * @param Title &$title the Title object for the current article
3589          * @param User &$user the User object describing the current user
3590          * @param ParserOptions $options parsing options
3591          * @param bool $clearState whether to clear the parser state first
3592          * @return string the altered wiki markup
3593          * @public
3594          */
3595         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3596                 $this->mOptions = $options;
3597                 $this->mTitle =& $title;
3598                 $this->setOutputType( OT_WIKI );
3599
3600                 if ( $clearState ) {
3601                         $this->clearState();
3602                 }
3603
3604                 $stripState = false;
3605                 $pairs = array(
3606                         "\r\n" => "\n",
3607                 );
3608                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3609                 $text = $this->strip( $text, $stripState, true, array( 'gallery' ) );
3610                 $text = $this->pstPass2( $text, $stripState, $user );
3611                 $text = $this->unstrip( $text, $stripState );
3612                 $text = $this->unstripNoWiki( $text, $stripState );
3613                 return $text;
3614         }
3615
3616         /**
3617          * Pre-save transform helper function
3618          * @private
3619          */
3620         function pstPass2( $text, &$stripState, $user ) {
3621                 global $wgContLang, $wgLocaltimezone;
3622
3623                 /* Note: This is the timestamp saved as hardcoded wikitext to
3624                  * the database, we use $wgContLang here in order to give
3625                  * everyone the same signature and use the default one rather
3626                  * than the one selected in each user's preferences.
3627                  */
3628                 if ( isset( $wgLocaltimezone ) ) {
3629                         $oldtz = getenv( 'TZ' );
3630                         putenv( 'TZ='.$wgLocaltimezone );
3631                 }
3632                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3633                   ' (' . date( 'T' ) . ')';
3634                 if ( isset( $wgLocaltimezone ) ) {
3635                         putenv( 'TZ='.$oldtz );
3636                 }
3637
3638                 # Variable replacement
3639                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3640                 $text = $this->replaceVariables( $text );
3641
3642                 # Strip out <nowiki> etc. added via replaceVariables
3643                 $text = $this->strip( $text, $stripState, false, array( 'gallery' ) );
3644
3645                 # Signatures
3646                 $sigText = $this->getUserSig( $user );
3647                 $text = strtr( $text, array(
3648                         '~~~~~' => $d,
3649                         '~~~~' => "$sigText $d",
3650                         '~~~' => $sigText
3651                 ) );
3652
3653                 # Context links: [[|name]] and [[name (context)|]]
3654                 #
3655                 global $wgLegalTitleChars;
3656                 $tc = "[$wgLegalTitleChars]";
3657                 $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3658
3659                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3660                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3661                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3662
3663                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3664                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3665                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3666
3667                 $t = $this->mTitle->getText();
3668                 $m = array();
3669                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3670                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3671                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3672                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3673                 } else {
3674                         # if there's no context, don't bother duplicating the title
3675                         $text = preg_replace( $p2, '[[\\1]]', $text );
3676                 }
3677
3678                 # Trim trailing whitespace
3679                 # __END__ tag allows for trailing
3680                 # whitespace to be deliberately included
3681                 $text = rtrim( $text );
3682                 $mw =& MagicWord::get( 'end' );
3683                 $mw->matchAndRemove( $text );
3684
3685                 return $text;
3686         }
3687
3688         /**
3689          * Fetch the user's signature text, if any, and normalize to
3690          * validated, ready-to-insert wikitext.
3691          *
3692          * @param User $user
3693          * @return string
3694          * @private
3695          */
3696         function getUserSig( &$user ) {
3697                 $username = $user->getName();
3698                 $nickname = $user->getOption( 'nickname' );
3699                 $nickname = $nickname === '' ? $username : $nickname;
3700
3701                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3702                         # Sig. might contain markup; validate this
3703                         if( $this->validateSig( $nickname ) !== false ) {
3704                                 # Validated; clean up (if needed) and return it
3705                                 return $this->cleanSig( $nickname, true );
3706                         } else {
3707                                 # Failed to validate; fall back to the default
3708                                 $nickname = $username;
3709                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3710                         }
3711                 }
3712
3713                 // Make sure nickname doesnt get a sig in a sig
3714                 $nickname = $this->cleanSigInSig( $nickname );
3715
3716                 # If we're still here, make it a link to the user page
3717                 $userpage = $user->getUserPage();
3718                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3719         }
3720
3721         /**
3722          * Check that the user's signature contains no bad XML
3723          *
3724          * @param string $text
3725          * @return mixed An expanded string, or false if invalid.
3726          */
3727         function validateSig( $text ) {
3728                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3729         }
3730
3731         /**
3732          * Clean up signature text
3733          *
3734          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3735          * 2) Substitute all transclusions
3736          *
3737          * @param string $text
3738          * @param $parsing Whether we're cleaning (preferences save) or parsing
3739          * @return string Signature text
3740          */
3741         function cleanSig( $text, $parsing = false ) {
3742                 global $wgTitle;
3743                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
3744
3745                 $substWord = MagicWord::get( 'subst' );
3746                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3747                 $substText = '{{' . $substWord->getSynonym( 0 );
3748
3749                 $text = preg_replace( $substRegex, $substText, $text );
3750                 $text = $this->cleanSigInSig( $text );
3751                 $text = $this->replaceVariables( $text );
3752
3753                 $this->clearState();
3754                 return $text;
3755         }
3756
3757         /**
3758          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3759          * @param string $text
3760          * @return string Signature text with /~{3,5}/ removed
3761          */
3762         function cleanSigInSig( $text ) {
3763                 $text = preg_replace( '/~{3,5}/', '', $text );
3764                 return $text;
3765         }
3766
3767         /**
3768          * Set up some variables which are usually set up in parse()
3769          * so that an external function can call some class members with confidence
3770          * @public
3771          */
3772         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3773                 $this->mTitle =& $title;
3774                 $this->mOptions = $options;
3775                 $this->setOutputType( $outputType );
3776                 if ( $clearState ) {
3777                         $this->clearState();
3778                 }
3779         }
3780
3781         /**
3782          * Transform a MediaWiki message by replacing magic variables.
3783          *
3784          * @param string $text the text to transform
3785          * @param ParserOptions $options  options
3786          * @return string the text with variables substituted
3787          * @public
3788          */
3789         function transformMsg( $text, $options ) {
3790                 global $wgTitle;
3791                 static $executing = false;
3792
3793                 $fname = "Parser::transformMsg";
3794
3795                 # Guard against infinite recursion
3796                 if ( $executing ) {
3797                         return $text;
3798                 }
3799                 $executing = true;
3800
3801                 wfProfileIn($fname);
3802
3803                 if ( $wgTitle ) {
3804                         $this->mTitle = $wgTitle;
3805                 } else {
3806                         $this->mTitle = Title::newFromText('msg');
3807                 }
3808                 $this->mOptions = $options;
3809                 $this->setOutputType( OT_MSG );
3810                 $this->clearState();
3811                 $text = $this->replaceVariables( $text );
3812
3813                 $executing = false;
3814                 wfProfileOut($fname);
3815                 return $text;
3816         }
3817
3818         /**
3819          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3820          * The callback should have the following form:
3821          *    function myParserHook( $text, $params, &$parser ) { ... }
3822          *
3823          * Transform and return $text. Use $parser for any required context, e.g. use
3824          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3825          *
3826          * @public
3827          *
3828          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3829          * @param mixed $callback The callback function (and object) to use for the tag
3830          *
3831          * @return The old value of the mTagHooks array associated with the hook
3832          */
3833         function setHook( $tag, $callback ) {
3834                 $tag = strtolower( $tag );
3835                 $oldVal = @$this->mTagHooks[$tag];
3836                 $this->mTagHooks[$tag] = $callback;
3837
3838                 return $oldVal;
3839         }
3840
3841         /**
3842          * Create a function, e.g. {{sum:1|2|3}}
3843          * The callback function should have the form:
3844          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3845          *
3846          * The callback may either return the text result of the function, or an array with the text
3847          * in element 0, and a number of flags in the other elements. The names of the flags are
3848          * specified in the keys. Valid flags are:
3849          *   found                     The text returned is valid, stop processing the template. This
3850          *                             is on by default.
3851          *   nowiki                    Wiki markup in the return value should be escaped
3852          *   noparse                   Unsafe HTML tags should not be stripped, etc.
3853          *   noargs                    Don't replace triple-brace arguments in the return value
3854          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3855          *
3856          * @public
3857          *
3858          * @param string $id The magic word ID
3859          * @param mixed $callback The callback function (and object) to use
3860          * @param integer $flags a combination of the following flags:
3861          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
3862          *
3863          * @return The old callback function for this name, if any
3864          */
3865         function setFunctionHook( $id, $callback, $flags = 0 ) {
3866                 $oldVal = @$this->mFunctionHooks[$id];
3867                 $this->mFunctionHooks[$id] = $callback;
3868
3869                 # Add to function cache
3870                 $mw = MagicWord::get( $id );
3871                 if( !$mw )
3872                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
3873
3874                 $synonyms = $mw->getSynonyms();
3875                 $sensitive = intval( $mw->isCaseSensitive() );
3876
3877                 foreach ( $synonyms as $syn ) {
3878                         # Case
3879                         if ( !$sensitive ) {
3880                                 $syn = strtolower( $syn );
3881                         }
3882                         # Add leading hash
3883                         if ( !( $flags & SFH_NO_HASH ) ) {
3884                                 $syn = '#' . $syn;
3885                         }
3886                         # Remove trailing colon
3887                         if ( substr( $syn, -1, 1 ) == ':' ) {
3888                                 $syn = substr( $syn, 0, -1 );
3889                         }
3890                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
3891                 }
3892                 return $oldVal;
3893         }
3894
3895         /**
3896          * Get all registered function hook identifiers
3897          *
3898          * @return array
3899          */
3900         function getFunctionHooks() {
3901                 return array_keys( $this->mFunctionHooks );
3902         }
3903
3904         /**
3905          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3906          * Placeholders created in Skin::makeLinkObj()
3907          * Returns an array of links found, indexed by PDBK:
3908          *  0 - broken
3909          *  1 - normal link
3910          *  2 - stub
3911          * $options is a bit field, RLH_FOR_UPDATE to select for update
3912          */
3913         function replaceLinkHolders( &$text, $options = 0 ) {
3914                 global $wgUser;
3915                 global $wgOutputReplace;
3916                 global $wgContLang;
3917
3918                 $fname = 'Parser::replaceLinkHolders';
3919                 wfProfileIn( $fname );
3920
3921                 $pdbks = array();
3922                 $colours = array();
3923                 $sk =& $this->mOptions->getSkin();
3924                 $linkCache =& LinkCache::singleton();
3925
3926                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3927                         wfProfileIn( $fname.'-check' );
3928                         $dbr =& wfGetDB( DB_SLAVE );
3929                         $page = $dbr->tableName( 'page' );
3930                         $threshold = $wgUser->getOption('stubthreshold');
3931
3932                         # Sort by namespace
3933                         asort( $this->mLinkHolders['namespaces'] );
3934
3935                         # Generate query
3936                         $query = false;
3937                         $current = null;
3938                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3939                                 # Make title object
3940                                 $title = $this->mLinkHolders['titles'][$key];
3941
3942                                 # Skip invalid entries.
3943                                 # Result will be ugly, but prevents crash.
3944                                 if ( is_null( $title ) ) {
3945                                         continue;
3946                                 }
3947                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3948
3949                                 # Check if it's a static known link, e.g. interwiki
3950                                 if ( $title->isAlwaysKnown() ) {
3951                                         $colours[$pdbk] = 1;
3952                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3953                                         $colours[$pdbk] = 1;
3954                                         $this->mOutput->addLink( $title, $id );
3955                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
3956                                         $colours[$pdbk] = 0;
3957                                 } else {
3958                                         # Not in the link cache, add it to the query
3959                                         if ( !isset( $current ) ) {
3960                                                 $current = $ns;
3961                                                 $query =  "SELECT page_id, page_namespace, page_title";
3962                                                 if ( $threshold > 0 ) {
3963                                                         $query .= ', page_len, page_is_redirect';
3964                                                 }
3965                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
3966                                         } elseif ( $current != $ns ) {
3967                                                 $current = $ns;
3968                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
3969                                         } else {
3970                                                 $query .= ', ';
3971                                         }
3972
3973                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
3974                                 }
3975                         }
3976                         if ( $query ) {
3977                                 $query .= '))';
3978                                 if ( $options & RLH_FOR_UPDATE ) {
3979                                         $query .= ' FOR UPDATE';
3980                                 }
3981
3982                                 $res = $dbr->query( $query, $fname );
3983
3984                                 # Fetch data and form into an associative array
3985                                 # non-existent = broken
3986                                 # 1 = known
3987                                 # 2 = stub
3988                                 while ( $s = $dbr->fetchObject($res) ) {
3989                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
3990                                         $pdbk = $title->getPrefixedDBkey();
3991                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
3992                                         $this->mOutput->addLink( $title, $s->page_id );
3993
3994                                         if ( $threshold >  0 ) {
3995                                                 $size = $s->page_len;
3996                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
3997                                                         $colours[$pdbk] = 1;
3998                                                 } else {
3999                                                         $colours[$pdbk] = 2;
4000                                                 }
4001                                         } else {
4002                                                 $colours[$pdbk] = 1;
4003                                         }
4004                                 }
4005                         }
4006                         wfProfileOut( $fname.'-check' );
4007
4008                         # Do a second query for different language variants of links (if needed)
4009                         if($wgContLang->hasVariants()){
4010                                 $linkBatch = new LinkBatch();
4011                                 $variantMap = array(); // maps $pdbkey_Variant => $pdbkey_original
4012
4013                                 // Add variants of links to link batch
4014                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4015                                         $title = $this->mLinkHolders['titles'][$key];
4016                                         if ( is_null( $title ) )
4017                                                 continue;
4018
4019                                         $pdbk = $title->getPrefixedDBkey();
4020
4021                                         // generate all variants of the link title text
4022                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($title->getText());
4023
4024                                         // if link was not found (in first query), add all variants to query
4025                                         if ( !isset($colours[$pdbk]) ){
4026                                                 foreach($allTextVariants as $textVariant){
4027                                                         $variantTitle = Title::makeTitle( $ns, $textVariant );
4028                                                         if(is_null($variantTitle)) continue;
4029                                                         $linkBatch->addObj( $variantTitle );
4030                                                         $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4031                                                 }
4032                                         }
4033                                 }
4034
4035
4036                                 if(!$linkBatch->isEmpty()){
4037                                         // construct query
4038                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4039
4040                                         $variantQuery =  "SELECT page_id, page_namespace, page_title";
4041                                         if ( $threshold > 0 ) {
4042                                                 $variantQuery .= ', page_len, page_is_redirect';
4043                                         }
4044
4045                                         $variantQuery .= " FROM $page WHERE $titleClause";
4046                                         if ( $options & RLH_FOR_UPDATE ) {
4047                                                 $variantQuery .= ' FOR UPDATE';
4048                                         }
4049
4050                                         $varRes = $dbr->query( $variantQuery, $fname );
4051
4052                                         // for each found variants, figure out link holders and replace
4053                                         while ( $s = $dbr->fetchObject($varRes) ) {
4054
4055                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4056                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4057                                                 $linkCache->addGoodLinkObj( $s->page_id, $variantTitle );
4058                                                 $this->mOutput->addLink( $variantTitle, $s->page_id );
4059
4060                                                 $holderKeys = $variantMap[$varPdbk];
4061
4062                                                 // loop over link holders
4063                                                 foreach($holderKeys as $key){
4064                                                         $title = $this->mLinkHolders['titles'][$key];
4065                                                         if ( is_null( $title ) ) continue;
4066
4067                                                         $pdbk = $title->getPrefixedDBkey();
4068
4069                                                         if(!isset($colours[$pdbk])){
4070                                                                 // found link in some of the variants, replace the link holder data
4071                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4072                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4073
4074                                                                 // set pdbk and colour
4075                                                                 $pdbks[$key] = $varPdbk;
4076                                                                 if ( $threshold >  0 ) {
4077                                                                         $size = $s->page_len;
4078                                                                         if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
4079                                                                                 $colours[$varPdbk] = 1;
4080                                                                         } else {
4081                                                                                 $colours[$varPdbk] = 2;
4082                                                                         }
4083                                                                 }
4084                                                                 else {
4085                                                                         $colours[$varPdbk] = 1;
4086                                                                 }
4087                                                         }
4088                                                 }
4089                                         }
4090                                 }
4091                         }
4092
4093                         # Construct search and replace arrays
4094                         wfProfileIn( $fname.'-construct' );
4095                         $wgOutputReplace = array();
4096                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4097                                 $pdbk = $pdbks[$key];
4098                                 $searchkey = "<!--LINK $key-->";
4099                                 $title = $this->mLinkHolders['titles'][$key];
4100                                 if ( empty( $colours[$pdbk] ) ) {
4101                                         $linkCache->addBadLinkObj( $title );
4102                                         $colours[$pdbk] = 0;
4103                                         $this->mOutput->addLink( $title, 0 );
4104                                         $wgOutputReplace[$searchkey] = $sk->makeBrokenLinkObj( $title,
4105                                                                         $this->mLinkHolders['texts'][$key],
4106                                                                         $this->mLinkHolders['queries'][$key] );
4107                                 } elseif ( $colours[$pdbk] == 1 ) {
4108                                         $wgOutputReplace[$searchkey] = $sk->makeKnownLinkObj( $title,
4109                                                                         $this->mLinkHolders['texts'][$key],
4110                                                                         $this->mLinkHolders['queries'][$key] );
4111                                 } elseif ( $colours[$pdbk] == 2 ) {
4112                                         $wgOutputReplace[$searchkey] = $sk->makeStubLinkObj( $title,
4113                                                                         $this->mLinkHolders['texts'][$key],
4114                                                                         $this->mLinkHolders['queries'][$key] );
4115                                 }
4116                         }
4117                         wfProfileOut( $fname.'-construct' );
4118
4119                         # Do the thing
4120                         wfProfileIn( $fname.'-replace' );
4121
4122                         $text = preg_replace_callback(
4123                                 '/(<!--LINK .*?-->)/',
4124                                 "wfOutputReplaceMatches",
4125                                 $text);
4126
4127                         wfProfileOut( $fname.'-replace' );
4128                 }
4129
4130                 # Now process interwiki link holders
4131                 # This is quite a bit simpler than internal links
4132                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4133                         wfProfileIn( $fname.'-interwiki' );
4134                         # Make interwiki link HTML
4135                         $wgOutputReplace = array();
4136                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4137                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4138                                 $wgOutputReplace[$key] = $sk->makeLinkObj( $title, $link );
4139                         }
4140
4141                         $text = preg_replace_callback(
4142                                 '/<!--IWLINK (.*?)-->/',
4143                                 "wfOutputReplaceMatches",
4144                                 $text );
4145                         wfProfileOut( $fname.'-interwiki' );
4146                 }
4147
4148                 wfProfileOut( $fname );
4149                 return $colours;
4150         }
4151
4152         /**
4153          * Replace <!--LINK--> link placeholders with plain text of links
4154          * (not HTML-formatted).
4155          * @param string $text
4156          * @return string
4157          */
4158         function replaceLinkHoldersText( $text ) {
4159                 $fname = 'Parser::replaceLinkHoldersText';
4160                 wfProfileIn( $fname );
4161
4162                 $text = preg_replace_callback(
4163                         '/<!--(LINK|IWLINK) (.*?)-->/',
4164                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4165                         $text );
4166
4167                 wfProfileOut( $fname );
4168                 return $text;
4169         }
4170
4171         /**
4172          * @param array $matches
4173          * @return string
4174          * @private
4175          */
4176         function replaceLinkHoldersTextCallback( $matches ) {
4177                 $type = $matches[1];
4178                 $key  = $matches[2];
4179                 if( $type == 'LINK' ) {
4180                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4181                                 return $this->mLinkHolders['texts'][$key];
4182                         }
4183                 } elseif( $type == 'IWLINK' ) {
4184                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4185                                 return $this->mInterwikiLinkHolders['texts'][$key];
4186                         }
4187                 }
4188                 return $matches[0];
4189         }
4190
4191         /**
4192          * Tag hook handler for 'pre'.
4193          */
4194         function renderPreTag( $text, $attribs ) {
4195                 // Backwards-compatibility hack
4196                 $content = preg_replace( '!<nowiki>(.*?)</nowiki>!is', '\\1', $text );
4197
4198                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4199                 return wfOpenElement( 'pre', $attribs ) .
4200                         wfEscapeHTMLTagsOnly( $content ) .
4201                         '</pre>';
4202         }
4203
4204         /**
4205          * Renders an image gallery from a text with one line per image.
4206          * text labels may be given by using |-style alternative text. E.g.
4207          *   Image:one.jpg|The number "1"
4208          *   Image:tree.jpg|A tree
4209          * given as text will return the HTML of a gallery with two images,
4210          * labeled 'The number "1"' and
4211          * 'A tree'.
4212          */
4213         function renderImageGallery( $text, $params ) {
4214                 $ig = new ImageGallery();
4215                 $ig->setShowBytes( false );
4216                 $ig->setShowFilename( false );
4217                 $ig->setParsing();
4218                 $ig->useSkin( $this->mOptions->getSkin() );
4219
4220                 if( isset( $params['caption'] ) )
4221                         $ig->setCaption( $params['caption'] );
4222
4223                 $lines = explode( "\n", $text );
4224                 foreach ( $lines as $line ) {
4225                         # match lines like these:
4226                         # Image:someimage.jpg|This is some image
4227                         $matches = array();
4228                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4229                         # Skip empty lines
4230                         if ( count( $matches ) == 0 ) {
4231                                 continue;
4232                         }
4233                         $tp = Title::newFromText( $matches[1] );
4234                         $nt =& $tp;
4235                         if( is_null( $nt ) ) {
4236                                 # Bogus title. Ignore these so we don't bomb out later.
4237                                 continue;
4238                         }
4239                         if ( isset( $matches[3] ) ) {
4240                                 $label = $matches[3];
4241                         } else {
4242                                 $label = '';
4243                         }
4244
4245                         $pout = $this->parse( $label,
4246                                 $this->mTitle,
4247                                 $this->mOptions,
4248                                 false, // Strip whitespace...?
4249                                 false  // Don't clear state!
4250                         );
4251                         $html = $pout->getText();
4252
4253                         $ig->add( new Image( $nt ), $html );
4254
4255                         # Only add real images (bug #5586)
4256                         if ( $nt->getNamespace() == NS_IMAGE ) {
4257                                 $this->mOutput->addImage( $nt->getDBkey() );
4258                         }
4259                 }
4260                 return $ig->toHTML();
4261         }
4262
4263         /**
4264          * Parse image options text and use it to make an image
4265          */
4266         function makeImage( $nt, $options ) {
4267                 global $wgUseImageResize, $wgDjvuRenderer;
4268
4269                 $align = '';
4270
4271                 # Check if the options text is of the form "options|alt text"
4272                 # Options are:
4273                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4274                 #  * left               no resizing, just left align. label is used for alt= only
4275                 #  * right              same, but right aligned
4276                 #  * none               same, but not aligned
4277                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4278                 #  * center             center the image
4279                 #  * framed             Keep original image size, no magnify-button.
4280
4281                 $part = explode( '|', $options);
4282
4283                 $mwThumb  =& MagicWord::get( 'img_thumbnail' );
4284                 $mwManualThumb =& MagicWord::get( 'img_manualthumb' );
4285                 $mwLeft   =& MagicWord::get( 'img_left' );
4286                 $mwRight  =& MagicWord::get( 'img_right' );
4287                 $mwNone   =& MagicWord::get( 'img_none' );
4288                 $mwWidth  =& MagicWord::get( 'img_width' );
4289                 $mwCenter =& MagicWord::get( 'img_center' );
4290                 $mwFramed =& MagicWord::get( 'img_framed' );
4291                 $mwPage   =& MagicWord::get( 'img_page' );
4292                 $caption = '';
4293
4294                 $width = $height = $framed = $thumb = false;
4295                 $page = null;
4296                 $manual_thumb = '' ;
4297
4298                 foreach( $part as $val ) {
4299                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
4300                                 $thumb=true;
4301                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
4302                                 # use manually specified thumbnail
4303                                 $thumb=true;
4304                                 $manual_thumb = $match;
4305                         } elseif ( ! is_null( $mwRight->matchVariableStartToEnd($val) ) ) {
4306                                 # remember to set an alignment, don't render immediately
4307                                 $align = 'right';
4308                         } elseif ( ! is_null( $mwLeft->matchVariableStartToEnd($val) ) ) {
4309                                 # remember to set an alignment, don't render immediately
4310                                 $align = 'left';
4311                         } elseif ( ! is_null( $mwCenter->matchVariableStartToEnd($val) ) ) {
4312                                 # remember to set an alignment, don't render immediately
4313                                 $align = 'center';
4314                         } elseif ( ! is_null( $mwNone->matchVariableStartToEnd($val) ) ) {
4315                                 # remember to set an alignment, don't render immediately
4316                                 $align = 'none';
4317                         } elseif ( isset( $wgDjvuRenderer ) && $wgDjvuRenderer
4318                                    && ! is_null( $match = $mwPage->matchVariableStartToEnd($val) ) ) {
4319                                 # Select a page in a multipage document
4320                                 $page = $match;
4321                         } elseif ( $wgUseImageResize && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
4322                                 wfDebug( "img_width match: $match\n" );
4323                                 # $match is the image width in pixels
4324                                 $m = array();
4325                                 if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
4326                                         $width = intval( $m[1] );
4327                                         $height = intval( $m[2] );
4328                                 } else {
4329                                         $width = intval($match);
4330                                 }
4331                         } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
4332                                 $framed=true;
4333                         } else {
4334                                 $caption = $val;
4335                         }
4336                 }
4337                 # Strip bad stuff out of the alt text
4338                 $alt = $this->replaceLinkHoldersText( $caption );
4339
4340                 # make sure there are no placeholders in thumbnail attributes
4341                 # that are later expanded to html- so expand them now and
4342                 # remove the tags
4343                 $alt = $this->unstrip($alt, $this->mStripState);
4344                 $alt = Sanitizer::stripAllTags( $alt );
4345
4346                 # Linker does the rest
4347                 $sk =& $this->mOptions->getSkin();
4348                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb, $page );
4349         }
4350
4351         /**
4352          * Set a flag in the output object indicating that the content is dynamic and
4353          * shouldn't be cached.
4354          */
4355         function disableCache() {
4356                 wfDebug( "Parser output marked as uncacheable.\n" );
4357                 $this->mOutput->mCacheTime = -1;
4358         }
4359
4360         /**#@+
4361          * Callback from the Sanitizer for expanding items found in HTML attribute
4362          * values, so they can be safely tested and escaped.
4363          * @param string $text
4364          * @param array $args
4365          * @return string
4366          * @private
4367          */
4368         function attributeStripCallback( &$text, $args ) {
4369                 $text = $this->replaceVariables( $text, $args );
4370                 $text = $this->unstripForHTML( $text );
4371                 return $text;
4372         }
4373
4374         function unstripForHTML( $text ) {
4375                 $text = $this->unstrip( $text, $this->mStripState );
4376                 $text = $this->unstripNoWiki( $text, $this->mStripState );
4377                 return $text;
4378         }
4379         /**#@-*/
4380
4381         /**#@+
4382          * Accessor/mutator
4383          */
4384         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4385         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4386         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4387         /**#@-*/
4388
4389         /**#@+
4390          * Accessor
4391          */
4392         function getTags() { return array_keys( $this->mTagHooks ); }
4393         /**#@-*/
4394
4395
4396         /**
4397          * Break wikitext input into sections, and either pull or replace
4398          * some particular section's text.
4399          *
4400          * External callers should use the getSection and replaceSection methods.
4401          *
4402          * @param $text Page wikitext
4403          * @param $section Numbered section. 0 pulls the text before the first
4404          *                 heading; other numbers will pull the given section
4405          *                 along with its lower-level subsections.
4406          * @param $mode One of "get" or "replace"
4407          * @param $newtext Replacement text for section data.
4408          * @return string for "get", the extracted section text.
4409          *                for "replace", the whole page with the section replaced.
4410          */
4411         private function extractSections( $text, $section, $mode, $newtext='' ) {
4412                 # strip NOWIKI etc. to avoid confusion (true-parameter causes HTML
4413                 # comments to be stripped as well)
4414                 $striparray = array();
4415
4416                 $oldOutputType = $this->mOutputType;
4417                 $oldOptions = $this->mOptions;
4418                 $this->mOptions = new ParserOptions();
4419                 $this->setOutputType( OT_WIKI );
4420
4421                 $striptext = $this->strip( $text, $striparray, true );
4422
4423                 $this->setOutputType( $oldOutputType );
4424                 $this->mOptions = $oldOptions;
4425
4426                 # now that we can be sure that no pseudo-sections are in the source,
4427                 # split it up by section
4428                 $uniq = preg_quote( $this->uniqPrefix(), '/' );
4429                 $comment = "(?:$uniq-!--.*?QINU)";
4430                 $secs = preg_split(
4431                 /*
4432                         "/
4433                         ^(
4434                         (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4435                         (?:
4436                                 (=+) # Should this be limited to 6?
4437                                 .+?  # Section title...
4438                                 \\2  # Ending = count must match start
4439                         |
4440                                 ^
4441                                 <h([1-6])\b.*?>
4442                                 .*?
4443                                 <\/h\\3\s*>
4444                         )
4445                         (?:$comment|<\/?noinclude>|\s+)* # Trailing whitespace ok
4446                         )$
4447                         /mix",
4448                 */
4449                         "/
4450                         (
4451                                 ^
4452                                 (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4453                                 (=+) # Should this be limited to 6?
4454                                 .+?  # Section title...
4455                                 \\2  # Ending = count must match start
4456                                 (?:$comment|<\/?noinclude>|[ \\t]+)* # Trailing whitespace ok
4457                                 $
4458                         |
4459                                 <h([1-6])\b.*?>
4460                                 .*?
4461                                 <\/h\\3\s*>
4462                         )
4463                         /mix",
4464                         $striptext, -1,
4465                         PREG_SPLIT_DELIM_CAPTURE);
4466
4467                 if( $mode == "get" ) {
4468                         if( $section == 0 ) {
4469                                 // "Section 0" returns the content before any other section.
4470                                 $rv = $secs[0];
4471                         } else {
4472                                 $rv = "";
4473                         }
4474                 } elseif( $mode == "replace" ) {
4475                         if( $section == 0 ) {
4476                                 $rv = $newtext . "\n\n";
4477                                 $remainder = true;
4478                         } else {
4479                                 $rv = $secs[0];
4480                                 $remainder = false;
4481                         }
4482                 }
4483                 $count = 0;
4484                 $sectionLevel = 0;
4485                 for( $index = 1; $index < count( $secs ); ) {
4486                         $headerLine = $secs[$index++];
4487                         if( $secs[$index] ) {
4488                                 // A wiki header
4489                                 $headerLevel = strlen( $secs[$index++] );
4490                         } else {
4491                                 // An HTML header
4492                                 $index++;
4493                                 $headerLevel = intval( $secs[$index++] );
4494                         }
4495                         $content = $secs[$index++];
4496
4497                         $count++;
4498                         if( $mode == "get" ) {
4499                                 if( $count == $section ) {
4500                                         $rv = $headerLine . $content;
4501                                         $sectionLevel = $headerLevel;
4502                                 } elseif( $count > $section ) {
4503                                         if( $sectionLevel && $headerLevel > $sectionLevel ) {
4504                                                 $rv .= $headerLine . $content;
4505                                         } else {
4506                                                 // Broke out to a higher-level section
4507                                                 break;
4508                                         }
4509                                 }
4510                         } elseif( $mode == "replace" ) {
4511                                 if( $count < $section ) {
4512                                         $rv .= $headerLine . $content;
4513                                 } elseif( $count == $section ) {
4514                                         $rv .= $newtext . "\n\n";
4515                                         $sectionLevel = $headerLevel;
4516                                 } elseif( $count > $section ) {
4517                                         if( $headerLevel <= $sectionLevel ) {
4518                                                 // Passed the section's sub-parts.
4519                                                 $remainder = true;
4520                                         }
4521                                         if( $remainder ) {
4522                                                 $rv .= $headerLine . $content;
4523                                         }
4524                                 }
4525                         }
4526                 }
4527                 # reinsert stripped tags
4528                 $rv = $this->unstrip( $rv, $striparray );
4529                 $rv = $this->unstripNoWiki( $rv, $striparray );
4530                 $rv = trim( $rv );
4531                 return $rv;
4532         }
4533
4534         /**
4535          * This function returns the text of a section, specified by a number ($section).
4536          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4537          * the first section before any such heading (section 0).
4538          *
4539          * If a section contains subsections, these are also returned.
4540          *
4541          * @param $text String: text to look in
4542          * @param $section Integer: section number
4543          * @return string text of the requested section
4544          */
4545         function getSection( $text, $section ) {
4546                 return $this->extractSections( $text, $section, "get" );
4547         }
4548
4549         function replaceSection( $oldtext, $section, $text ) {
4550                 return $this->extractSections( $oldtext, $section, "replace", $text );
4551         }
4552
4553 }
4554
4555 /**
4556  * @todo document
4557  * @package MediaWiki
4558  */
4559 class ParserOutput
4560 {
4561         var $mText,             # The output text
4562                 $mLanguageLinks,    # List of the full text of language links, in the order they appear
4563                 $mCategories,       # Map of category names to sort keys
4564                 $mContainsOldMagic, # Boolean variable indicating if the input contained variables like {{CURRENTDAY}}
4565                 $mCacheTime,        # Time when this object was generated, or -1 for uncacheable. Used in ParserCache.
4566                 $mVersion,          # Compatibility check
4567                 $mTitleText,        # title text of the chosen language variant
4568                 $mLinks,            # 2-D map of NS/DBK to ID for the links in the document. ID=zero for broken.
4569                 $mTemplates,        # 2-D map of NS/DBK to ID for the template references. ID=zero for broken.
4570                 $mImages,           # DB keys of the images used, in the array key only
4571                 $mExternalLinks,    # External link URLs, in the key only
4572                 $mHTMLtitle,            # Display HTML title
4573                 $mSubtitle,                     # Additional subtitle
4574                 $mNewSection,           # Show a new section link?
4575                 $mNoGallery;            # No gallery on category page? (__NOGALLERY__)
4576
4577         function ParserOutput( $text = '', $languageLinks = array(), $categoryLinks = array(),
4578                 $containsOldMagic = false, $titletext = '' )
4579         {
4580                 $this->mText = $text;
4581                 $this->mLanguageLinks = $languageLinks;
4582                 $this->mCategories = $categoryLinks;
4583                 $this->mContainsOldMagic = $containsOldMagic;
4584                 $this->mCacheTime = '';
4585                 $this->mVersion = MW_PARSER_VERSION;
4586                 $this->mTitleText = $titletext;
4587                 $this->mLinks = array();
4588                 $this->mTemplates = array();
4589                 $this->mImages = array();
4590                 $this->mExternalLinks = array();
4591                 $this->mHTMLtitle = "" ;
4592                 $this->mSubtitle = "" ;
4593                 $this->mNewSection = false;
4594                 $this->mNoGallery = false;
4595         }
4596
4597         function getText()                   { return $this->mText; }
4598         function &getLanguageLinks()          { return $this->mLanguageLinks; }
4599         function getCategoryLinks()          { return array_keys( $this->mCategories ); }
4600         function &getCategories()            { return $this->mCategories; }
4601         function getCacheTime()              { return $this->mCacheTime; }
4602         function getTitleText()              { return $this->mTitleText; }
4603         function &getLinks()                 { return $this->mLinks; }
4604         function &getTemplates()             { return $this->mTemplates; }
4605         function &getImages()                { return $this->mImages; }
4606         function &getExternalLinks()         { return $this->mExternalLinks; }
4607         function getNoGallery()              { return $this->mNoGallery; }
4608         function getSubtitle()               { return $this->mSubtitle; }
4609
4610         function containsOldMagic()          { return $this->mContainsOldMagic; }
4611         function setText( $text )            { return wfSetVar( $this->mText, $text ); }
4612         function setLanguageLinks( $ll )     { return wfSetVar( $this->mLanguageLinks, $ll ); }
4613         function setCategoryLinks( $cl )     { return wfSetVar( $this->mCategories, $cl ); }
4614         function setContainsOldMagic( $com ) { return wfSetVar( $this->mContainsOldMagic, $com ); }
4615         function setCacheTime( $t )          { return wfSetVar( $this->mCacheTime, $t ); }
4616         function setTitleText( $t )          { return wfSetVar($this->mTitleText, $t); }
4617         function setSubtitle( $st )          { return wfSetVar( $this->mSubtitle, $st ); }
4618
4619         function addCategory( $c, $sort )    { $this->mCategories[$c] = $sort; }
4620         function addImage( $name )           { $this->mImages[$name] = 1; }
4621         function addLanguageLink( $t )       { $this->mLanguageLinks[] = $t; }
4622         function addExternalLink( $url )     { $this->mExternalLinks[$url] = 1; }
4623
4624         function setNewSection( $value ) {
4625                 $this->mNewSection = (bool)$value;
4626         }
4627         function getNewSection() {
4628                 return (bool)$this->mNewSection;
4629         }
4630
4631         function addLink( $title, $id = null ) {
4632                 $ns = $title->getNamespace();
4633                 $dbk = $title->getDBkey();
4634                 if ( !isset( $this->mLinks[$ns] ) ) {
4635                         $this->mLinks[$ns] = array();
4636                 }
4637                 if ( is_null( $id ) ) {
4638                         $id = $title->getArticleID();
4639                 }
4640                 $this->mLinks[$ns][$dbk] = $id;
4641         }
4642
4643         function addTemplate( $title, $id ) {
4644                 $ns = $title->getNamespace();
4645                 $dbk = $title->getDBkey();
4646                 if ( !isset( $this->mTemplates[$ns] ) ) {
4647                         $this->mTemplates[$ns] = array();
4648                 }
4649                 $this->mTemplates[$ns][$dbk] = $id;
4650         }
4651
4652         /**
4653          * Return true if this cached output object predates the global or
4654          * per-article cache invalidation timestamps, or if it comes from
4655          * an incompatible older version.
4656          *
4657          * @param string $touched the affected article's last touched timestamp
4658          * @return bool
4659          * @public
4660          */
4661         function expired( $touched ) {
4662                 global $wgCacheEpoch;
4663                 return $this->getCacheTime() == -1 || // parser says it's uncacheable
4664                        $this->getCacheTime() < $touched ||
4665                        $this->getCacheTime() <= $wgCacheEpoch ||
4666                        !isset( $this->mVersion ) ||
4667                        version_compare( $this->mVersion, MW_PARSER_VERSION, "lt" );
4668         }
4669 }
4670
4671 /**
4672  * Set options of the Parser
4673  * @todo document
4674  * @package MediaWiki
4675  */
4676 class ParserOptions
4677 {
4678         # All variables are supposed to be private in theory, although in practise this is not the case.
4679         var $mUseTeX;                    # Use texvc to expand <math> tags
4680         var $mUseDynamicDates;           # Use DateFormatter to format dates
4681         var $mInterwikiMagic;            # Interlanguage links are removed and returned in an array
4682         var $mAllowExternalImages;       # Allow external images inline
4683         var $mAllowExternalImagesFrom;   # If not, any exception?
4684         var $mSkin;                      # Reference to the preferred skin
4685         var $mDateFormat;                # Date format index
4686         var $mEditSection;               # Create "edit section" links
4687         var $mNumberHeadings;            # Automatically number headings
4688         var $mAllowSpecialInclusion;     # Allow inclusion of special pages
4689         var $mTidy;                      # Ask for tidy cleanup
4690         var $mInterfaceMessage;          # Which lang to call for PLURAL and GRAMMAR
4691         var $mMaxIncludeSize;            # Maximum size of template expansions, in bytes
4692         var $mRemoveComments;            # Remove HTML comments. ONLY APPLIES TO PREPROCESS OPERATIONS
4693
4694         var $mUser;                      # Stored user object, just used to initialise the skin
4695
4696         function getUseTeX()                        { return $this->mUseTeX; }
4697         function getUseDynamicDates()               { return $this->mUseDynamicDates; }
4698         function getInterwikiMagic()                { return $this->mInterwikiMagic; }
4699         function getAllowExternalImages()           { return $this->mAllowExternalImages; }
4700         function getAllowExternalImagesFrom()       { return $this->mAllowExternalImagesFrom; }
4701         function getEditSection()                   { return $this->mEditSection; }
4702         function getNumberHeadings()                { return $this->mNumberHeadings; }
4703         function getAllowSpecialInclusion()         { return $this->mAllowSpecialInclusion; }
4704         function getTidy()                          { return $this->mTidy; }
4705         function getInterfaceMessage()              { return $this->mInterfaceMessage; }
4706         function getMaxIncludeSize()                { return $this->mMaxIncludeSize; }
4707         function getRemoveComments()                { return $this->mRemoveComments; }
4708
4709         function &getSkin() {
4710                 if ( !isset( $this->mSkin ) ) {
4711                         $this->mSkin = $this->mUser->getSkin();
4712                 }
4713                 return $this->mSkin;
4714         }
4715
4716         function getDateFormat() {
4717                 if ( !isset( $this->mDateFormat ) ) {
4718                         $this->mDateFormat = $this->mUser->getDatePreference();
4719                 }
4720                 return $this->mDateFormat;
4721         }
4722
4723         function setUseTeX( $x )                    { return wfSetVar( $this->mUseTeX, $x ); }
4724         function setUseDynamicDates( $x )           { return wfSetVar( $this->mUseDynamicDates, $x ); }
4725         function setInterwikiMagic( $x )            { return wfSetVar( $this->mInterwikiMagic, $x ); }
4726         function setAllowExternalImages( $x )       { return wfSetVar( $this->mAllowExternalImages, $x ); }
4727         function setAllowExternalImagesFrom( $x )   { return wfSetVar( $this->mAllowExternalImagesFrom, $x ); }
4728         function setDateFormat( $x )                { return wfSetVar( $this->mDateFormat, $x ); }
4729         function setEditSection( $x )               { return wfSetVar( $this->mEditSection, $x ); }
4730         function setNumberHeadings( $x )            { return wfSetVar( $this->mNumberHeadings, $x ); }
4731         function setAllowSpecialInclusion( $x )     { return wfSetVar( $this->mAllowSpecialInclusion, $x ); }
4732         function setTidy( $x )                      { return wfSetVar( $this->mTidy, $x); }
4733         function setSkin( $x )                      { $this->mSkin = $x; }
4734         function setInterfaceMessage( $x )          { return wfSetVar( $this->mInterfaceMessage, $x); }
4735         function setMaxIncludeSize( $x )            { return wfSetVar( $this->mMaxIncludeSize, $x ); }
4736         function setRemoveComments( $x )            { return wfSetVar( $this->mRemoveComments, $x ); }
4737
4738         function ParserOptions( $user = null ) {
4739                 $this->initialiseFromUser( $user );
4740         }
4741
4742         /**
4743          * Get parser options
4744          * @static
4745          */
4746         static function newFromUser( $user ) {
4747                 return new ParserOptions( $user );
4748         }
4749
4750         /** Get user options */
4751         function initialiseFromUser( $userInput ) {
4752                 global $wgUseTeX, $wgUseDynamicDates, $wgInterwikiMagic, $wgAllowExternalImages;
4753                 global $wgAllowExternalImagesFrom, $wgAllowSpecialInclusion, $wgMaxArticleSize;
4754                 $fname = 'ParserOptions::initialiseFromUser';
4755                 wfProfileIn( $fname );
4756                 if ( !$userInput ) {
4757                         global $wgUser;
4758                         if ( isset( $wgUser ) ) {
4759                                 $user = $wgUser;
4760                         } else {
4761                                 $user = new User;
4762                         }
4763                 } else {
4764                         $user =& $userInput;
4765                 }
4766
4767                 $this->mUser = $user;
4768
4769                 $this->mUseTeX = $wgUseTeX;
4770                 $this->mUseDynamicDates = $wgUseDynamicDates;
4771                 $this->mInterwikiMagic = $wgInterwikiMagic;
4772                 $this->mAllowExternalImages = $wgAllowExternalImages;
4773                 $this->mAllowExternalImagesFrom = $wgAllowExternalImagesFrom;
4774                 $this->mSkin = null; # Deferred
4775                 $this->mDateFormat = null; # Deferred
4776                 $this->mEditSection = true;
4777                 $this->mNumberHeadings = $user->getOption( 'numberheadings' );
4778                 $this->mAllowSpecialInclusion = $wgAllowSpecialInclusion;
4779                 $this->mTidy = false;
4780                 $this->mInterfaceMessage = false;
4781                 $this->mMaxIncludeSize = $wgMaxArticleSize * 1024;
4782                 $this->mRemoveComments = true;
4783                 wfProfileOut( $fname );
4784         }
4785 }
4786
4787 /**
4788  * Callback function used by Parser::replaceLinkHolders()
4789  * to substitute link placeholders.
4790  */
4791 function &wfOutputReplaceMatches( $matches ) {
4792         global $wgOutputReplace;
4793         return $wgOutputReplace[$matches[1]];
4794 }
4795
4796 /**
4797  * Return the total number of articles
4798  */
4799 function wfNumberOfArticles() {
4800         global $wgNumberOfArticles;
4801
4802         wfLoadSiteStats();
4803         return $wgNumberOfArticles;
4804 }
4805
4806 /**
4807  * Return the number of files
4808  */
4809 function wfNumberOfFiles() {
4810         $fname = 'wfNumberOfFiles';
4811
4812         wfProfileIn( $fname );
4813         $dbr =& wfGetDB( DB_SLAVE );
4814         $numImages = $dbr->selectField('site_stats', 'ss_images', array(), $fname );
4815         wfProfileOut( $fname );
4816
4817         return $numImages;
4818 }
4819
4820 /**
4821  * Return the number of user accounts
4822  * @return integer
4823  */
4824 function wfNumberOfUsers() {
4825         wfProfileIn( 'wfNumberOfUsers' );
4826         $dbr =& wfGetDB( DB_SLAVE );
4827         $count = $dbr->selectField( 'site_stats', 'ss_users', array(), 'wfNumberOfUsers' );
4828         wfProfileOut( 'wfNumberOfUsers' );
4829         return (int)$count;
4830 }
4831
4832 /**
4833  * Return the total number of pages
4834  * @return integer
4835  */
4836 function wfNumberOfPages() {
4837         wfProfileIn( 'wfNumberOfPages' );
4838         $dbr =& wfGetDB( DB_SLAVE );
4839         $count = $dbr->selectField( 'site_stats', 'ss_total_pages', array(), 'wfNumberOfPages' );
4840         wfProfileOut( 'wfNumberOfPages' );
4841         return (int)$count;
4842 }
4843
4844 /**
4845  * Return the total number of admins
4846  *
4847  * @return integer
4848  */
4849 function wfNumberOfAdmins() {
4850         static $admins = -1;
4851         wfProfileIn( 'wfNumberOfAdmins' );
4852         if( $admins == -1 ) {
4853                 $dbr =& wfGetDB( DB_SLAVE );
4854                 $admins = $dbr->selectField( 'user_groups', 'COUNT(*)', array( 'ug_group' => 'sysop' ), 'wfNumberOfAdmins' );
4855         }
4856         wfProfileOut( 'wfNumberOfAdmins' );
4857         return (int)$admins;
4858 }
4859
4860 /**
4861  * Count the number of pages in a particular namespace
4862  *
4863  * @param $ns Namespace
4864  * @return integer
4865  */
4866 function wfPagesInNs( $ns ) {
4867         static $pageCount = array();
4868         wfProfileIn( 'wfPagesInNs' );
4869         if( !isset( $pageCount[$ns] ) ) {
4870                 $dbr =& wfGetDB( DB_SLAVE );
4871                 $pageCount[$ns] = $dbr->selectField( 'page', 'COUNT(*)', array( 'page_namespace' => $ns ), 'wfPagesInNs' );
4872         }
4873         wfProfileOut( 'wfPagesInNs' );
4874         return (int)$pageCount[$ns];
4875 }
4876
4877 /**
4878  * Get various statistics from the database
4879  * @private
4880  */
4881 function wfLoadSiteStats() {
4882         global $wgNumberOfArticles, $wgTotalViews, $wgTotalEdits;
4883         $fname = 'wfLoadSiteStats';
4884
4885         if ( -1 != $wgNumberOfArticles ) return;
4886         $dbr =& wfGetDB( DB_SLAVE );
4887         $s = $dbr->selectRow( 'site_stats',
4888                 array( 'ss_total_views', 'ss_total_edits', 'ss_good_articles' ),
4889                 array( 'ss_row_id' => 1 ), $fname
4890         );
4891
4892         if ( $s === false ) {
4893                 return;
4894         } else {
4895                 $wgTotalViews = $s->ss_total_views;
4896                 $wgTotalEdits = $s->ss_total_edits;
4897                 $wgNumberOfArticles = $s->ss_good_articles;
4898         }
4899 }
4900
4901 /**
4902  * Get revision timestamp from the database considering timecorrection
4903  *
4904  * @param $id Int: page revision id
4905  * @return integer
4906  */
4907 function wfRevisionTimestamp( $id ) {
4908         global $wgContLang;
4909         $fname = 'wfRevisionTimestamp';
4910
4911         wfProfileIn( $fname );
4912         $dbr =& wfGetDB( DB_SLAVE );
4913         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4914                         array( 'rev_id' => $id ), __METHOD__ );
4915         $timestamp = $wgContLang->userAdjust( $timestamp );
4916         wfProfileOut( $fname );
4917
4918         return $timestamp;
4919 }
4920
4921 /**
4922  * Escape html tags
4923  * Basically replacing " > and < with HTML entities ( &quot;, &gt;, &lt;)
4924  *
4925  * @param $in String: text that might contain HTML tags.
4926  * @return string Escaped string
4927  */
4928 function wfEscapeHTMLTagsOnly( $in ) {
4929         return str_replace(
4930                 array( '"', '>', '<' ),
4931                 array( '&quot;', '&gt;', '&lt;' ),
4932                 $in );
4933 }
4934
4935 ?>